P57 Value of circulating tumor DNA in the differentiation between malignant and benign ovarian tumors

Introduction/Background In the Netherlands, yearly 7600 women are diagnosed with an ovarian tumor. Only five percent of these tumors is malignant. It is challenging to identify all malignant tumors preoperatively. An accurate pre-operative diagnosis is important, because patients with ovarian cancer should undergo surgical staging in an oncology center performed by a gynecologic oncologist. Tissue biopsies could confirm or exclude malignancy preoperatively, but may cause spreading of tumor cells in case of malignancy. ‘Liquid’ biopsies overcome the important limitations of current tests as it is minimally invasive and holds the promise of increasing sensitivity and specificity in detecting ovarian cancer. The aim of this study is to find accurate molecular biomarkers in liquid biopsies to identify malignant tumors in patients with an ovarian tumor. Methodology In this prospective case-control study, we collected liquid biopsies from patients with an ovarian tumor (n=64). We analyzed copy number aberrations in circulating tumor DNA (ctDNA) with low-coverage whole genome sequencing (0.5x). Sequencing data were analyzed using WISECONDOR (WIthin-SamplE COpy Number aberration DetectOR). Results We included 32 patients with a benign ovarian tumor and 32 patients with a clinically low-stage malignant ovarian tumor. Nine patients showed a highly abnormal genetic profile. They were all diagnosed with ovarian cancer (specificity 100%). The rest of the patients had a normal to slightly abnormal genetic profile, including 23 patients with a malignant tumor (sensitivity 28%), of which five turned out to have advanced ovarian cancer after surgery. Conclusion Low-coverage whole genome sequencing of ctDNA to identify early stage ovarian cancer has an extremely high specificity. However, the sensitivity is low, which makes it not implementable as a single test in clinical practice yet. Further training of the algorithm and development of specific software for ovarian tumor analysis could improve the performance. Also combination with other biomarkers or ultrasound models should be investigated. Disclosure Nothing to disclose.