Tissue-free liquid biopsies combining genomic and methylation signals for minimal residual disease detection in patients with early colorectal cancer from the UK TRACC Part B study

Abstract Purpose: The absence of postoperative circulating tumor DNA (ctDNA) identifies patients with resected colorectal cancer (CRC) with low recurrence risk for adjuvant chemotherapy (ACT) de-escalation. Our study presents the largest resected CRC cohort to date with tissue-free minimal residual disease (MRD) detection. Experimental Design: TRACC (tracking mutations in cell-free tumor DNA to predict relapse in early colorectal cancer) included patients with stage I to III resectable CRC. Prospective longitudinal plasma collection for ctDNA occurred pre- and postsurgery, post-ACT, every 3 months for year 1 and every 6 months in years 2 and 3 with imaging annually. The Guardant Reveal assay evaluated genomic and methylation signals. The primary endpoint was 2-year recurrence-free survival (RFS) by postoperative ctDNA detection (NCT04050345). Results: Between December 2016 and August 2022, 1,203 were patients enrolled. Plasma samples (n = 997) from 214 patients were analyzed. One hundred forty-three patients were evaluable for the primary endpoint; 92 (64.3%) colon, 51 (35.7%) rectal; two (1.4%) stage I, 64 (44.8%) stage II, and 77 (53.8%) stage III. Median follow-up was 30.3 months (95% CI, 29.5–31.3). Two-year RFS was 91.1% in patients with ctDNA not detected postoperatively and 50.4% in those with ctDNA detected HR, 6.5 (2.96–14.5); P 0.0001. Landmark negative predictive value (NPV) was 91.2% (95% CI, 83.9–95.9). Longitudinal sensitivity and specificity were 62.1% (95% CI, 42.2–79.3) and 85.9% (95% CI, 78.9–91.3), respectively. The median lead time from ctDNA detection to radiological recurrence was 7.3 months (IQR, 3.3–12.5; n = 9). Conclusions: Tissue-free MRD detection with longitudinal sampling predicts recurrence in patients with stage I to III CRC without the need for tissue sequencing. The UK TRACC Part C study is currently investigating the potential for ACT de-escalation in patients with undetectable postoperative ctDNA, given the high NPV indicating a low likelihood of residual disease.