Small Extracellular Vesicle and Total Lipid Profiles Provide a Diagnostic Advantage Over CA-125 Across Ovarian Cancer Subtypes.

Ovarian cancer remains the most lethal gynaecological malignancy, largely due to late-stage diagnosis and the limited sensitivity of CA-125 for borderline and low-grade tumours. Dysregulated lipid metabolism may provide complementary diagnostic information. This study aimed to develop lipid-based biomarkers to improve the differential diagnosis of adnexal masses. We profiled total plasma and small extracellular vesicle (sEV) lipids in women with benign, borderline and malignant ovarian tumours, using healthy women as controls. Solid-phase extraction demonstrated superior reproducibility, lipid-class specificity and sEV lipid enrichment compared with liquid-liquid extraction. Combined with CA-125, both plasma and sEV lipid biomarkers outperformed CA-125 alone for distinguishing borderline and low-grade serous ovarian cancer from benign tumours (accuracy up to 0.89 vs. 0.67). In patients with normal CA-125 levels, sEV lipids provided the greatest diagnostic advantage, distinguishing invasive and borderline tumours from benign adnexal masses (AUROC up to 0.82). Generative artificial intelligence was used to simulate biomarker performance in synthetic cohorts for differential diagnosis (n = 10,000) and screening (n = 100,000). For differential diagnosis, sEV lipids combined with CA-125 outperformed plasma lipids for identifying endometrioid and mucinous ovarian cancer from benign adnexal masses. For screening, sEV lipids alone achieved accuracies of 0.91 and 0.94 for benign conditions and high-grade serous ovarian cancer, respectively. Simulations restricted to patients with normal CA-125 levels confirmed that sEV lipids provided the highest discriminative performance for low-grade serous ovarian cancer (AUROC up to 0.74), demonstrating their potential where CA-125 alone fails. To the best of our knowledge, this is the first study to directly compare matched plasma and sEV lipid profiles across ovarian cancer histotypes, demonstrating a diagnostic advantage for sEV lipids-particularly for borderline, low-grade serous and mucinous tumours, supporting their integration alongside CA-125 in future differential diagnosis and screening strategies.