What ctDNA Minimal Residual Disease Testing Can and Cannot Do

The short answer

Circulating tumor DNA (ctDNA) testing for minimal residual disease (MRD) is a genuinely powerful prognostic tool. When fragments of tumor DNA remain detectable in the blood after cancer surgery, the risk of recurrence rises sharply, often flagged before imaging shows anything. What the test cannot yet do is prove that changing treatment based on that result improves survival. Predicting who is likely to relapse and demonstrating that acting on the prediction helps are two different claims, and only the first is well established today. This piece is educational and not medical advice.

What ctDNA MRD testing actually measures

Tumors shed small pieces of DNA into the bloodstream. After a cancer is surgically removed, most of that signal should disappear. If sensitive sequencing still finds tumor-specific DNA weeks later, it suggests that microscopic disease survived the operation, below the resolution of any scan. That residual signal is what MRD refers to in this setting.

Two broad assay designs exist. Tumor-informed tests first sequence the resected tumor, build a personalized panel of that patient's mutations, and then hunt for exactly those variants in blood. Tumor-agnostic tests look for cancer-associated changes without a bespoke panel. The distinction matters for performance, and the evidence favors the tumor-informed approach. A 2025 systematic review and meta-analysis in the International Journal of Molecular Sciences reported that tumor-informed methods carried a stronger association with recurrence than tumor-agnostic ones, and it raised concerns about false-positive results with some tumor-agnostic assays. No assay is perfect, and the type of test shapes how confidently a single result can be read.

What the evidence shows it can do

The prognostic signal is large and consistent. In the CIRCULATE-Japan GALAXY study, an observational cohort led by Nakamura and colleagues and published in Nature Medicine in 2024, patients with detectable ctDNA during the postoperative MRD window had markedly worse outcomes than those without. Across a population that included stage II to IV colorectal cancer, ctDNA positivity during that window was associated with substantially inferior disease-free survival and overall survival, with hazard ratios far above what most single biomarkers produce. These are among the steepest risk gradients reported for a blood test in solid tumors.

Timing is the second strength. The 2025 meta-analysis reported that ctDNA identified recurrence with roughly 85 percent sensitivity, compared with about 41 percent for the standard blood marker CEA, and that it tended to flag relapse before clinical or radiologic recurrence became apparent. Detecting relapse earlier, and catching it in patients a protein marker would miss, is a real analytical advance.

The GALAXY data add a further observation with intuitive appeal. Patients whose ctDNA cleared and stayed cleared after adjuvant chemotherapy fared far better than those in whom clearance was only transient. That pattern suggests the test tracks something biologically meaningful about treatment response, not merely baseline risk.

What it cannot yet do

Here is the distinction that gets blurred in marketing and headlines. A biomarker can be strongly prognostic while still failing to guide treatment. Prognostic means the result predicts the future course. Actionable, in the rigorous sense, means that a randomized trial has shown that changing therapy because of the result leads to better outcomes than not doing so. Those are separate evidentiary bars, and ctDNA MRD has cleared the first far more convincingly than the second.

Consider a positive test after surgery. It reliably identifies elevated recurrence risk. It does not, on its own, prove that adding or intensifying chemotherapy in response will change the trajectory. The patient may relapse regardless, in which case earlier knowledge carries cost without benefit. Consider a negative test. It is reassuring, but sensitivity is not 100 percent, and a clear result does not by itself justify withholding a therapy that guidelines would otherwise recommend. De-escalating real treatment on the strength of a negative ctDNA result is exactly the decision that requires randomized proof before it becomes standard.

The GALAXY study, by design, cannot settle these questions. It is observational. It shows association, tracks who fared better or worse, and generates hypotheses. It was not built to prove that ctDNA-guided treatment changes beat usual care. Its own framing points toward the randomized trials still maturing to test that.

Surrogate endpoint is a separate, higher bar

There is a parallel debate in drug development. A surrogate endpoint is a measurable stand-in, like a lab value, accepted as a proxy for outcomes that matter, such as survival. Qualifying ctDNA clearance as a validated surrogate would let trials read out faster by measuring the marker instead of waiting years for survival data.

Regulators are engaging carefully rather than declaring the question closed. The FDA finalized guidance in late 2024 on using ctDNA in early-stage solid tumor drug development. It treats the marker as a promising candidate for patient selection, enrichment, and early signals, while emphasizing rigorous assay validation and stopping short of endorsing ctDNA response as an established surrogate sufficient on its own to support approval. Promising candidate and validated surrogate are not the same status, and the honest current answer is that the marker sits in the former category.

How to hold both truths at once

None of this diminishes the technology. A test that identifies high recurrence risk before imaging, and outperforms the incumbent blood marker on sensitivity, is a meaningful step. The disciplined reading is narrow. Use the strong evidence for what it supports, which is prognostic information, and wait for randomized results before treating a ctDNA result as a command to escalate or abandon therapy. The ongoing interventional trials, several referenced across the colorectal literature, are precisely the studies that will convert a powerful predictor into a validated decision tool, or reveal where acting on it does not help. Until they report, the most accurate statement is also the most useful. ctDNA MRD testing tells you a great deal about risk, and comparatively little, so far, about what to do next.