Monitoring molecular residual disease in colorectal cancer using tumor-informed ctDNA analysis

Background: Circulating tumor DNA (ctDNA) has become a critical tool for detecting molecular residual disease (MRD) and predicting recurrence in colorectal cancer (CRC). Tumor-informed ctDNA assays depend on a reference mutation profile from tumor tissue, commonly sourced from formalin-fixed, paraffin-embedded (FFPE) or fresh-frozen (FF) samples. Although FFPE tissues are widely available, they are susceptible to formalin-induced artifacts that may affect mutation calling accuracy. This study evaluated the sensitivity and specificity of ctDNA MRD assays, compared the mutation profiles and ctDNA MRD performance between FFPE and FF tissues, and evaluated the prognostic value of ctDNA. Methods: We analyzed 28 CRC patients after curative-intent surgery. Whole-genome sequencing of 18 FFPE and 16 FF tumor samples (including six matched pairs) was conducted to assess variant burden, allele frequency, and mutational signatures. Plasma ctDNA was measured pre- and post-operatively using the tandem error-correction sequencing platform (AccuScan). Performance metrics for recurrence were evaluated at a landmark period (2-6 weeks post-surgery) and during longitudinal follow-up. Disease-free survival (DFS) was analyzed with Kaplan-Meier methods. Results: FFPE samples exhibited a higher median variant count than FF tissues (8,774 vs. 4,906, p = 0.001) and lower variant allele frequency (17.54% vs. 21.38%, p = 0.033), along with enriched A>T transversions ( p = 0.031). Despite these differences, ctDNA MRD detection showed 100% specificity and positive predictive value for recurrence across tissue types. FFPE-informed testing showed higher sensitivity during longitudinal monitoring (88.9% vs. 71.4%). MRD positivity was strongly associated with reduced DFS ( p < 0.0001). In two cases, recurrence detected by FFPE-informed testing was missed with FF-based analysis, suggesting potential false negatives with FF samples. Discussion: These findings support that FFPE tissues, despite more artifacts, are a clinically practical and often more sensitive source for ctDNA MRD assays after bioinformatic processing. The high specificity of ctDNA underscores its utility in guiding treatment decisions. This study confirms the prognostic value of ctDNA and promotes the use of archival FFPE tissues to improve test accessibility without compromising accuracy, facilitating the integration of liquid biopsy into personalized CRC management.