Monitoring Molecular Residual Disease in Colorectal Cancer Using Tumor-Informed ctDNA Analysis

Abstract

: Background: The early detection of molecular residual disease (MRD) is critical for predicting recurrence and guiding management in colorectal cancer (CRC). We aimed to evaluate the performance of tumor-informed circulating tumor DNA (ctDNA) analysis in monitoring MRD after curative-intent surgery. Methods: In this cohort study of 28 resected CRC patients, tumor-informed variants were identified from formalin-fixed paraffin-embedded (FFPE) or fresh-frozen (FF) tissues using whole-genome sequencing. Post-operative plasma ctDNA was analyzed with the next-generation sequencing-based AccuScan platform at landmark (2–6 weeks) and longitudinal time points. Results: ctDNA-based MRD detection achieved 100% specificity (95% CI: 78.2–100%) and positive predictive value for recurrence in this cohort of 28 patients, which included 13 recurrence events with a median follow-up of 36.2 months. Sensitivity was 83.3% (95% CI: 35.9–99.6%) for patients with FFPE-guided analysis ( n = 13) and 85.7% (95% CI: 42.1–99.6%) for the combined cohort (FFPE-and/or FF-guided analysis ( n = 20) at the landmark time point, improving to 88.9% and 92.3%, respectively, with longitudinal monitoring. MRD-positive status was associated with significantly reduced disease-free survival ( p < 0.0001) in unadjusted analyses. Although FFPE-derived sequencing introduced more artifacts, bioinformatic filtering preserved assay accuracy, and the concurrent use of both reference types in a subset of patients highlighted the potential to improve overall detection sensitivity. Conclusions: These findings suggest that tumor-informed ctDNA analysis provides a promising approach for post-operative MRD surveillance and recurrence risk stratification in CRC. However, the small sample size, lack of external validation, and absence of neoadjuvant-treated patients preclude definitive conclusions. Independent prospective validation is required.