Highlight
• DYNAMIC-III (Tie et al., Nat Med 2025) enrolled 968 evaluable stage III colon cancer patients and demonstrates that plasma ctDNA measured 5–6 weeks after surgery is a powerful prognostic classifier: 3-year recurrence-free survival (RFS) was 87% for ctDNA-negative versus 49% for ctDNA-positive patients (P < 0.001).
• In ctDNA-negative patients, a ctDNA-guided de-escalation strategy markedly reduced oxaliplatin exposure (34.8% vs 88.6%) and hospitalizations (8.5% vs 13.2%) but did not meet the non-inferiority margin for RFS (85.3% vs 88.1%).
• In ctDNA-positive patients, higher ctDNA burden predicted recurrence risk, but chemotherapy intensification did not improve 2-year RFS (51% with escalation vs 61% with standard management).
Background
Adjuvant chemotherapy after curative resection reduces recurrence risk in stage III colon cancer at the population level, but benefit varies substantially among individuals and must be balanced against treatment toxicities—most notably oxaliplatin-induced peripheral neuropathy. Circulating tumor DNA (ctDNA), fragments of tumor-derived DNA detectable in plasma, reflects minimal residual disease (MRD) and has emerged as a promising biomarker to refine adjuvant treatment decisions. Prior nonrandomized and small randomized studies suggested that ctDNA positivity after surgery identifies patients at high risk of recurrence, and that ctDNA-guided approaches might safely reduce overtreatment. DYNAMIC-III is a large randomized phase 2/3 trial designed to evaluate whether ctDNA-directed de-escalation or escalation strategies can improve the risk–benefit balance of adjuvant therapy in stage III disease.
Study design
DYNAMIC-III was a multicenter, randomized, phase 2/3 trial conducted by the Australasian Gastro-Intestinal Trials Group (AGITG) in collaboration with the Canadian Cancer Trials Group. Patients with stage III colon cancer underwent plasma ctDNA testing 5–6 weeks after curative surgery and were randomized 1:1 to either ctDNA-guided management or standard management. In the ctDNA-guided arm, ctDNA-negative patients received de-escalated therapy whereas ctDNA-positive patients received escalated therapy; clinicians prespecified the standard regimen per local practice. Primary endpoints were 3-year RFS for ctDNA-negative patients and 2-year RFS for ctDNA-positive patients. Secondary endpoints included treatment-related hospitalization, toxicity, and ctDNA clearance after adjuvant therapy. The trial enrolled 968 evaluable patients; 702 (72.5%) were ctDNA negative at baseline. Median follow-up was 47 months.
Key findings
Prognostic performance of ctDNA
The trial robustly confirmed ctDNA as a potent prognostic classifier. Overall, ctDNA-negative patients had a 3-year RFS of 87% compared with 49% for ctDNA-positive patients (P < 0.001). Moreover, among ctDNA-positive patients a higher quantitative ctDNA burden strongly correlated with recurrence risk: 3-year RFS varied across quartiles from approximately 77% down to 23% (P < 0.001), indicating a graded relationship between ctDNA level and residual tumor burden.
ctDNA-guided de-escalation (ctDNA-negative cohort)
Among the 702 ctDNA-negative patients, clinicians in the ctDNA-guided arm reduced exposure to oxaliplatin-based therapy: 34.8% received oxaliplatin versus 88.6% in the standard-management arm. Treatment-related hospitalizations were also fewer with de-escalation (8.5% vs 13.2%). Three-year RFS was 85.3% in the ctDNA-guided (de-escalated) group versus 88.1% with standard management. Although the absolute difference in RFS was modest, the de-escalation strategy did not meet the prespecified non-inferiority margin for RFS.
ctDNA-guided escalation (ctDNA-positive cohort)
In the 266 ctDNA-positive patients, escalation of systemic therapy (for example longer duration or more intense regimens as per protocol choices) did not improve outcomes. Two-year RFS was 51% in the escalated-treatment group versus 61% in the standard-management group—no benefit observed. There was no signal that intensification reduced early recurrence or improved ctDNA clearance rates compared with standard approaches.
ctDNA dynamics and persistence
Of note, persistent ctDNA after completion of adjuvant therapy identified an extremely high-risk subgroup: patients with persistent ctDNA had a 3-year RFS of only 14%, compared with 79% for those who cleared ctDNA—highlighting the potential of serial ctDNA monitoring to identify patients with treatment-refractory micrometastatic disease.
Safety
No unexpected toxicities were reported. The de-escalation arm reduced oxaliplatin exposure and associated healthcare utilization. Standard chemotherapy-related adverse events were otherwise similar to historical expectations, and escalated therapy did not show novel safety signals that would explain the lack of efficacy.
Expert commentary and interpretation
DYNAMIC-III provides high-quality randomized evidence that ctDNA is a powerful prognostic biomarker in stage III colon cancer. The trial supports two principal conclusions: (1) ctDNA-negative status after surgery identifies patients with substantially lower recurrence risk, and a ctDNA-guided reduction in chemotherapy intensity can meaningfully reduce oxaliplatin exposure and hospitalizations; and (2) simply intensifying conventional chemotherapy for ctDNA-positive patients does not translate into improved RFS, indicating that standard cytotoxic escalation is insufficient to eradicate MRD detected by ctDNA in many cases.
Why might escalation have failed? Several, non-mutually exclusive explanations deserve consideration. First, ctDNA positivity may reflect biologically aggressive or chemotherapy-resistant clones—so higher doses or prolonged duration of oxaliplatin-based regimens may not overcome intrinsic resistance. Second, ctDNA-positive disease may represent systemic micrometastases that require alternative systemic strategies (e.g., targeted agents guided by molecular profiling, or incorporation of immunotherapy where appropriate) rather than conventional cytotoxic intensification. Third, heterogeneity in the prespecified ‘escalation’ interventions and limited statistical power for subgroup comparisons may have diluted a potential benefit.
Importantly, the de-escalation results should be interpreted cautiously. Although the absolute RFS difference between de-escalated and standard care was small and clinical benefits in reduced toxicity are compelling, the non-inferiority margin was not met. Clinicians and guideline panels will need to weigh patient priorities (e.g., preventing neuropathy) against a small potential absolute reduction in RFS when considering ctDNA-guided de-escalation outside of clinical trials.
Limitations and generalizability
Key limitations include the open-label design and heterogeneity in prespecified standard and escalated regimens determined by treating clinicians, which reflects real-world practice but complicates interpretation of which specific intensification strategies (if any) might be effective. The ctDNA assay, its sensitivity and limits of detection, timing of sampling (5–6 weeks post-op), and the single pre-adjuvant assay used for initial stratification also influence generalizability—different platforms or sampling schedules might yield different performance. Finally, although the trial is large, subgroup analyses (particularly within the ctDNA-positive cohort) may be underpowered to detect modest benefits from specific escalated regimens.
Clinical implications and next steps
For practicing clinicians, DYNAMIC-III strengthens the evidence that postoperative ctDNA is a powerful prognostic tool that can help risk-stratify patients with stage III colon cancer. The data support using ctDNA to inform shared decision-making about adjuvant chemotherapy intensity, especially for patients prioritizing avoidance of long-term oxaliplatin toxicity. However, because de-escalation did not meet non-inferiority thresholds, routine adoption outside of trial contexts should be cautious and ideally within pathways that include informed consent and careful follow-up.
For ctDNA-positive patients, the lack of benefit with chemotherapy intensification indicates an unmet need. Future strategies should prioritize enrolling ctDNA-positive patients into biomarker-directed clinical trials that explore targeted therapies, immunotherapy combinations, novel systemic agents, or earlier therapeutic switching based on ctDNA dynamics. Persistent post-treatment ctDNA identifies an extremely high-risk subgroup that could be prioritized for such experimental approaches and for trials using ctDNA clearance as an early surrogate endpoint.
Conclusion
DYNAMIC-III validates postoperative ctDNA as a strong prognostic classifier in stage III colon cancer and demonstrates that ctDNA-guided de-escalation can reduce oxaliplatin use and hospitalizations with outcomes that approach standard care. However, ctDNA-guided escalation of conventional chemotherapy does not improve recurrence-free survival, highlighting the need for novel therapeutic strategies for ctDNA-positive patients and for prospective trials that evaluate targeted or immune-based interventions in this high-risk group. Integration of ctDNA into routine practice should be guided by patient values and ideally occur alongside prospective studies designed to improve outcomes for MRD-positive disease.
Funding and clinicaltrials.gov
Trial registration: Australian New Zealand Clinical Trials Registry Identifier: ACTRN12617001566325. The study was performed by the AGITG DYNAMIC-III Study Group in collaboration with the Canadian Cancer Trials Group. Full funding disclosures are provided in the primary manuscript (Tie J et al., Nat Med. 2025).
References
1. Tie J, Wang Y, Loree JM, Cohen JD, Wong R, Price T, Tebbutt NC, Gebski V, Espinoza D, Burge M, Harris S, Lynam J, Lee B, Lee MM, Breadner D, Debrincat M, Foroughi S, Chantrill L, Lim SH, Gill S, O’Callaghan C, Ptak J, Silliman N, Dobbyn L, Popoli M, Bettegowda C, Papadopoulos N, Kinzler KW, Vogelstein B, Gibbs P; AGITG DYNAMIC-III Study Group. Circulating tumor DNA-guided adjuvant therapy in locally advanced colon cancer: the randomized phase 2/3 DYNAMIC-III trial. Nat Med. 2025 Oct 20. doi: 10.1038/s41591-025-04030-w. Epub ahead of print. PMID: 41115959.