ctDNA has become a powerful marker of molecular residual disease in colorectal cancer, but treatment decisions still outrun the evidence
Circulating tumour DNA ctDNA detected after curative-intent therapy now sits near the center of colorectal cancer surveillance discussions. Its prognostic value is difficult to dispute: when ctDNA remains detectable after surgery or other definitive treatment, recurrence risk rises sharply, and when it clears, the outlook improves. The problem is not whether the biomarker matters biologically. The problem is whether a positive or negative result should directly change treatment in routine care.
Randomised trials have moved the field beyond speculation, and they have done so unevenly. In selected stage II colon cancer, ctDNA-informed management has reduced adjuvant chemotherapy exposure. In stage III disease, non-inferiority has not been established. Escalating cytotoxic therapy for ctDNA-positive molecular residual disease has also failed to improve recurrence-free survival. Those results leave clinicians with a marker that predicts relapse extremely well, yet still does not reliably identify a treatment strategy that improves outcomes when acted on.
Prognosis is established; clinical utility is still being defined
The central tension in molecular residual disease testing is straightforward. A biomarker can be strongly associated with outcome without being ready to dictate therapy. ctDNA after curative-intent treatment captures the presence of microscopic disease that standard imaging cannot see, and that makes it biologically compelling. Residual malignant clones shed fragments of DNA into the circulation long before a lesion becomes radiographically obvious, which is why ctDNA can identify risk earlier than conventional follow-up.
A positive test, however, is not the same as proof that one more drug, a higher-intensity regimen, or earlier initiation of therapy will change the course of disease. The current interventional data show that the biomarker’s prognostic strength has outpaced the evidence for treatment guidance. In practice, that means ctDNA can refine risk estimation now, but it should not automatically function as a stand-alone trigger for omission, intensification, or initiation of chemotherapy outside a trial.
Serial testing adds another layer of meaning. Persistent positivity, transient clearance, stable clearance, and re-emergence are not interchangeable biological states. A single snapshot may miss the kinetics of microscopic disease, while a longitudinal pattern may better reflect whether treatment has suppressed residual clones or whether those clones are expanding again. The biology is dynamic, and the assay must be interpreted with that in mind.
ctDNA is a clock, but it is not yet a treatment algorithm.
Timing, assay design, and sampling drive much of the uncertainty
The usefulness of ctDNA depends on four interacting elements: the biological state represented by positivity, the timing of testing and intervention, assay and sampling heterogeneity, and trial design. Each can shift interpretation. A test drawn too early after surgery may reflect postoperative DNA noise rather than true residual disease. A test drawn too late may identify recurrence risk after the ideal intervention window has already narrowed. Different assays also vary in sensitivity, and sampling schedules differ across studies, which complicates cross-trial comparison.
This matters because ctDNA is being used to answer two different clinical questions at once. One question is prognostic: who is likely to relapse? The other is predictive: who benefits from acting on the result? The first question has a much firmer answer than the second. Trials that reduce chemotherapy exposure in lower-risk settings do not prove that all ctDNA-negative patients can safely avoid treatment, and trials that intensify therapy in ctDNA-positive patients do not yet prove that a positive result can be translated into better survival with escalation alone.
These distinctions become even more important in colorectal cancer, where recurrence biology differs across stages and treatment contexts. Stage II disease and stage III disease do not behave identically, and a strategy that appears safe in one setting may fail in another. The evidence already shows that a one-size-fits-all approach to ctDNA-guided care is not defensible.
Western Pacific studies have driven the field, but implementation remains uneven
Much of the development work has come from the Western Pacific region. Studies including GALAXY and ALTAIR in Japan and Taiwan, DYNAMIC in Australia, FIND and CINTS-R in China, and CLAUDIA in Korea have shaped current thinking about ctDNA in colorectal cancer. Together, they have anchored the biomarker in real-world and prospective trial settings and have helped define where it performs best and where uncertainty remains.
Regional adoption, however, is not uniform. Laboratory infrastructure is uneven, costs remain substantial, regulatory access is inconsistent, and evidence from Southeast Asia and Pacific lower-resource settings is limited. Those barriers are not administrative footnotes; they affect whether ctDNA can be measured reliably, repeated on schedule, and interpreted in a system that can act on the result. A biomarker that requires high analytic rigor and frequent serial sampling can be difficult to scale without coordinated pathways for specimen handling, reporting, and clinical response.
The regional evidence base also shapes which patients are most likely to benefit from testing. Populations represented in well-resourced trial networks may not mirror the case mix, access patterns, or treatment constraints of lower-resource settings. Before ctDNA can be integrated broadly, the operational question has to be answered alongside the biological one: can the test be delivered consistently enough to support care decisions that affect outcomes?
Future trials will need better endpoints than biomarker movement alone
The next generation of studies has to do more than prove that ctDNA status changes. They must show that changing treatment because of ctDNA improves what patients actually experience. That means aligning assay performance with the biological timing of residual disease, matching the intervention mechanism to the risk state being measured, and selecting patient-centred endpoints that capture recurrence, survival, and the real burden of treatment.
A trial that measures only ctDNA clearance risks confusing biomarker response with clinical benefit. Clearance may be encouraging, but it is not the same as durable freedom from relapse. A study that escalates chemotherapy for every ctDNA-positive result may also miss the core issue if the intervention is too blunt for the biology being targeted. Residual microscopic disease may require more tailored strategies than dose escalation alone can provide.
For now, ctDNA has earned a place as one of the strongest prognostic tools in colorectal cancer. It has not yet earned blanket authority to dictate adjuvant treatment decisions outside research settings. Clinicians can use it to sharpen recurrence risk estimates, but the evidence still supports restraint when a result is being asked to do more than prognosis can justify.