Minimal residual disease (MRD) assessment by circulating tumor DNA (ctDNA) to stratify recurrence risk and adjuvant treatment benefit in a real-world colorectal cancer (CRC) cohort.

3616 Background: ctDNA detection following curative surgery is known to predict recurrence in patients (pts) with stage II-III CRC. We investigate ctDNA dynamics in relation to clinical outcomes and their real-world impact on treatment decisions among pts with resected CRC. Methods: Based in a large community hospital system, we conducted a retrospective analysis of a prospectively maintained database inclusive of 407 pts with resected stage II-IV CRC. ctDNA was longitudinally quantified at MRD (2-12 weeks post-op) and surveillance timepoints (tp) using a clinically validated, personalized, tumor-informed 16-plex mPCR-NGS assay. Primary endpoint was the correlation between ctDNA status and disease-free survival (DFS) and overall survival (OS). Secondary objective was to evaluate its impact on treatment decisions. Survival distribution was analyzed by Kaplan-Meier Method. Multivariable Cox proportional hazards models were applied to adjust for clinicopathological covariates: age, gender, T, N, high risk features, obstruction, perforation, KRAS, BRAF, MMR. Results: Of 407 pts, 303 had complete evaluable data for analysis at cutoff 12/5/2025. After median follow-up of 29.6 months (mo), N=2342 ctDNA plasma levels were analyzed: 2 stage I, 119 stage II, 170 stage III, 12 stage IV. ctDNA at MRD tp was available for 259 pts (86%): 99 stage II, 144 stage III. ctDNA was negative at MRD tp in 211/259 (MRD-), positive in 48/259 (MRD+). Of 303 pts,180 (60%) received adjuvant chemotherapy (ACT): 38 MRD+, 118 MRD-. 123 (40%) were observed: 9MRD+, 93 MRD-. MRD positivity was significantly associated with inferior DFS in all stages combined (HR = 13.4, 95% CI: 7.64–23.64, p<0.0001) and in all stage-stratified subgroups. Multivariate analysis confirmed ctDNA-positivity to be the most significant prognostic factor associated with DFS when compared with clinicopathologic factors. In MRD+ subgroup, pts with transient ctDNA clearance had significantly worse DFS than those with sustained clearance (HR = 5.27, 95% CI: 2.49–11.14, p<0.0001). Among MRD- pts, ACT conferred no significant DFS benefit vs observation (HR=1.52, p<0.05). Among MRD+ pts, ACT did not reach statistical significance. ctDNA positivity predicted recurrence with a lead time up to 23.9 mo (median 3.8 mo) before imaging, including in low ctDNA shedding sites. 7% had metastatic disease diagnosed earlier because of a positive test, prompting timely therapy. ctDNA testing was associated with higher than historical rates of oligometastasis-directed therapy (17%). Conclusions: Our results highlight the value of post-surgical ctDNA monitoring as a prognostic biomarker in CRC and its potential value for treatment decision making. Clinical utility of ctDNA-guided AC will further be established by ongoing trials (CIRCULATE-NA).