Liquid biopsy detection of structural variant breakpoints to monitor ovarian cancer clonal evolution.

e17538 Background: Chemotherapy is effective at inducing remission for most patients with high grade serous ovarian cancer (HGSOC), yet relapse is frequent. Methods to interrogate the clonal composition and molecular properties of minimal residual disease are needed to better understand the mechanisms of HGSOC relapse. To address this unmet need, we developed and validated a novel cfDNA-based approach for tracking on-treatment clonal evolution using tumor-informed structural variants (SVs). Methods: Genomic DNA (gDNA) was isolated from cryopreserve multisite pre-treatment HGSOC biopsies and matched germline samples and whole genome sequencing (WGS) was performed. cfDNA was isolated from contemporary pre-treatment plasma. High confidence SVs were identified from multisite WGS using a consensus method incorporating five published SV calling algorithms. Tumor-informed bespoke quantitative PCR primer/probe sets were designed to interrogate SVs breakpoints in tumor gDNA and matched cfDNA. Results: Optimization of the methodology was first performed using synthetic cfDNA generated by ultrasonication of DNA from ovarian cancer cell lines. WGS was used to identify SVs and SV breakpoint-spanning PCR assay exhibited 90% sensitivity and 89% specificity in detection of WGS-informed SVs. 11 validated SVs breakpoint-spanning PCR assays had specific amplifications in synthetic cfDNA. 8 SVs breakpoint-spanning quantitative PCR primer/probe sets were synthesized for qPCR (real-time PCR) and ddPCR (digital droplet), the custom TaqMan probe assays had significant increased sensitivity and superior specificity. When cutoff value of negative was set to “undetected” in qPCR and 0 copies/ul in ddPCR, the sensitivity and specificity of custom TaqMan probe assays can still be 100%. . After screening 18 SVs by tumor-informed SV breakpoint-spanning PCR assays, 8 SVs custom TaqMan probe qPCR assays demonstrated 100 % sensitivity and specificity in multisite biopsy samples. 3 shared and 3 private SVs custom TaqMan probe ddPCR assays were selected for absolute quantification, none of these assays had false positive or false negative when 0 copies/ul was used for cutoff value. 1 shared SV custom TaqMan probe ddPCR assay measured 0.947 copies/ul in cfDNA and 0 copies/ul in matched germline control. Altogether, our multi-layer screening approach yields the “perfect”- high sensitivity and specificity- SV breakpoint-spanning assays that are ready to apply for SVs detection in cfDNA. Conclusions: Detection of SVs from pre-treatment cfDNA using tumor-informed breakpoint-spanning ddPCR is feasible and may enable a novel and sensitive method for monitoring on-treatment disease burden. Work is ongoing to extend this method to private SVs with a goal of tracking treatment-induced tumor clonal dynamics using cfDNA.