AI Tool Beats Gene Test at Predicting Late Breast Cancer Return

A new study reports the IICM+ AI model predicted distant breast cancer recurrence beyond five years more accurately than the 21-gene Recurrence Score test.

@sharedot

AI Tool Beats Gene Test at Predicting Late Breast Cancer Return

By @sharedot · · 6 pages

  • Health
  • Breast Cancer
  • Oncology
  • AI

A new study reports the IICM+ AI model predicted distant breast cancer recurrence beyond five years more accurately than the 21-gene Recurrence Score test.

What the study found

According to Medical News Today, researchers have developed an AI model called IICM+ to predict distant recurrence in early-stage, hormone receptor-positive, HER2-negative (HR+/HER2–) node-negative breast cancer, the most common type of the disease.

The model was built using data from 2,808 patients in the TAILORx breast cancer trial and then tested on a separate group of 1,621 patients.

Using the C-index metric, which measures how well a model ranks recurrence risk, IICM+ scored 0.735 for predicting overall distant recurrence, while the existing 21-gene Recurrence Score (Oncotype DX) scored 0.578. The study was published in August in npj Breast Cancer.

Why the gap matters

The size of the performance gap is what makes the result striking.

Debra Patt, a practicing medical oncologist at Texas Oncology who was not involved in the study, called it 'a very meaningful difference,' explaining that a C-index of 1 represents perfect ranking while 0.5 performs only better than random guessing.

At 0.735, she said, the model 'shows good discrimination' and acts as 'a more reliable crystal ball' for estimating risk.

Dr. Richard Reitherman, a radiologist and medical director of breast imaging at MemorialCare Breast Center, agreed the difference is 'potentially quite meaningful,' suggesting the tool may better distinguish which patients face higher recurrence risk.

The late-recurrence problem

More than half of distant recurrences in HR+/HER2– node-negative breast cancer occur five or more years after treatment ends, and Medical News Today reports the study authors attribute roughly 60% of all breast cancer deaths to this type, given its high prevalence and persistent late recurrence risk.

The 21-gene Recurrence Score is most effective at predicting recurrence within five years, leaving the later window poorly covered.

Reitherman noted that cancer cells can travel to another organ and stay dormant for years before becoming apparent, so better identification of patients at risk five, ten, or even fifteen years out could let doctors tailor long-term treatment.

The authors also suggest IICM+ could flag higher-risk patients among those with Recurrence Scores of 0 to 25 who would otherwise decline adjuvant chemotherapy.

What comes next

Both experts urged caution before any change in clinical practice.

Reitherman told Medical News Today that results must be reproduced in large, independent, and diverse patient populations, and that Oncotype DX itself took years of research before wide acceptance — IICM+ must prove it is reproducible, accessible, cost-effective, and easy to fit into existing workflows.

He stressed the tool 'would inform the conversation; it would not replace clinical judgment or shared decision making.'

The finding also lands in a year of active prevention research: ABC News reports a British Journal of Sports Medicine study of 11,484 participants found five-minute hourly walking breaks improved mood and reduced fatigue without hurting work performance, underscoring how much of 2026's health news turns on evidence-based, low-cost interventions.

Sources

  1. medicalnewstoday.com › AI could offer more precise breast cancer recurrence risk, study finds
  2. abcnews.com › Hourly movement breaks can help counter effects of prolonged sitting, study finds

More on Health