Cancer surgery can remove everything a surgeon and radiologist can see while still leaving behind a tiny population of malignant cells capable of causing recurrence months or years later. This invisible disease is generally described as minimal or molecular residual disease, and circulating tumor DNA has emerged as one of the most promising ways of detecting it. Instead of waiting for a recurrent tumor to become large enough to appear on CT or MRI, a liquid-biopsy assay searches blood for fragments of DNA carrying molecular features derived from cancer cells.
Across colorectal, lung, breast and other solid tumors, detectable ctDNA following curative-intent treatment is consistently associated with a substantially higher risk of recurrence. That has made MRD one of the most commercially competitive areas in diagnostics, but it has also exposed a fundamental difference between a prognostic test and a predictive clinical tool. A test can identify that a patient is at high risk and still fail to prove that changing treatment because of the test improves survival.
Where does circulating tumor DNA actually come from?
Tumor cells release fragments of DNA into the circulation as they die or through other biological processes. These fragments exist among a much larger background of cell-free DNA produced by normal tissues, meaning an MRD assay is trying to identify an extremely small cancer-derived signal inside a large volume of ordinary human DNA.
The challenge becomes greatest after successful surgery or chemotherapy because the entire purpose of the test is to detect disease when very little tumor remains. Metastatic cancer may shed abundant ctDNA, while microscopic residual disease can produce concentrations near or below the technical detection limits of even highly sensitive assays.
Shedding also differs according to cancer type, anatomical location and biological behavior. A negative test therefore does not prove that every malignant cell has disappeared; it may mean the residual tumor is not releasing enough detectable DNA into the sampled blood at that moment.
What is the difference between tumor-informed and tumor-agnostic MRD testing?
A tumor-informed assay begins by sequencing or otherwise profiling the patient’s actual tumor tissue to identify mutations or other molecular features unique to that cancer. The subsequent blood test searches specifically for those personalized markers. This can improve specificity because the assay knows exactly which genomic signals to seek, but it requires access to tumor tissue and additional processing before surveillance can begin.
Tumor-agnostic approaches attempt to detect cancer-derived signals directly from blood without building an individualized panel from the resected tumor. They may analyze mutations, methylation patterns or combinations of genomic and epigenomic features. This can simplify logistics and potentially provide results when tumor tissue is unavailable, but sensitivity and specificity depend heavily on the assay architecture.
Neither approach is automatically superior in every cancer. Tumor-informed tests can offer exceptionally deep personalized tracking, while tumor-agnostic systems may scale more easily and capture cancer evolution that no longer matches the original tumor specimen.
How much earlier can ctDNA detect recurrence than conventional imaging?
Multiple observational studies have found that ctDNA positivity can precede radiological recurrence by months in some cancers. This lead time is one of the most attractive features of MRD monitoring because clinicians may theoretically have an opportunity to intervene while recurrent disease remains microscopic rather than waiting for a measurable metastatic lesion.
Earlier detection is not automatically beneficial, however. Screening history contains many examples where identifying disease sooner increased the period during which a patient knew about the disease without changing the eventual outcome. For ctDNA to improve survival, the earlier treatment triggered by the test must be more effective than waiting until standard evidence of recurrence appears or must allow clinicians to avoid unnecessary therapy safely when ctDNA remains negative.
That is why randomized trials are so important. They test not merely whether ctDNA predicts recurrence but whether using the information to escalate, de-escalate or change therapy produces better clinical outcomes. Commentary around DYNAMIC-III in colon cancer has highlighted how difficult that transition from prognostic marker to practice-changing treatment strategy remains.
Why could ctDNA eventually reduce unnecessary chemotherapy?
Adjuvant chemotherapy is often administered because clinicians know that a proportion of patients harbor microscopic disease after surgery but cannot determine exactly which individuals do. Treatment therefore exposes many patients who may already have been cured surgically to chemotherapy toxicity in order to reduce recurrence among those with hidden cancer.
An accurate MRD test could theoretically separate these populations. Patients who remain ctDNA negative might be candidates for less treatment or closer surveillance, while those with detectable ctDNA could receive intensified therapy or enter trials testing novel interventions.
This de-escalation concept can be just as valuable as identifying high-risk patients. Avoiding unnecessary chemotherapy can reduce neuropathy, marrow suppression, infection, financial toxicity and long-term treatment burden. But de-escalation demands extremely strong evidence because a false-negative test followed by undertreatment could sacrifice a patient’s opportunity for cure.
Why do timing and repeated testing matter?
The bloodstream changes rapidly after surgery. Tissue injury releases large amounts of ordinary cell-free DNA, potentially diluting the proportion of tumor-derived fragments and making immediate postoperative testing difficult to interpret. Waiting too long, however, could delay treatment decisions.
Serial monitoring can partly solve the problem by converting one binary result into a longitudinal signal. A patient who remains negative over repeated samples may represent a different risk category from someone whose ctDNA becomes detectable and rises steadily over time. Reviews increasingly frame MRD as a dynamic biomarker rather than one yes-or-no blood test.
Longitudinal testing can also expose tumor evolution. Mutations that appear at recurrence may differ from the original tumor, potentially creating an opportunity to select targeted therapies according to the molecular characteristics of the emerging disease rather than a years-old surgical specimen.
What are the biggest risks of overinterpreting ctDNA?
False negatives are one concern, particularly for tumors that shed relatively little DNA. False positives can also occur through technical noise or biological phenomena unrelated to the original cancer. Clonal hematopoiesis, in which blood-forming cells acquire age-related mutations, can create genomic signals that resemble tumor mutations unless assays and bioinformatics are designed to distinguish them.
Another problem is actionability. A patient can receive a repeatedly positive MRD result while CT imaging remains normal and no proven treatment exists for that molecular-only disease state. The information may be prognostically accurate while generating considerable anxiety and uncertainty about what treatment should follow.
Standardization remains incomplete as well. Different assays use different sequencing depths, marker numbers, positivity thresholds and sampling schedules, making one company’s negative result difficult to assume equivalent to another company’s negative result.
Could MRD become a routine endpoint in drug development?
Potentially, and this may become one of its largest industry impacts. Traditional adjuvant oncology trials can require years of follow-up before enough patients recur for investigators to measure disease-free or overall survival. If validated MRD clearance or recurrence reliably predicts long-term benefit in a defined setting, molecular endpoints could potentially help identify active therapies earlier.
Regulators and trial designers still need disease-specific validation before ctDNA can substitute routinely for established clinical outcomes. The relationship between molecular response and survival may differ between cancer types, therapies and assay technologies.
The enduring promise of ctDNA lies in its ability to observe a stage of cancer that conventional imaging cannot see. The harder task is determining what medicine should do with that information. Detecting recurrence months earlier sounds inherently valuable, but the true clinical breakthrough occurs only when acting on that early signal either saves more patients or safely spares treatment. That distinction will determine whether MRD testing becomes one of oncology’s most important routine diagnostics or remains primarily an extraordinarily powerful prognostic technology.
