Correlation between the timing of recurrence and circulating tumor DNA (ctDNA) doubling time in patients (pts) with resected colon cancer.

Abstract

220 Background: Previous studies have demonstrated that persistent ctDNA positivity in pts with colon cancer is associated with significantly worse outcomes compared to persistent ctDNA negativity. However, the association between outcomes and the rate of change in ctDNA levels has yet to be established. Here, we evaluated lead time from ctDNA positivity to recurrence based on rate of ctDNA increase. Methods: Pts with resected stage I-IV colon cancer enrolled in the prospective, observational GALAXY study (UMIN000039205) who had ctDNA testing (Signatera) with ≥2 consecutive ctDNA-positive post-definitive treatment results and clinical relapse information were included (N=164; median 3 timepoints per pt, range 2-7). The rate of change in ctDNA levels (mean tumor molecules/mL) was calculated; a slope of 0.3, corresponding to a 1-month doubling time, was used as a cutoff for fast (≥0.3) and slow (<0.3) change. Recurrence-free survival (RFS) from the first positive ctDNA sample to relapse (lead time) was calculated using the Kaplan-Meier method, and comparisons were made using log-rank tests. Restricted mean survival time (RMST) was used to compare lead time at 3-month intervals. Analyses were performed in all pts and by adjuvant chemotherapy (ACT) status (ACT N=39; no ACT N=125). Results: Overall, ctDNA increase was fast for 58 (35.4%) pts and slow for 106 (64.6%) pts. Compared to pts with slow ctDNA increase, pts with fast ctDNA increase had a significantly inferior RFS (HR: 2.4, 95% CI: 1.7-3.4, p<0.0001). The median lead time from the first ctDNA-positive result to relapse was significantly shorter in fast versus slow increase pts (91 vs 155 days, p<0.0001); RMST demonstrated that the difference in lead time was significantly different at all 3-month time intervals. This trend of inferior survival for fast vs. slow was consistent when analyzing the cohort by ACT status. Among pts who did not receive ACT (fast N=46, slow N=79), and who received ACT (fast N=12, slow N=27), pts with fast ctDNA increase had a significantly inferior RFS compared to pts with slow ctDNA increase (no ACT HR: 2.0, 95% CI: 1.3-2.9, p=0.0008; ACT HR: 7.2, 95% CI: 2.9-17.9, p<0.0001). The median lead time was significantly shorter in the fast versus slow groups, both overall (no ACT 91 vs 147 days, p=0.0003; with ACT 111 vs 229 days, p=0.0005) and at each 3-month interval. Conclusions: Our findings suggest that the rate of ctDNA doubling time is a strong predictor of time to clinical progression, reflecting the tumor’s underlying biology. This may help guide risk assessment and surveillance strategies, such as performing earlier and/or more frequent imaging in those with rapidly increasing ctDNA levels. Clinical trial information: UMIN000039205 .