Circulating tumor DNA to guide adjuvant therapy in colon cancer: what the trials show

Circulating tumor DNA (ctDNA) is tumor-derived DNA in plasma ^[1]. Han and colleagues, in a 2026 meta-analysis, described postoperative ctDNA as a strong marker of molecular residual disease after curative-intent colorectal cancer resection ^[2].

What changed in 2025?

• Stage II, 5-year update (March 2025 online): Tie and colleagues reported 5-year recurrence-free survival of 88% with ctDNA-guided management and 87% with standard management (difference 1.1 percentage points, 95% CI −5.8 to 8.0) ^[3].
• Stage II overall survival: the same 2025 update by Tie and colleagues found 5-year overall survival of 93.8% with ctDNA-guided management versus 93.3% with standard management (hazard ratio 1.05) ^[3].
• Stage III, phase 2/3 (October 2025 online): Tie and colleagues reported that in ctDNA-negative patients, de-escalation did not meet the non-inferiority margin for 3-year recurrence-free survival ^[4].

What do the randomized trials show?

Stage II. In the original 2022 stage II report (n=455, randomized 2:1), Tie and colleagues found adjuvant chemotherapy use of 15% with ctDNA guidance versus 28% with standard management ^[5]. That 2022 Tie trial found 2-year recurrence-free survival of 93.5% versus 92.4%, within its non-inferiority margin of −8.5 percentage points ^[5]. In the 2022 Tie trial, ctDNA-negative patients managed without chemotherapy had 3-year recurrence-free survival of 92.5% ^[5].

Stage III. In the 2025 stage III trial, Tie and colleagues found 3-year recurrence-free survival of 85.3% with de-escalation versus 88.1% with standard management among ctDNA-negative patients, which did not meet the non-inferiority margin ^[4]. In the same trial, 702 of 968 evaluable patients (72.5%) were ctDNA negative 5 to 6 weeks after surgery, and their 3-year recurrence-free survival was 87% versus 49% in ctDNA-positive patients ^[4]. In the 2025 stage III trial, Tie and colleagues found that in ctDNA-negative patients, de-escalation lowered oxaliplatin use from 88.6% to 34.8% and treatment-related hospitalizations from 13.2% to 8.5% ^[4]. Among ctDNA-positive patients, the 2025 Tie analysis found that higher ctDNA burden correlated with recurrence risk (3-year recurrence-free survival 77% to 23% across quartiles) ^[4]. The 2025 Tie analysis also found that escalated therapy in ctDNA-positive patients did not improve 2-year recurrence-free survival over standard management (51% with escalated versus 61% with standard management), a primary endpoint ^[4]. The 2025 Tie analysis reported that persistent ctDNA after treatment predicted markedly worse 3-year recurrence-free survival (14% versus 79%) ^[4].

Where is the evidence uncertain?

In 2024, Nakamura and colleagues analyzed 2,240 patients with resectable stage II-III colon cancer or stage IV colorectal cancer in a Japanese observational study ^[6]. In that 2024 analysis, postoperative ctDNA positivity was linked to shorter disease-free survival (hazard ratio 11.99), and spontaneous ctDNA clearance without clinical recurrence occurred in 1.9% (2 of 105) ^[6]. In a 2025 meta-analysis of seven stage II colorectal studies, Negro and colleagues found a pooled recurrence risk ratio of 3.66 (95% CI 1.25 to 10.72) for postoperative ctDNA positivity ^[7]. Negro and colleagues judged that assay standardization and large validation studies are still needed before routine clinical integration ^[7].

In a 2026 meta-analysis of seven prospective cohorts in resected colorectal cancer, Han and colleagues found a pooled recurrence hazard ratio of 7.60 (95% CI 4.70 to 12.31) for postoperative ctDNA positivity, with substantial heterogeneity (I² 80.6%) and a wide prediction interval (1.60 to 36.25) ^[2]. These three estimates are not comparable: Nakamura's 11.99 is a hazard ratio for disease-free survival ^[6], Negro's 3.66 is a risk ratio for recurrence in stage II disease only ^[7], and Han's 7.60 is a hazard ratio for recurrence ^[2]. In 2026, Han and colleagues concluded that this prognostic signal alone does not establish treatment-decision utility, and that current evidence does not support routine empiric escalation for ctDNA-positive disease ^[2]. A 2024 review by Caughey and Parikh, written before the stage III trial reported, judged the data insufficient for routine ctDNA-guided adjuvant decisions in resectable stage III disease ^[1].

At the point of care: what to check

• Timing of the draw: the 2025 stage III trial sampled 5 to 6 weeks after surgery, and the 2022 stage II trial at 4 or 7 weeks ^[4][5].
• Stage: the 2022 stage II trial met its non-inferiority margin at 2 years, and the 2025 stage III trial did not meet it in ctDNA-negative patients at 3 years ^[4][5].
• ctDNA burden: in the 2025 stage III trial, higher ctDNA burden correlated with recurrence risk in ctDNA-positive patients ^[4].
• End-of-treatment status: in the 2025 stage III trial, persistent ctDNA after therapy predicted markedly worse 3-year recurrence-free survival ^[4].