AI Tool Outpredicts Oncotype DX for Late Breast Cancer Risk

An AI model called IICM+ predicted distant breast cancer recurrence after five years far better than the standard 21-gene Recurrence Score, per a new study.

MedicalNewsToday reports that researchers have developed an AI tool called IICM+ to predict distant recurrence in early-stage, hormone receptor-positive, HER2-negative (HR+/HER2–) node-negative breast cancer — the most common breast cancer type. According to MedicalNewsToday, the study, published in August in *npj Breast Cancer*, found that more than half of distant recurrences in this subtype occur five or more years after treatment ends, and that HR+/HER2– disease "accounts for about 60% of all breast cancer deaths due to high prevalence and persistent recurrence risk beyond 5 years." The existing 21-gene Recurrence Score (Oncotype DX) is strongest at predicting recurrence within five years, so the team designed IICM+ specifically to look further out.

The researchers built the model using data from 2,808 patients in the TAILORx trial, then tested it on a separate group of 1,621 patients. Using the C-index metric — where 1.0 is perfect ranking and 0.5 is no better than random guessing — IICM+ scored 0.735 for predicting overall distant recurrence, versus 0.578 for the 21-gene Recurrence Score. Dr. Debra Patt, a medical oncologist at Texas Oncology not involved in the study, told MedicalNewsToday the difference is "very meaningful," calling the model "a more reliable crystal ball." Dr. Richard Reitherman, a radiologist at MemorialCare Breast Center, likewise said the gap is "potentially quite meaningful" for distinguishing higher-risk patients.

MedicalNewsToday reports that IICM+ could change adjuvant chemotherapy decision-making for patients with a 21-gene Recurrence Score of 0 to 25, who would typically decline chemotherapy. According to the study authors, a high IICM+ risk assessment could prompt such patients to opt in to appropriate adjuvant treatment. Reitherman told MedicalNewsToday that IICM+ "may add another layer of information to the Oncotype DX result," surfacing higher risk in patients who otherwise appear low-risk — while stressing it "would inform the conversation; it would not replace clinical judgment or shared decision making." Because late-recurrence cancer cells can stay dormant for years, better risk identification could let doctors tailor long-term therapy.

According to MedicalNewsToday, experts caution IICM+ is not ready for the clinic: Reitherman wants results reproduced in large, independent and diverse populations, proof it adds information beyond existing genomic tests, and confirmation that it improves outcomes, noting Oncotype DX itself took years to gain acceptance. Reproducibility, accessibility, cost-effectiveness and consistent performance across populations must also be established. Separately, ABC News reports that a British Journal of Sports Medicine study of 11,484 "Body Electric Challenge" participants found five-minute hourly walking breaks improved mood and reduced fatigue among people sitting much of the day — with breaks every 30 minutes best but hourly breaks the most practical balance — without undermining work performance.