In non-metastatic colorectal cancer (CRC), we evaluated prospectively the pertinence of longitudinal detection and quantification of circulating tumor DNA (ctDNA) as a prognostic marker of recurrence.

The presence of ctDNA was assessed from plasma collected before and after surgery for 184 patients classified as stage II or III and at each visit during 3–4 years of follow-up. The ctDNA analysis was performed by droplet-based digital polymerase chain reaction, targeting mutation and methylation markers, blindly from the clinical outcomes. Multivariate analyses were adjusted on age, gender, stage, and adjuvant chemotherapy.

Before surgery, 27.5% of patients were positive for ctDNA detection. The rate of recurrence was 32.7% and 11.6% in patients with or without detectable ctDNA respectively (P = 0.001). Time to recurrence (TTR) was significantly shorter in patients with detectable ctDNA before (adjusted hazard ratio [HR] = 3.58, 95% confidence interval [CI] 1.71–7.47) or immediately after surgery (adjusted HR = 3.22, 95% CI 1.32–7.89). The TTR was significantly shorter in patients with detectable ctDNA during the early postoperative follow-up (1–6 months) (adjusted HR = 5, 95% CI 1.9–12.9). Beyond this period, ctDNA remained a prognostic marker with a median anticipated diagnosis of recurrence of 13.1 weeks (interquartile range 28 weeks) when compared to imaging follow-up. The rate of ctDNA+ might be underestimated knowing that consensus pre-analytical conditions were not described at initiation of the study.

This prospective study confirms the relevance of ctDNA as a recurrence risk factor in stage II and III CRC before surgery and as a marker of minimal residual disease after surgery that may predict recurrence several months before imaging techniques.