Circulating tumor DNA to detect minimal residual disease, response to adjuvant therapy, and identify patients at high risk of recurrence in patients with stage I-III CRC.

4009 Background: The clinical utility of tracking circulating tumor DNA (ctDNA) as a non-invasive biomarker for detecting minimal residual disease (MRD) and stratifying patients based on their risk of developing relapse has been well established in colorectal cancer (CRC). This study evaluates the detection and longitudinal monitoring of ctDNA in CRC patients pre- and post-operatively, during and after adjuvant chemotherapy (ACT). Methods: The prospective, multicenter cohort study recruited patients (n = 193) diagnosed with resected stage I-III CRC. Plasma samples (n = 1052) were collected at various timepoints with a median follow up of 21.6 months (4.6-38.5 months). Individual tumors and matched germline DNA were whole-exome sequenced and somatic mutations identified. Multiplex PCR assays were designed to 16 tumor-specific single-nucleotide variants to track ctDNA in plasma samples. The study evaluated the relationship between ctDNA status and clinical outcomes including radiologic imaging. Cox regression was used to calculate recurrence-free survival (RFS) in patients stratified by ctDNA status postoperatively and post-ACT. Multivariable analysis was performed with all clinical variables. Best model was selected according to Akaike Information Criterion. Results: Pre-operatively ctDNA was detected in 90% (n = 166/185) of the patients. Post-operative ctDNA status prior to ACT was assessed in 152 patients, of which 9.2% (14/152) were identified to be MRD-positive and 78.5% (11/14) eventually relapsed. In contrast, 10.1% (14/138) of MRD-negative cases relapsed (HR: 16.53; 95% CI: 7.19-38.02; p < 0.001). Longitudinal ctDNA-positive status, post-ACT (n = 84) and post definitive therapy (n = 139) was associated with a 27.92 HR (95% CI: 9.16-85.11; p < 0.001) and a 47.52 HR (95% CI: 17.34-130.3.; p < 0.001), respectively. In the multivariable analysis, longitudinal ctDNA status was the only significant prognostic factor associated with RFS (HR: 53.19, 95% CI: 18.87-149.90; p < 0.001). Serial ctDNA analysis detected MRD up to a median of 9.08 months (0.56-16.5 months) ahead of radiologic relapse with a sensitivity of 79.1% and specificity of 99%. Conclusions: Postoperative ctDNA analyses detect patients with high-risk of recurrence, with near 100% specificity. Early detection of MRD and longitudinal monitoring of ctDNA could guide treatment decisions. Intervention trials to assess the clinical benefit of ctDNA use are underway.