Early ovarian cancer detection with a multiomic liquid biopsy.

Abstract

e17563 Background: Ovarian cancer is associated with one of the highest mortality rates among gynecologic malignancies. However, detecting ovarian cancer early can significantly improve patient prognosis; early-stage disease is associated with five-year survival rates exceeding 90%, compared to approximately 30% for advanced-stage disease. Despite this marked survival benefit, fewer than 25% of ovarian cancers are diagnosed at an early stage, reflecting the limitations of current detection strategies. Methods: This study explores the ability of a multiomic liquid biopsy as an alternative strategy for ovarian cancer detection. This approach utilizes infrared (IR) spectroscopy to interrogate blood samples, producing disease-specific spectral signatures that capture cancer-associated biochemical alterations. 201 patients were included in this proof-of-concept study, 50 with ovarian cancer and 151 with a non-cancer diagnosis. Blood plasma samples were analyzed by the Dxcover Liquid Biopsy Platform and classified with machine learning algorithms. Levels of CA-125 and HE4 were obtained for each patient, enabling a comparison and combination of biomarker and spectral data. Results: The receiver operating characteristic (ROC) curve reported an area under the curve (AUC) value of 0.85. The sensitivity-tuned algorithm reported 90% sensitivity with 62% specificity, and the specificity-tuned model reported 62% sensitivity with 90% specificity. Significantly, the diagnostic performance was unaffected by cancer stage showing enhanced utility as an early detection test. The addition of spectral data to biomarker-only models improved the diagnostic performance overall. Conclusions: Earlier detection of ovarian cancer is associated with improved prognosis and survival. The blood-based test described here offers a low barrier to clinical integration, as it is simple to operate, requires only minimal sample volumes, and delivers rapid results.