Circulating tumor DNA (ctDNA) tumor fraction (TF) dynamics to refine progression-free survival and radiographic response during anti-EGFR rechallenge in metastatic colorectal cancer.

3572 Background: Reliable biomarkers of benefit from anti-EGFR rechallenge are needed in anti-EGFR-refractory metastatic colorectal cancer (mCRC), where prior studies have shown modest progression-free survival (PFS) and radiographic response may incompletely capture biological sensitivity. We evaluated longitudinal ctDNA dynamics as a real-time biomarker of treatment effect. Methods: This analysis was conducted within a phase II enrichment study of panitumumab with or without trametinib in anti-EGFR-refractory mCRC; crossover was allowed after progression on monotherapy. Plasma was collected at baseline, prior to each infusion, and at progression or end of treatment. Patients with a baseline and ≥ 1 subsequent on-treatment sample were evaluable. ctDNA TF was estimated with the Guardant Reveal assay by normalizing cancer-specific differentially methylated regions with matched controls in each sample. In samples with detectable ctDNA, Guardant360 Liquid was used for genotyping of > 700 genes and tumor mutational burden (TMB). ctDNA TF dynamics were defined by best % change from baseline. Baseline TF, TMB, and focal amplifications in the MAPK pathway were also evaluated. PFS was the primary endpoint and best radiographic response (RECIST 1.1) was the secondary endpoint. Differences were assessed using Kaplan-Meier analyses, Cox proportional hazards models adjusted by age, sex, line of therapy, and treatment regimen, and the Kruskal-Wallis test. Results: A total of 38 patients were included. Consistent with the advanced clinical setting, median TF at baseline was 11.9%. Following adjusted Cox modeling, baseline ctDNA TF was prognostic when dichotomized at 2% (median PFS (mPFS): 3.55 vs 1.97 months, HR 0.32, 95% CI 0.13-0.77), whereas baseline TMB and focal amplifications in the MAPK pathway were not associated with PFS. ctDNA TF % change from baseline to best on-treatment value was strongly associated with PFS in patients with any decrease vs no decrease (mPFS: 3.16 vs 1.77 months, HR 0.24, 95% CI 0.08-0.69, p = 0.008), and the longest PFS was observed in patients achieving ≥ 80% TF decrease vs all others (mPFS: 3.61 vs 1.87 months, HR 0.20, 95% CI 0.08-0.49, p 80% TF decrease, while in patients with PD as best response, 74% had any TF decrease and 5% had a > 80% TF decrease. No impact of ctDNA-related variables was observed in the crossover setting. Conclusions: ctDNA TF dynamics are strongly associated with PFS and provide clinically meaningful information beyond conventional imaging in mCRC patients undergoing anti-EGFR rechallenge, highlighting ctDNA TF as a powerful tool for refining response assessment and personalizing treatment decisions.