Risk stratification by minimal residual disease detection and ctDNA dynamics from personalised assays in stage II-III colorectal cancer in a UK multi-centre prospective study (TRACC Part B)

Background Circulating tumour DNA (ctDNA) detects minimal residual disease (MRD) and predicts recurrence in resected colorectal cancer (CRC), with implications for adjuvant chemotherapy (ACT) decision-making. Longitudinal ctDNA quantification and dynamics are less well studied. We present results from the ongoing prospective multi-centre TRACC study using a tumour-informed ctDNA assay. Methods TRACC Part B recruited stage I-III CRC patients undergoing curative surgery +/- neo-adjuvant (chemo)radiotherapy and ACT. Blood samples were collected pre-treatment, post-operatively, 3-monthly for year 1 and 6 monthly for years 2–3, with annual imaging. We report a cohort with stage II-III CRC ( n  = 122), including a subset with locally advanced rectal cancers ( n  = 23). Plasma samples were analysed using a personalisedmultiplex polymerase chain reaction next-generation sequencing (mPCR-NGS) ctDNA assay (Signatera™) targeting mutated loci identified in tumour tissue. The primary endpoint was recurrence-free survival (RFS) by ctDNA status. Exploratory objectives included ctDNA quantification and dynamics. Median follow-up was 35 months (range: 2–64). (NCT04050345) Results Of 122 patients, library preparation for whole exome sequencing of primary tumour tissue was possible in 116. Pre-treatment ctDNA detection rate was 94.0% (109/116). ctDNA detection rate 2–8 weeks post-operatively (MRDpos) was 12.1% (14/116); 10.9% in stage III (23.8% high-risk, 4.7% low-risk) and 13.5% in stage II (23.5% high-risk, 8.6% low-risk). Of MRDpos patients, 78.6% (11/14) recurred. Overall, 18.6% (19/102) MRD negative (MRDneg) patients recurred. Twenty-four-month RFS was 30.8% in MRDpos versus 88.1% in MRDneg patients (HR 8.4, 95% CI 4–18; p < 0.001). In a multivariate analysis, ctDNA detection remained the most significant prognostic factor (HR 7.2, 95% CI 2.94–17.6; p < 0.001)followed by nodal stage (HR 2.58, 95% CI 1.01–6.6; p = 0.047). Among high-risk CRC, ACT conferred survival benefit only in MRDpos patients. Longitudinal ctDNA sampling identified relapse with 80% sensitivity. ctDNA concentration increased exponentially prior to recurrence ( n  = 11). ctDNA growth rate was prognostic of RFS (p = 0.021). ctDNA detection was lower in patients with lung-only metastases. Conclusions Post-operative and longitudinal tumour-informed ctDNA testing identifies patients at high CRC recurrence risk and may guide ACT decisions. Further studies are needed to establish the interpretation and utility of longitudinal ctDNA quantification and dynamics to personalise patient management.