ctDNA and MRD in Colorectal Cancer: What Physicians Need to Know

For physicians who treat colorectal cancer, circulating tumor deoxyribonucleic acid (ctDNA) has moved well beyond a theoretical biomarker. In 2026, the question is no longer whether ctDNA has prognostic value in colorectal cancer. It does. The real question is where ctDNA colorectal cancer testing is ready to influence management, where it should still be considered investigational, and how to interpret results without overreacting to a single blood draw.

The answer is nuanced. The strongest current evidence supports ctDNA as a marker of molecular residual disease after curative-intent treatment, especially in stage II colon cancer. The weakest use case is broad, unsupervised treatment selection across all colorectal settings without regard to assay type, timing, or the limits of the evidence base. 

That distinction matters because adjuvant decision-making in colorectal cancer has always been imprecise. Pathologic stage, T4 status, lymphovascular invasion, inadequate nodal sampling, perforation, obstruction, tumor budding, and mismatch repair (MMR) status remain clinically important, but they do not perfectly identify who actually harbors occult systemic disease after resection. ctDNA offers something different: a direct molecular signal that residual tumor is still present, even when imaging and routine laboratory studies remain negative. That is why colorectal cancer has become one of the leading disease settings for measurable residual disease (MRD) development. 

The practical message for oncologists is this: ctDNA in colorectal cancer is already a high-value prognostic biomarker. It is increasingly useful in stage II colon cancer treatment discussions, promising but less settled in stage III disease, and still investigational in many rectal cancer and surveillance workflows. The best way to use it today is not as a replacement for pathology or clinical judgment, but as an additional layer of risk stratification that must be interpreted in context. 

What ctDNA and MRD mean in colorectal cancer

ctDNA refers to tumor-derived DNA fragments released into the bloodstream. In colorectal cancer, ctDNA can be used for several purposes across the disease continuum, including molecular profiling in advanced disease, tracking clonal evolution, monitoring treatment response, and detecting molecular residual disease after curative-intent therapy. When ctDNA remains detectable after surgery or after completion of adjuvant therapy, it is generally interpreted as evidence of MRD, meaning residual disease below the threshold of conventional radiographic detection. 

This distinction between ctDNA as a general liquid biopsy tool and ctDNA as an MRD tool is important. In metastatic disease, physicians may order plasma-based profiling to look for RAS, BRAF, HER2, NTRK, or resistance alterations. In the postoperative setting, the physician is not primarily asking what target is present. The physician is asking whether enough residual malignant burden remains to predict recurrence or justify additional therapy. That is why MRD discussions should be framed around recurrence risk, assay performance, and actionability, not just biomarker detection. 

The appeal in colorectal cancer is obvious. Standard postoperative imaging is relatively insensitive to microscopic disease, and conventional clinicopathologic risk features remain blunt tools. ctDNA offers a dynamic molecular readout of postoperative disease biology. In the NCI white paper on ctDNA applications in colorectal cancer, the panel identified MRD detection as one of the four key areas most likely to change practice, alongside rectal cancer management, treatment-response monitoring, and clonal tracking in metastatic disease. 

Why this topic matters now

The current conversation around ctDNA colorectal cancer testing is more urgent than it was even a few years ago because the field has matured from retrospective associations to randomized and prospective clinical studies. In stage II colon cancer, the DYNAMIC trial established that a ctDNA-guided approach could reduce the proportion of patients receiving adjuvant chemotherapy compared with standard management, without compromising recurrence outcomes. Mature five-year follow-up has reinforced the durability of that signal. 

At the same time, later-stage and broader treatment-selection questions remain unsettled. In stage III colon cancer, postoperative ctDNA is clearly prognostic, but the 2025 DYNAMIC-III results showed that de-escalated chemotherapy in ctDNA-negative patients did not meet non-inferiority against standard care, and exploratory escalation in ctDNA-positive patients did not improve outcomes over standard management. That result is clinically important because it reminds physicians that a powerful prognostic biomarker does not automatically become a reliable treatment-selection biomarker in every setting. 

That is also why the field still sits in a middle ground between promise and routine use. A 2024 JNCCN review described ctDNA/MRD testing in colon cancer as a work in progress rather than standard prime-time care, while a 2025 authors’ reply in the same journal argued that stage II data now support both strong prognostic utility and predictive relevance for adjuvant chemotherapy benefit in that narrower setting. The tension between those views is not a contradiction. It reflects a field in which stage II colon cancer is becoming more actionable, while broader colorectal adoption still requires discipline and setting-specific caution. 

How current assays differ in practice

One reason interpretation remains challenging is that not all ctDNA assays are the same. Broadly, MRD platforms in colorectal cancer can be divided into tumor-informed and tumor-agnostic approaches. Tumor-informed assays first sequence the patient’s tumor and then track a personalized set of variants in plasma. Tumor-agnostic assays do not require prior tumor sequencing and instead use preset genomic or epigenomic features. In a 2025 diagnostic meta-analysis, tumor-informed assays showed markedly higher sensitivity than tumor-agnostic assays in serial monitoring after colorectal cancer resection, with no significant worsening of false-positive rates. 

That difference matters in the clinic. Tumor-informed assays generally offer stronger specificity and may reduce false positives from clonal hematopoiesis, but they require tissue availability and can delay turnaround because the tumor must first be sequenced. Tumor-agnostic assays may be faster and more operationally convenient, but sensitivity and specificity depend heavily on platform design, sampling strategy, and the biological context. An NCI white paper explicitly emphasizes that rapid turnaround is crucial in curative-intent settings because postoperative adjuvant decisions are time-sensitive, and it notes that false positives from CHIP can be mitigated by tumor-informed design, molecular barcoding, and comparison with germline or leukocyte DNA. 

Sampling strategy also matters as much as platform selection. A single postoperative time point can be informative, but serial testing is often more clinically useful because it captures dynamics rather than a static snapshot. The NCI white paper notes that serial and/or larger plasma-volume approaches currently provide the best sensitivity for MRD detection, and more recent colorectal data have consistently shown that serial monitoring strengthens prognostic discrimination. 

Physicians should also remember that negative ctDNA does not mean zero risk. The NCI consensus notes that blood-based ctDNA assays will likely hit a ceiling because some tumors are low shedders and certain metastatic sites, especially peritoneal and nervous system disease, may be less reliably detected in plasma. That biological reality is one reason ctDNA-negative results are reassuring but not absolute, especially in high-risk clinicopathologic scenarios. 

Where the evidence is strongest today

Stage II colon cancer

Stage II colon cancer is where ctDNA has the clearest near-term clinical relevance. The original randomized DYNAMIC trial asked whether postoperative ctDNA could guide adjuvant chemotherapy decisions more precisely than conventional clinicopathologic criteria. The study showed a meaningful reduction in chemotherapy use, 15% in the ctDNA-guided arm versus 28% with standard management, without compromising recurrence outcomes. Subsequent mature follow-up continued to support comparable long-term outcomes with this biomarker-guided approach. 

Why is this clinically important? Because stage II disease has always been the gray zone of adjuvant decision-making. Many patients are cured by surgery alone, but a minority recur despite “low-risk” pathology, while others with high-risk features may be overtreated. ctDNA helps narrow that uncertainty. The 2025 DYNAMIC follow-up further suggested that ctDNA burden and end-of-treatment ctDNA may provide additional refinement within the ctDNA-positive subgroup, which raises the possibility that ctDNA may eventually do more than simply separate positive from negative. 

That said, ctDNA should not erase the rest of the pathology report. DYNAMIC five-year reporting noted continued interest in T4 subgroup outcomes, and current adjuvant decisions still need to integrate MMR status, T stage, nodal yield, and other risk features. The same long-term report also reiterated that current guidelines do not recommend adjuvant therapy for the roughly one-fifth of stage II colon cancers that are dMMR, given their favorable biology, even though recurrence can still occur in a subset. For practicing physicians, that means ctDNA is best used as an added layer of individualized risk assessment, not as a stand-alone override of biology and stage. 

Stage III colon cancer

In stage III colon cancer, the evidence is both compelling and cautionary. Long before the randomized de-escalation and escalation studies matured, observational work had already shown that postoperative ctDNA is strongly associated with recurrence risk. In the 2019 JAMA Oncology cohort study, three-year recurrence-free interval was substantially lower in patients with detectable ctDNA after surgery than in those without detectable ctDNA, and the same pattern held after completion of chemotherapy, where post-chemotherapy ctDNA positivity identified a group at especially high risk of recurrence. 

Those data helped establish ctDNA as a strong prognostic classifier in stage III disease, but they did not prove that changing treatment based on ctDNA would improve outcomes. That is why the randomized DYNAMIC-III trial matters. The published 2025 data showed that ctDNA-negative patients had excellent outcomes overall, but ctDNA-guided de-escalation did not meet non-inferiority compared with standard adjuvant management. In ctDNA-positive patients, exploratory treatment escalation did not improve recurrence-free survival over standard management, even though ctDNA burden itself remained highly prognostic. 

This is the key lesson for physicians: in stage III colon cancer, ctDNA is already clinically informative, but it is not yet a universal license to shorten therapy in ctDNA-negative patients or intensify therapy in ctDNA-positive patients outside carefully defined contexts. It can refine prognosis, guide counseling, and help identify patients for clinical trials. It cannot yet be treated as a fully validated replacement for the evidence base that supports oxaliplatin-based adjuvant therapy in node-positive disease.  

Rectal cancer

Rectal cancer is often included in colorectal cancer ctDNA discussions, but the data remain less mature than in colon cancer. A 2025 review in The Oncologist concluded that the utility of ctDNA in nonmetastatic rectal cancer is still under investigation, and the older NCI white paper had already flagged rectal cancer as an area of major promise but limited data. That makes sense biologically. Total neoadjuvant therapy, nonoperative management, lower tumor burden after chemoradiation, and the watch-and-wait paradigm all make rectal ctDNA interpretation more complicated than straightforward postoperative stage II colon disease.  

What is encouraging is that newer prospective work is beginning to address this gap. Recent rectal cancer studies have linked persistent ctDNA positivity after chemoradiation or during watch-and-wait surveillance with worse outcomes, but the evidence is not yet strong enough to make ctDNA the central determinant of TNT tailoring or nonoperative management decisions in routine practice. For now, physicians should view rectal ctDNA as promising and research-active, but still secondary to established clinical, radiographic, and endoscopic assessments. 

Resected metastatic disease

MRD is not only an early-stage question. In patients with resected colorectal liver metastases, serial ctDNA analysis has also shown meaningful prognostic value. A 2021 prospective cohort study in PLOS Medicine confirmed that postoperative ctDNA in resectable colorectal liver metastases is strongly associated with recurrence risk and suggested that serial analysis could help inform perioperative strategy. Still, that is not the same as saying ctDNA-guided intervention is standard of care after metastasectomy. The biomarker is informative, but the optimal treatment response to a positive result remains unsettled. 

How to interpret common results in clinic

A positive postoperative ctDNA result

A positive ctDNA result after curative-intent resection should be treated as a serious biologic warning sign. Across stage II, stage III, and broader resectable colorectal cohorts, postoperative ctDNA positivity has consistently been associated with markedly higher recurrence risk and shorter disease-free survival. In the GALAXY study, postoperative ctDNA positivity was the single strongest prognostic factor for inferior DFS and OS in multivariable analysis, and patients with ctDNA positivity at any surveillance time point were dramatically more likely to recur than those who remained serially negative.  

But positive does not mean “radiographically recurrent today,” and it does not automatically dictate the same action in every patient. In stage II colon cancer, a positive postoperative result makes the argument for adjuvant therapy much stronger. In stage III disease, it identifies a high-risk subgroup, but current randomized data do not yet prove that empirical escalation improves outcomes. In postoperative surveillance, a positive result may justify earlier imaging, multidisciplinary review, or clinical trial discussion, but starting systemic therapy solely on a biomarker result without a defined evidence-based pathway remains controversial. The NCI has explicitly highlighted this uncertainty in real-world use. 

A negative postoperative ctDNA result

A negative postoperative ctDNA result is reassuring, but it is not equivalent to cure. In stage II colon cancer, the DYNAMIC data make a ctDNA-negative result particularly meaningful because outcomes remained excellent even with less chemotherapy use. In observational surveillance cohorts, serially negative results are associated with very low recurrence risk. However, false negatives remain possible in low-shedding disease, in certain metastatic distributions, or when assay sensitivity is not high enough for true de-escalation use. 

In practice, a negative result should lower risk, not erase it. A ctDNA-negative T3N0 patient with favorable biology is very different from a ctDNA-negative T4 lesion with limited nodal sampling, positive tumor budding, or other high-risk features. Physicians should resist converting a probabilistic biomarker into a binary reassurance test. ctDNA is a risk modifier, not a replacement for the rest of the case. 

ctDNA clearance during or after adjuvant therapy

ctDNA clearance is one of the most clinically interesting signals in the field. In the GALAXY data, patients with detectable postoperative ctDNA who received adjuvant therapy and subsequently cleared ctDNA had a lower risk of recurrence than those who remained positive. Likewise, in stage III colon cancer, persistent ctDNA positivity after completion of adjuvant chemotherapy in the 2019 JAMA Oncology cohort identified a group with particularly poor prognosis. 

This raises an important clinical question: should ctDNA clearance become a surrogate endpoint or a decision point during adjuvant therapy? The NCI consensus argues that ctDNA clearance deserves evaluation as a surrogate endpoint in adjuvant trials. That is a reasonable research direction, but physicians should be careful not to overextend it in routine practice. Clearance is biologically encouraging, but outside a trial, it has not yet been fully validated as a reason to stop, extend, or switch therapy in a standardized way across colorectal cancer settings. 

Conversion from negative to positive during surveillance

A patient who converts from ctDNA-negative to ctDNA-positive during surveillance represents one of the most challenging real-world scenarios. The signal is often clinically meaningful. In BESPOKE CRC, patients who converted after an initially negative result were more likely to recur, and the NCI report noted that ctDNA was detecting recurrent disease months before scans in many cases. Earlier JNCCN review data similarly found that ctDNA turned positive before imaging more often than CEA and with a longer median lead time. 

The unresolved question is what to do with that lead time. Earlier molecular detection is not automatically equivalent to better survival unless it changes management in a way that benefits patients. For some patients, especially those who may become candidates for metastasis-directed treatment of oligometastatic relapse, earlier detection could matter a great deal. For others, it may mostly intensify anxiety and imaging without changing outcome. That is one reason serial ctDNA surveillance is promising, but still not a simple plug-and-play substitute for standard surveillance algorithms. 

Where ctDNA helps in surveillance, and where it still falls short

There is little doubt that ctDNA can identify recurrence earlier than conventional methods in at least some colorectal cancer populations. The BESPOKE report cited detection roughly six to nine months before imaging in recurrent cases, and review data in JNCCN found a median lead time for ctDNA positivity of 167 days versus 61 days for CEA. A 2024 epigenomic study in resected colorectal cancer also reported a median lead time of 5.3 months before clinical detection of recurrence.  

However, surveillance is where physicians need to be especially disciplined. Earlier recurrence detection is only valuable if there is a clinically actionable pathway. At present, the strongest rationale exists for patients who could benefit from curative-intent local treatment of limited metastatic disease, earlier trial referral, or closer imaging to localize molecular relapse. The weakest rationale is routine biomarker surveillance without a defined plan for what a positive result will trigger. ctDNA should enhance surveillance strategy, not create a new source of unmanaged ambiguity. 

Practical barriers and clinical pitfalls

The first pitfall is assuming that all positive tests mean the same thing. Assay design, timing, blood volume, prior therapy, tumor shedding biology, and CHIP can all affect results. The NCI white paper specifically warns that false positives may arise from clonal hematopoiesis, including alterations in genes such as TP53 and KRAS, and recommends approaches such as leukocyte matching and tumor-informed design to reduce this problem. 

The second pitfall is assuming that a negative result gives permission to disregard everything else. The same white paper notes that assay sensitivity in MRD detection is constrained by tumor burden, input molecules, and the biology of non-shedding tumors, and it recommends very high sensitivity thresholds for true de-escalation trials. This is especially important in rectal cancer and in settings where undertreatment could sacrifice cure. 

The third pitfall is operational. Tumor-informed assays can be analytically attractive, but they require adequate tissue, sequencing infrastructure, and turnaround fast enough to inform postoperative decisions. Access also remains uneven across practices and payers. The ASCO Post’s 2025 summary of ctDNA risk stratification in colorectal cancer highlighted that equitable access and optimal real-world application remain unresolved questions. That matters because a biomarker that only some patients can obtain on time is not just a scientific issue. It is a systems-of-care issue. 

The fourth pitfall is overreacting outside evidence-based boundaries. NCI noted in 2024 that ctDNA assays were already being used in some clinics, while also acknowledging real uncertainty about how to act when ctDNA is positive but imaging is negative. That remains a live problem. A physician who sends a test must know in advance what actions are reasonable, what actions are investigational, and how the result will be discussed with the patient. 

What physicians should do now

The most defensible current approach is practical rather than doctrinaire.