Extracellular vesicle–based ovarian cancer score combined with O-RADS ultrasound for preoperative risk stratification of adnexal masses

Ultrasound-based risk stratification is central to the preoperative evaluation of adnexal masses, yet diagnostic uncertainty persists when morphologic findings are not clearly benign or malignant. This uncertainty is particularly relevant for lesions assigned to intermediate O-RADS categories, for which management decisions often require additional clinical judgment. We examined whether integrating an extracellular vesicle–based Ovarian Cancer Score (OCS) with O-RADS ultrasound and selected clinical or imaging features could improve the preoperative classification of adnexal masses as benign or malignant. This prospective single-center diagnostic study included 250 women with surgically confirmed adnexal masses; 185 had benign disease and 65 had malignant tumors. All participants underwent preoperative transvaginal ultrasound with O-RADS classification, serum CA125 testing, extracellular vesicle–based OCS testing, and postoperative histopathologic confirmation. For model development and internal evaluation, participants were randomly assigned to a training set and an internal held-out test set. In the overall cohort, OCS demonstrated better diagnostic performance than CA125. The largest difference in accuracy was observed among O-RADS 3–4 lesions, for which OCS achieved an accuracy of 86.09% compared with 69.30% for CA125. Both integrated models showed strong discrimination in the internal test-set evaluation. Model 1 algorithms achieved test-set AUCs of up to 0.980, whereas the best-performing Model 2 algorithm achieved a test-set AUC of 0.994. Exploratory decision-curve analysis in the full cohort suggested a positive net benefit across much of the evaluated threshold range. Exploratory subgroup analyses showed numerically higher accuracy in some stage-defined strata; however, the small numbers of malignant cases require cautious interpretation. OCS may be most useful as a complementary molecular layer within ultrasound-led risk assessment rather than as a stand-alone alternative to O-RADS. In this single-center cohort of surgically managed patients, integrating OCS with O-RADS or structured ultrasound features improved internal discrimination between benign and malignant adnexal masses, particularly for lesions with indeterminate imaging findings. Multicenter external validation in broader outpatient populations with adnexal masses is required before this extracellular vesicle–based imaging–molecular strategy can be considered for routine preoperative triage. ClinicalTrials.gov, NCT06366997.