Whole-genome cell-free DNA (cfDNA) changes as a dynamic blood-based biomarker for early response assessment of advanced tumors.

Abstract

3041 Background: Liquid biopsies have potential clinical utility as dynamic biomarkers for treatment response. We analyzed serial changes in whole-genome (WG) cfDNA to identify patients with disease progression prior to routine imaging. Methods: We prospectively collected clinical data and blood from 69 advanced cancer patients (28 lung, 25 breast, 16 other). Blood was collected at baseline prior to initiation of a new treatment and at one or two additional timepoints (median 21 and 42 days). We isolated plasma cfDNA and prepared sequencing libraries for WG sequencing or WG bisulfite sequencing (median depth 20X). We quantified changes in the fraction of tumor-derived cfDNA over the initial course of treatment to predict progression vs. no progression. Treatment response at first post-treatment imaging was determined by RECIST 1.1 and clinical assessment. Study endpoints were agreement with first post-treatment imaging and progression-free survival (PFS) by cfDNA prediction. Results: Median age of patients was 70 and 59% were female. Patients were treated with the following therapies: chemotherapy (37), immunotherapy (17), endocrine (9), or targeted therapy (6). Patients with predicted progression by cfDNA (14), indicated by an increase in tumor fraction at either post-treatment blood collection, had shorter PFS (median 63 days) compared to patients without an increase (N = 55; median 255 days), with hazard ratio of 10.3 (95% confidence interval 4.6-23.4, log-rank P = 1x10-11). Positive predictive value was 100% for disease progression and negative predictive value was 78%. These findings were consistent in subset analyses of patients with lung (log-rank P = 2x10-5), breast (log-rank P = 3x10-4), and those treated with immunotherapy (log-rank P = 5x10-6). Conclusions: Our results show the ability to detect early disease progression with high fidelity using WG cfDNA prior to first imaging. These findings were consistent across multiple tumor types and treatments, including immunotherapy patients. Once validated, this dynamic, predictive, blood-based biomarker could aid in clinical decision making for early treatment change as a novel and cost-effective approach.