P1.01-46 Response Assessment Using Plasma Cell-Free DNA (cfDNA) – When Is the Optimal Time to Assess Response?

Abstract

Plasma cfDNA analysis is routine for non-invasive genotyping of advanced NSCLC, however response assessment using plasma cfDNA is not well characterized. We hypothesized that response in cfDNA would be an early process occurring well before routine imaging timepoints. We retrospectively analyzed a total of 48 baseline and serial on-treatment plasma samples collected from 16 patients enrolled across three Experimental Therapeutics Clinical Trials Network (ETCTN) phase I trials of osimertinib combinations in advanced EGFR-mutant NSCLC. For validation, we also retrospectively analyzed a total of 201 baseline and serial on-treatment samples from an institutional cohort of 67 advanced NSCLC patients receiving systemic treatment. Using droplet digital PCR (ddPCR) of key EGFR or KRAS driver mutations, plasma response was defined as any decrease in mutation concentration to below baseline levels. We compared the magnitude of initial (baseline to day 11-30) and subsequent (day 11-30 to day 36-84) plasma response. Finally, we prospectively assessed response using serial amplicon-based plasma next-generation sequencing (NGS) in a pilot cohort of 8 NSCLC patients starting systemic therapy. Of 15 ETCTN patients with any plasma response, best plasma response was seen at the initial response timepoint in 12 patients (80.0%) and ≥90% of the total plasma response was seen at the initial response timepoint in 14 patients (93.3%). In the validation cohort of 61 patients with any plasma response (Figure), best plasma response was seen at the initial response timepoint in 39 patients (63.9%) and ≥90% of the total plasma response was seen at the initial response timepoint in 52 patients (85.2%). Complete plasma responses (-100%) were seen as early as 11 days after initiating therapy. In the prospective clinical cohort, plasma NGS detected genomic alterations and enabled monitoring of changes in mutant allele fraction in all 8 patients. The median turnaround time of the assay was 8 days. Plasma response is an early phenomenon, with the vast majority of plasma response seen within 30 days, and as early as 11 days. These findings suggest that early plasma cfDNA analysis may permit response assessment well before standard imaging timepoints, with potential as an early marker of drug effect. Additional investigation to understand the relationship between early plasma response, radiographic response, and durability of treatment effect is still needed.