Primary Prevention of Breast Cancer Based on AI Risk Model

By: Victoria Kuhr, MS
Posted: Wednesday, February 5, 2025

A 10-year image-derived artificial intelligence (AI) risk model for breast cancer portrayed a significantly higher discriminatory performance than the clinical Tyrer-Cuzick v8 risk model, based on findings presented at the 2024 San Antonio Breast Cancer Symposium (SABCS; Abstract GS2-10). Mikael Eriksson, PhD, of Karolinska Institutet, Stockholm, and colleagues believe the image-derived AI risk model has the potential for clinical use in primary prevention of perhaps one-third of breast cancers.

The two-site case-cohort study included women (aged 30–90) from two screening settings in Olmsted County, Minnesota and the KARMA cohort (Sweden), recruited between 2009 and 2017. Absolute 10-year risks were calculated at study entry. Time-dependent AUC and the ratio of expected vs observed events were estimated. Analyses were performed for risk of all breast cancers and restricted to invasive cancer alone.

The Olmsted/KARMA case-cohorts included 8,721 women (with a mean age of 54.4 years) in the subcohort and 1,633 incident breast cancer cases (with a mean age 57.0 years). Overall, the median follow-up was 10 years.

The image-derived AI 10-year average risk was estimated as 3.85% in Olmsted and 3.16% in KARMA. The expected vs observed events ratio was 1.01 in Olmsted and 0.98 in KARMA. The 10-year time-dependent AUC was 0.71 in Mayo and 0.72 in KARMA.

Based on the National Institute for Health and Care Excellence guidelines for women at a high risk of cancer, considering 8% of women at high risk, 32% of breast cancers may be subject to preventive strategies in the 9.7% of women at high 10-year risk based on the AI risk model. The corresponding findings were 7.2% and 2.2% for Tyrer-Cuzick version 8 risk model in KARMA. These results were similar when restricted to invasive cancers alone.

Disclosure: For full disclosures of the study authors, visit sabcs.org.

2024 San Antonio Breast Cancer Symposium

---