Serial circulating-tumor DNA analysis with a tumor naïve next-generation sequencing panel detects minimal residual disease and predicts outcome in ovarian cancer

Abstract Circulating tumor DNA (ctDNA) may aid in personalizing ovarian cancer therapeutic options. Here, we aimed to assess the clinical utility of serial ctDNA testing using tumor-naïve, small-sized next-generation sequencing (NGS) panels. A total of 296 patients, including 201 with ovarian cancer and 95 with benign or borderline disease, were enrolled. Samples were collected at baseline (initial diagnosis or surgery) and every 3 months after that, resulting in a total of 811 blood samples. Patients received adjuvant therapy based on the current standard of care. Cell-free DNA was extracted and sequenced using an NGS panel of 9 genes: TP53, BRCA1, BRCA2, ARID1A, CCNE1, KRAS, MYC, PIK3CA, and PTEN. Pathogenic somatic mutations were identified in 69.2% (139/201) of patients with ovarian cancer at baseline but not in those with benign or borderline disease. Detection of ctDNA at baseline and/or at 6 months follow-up was predictive of progression-free survival (PFS). PFS was significantly poorer in patients with detectable pathogenic mutations at baseline that persisted at follow-up than in patients that converted from having detectable ctDNA at baseline to being undetectable at follow-up; survival did not differ between patients without pathogenic ctDNA mutations in baseline or follow-up samples and those that converted from ctDNA positive to negative. Disease recurrence was also detected earlier with ctDNA than with conventional radiologic assessment or CA125 monitoring. These findings demonstrate that serial ctDNA testing could effectively monitor patients and detect minimal residual disease, facilitating early detection of disease progression and tailoring of adjuvant therapies for ovarian cancer treatment. Significance: In ovarian cancer, serial circulating tumor DNA testing is a highly predictive marker of patient survival, with a significantly improved recurrence detection lead time compared with conventional monitoring tools.