ctDNA is becoming a powerful marker of residual disease in colorectal cancer
Circulating tumor DNA, or ctDNA, has moved from a research curiosity to one of the most closely watched biomarkers in colorectal cancer. In resectable disease, it offers a molecular readout of minimal residual disease after surgery or local therapy. In advanced settings, it may help identify recurrence earlier than conventional surveillance. Even so, the central clinical question remains unchanged: does a positive ctDNA result only predict risk, or can it safely guide treatment decisions in routine care?
This narrative review brings together evidence from stage I through III colorectal cancer and selected patients with resected or ablated oligometastatic stage IV disease. It focuses on tumor-informed and tumor-agnostic assays, postoperative sampling, serial testing, landmark testing, and the technical pitfalls that can distort interpretation. The data are consistent on one point. A positive postoperative ctDNA result identifies patients at substantially higher risk of recurrence.
The harder question is what to do with that information at the bedside.
Prognostic power is strong, but clinical action is still uneven
Across prospective studies, postoperative ctDNA positivity repeatedly tracked with a markedly worse prognosis. Serial testing added another layer of information, since rising or persistently detectable ctDNA often signaled molecular relapse before conventional imaging or symptoms. In practical terms, ctDNA can identify patients who are already on a path toward recurrence long before the relapse becomes radiographically obvious.
The biologic rationale is straightforward. Surgery may remove the visible tumor while leaving behind microscopic deposits that continue to shed DNA into the bloodstream. ctDNA therefore serves as a molecular footprint of residual disease. In low-shedding tumors, however, a negative test does not guarantee cure, and an assay that misses small-volume disease can create false reassurance. Clonal hematopoiesis can also complicate interpretation, especially when variants arising from blood cells are mistaken for tumor-derived signals.
A positive result is informative. A negative result is not always definitive.
DYNAMIC changed the conversation, but not the entire standard of care
The randomized DYNAMIC trial in stage II colon cancer remains one of the strongest examples of how ctDNA can influence management without obviously worsening outcomes. A ctDNA-guided approach reduced adjuvant chemotherapy use while preserving long-term recurrence outcomes. For a disease subset in which overtreatment has long been a concern, that result carried real clinical weight. It suggested that molecular risk stratification could spare many patients unnecessary chemotherapy exposure while maintaining oncologic safety.
Yet DYNAMIC should not be stretched beyond what it showed. The trial supports ctDNA-guided de-escalation in a defined setting, not a blanket recommendation for all resected colorectal cancers. It also does not settle whether ctDNA-negative patients across other stages can safely omit treatment, or whether ctDNA-positive patients uniformly benefit from escalation.
That distinction matters in stage III disease, where the evidence is less favorable for routine treatment tailoring based on ctDNA alone.
Stage III escalation remains unproven despite molecular risk selection
The DYNAMIC-III study did not establish non-inferiority for ctDNA-guided de-escalation in stage III disease. It also failed to show a recurrence-free survival benefit from conventional chemotherapy escalation in ctDNA-positive patients. Those findings are important because they cut against a simplistic assumption that molecular positivity automatically identifies a subgroup that benefits from more treatment.
In stage III colon cancer, the appeal of ctDNA-based escalation is obvious. A patient who remains ctDNA-positive after surgery appears biologically high risk, and more chemotherapy seems intuitively appropriate. But intuition is not enough. The available trial data show that risk enrichment does not automatically translate into improved outcomes when therapy is intensified on the basis of ctDNA alone.
For clinicians, the implication is immediate. ctDNA can sharpen prognostic counseling, but it cannot yet be used as a stand-alone trigger for chemotherapy escalation in routine practice.
Molecular relapse detection is earlier than benefit from treatment
The ALTAIR phase III trial added another cautionary note. Treatment of molecular relapse with trifluridine/tipiracil did not meet its prespecified primary disease-free survival endpoint. Earlier molecular detection, by itself, did not guarantee that intervening at that point would improve outcomes. This is one of the most important lessons in the field: a biomarker can detect recurrence before imaging does and still fail to demonstrate that earlier treatment changes the clinical course.
That gap between detection and benefit is where ctDNA currently lives. It can identify risk with impressive sensitivity in many postoperative contexts, yet the downstream therapeutic strategy remains unresolved in several settings. A biomarker that improves surveillance precision is not automatically a biomarker that improves survival.
Assay design, sampling timing, and tumor biology shape interpretation
The review also emphasizes why results can differ across platforms and disease states. Tumor-informed assays track mutations known to exist in the patient’s tumor and can offer higher specificity. Tumor-agnostic approaches avoid the need for tumor tissue but may be more vulnerable to noise in low-level disease. Postoperative timing matters because testing too early can capture transient DNA release from surgical manipulation, while testing too late may delay actionable decisions.
Low-shedding disease remains a major obstacle. Some tumors release very little ctDNA even when residual disease is present, limiting sensitivity. Serial testing helps, but repeated measurements increase complexity and cost. The result is a field in which assay performance is highly context dependent, and interpretation depends on biology as much as on technology.
Current practice supports prognosis and trial enrollment, not routine replacement of surveillance
Professional guidance now draws a careful line between clinical validity and clinical utility. ctDNA has clear prognostic value and can help with recurrence risk counseling, postoperative planning, and clinical-trial eligibility. It can also help identify patients most likely to be considered for research protocols testing intensified surveillance or adaptive therapy.
What it should not yet do is routinely replace standard follow-up or independently determine adjuvant escalation or de-escalation outside evidence-based indications. Imaging, pathology, and conventional clinical risk factors still matter. The available evidence does not support using ctDNA alone to override established standards across all stages of colorectal cancer.
Harmonized platforms and combined risk models are the next hurdle
Future progress will depend on more than additional observational data. Harmonized assays are needed so results can be compared across institutions and trials. Randomized studies must show not only that ctDNA detects risk, but that ctDNA-guided action improves outcomes patients actually care about. Integrated models that combine ctDNA with pathology, transcriptomic data, and AI-derived risk prediction may eventually improve precision beyond any single marker.
For now, the field has reached an uncomfortable but useful point. ctDNA is highly informative, sometimes early, and often prognostic. It is not yet a universal treatment compass for resectable or advanced colorectal cancer, and the trials that will define that role are still in motion.