PR002386 (Project)

Description:Background: Ovarian cancer (OC) ranks as the most lethal gynecological malignancy worldwide, with early diagnosis being crucial yet challenging. Current diagnostic methods like transvaginal ultrasound and blood biomarkers show limited sensitivity/specificity. This study aimed to identify and validate serum metabolic biomarkers for OC diagnosis using the largest cohort reported to date. Methods: We constructed a large-scale OC-associated cohort of 1,432 subjects, including 662 OC, 563 benign ovarian disease, and 207 healthy control subjects, across retrospective (n=1,073) and set-aside validation (n=359) cohorts. Serum metabolic fingerprints (SMFs) were recorded using nanoparticle-enhanced laser desorption/ionization mass spectrometry (NELDI-MS). A diagnostic panel was developed through machine learning of SMFs in the discovery cohort and validated in independent verification and set-aside validation cohorts. The identified metabolic biomarkers were further validated using liquid chromatography MS and their biological functions were assessed in OC cell lines. Findings: We identified a metabolic biomarker panel including glucose, histidine, pyrrole-2-carboxylic acid, and dihydrothymine. This panel achieved consistent areas under the curve (AUCs) of 0.87-0.89 for distinguishing between malignant and benign ovarian masses across all cohorts, and improved to AUCs of 0.95-0.99 when combined with risk of ovarian malignancy algorithm (ROMA). In vitro validation provided initial biological context for the metabolic alterations observed in our diagnostic panel. Interpretation: Our study established a reliable serum metabolic biomarker panel for OC diagnosis with potential clinical translations. The NELDI-MS based approach offers advantages of fast analytical speed (~30 seconds/sample) and low cost (~2-3 dollars/sample), making it suitable for large-scale clinical applications.