Liquid biopsies in ovarian cancer: clinical applications of cell-free DNA analysis

The analysis of cell-free DNA analysis has the potential to improve oncological care for women with ovarian cancer. In order to investigate this assumed potential, we delineated 3 translational settings in which cell-free DNA (cfDNA) analysis could demonstrate clinical utility. In a first part, we evaluated cfDNA as a tool to improve ovarian cancer diagnosis in high-risk patients. We explored the diagnostic value of baseline cfDNA analysis for detection of invasive ovarian malignancy in a preselected group of female patients with adnexal masses (chapter 3). The method of detection was based upon the detection of tumor-associated aneuploidy in cfDNA, which showed excellent performance for the detection of invasive malignancy. In this cohort, it outperformed clinically established methods including ultrasound imaging and serum CA-125 testing. This cohort was expanded in an extended analysis (chapter 4), where the focus was also put on the detection of ovarian cancer cases less characterized by aneuploidy. To this aim, we characterized DNA fragmentation patterns in patient cfDNA samples and quantified the deviation from expected fragmentation patterns in healthy individuals as a marker of malignancy, Using this method, we could also identify malignancy in patients with uncommon forms of epithelial ovarian cancer, less characterized by aneuploidy. In a second part, we aimed to further develop cfDNA testing by quantifying the tumor-derived fraction through the detection of tumor-specific genetic variants. We then investigated this quantitative read-out, under the form of a variant allelic frequency, as a biomarker for the load of tumoral disease. We analyzed sequential cfDNA from patients with platinum-resistant high-grade serous ovarian cancer (HGSOC) enrolled in the phase 2 randomized GANNET53 trial. The aim of this translational substudy (chapter 5) was to assess the clinical value of cell-free DNA analysis for early prediction of treatment response and it was shown that the level of tumor-derived cell free DNA present after 4 and 8 weeks of commencing therapy were an accurate and early marker of therapeutic response. Finally, in a third part, we elaborated on genomic biomarkers of deficient DNA repair by homologous recombination (HR) predictive of response to DNA damaging agents in ovarian cancer. In an appendix to chapter 1, we reviewed the current evidence on genomic biomarkers predictive of HR deficiency. In addition, we validated existing biomarkers such as HR gene mutations and genomic scars in a large collection of primary cases of HGSOC (chapter 6) and we investigated the evolution of these genomic biomarkers in paired samples of primary and recurrent HGSOC (chapter 7). The final aim was to assess these biomarkers using cfDNA as a substrate. In a first step to achieve this goal, we set up the ongoing CLIO clinical trial with olaparib monotherapy in relapsed ovarian cancer (chapter 8). The first efficacy results of this trial are presented in the current manuscript (chapter 9).status: publishe