Ultrasensitive circulating tumor DNA (ctDNA) testing may help identify women with triple-negative breast cancer who are at either a very high or very low risk for recurrence after surgery, new research suggested.

The study found that when patients with stage II or III disease had no detectable ctDNA after surgery, their outcomes were “exceptional” — with 94% free of invasive disease at 5 years. In contrast, those who tested ctDNA-positive were at a very high risk for recurrence.

Researchers said the findings, published in the Journal of Clinical Oncology, suggest that ctDNA could eventually be used to guide treatment escalation or de-escalation for patients with early-stage triple-negative disease.

“We have never had this kind of a marker that is so strongly prognostic,” said study co-author Heather Parsons, MD, MPH, head of the breast oncology program at Fred Hutchinson Cancer Center in Seattle.

She stressed, however, that interventional trials are needed to prove that ctDNA-guided treatment benefits patients.

“We need a study with probably larger numbers where we’re actually acting on the [test] results,” Parsons told Medscape Medical News, noting that such studies are being planned.

Parsons discussed the findings at Miami Breast Cancer Conference 2026 earlier this month, as part of a broader presentation on what’s known — and what remains to be proven — about the role of ctDNA in early-stage breast cancer.

For their study, PREDICT-DNA, Parsons and her colleagues enrolled 227 patients with stage II or III HER2-positive or triple-negative breast cancer who were planning to undergo neoadjuvant chemotherapy. Blood samples were collected at baseline, after neoadjuvant therapy but before surgery, at 6 months, and then yearly for up to 5 years.

The primary objective was to see how well ctDNA results corresponded with pathologic complete response (pCR) following neoadjuvant therapy — a commonly used surrogate for long-term outcomes in triple-negative breast cancer.

In total, 91 of 220 evaluable patients achieved pathologic complete response. But while all patients with pCR were ctDNA-negative after neoadjuvant therapy, so were 40% of patients without pCR. That negative predictive value of 60% fell far short of the prespecified 90% threshold.

However, the findings were more encouraging when it came to the trial’s secondary objective: assessing the association between ctDNA and 5-year invasive disease-free survival in the subset of patients with triple-negative disease.

Overall, Parson’s team found, patients with detectable ctDNA after neoadjuvant therapy had a nearly ninefold higher risk for recurrence vs those who were ctDNA-negative [hazard ratio [HR], 8.9; P = .001] — independent of their pCR status.

In addition, detectable ctDNA after surgery was tied to an “extremely high” recurrence risk [HR, 128; P < .001]. Conversely, patients who were ctDNA-negative post-surgery had markedly different outcomes: 94% remained free of invasive disease at 5 years.

The fact that a negative ctDNA result points to a very low recurrence risk, even in some patients with residual disease, is intriguing, according to Kathy Miller, MD, co-leader of the breast cancer program at Indiana University Health in Indianapolis.

“What we ultimately hope to do is get the right amount of therapy to the right people — not too much, not too little,” Miller told Medscape Medical News. “So the more effectively we can separate groups into those that have a really phenomenally good outcome and those that have a much higher risk, the better we’ll be able to do that.”

But like Parsons, she stressed that we’re not there yet.

“Right now,” Miller said, “there is no data that says altering treatment decisions on the basis of ctDNA improves outcomes.”

In the absence of that evidence, she added, using ctDNA to guide treatment would risk causing patients real harm.

This study received support from Sysmex Inostics Inc., Personalis Inc., the Susan G. Komen Breast Cancer Foundation, the National Institutes of Health, and others. Parsons disclosed receiving honoraria, consulting roles, or research funding from Guardant Health, Illumina, Merck, Pfizer, and others. She reported being a member of the JCO editorial board but had no role in the peer review of the manuscript. Miller reported having financial relationships with Merck, Roche, Celcuity Inc., and others.