ctDNA spots cancer relapse months early

Clinical Takeaway: In stage II colon cancer and muscle-invasive bladder cancer, ctDNA can now inform decisions about treatment after surgery. Outside those settings, a positive result identifies higher-risk patients but does not yet indicate who benefits from earlier therapy.

A new JAMA review takes stock of circulating tumor DNA (ctDNA) testing across its clinical uses: selecting targeted therapy, detecting residual disease after surgery, monitoring for resistance during treatment, and screening people without symptoms. The evidence is uneven. It is strongest where a trial attached a specific decision to the test result, and thinnest where clinicians are left to interpret a result on their own.

What ctDNA detects is not in question. Among patients with colorectal cancer who underwent resection, ctDNA present four weeks after surgery was linked to recurrence in about six in 10, compared with one in 10 of those testing negative. In a separate cohort of 130 patients, molecular evidence of disease appeared a median of nearly nine months before standard surveillance found it.

Three trials cited by the review paired a result with a pre-specified action. In stage II colon cancer, DYNAMIC withheld adjuvant chemotherapy from ctDNA-negative patients and matched conventional decision-making on two-year recurrence-free survival, 93.5% vs 92.4%, while treating about half as many patients. In muscle-invasive bladder cancer, IMvigor011 added adjuvant atezolizumab for patients with ctDNA detected after cystectomy, extending median overall survival to 32.8 months from 21.1. The FDA approved a companion assay for that indication in May 2026.

The third trial, SERENA-6, tested every two to three months during endocrine therapy for advanced breast cancer and switched drugs when a resistance variant emerged, improving progression-free survival by about 7 months. All three fixed the moment of decision in advance. None addressed the situation clinicians meet most often, in which a patient under routine follow-up has a positive result and a clean scan.

A negative result carries less information than it appears to. ctDNA is detectable in more than half of patients with bladder, colorectal, gastroesophageal, ovarian, and pancreatic cancer, but in fewer than 10% of those with glioma, and the authors say it is not known why shedding varies this much. Screening asymptomatic adults fared worst of the uses reviewed, with most positive signals in one large cohort turning out to be false.

Where a trial has defined both the moment to test and the action to take, ctDNA has earned its place. Where it has not, a positive result relocates uncertainty rather than resolving it. Personalized residual-disease assays run $3,000 to $5,000 or more, so quarterly testing can cost more than imaging. The question worth asking before ordering is not what the test might find, but what would change if it found something. Where the answer is nothing, the result offers only expense and anxiety without direction.

Source: Mezzanotte-Sharpe J, et al. (2026 Aug 10) JAMA. Liquid biopsies for cancer: a translational science review