Abstract 2279: Serial changes in tumor-derived whole-genome cfDNA fraction to identify early disease progression prior to imaging

Abstract

Abstract Background: Response to cancer treatment is usually determined by clinical exam and imaging assessment. Here, we analyzed changes in tumor-derived whole-genome cell-free DNA (cfDNA) at baseline and after treatment initiation to determine response to treatment prior to routine imaging. Methods: We prospectively enrolled and serially collected blood from 54 patients with metastatic malignancies (21 lung, 20 breast, 13 other tumor types). Baseline blood samples were drawn prior to initiation of a new treatment and at one or two additional time points, after the first cycle (median 21 days) and the second cycle (median 42 days). 4 mL of plasma was separated from peripheral blood collected in Streck Cell-Free DNA Blood Collection Tubes, and cfDNA was isolated from plasma aliquots using Qiagen QIAmp extraction kits. To prepare sequencing libraries, a method optimized for whole genome sequencing (WGS) was used based on the Kapa HyperPrep chemistry. WGS was performed at approximately 25X depth on the Illumina HiSeq X. Based on a patient-specific profile of whole genome features, changes in the fraction of tumor-derived cfDNA were quantified over the initial course of treatment. Imaging was performed per standard practice with treatment response determined by RECIST. Results: Median number of prior treatment lines was 1 [range 0-6]. Patients were treated with the following therapies: chemotherapy (27), immunotherapy (14), hormone therapy (7), or targeted therapy (6). For the entire cohort, patients with predicted progression by cfDNA (n=11), indicated by an increase in tumor fraction at either post-treatment blood collection, had worse event-free survival compared to patients that did not show an increase (n=43) (hazard ratio 8.0, [95% CI 3.4-19.2], log-rank p=4.5 x 10-8). For the patients who were predicted to progress, the cfDNA assay preceded clinical evaluation by a median of 39 days. Median progression-free survival was 62 days for patients with predicted progression versus 232 days for others. All patients with predicted progression were later confirmed to progress at the first follow-up evaluation (11/11, 100% positive predictive value). For the remaining patients, 32 of 43 did not progress (74% negative predictive value). Therefore, sensitivity for the assay was 50% and specificity was 100%. Conclusions: Analyzing tumor-derived cfDNA early in the course of a new therapy holds promise to identify patients with early disease progression across a variety of tumor histologies and types of treatment. Early identification of patients who are not benefitting from treatment will enable initiation of other potentially effective therapies, and reduce unnecessary side effects and cost associated with these treatments. Further studies are warranted to validate these findings in larger cohorts and to confirm the histology and treatment-independent nature of the approach. Citation Format: Andrew A. Davis, Wade T. Iams, David Chan, Michael S. Oh, Robert W. Lentz, Neil Peterman, Alex Robertson, Abhik Shah, Rohith Srivas, Nicole Lambert, Tim Wilson, Peter George, Becky Wong, Ayse Tezcan, Ram Yalamanchili, Ken Nesmith, John C. Spinosa, Haluk Tezcan, Young Kwang Chae. Serial changes in tumor-derived whole-genome cfDNA fraction to identify early disease progression prior to imaging [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2279.