ASCO’s First ctDNA Guideline Found Few Instances of Clinical Utility. Residual Disease Got No Recommendation at All.

The headline writes itself: after nearly a decade of liquid biopsy entering routine practice, the American Society of Clinical Oncology has finally issued formal guidance. Read the document and a different story appears. Most of it concerns where ctDNA testing should not be used, and the application carrying the largest commercial expectations received no formal recommendation.

Eight years on, the answer is still largely not yet

In 2018 ASCO and the College of American Pathologists conducted a joint systematic review of ctDNA testing covering preanalytical variables, analytical validity, interpretation and reporting, and clinical validity and utility. Their conclusion was that a clinical practice guideline was not feasible, and they made no formal recommendations. That review has been cited ever since as the field’s foundational assessment.

The 2026 guideline is the sequel, and it is a guideline rather than a review, which is progress. But the panel’s own framing is unusually restrained for a document of this kind. It states that evidence on clinical utility is evolving rapidly, that its relevance is closely tied to the specifics of a patient’s cancer, and that at the time of writing there were few instances in which the panel identified evidence of clinical utility at all.

What the panel does recommend is narrow and conditional. Outside a clinical trial, ctDNA testing should only be offered where the result could actually change a clinical decision. It may be used as a first assessment, a replacement or an addition to tissue-based testing where tissue sampling is difficult, unfeasible, or carries unacceptable risk to the patient. Testing to determine trial eligibility is appropriate. Beyond genetic alteration testing, it may be offered where a specific evidence-based action follows from the result, or where it resolves a conflict or ambiguity in standard-of-care assessment.

Read as a whole, that is a permission structure built around exceptions rather than a positive indication.

The asymmetry between specificity and sensitivity is the operational constraint

The guideline’s co-chairs have described the central technical tension as balancing high specificity against moderate sensitivity. In practice that means a positive ctDNA result is informative and a negative one is not. The guidance follows directly: negative or inconclusive ctDNA results should be confirmed with tissue testing.

That single requirement constrains the substitution case severely. A test that can rule in but cannot rule out does not replace tissue biopsy; it front-runs it in a subset of patients and leaves the tissue pathway intact for everyone else. Where tissue is genuinely unobtainable, that is a real and valuable gain. Where tissue is merely inconvenient, the health economics look different, because the tissue procedure remains in the pathway for every negative result.

Payers read guidelines for indication language, and this one gives them a tightly bounded set. Coverage policies built on the 2026 guideline will reimburse ctDNA where tissue is unobtainable or hazardous, where a specific actionable alteration is being sought, and for trial eligibility. They will have no guideline basis for reimbursing surveillance testing in asymptomatic patients after curative-intent treatment, which is the highest-volume application the sector has been building toward. Anyone modelling reimbursed volume from this document should model the exceptions, not the population.

The commercial centre of the field received no recommendation

Molecular residual disease detection, testing patients after curative-intent treatment to find recurrence before imaging can, is the application that has attracted the most investment and the most aggressive volume forecasting. The guideline addresses it, discusses the emerging evidence, and declines to make formal recommendations at this time.

That is not a rejection. It is an assessment that the evidence has not yet reached the threshold at which a professional society will tell oncologists what to do. The distinction matters commercially, because it means MRD testing continues without guideline endorsement, and therefore without the coverage leverage that endorsement provides. The panel names standardising MRD definitions as an outstanding question, which is a plain statement that the field has not yet agreed what it is measuring.

The second omission is more specific and more damaging to a particular class of product. Fractional, percentage and concentration-based measures of ctDNA, and total cell-free DNA concentration, are not recommended as surrogate measures of disease burden for routine treatment decisions outside clinical trials. The panel accepts these measures may carry prognostic information while holding that they are not validated to guide management independently. Any product whose value proposition is quantitative response monitoring now sits outside the guideline by name.

Seven randomised trials is a thin base for a field this developed

The systematic review covered eight years of literature and surfaced 54 meta-analyses and 22 study reports, including 7 randomised controlled trials. Spread across every solid tumour and lymphoma, that is a small number of randomised comparisons for a technology already embedded in routine practice and reimbursed in multiple markets.

It also explains the guideline’s shape. A panel cannot recommend what has not been tested prospectively, and much of the ctDNA evidence base consists of retrospective correlations between molecular findings and outcomes. Those establish prognostic association. They do not establish that acting on the result improves anything, which is the question a guideline panel is obliged to ask.

What to watch

Three things. Whether payers rewrite coverage policy against this guideline, and specifically whether existing MRD coverage survives the absence of a recommendation. Whether the tumour-specific guidelines the co-chairs describe as necessary arrive with stronger MRD language, since a colorectal or lung-specific document can rest on a deeper evidence base than a pan-tumour one. And whether the randomised trials now running in MRD-guided treatment escalation report in time to change the next revision.

The uncomfortable reading is that a technology can be widely adopted, extensively reimbursed and commercially significant while the evidence that acting on it helps patients remains largely unbuilt. ASCO has not said ctDNA does not work. It has said, carefully and across four clinical domains, that it does not yet know where using it changes outcomes. Eight years after the first review reached almost the same conclusion, that is the finding worth sitting with.