Loading ...

Despite having a very low recurrence risk, the patient wondered if she was doing enough to prevent a relapse of triple-negative breast cancer. Circulating tumor DNA (ctDNA) tests, which can detect molecular evidence of recurrence, sometimes months before scans or symptoms occur, had all been negative.

But she wanted a second opinion from Kathy D. Miller, MD, a breast oncologist at Indiana University Health Simon Cancer Center in Indianapolis.

“Suppose the blood test is positive, but you’ve gotten every scan we can come up with and they’re all normal. Now what do you do?” Miller recalled telling the patient. “Start chemo? I have no clue.”

As ctDNA liquid biopsy testing becomes more common in clinical practice, oncologists like Miller are confronting a dilemma: how, or whether, to act on the results.

If ctDNA findings are negative, can patients skip or stop treatment? A negative ctDNA test suggests the absence of residual disease but does not rule it out. Cancers shed tumor DNA at different rates, so tumors that shed little DNA could still go undetected, as can mutations that evolve after neoadjuvant treatment and surgery. Both scenarios make false-negative ctDNA findings more likely.

A positive ctDNA test, however, presents a different conundrum: Should the oncologist start treatment immediately to target residual cancer cells or wait until recurrence is visible on a scan? While a growing body of evidence shows that ctDNA can flag a patient’s risk for recurrence before imaging and is associated with poorer outcomes, it’s less clear whether starting treatment sooner will improve a patient’s outcome.

“No study in breast cancer, for instance, has shown that if you intervene on a positive ctDNA test, you can improve the outcome,” said Anthony Lucci, MD, professor of breast surgical oncology at the University of Texas MD Anderson Cancer Center in Houston.

What ctDNA Can, and Can’t, Tell Us

ctDNA, the DNA fragments released from cancer cells that circulate in the bloodstream, can indicate the presence of molecular residual disease after treatment. The most sensitive ctDNA tests are tailored to each patient’s tumor, enabling highly specific detection of residual disease.

Looking for cancer biomarkers during treatment is not new, Lucci noted. It started with circulating tumor cells in the bone marrow, which were difficult to sample and monitor reliably, prompting researchers to search for these cells in the bloodstream instead.

Blood tests for circulating tumor cells “were proven to be prognostically important, but they were not predictive,” Lucci said. “Unfortunately, right now, I think we’re at the same point with ctDNA.”

Research has consistently shown that ctDNA is prognostic for recurrence and survival outcomes in a range of cancers.

In colorectal cancer (CRC), the 2024 CIRCULATE-Japan GALAXY trial found that postoperative ctDNA positivity was associated with significantly worse disease-free survival (hazard ratio [HR], 11.99) and overall survival (HR, 9.68) in patients with stage II or III colon cancer or stage IV CRC.

Several recent smaller trials have added weight to these findings. The 2025 BESPOKE CRC study found that compared to ctDNA negativity, ctDNA positivity after surgery was associated with significantly worse 2-year disease-free survival in stage II (45% vs 91%) and stage III (35% vs 87%) colon cancer. The 2024 INTERCEPT trial revealed that after surgery, patients with stage IV CRC who were ctDNA negative had much better 12-month disease-free survival outcomes than those with stage II disease who were ctDNA positive (87.7% vs 38.9%), underscoring that ctDNA status may outperform stage alone in prognosticating outcomes.

Even in early-stage non-small cell lung cancer (NSCLC), where recurrence is a significant concern and ctDNA detection remains challenging, ctDNA has shown prognostic promise with highly sensitive tests. In 2025 analysis of patients with NSCLC from the TRACERx study, researchers found that compared with no detectable ctDNA, persistent ultralow-level ctDNA detection (below 80 parts per million) after surgery was associated with significantly worse recurrence-free (HR, 4.9) and overall (HR, 3.08) survival.

Although ctDNA positivity is strongly prognostic for recurrence and survival, evidence that it can predict benefit from adjuvant therapy remains limited.

“People are hopeful that ctDNA tests become standards of care, but they’re by no means standards of care in 2026,” said Mark G. Kris, MD, a thoracic medical oncologist at Memorial Sloan Kettering Cancer Center in New York City.

Recent trials illustrate that challenge.

Take the 2025 analysis of the DYNAMIC-III trial. Among ctDNA-positive patients with stage III colon cancer, escalated therapy did not improve recurrence-free survival over standard of care. The ctDNA-informed approach was, in fact, associated with numerically worse recurrence-free survival at 2 years — 52% in the escalated therapy group vs 61% in the standard care group (HR, 1.11).

ctDNA can also have blind spots. False negatives occur, especially for cancers that don’t shed much tumor DNA or that mutate after neoadjuvant treatment or surgery.

The INTERCEPT analysis identified 15 false-negative ctDNA results, representing 1.3% of all patients in the cohort. An analysis of CIRCULATE-Japan found that 4.7% of patients with stage II-III CRC had false-negative ctDNA results before surgery.

Mark Lewis, MD, a gastrointestinal oncologist at Intermountain Health Cancer Center in Utah, noted that, in his practice, multiple Signatera ctDNA tests over several months missed colon cancer that had spread to a patient’s peritoneum, despite the patient “very clearly experiencing progression.”

Some mutations are simply harder to detect in the blood. The EGFR T790M mutation in advanced NSCLC, for instance, was only detected in 61% of patients whose tissue had already tested positive for the mutation.

Importantly, ctDNA testing also does not always provide enough lead time before clinical recurrence.

ZEST, the first phase 3 trial using ctDNA-guided therapy in breast cancer, was terminated in 2024 due to low rates of minimal residual disease detection. Of the 147 patients with ctDNA positivity after their first test following definitive therapy, 73 had already developed metastases.

“The ctDNA analyses do give us greater lead time for at least some of the patients, though the idea that it gives you 9-12 months lead time for most patients is cuckoo,” Miller said.

When ctDNA May Be Predictive

In certain scenarios, however, ctDNA findings can inform treatment decisions.

The strongest evidence for ctDNA’s predictive ability comes from studies in bladder and colon cancer.

The 2025 IMvigor011 trial is among the first studies to demonstrate an overall survival benefit with ctDNA-guided adjuvant immunotherapy. The trial included 607 patients with muscle-invasive bladder cancer and no radiographic evidence of disease within 6-24 weeks of radical cystectomy, with patients receiving serial ctDNA monitoring for up to 1 year after surgery. The 250 ctDNA-positive patients were randomly assigned to receive either atezolizumab or placebo for up to 1 year, and the 357 ctDNA-negative patients were surveilled without adjuvant therapy.

At a median follow-up of 16 months, median overall survival was nearly 1 year longer for ctDNA-positive patients receiving atezolizumab (32.8 vs 21.1 months; HR, 0.59) and median disease-free survival was about two times higher in the atezolizumab group (9.9 vs 4.8 months).

Notably, persistently ctDNA-negative patients did well without adjuvant treatment, with 88% remaining disease-free at 2 years. 

While ctDNA-positive patients are “very likely” to recur and likely to benefit from therapy, persistently ctDNA-negative patients have a “very small” recurrence risk and can generally be monitored without therapy, said lead study author Thomas Powles, MD, a professor of genitourinary oncology and director of Barts Cancer Centre at St. Bartholomew’s Hospital in London, England.

Knowing which patients will benefit from atezolizumab is critical because atezolizumab can cause severe or life-threatening side effects in about 10%-15% of patients. “We know that for over half the patients, the cancer is never going to come back, so you’re putting [that] half at risk” if everyone receives the drug, Powles said.

ctDNA has also demonstrated predictive utility in certain subsets of colon cancer.

A 2025 post hoc analysis of the CALGB/SWOG 80702 trial found that among patients with stage III colon cancer who were ctDNA positive after surgery, adding celecoxib to standard FOLFOX chemotherapy was associated with improved disease-free survival (adjusted HR, 0.61) and overall survival (adjusted HR, 0.62) compared to placebo.

And among ctDNA-negative patients, celecoxib was not associated with a significant disease-free or overall survival benefit.

An earlier analysis from the DYNAMIC trial found that patients with ctDNA-negative stage II colon cancer could safely forgo adjuvant chemotherapy without compromising recurrence-free survival. Among ctDNA-negative patients, 3-year recurrence-free survival was 92.5%.

In DYNAMIC, the evidence suggests “ctDNA can be a really good tiebreaker” when deciding whether to treat with chemotherapy, said Emil Lou, MD, PhD, professor of hematology, oncology, and transplantation at the University of Minnesota in Minneapolis.

What’s Next for ctDNA

Over the next few years, ctDNA will likely play an increasingly important role in how oncologists track patients’ disease progression. The shift may influence decision-making around adjuvant therapy as well as the timing of scans, treatment escalation, and whether to undergo surgery.

In the next 5 years, Lou said ctDNA could be used to tailor active surveillance for individual patients with stage II or III colon cancer. A patient could have ctDNA testing two to four times per year, which may mean “moving up CT scans if someone is ctDNA positive,” he said.

Lewis sees a role for ctDNA in determining whether patients with stage IV colon cancer could benefit from metastasis-directed therapy with surgery or radiation rather than going straight to chemotherapy.

In advanced lung cancer, ctDNA testing could help oncologists determine how long to continue immune checkpoint inhibitor therapy in long-term responders, said Kris. For instance, among long-term responders with no detectable ctDNA, it may be safe to stop treatment, and in stage IB to select IIIB NSCLC, becoming ctDNA-negative after neoadjuvant therapy and surgery could help discriminate those who are cured, alongside pathologic examination, Kris explained.

Because ctDNA can also be picked up in urine analysis, researchers are investigating if using ctDNA and urinary tumor DNA “will give us information about whether there’s local or metastatic disease,” potentially identifying recurrence earlier, Powles said.

Despite the progress, and potential, of ctDNA, it currently has limited utility for guiding treatment decisions for many cancer types and stages, which means conversations with patients need to be nuanced.

For stage II colon cancer, a negative ctDNA test can be “reassuring” for Lou to not recommend chemotherapy, but it’s not definitive — there’s a roughly 1 in 10 chance of recurrence in this population. For conversations with patients, it’s more of a question of “do you feel comfortable with those odds?” Lou said.

Lucci explains to his patients with breast cancer that ctDNA testing can be useful as a prognostic tool, “but we don’t know if intervening at that moment will make your outcomes better.”

If ctDNA could find disease early enough for current or future therapies to change the trajectory of disease, allowing people to be cured or live substantially longer, “that would be fantastic for my breast cancer patients,” said Miller. “I just don’t know if we’re there yet.”

Lou reported having financial interests in Ryght, Inc.; being a consultant and an advisor for Myer Research, Inc.; being a paid consultant for Primum and Impiricus; among others. Kris reported having relationships with AstraZeneca, Bristol Myers Squibb, Daiichi Sankyo, Pfizer, Merck, and Boehringer Ingelheim. Miller reported serving on the Independent Data Monitoring Committee for trials conducted by Celcuity, Roche/Genentech, and Merck and reviewing SERENA-6 data for AstraZeneca. Lucci and Lewis reported having no relevant disclosures.