Study: AI, genetics, and clinical data improve breast cancer risk prediction

Combining an artificial intelligence risk score with genetic and clinical data more accurately identifies women at high risk of developing breast cancer than using any risk score alone or in a 2-score combination, a new Kaiser Permanente study found.

The finding points to a broader health care transformation grounded in medical excellence: using advanced tools and richer data to personalize disease detection. It also helps move medicine closer to the goals of value-based care: delivering the right care to the right patient at the right time.

“Breast cancer risk tools can help identify high-risk women who are most likely to benefit from more frequent breast cancer screening or risk reduction with medications,” said lead author Vignesh Arasu, MD, PhD, a research scientist at the Kaiser Permanente Division of Research and a radiologist with The Permanente Medical Group. “Our study shows that each of the 3 tests identifies a unique group of women, and that when all 3 risk tests are used we increase our ability to differentiate high-risk and low-risk women and provide more personalized screening recommendations.”

Related health care innovation story: Why health care innovation is essential during uncertain times

Combining risk scores strengthens breast cancer risk prediction

Published in the Journal of the National Cancer Institute, the study is one of the largest and most diverse to evaluate how well 3 scores predict breast cancer risk. Those scores included:

• Mammography AI algorithm, which predicts 5-year breast cancer risk based on the presence of imaging biomarkers. The technology assists radiologists by scanning mammography images in seconds to identify tiny, subtle lesions or calcifications that might be missed by the human eye.
• Polygenic risk score, which is determined by the presence or absence of 313 single nucleotide polymorphisms (SNPs) that prior studies have found to be associated with breast cancer.
• Clinical risk score, which considered factors such as age, race or ethnicity, family history of breast cancer, breast density, and body mass index.

The study included 82,957 women enrolled between 2003 and 2020 in the Kaiser Permanente Research Bank, a national biobank that includes medical records, survey, and genetic data from more than 400,000 Kaiser Permanente members. All study participants had a recent mammogram with no signs of breast cancer and no known genetic mutation or prior diagnosis that increased their risk.

Over a decade, 2,471 or 3% of the women in the study were diagnosed with invasive breast cancer or abnormal, cancerous cells in the lining of the milk ducts, but that had not spread to surrounding breast tissue. The model that combined all 3 risk scores was the most accurate, with a Concordance Index score of .70, indicating the predictive model has good accuracy. A score of .5 is equivalent to a coin flip, while 1.0 represents perfect accuracy. By comparison, the clinical risk tool scored .62, the polygenic test scored .61, and the combined clinical and polygenic risk scores reached .66.

Among women at the highest risk of developing breast cancer, the clinical risk score alone identified 19% of the women who developed breast cancer over a decade while the combined model identified 26% of these women.

Related value-based care story: How value-based care improves breast cancer survival rates

Personalized screening advances health care transformation

That ability to better match screening intensity to individual risk reflects a meaningful shift in health care transformation — one that brings together new tools, prevention, evidence-based medicine, and a commitment to improving care quality. It is also central to value-based care, where innovation is used to make care more precise, proactive, and personal.

The new research builds on prior research led by Dr. Arasu showing that AI-based mammography risk assessment more accurately predicted a woman’s future breast cancer risk than a clinical risk model.

“Our previous study showed that an AI risk score was slightly more accurate than a clinical risk score,” said Dr. Arasu. “This new study shows that by combining them, and by adding a polygenic risk score, we make a substantial improvement in accurately assessing risk.”

Read the full story on the Kaiser Permanente Division of Research website.

The post Study: AI, genetics, and clinical data improve breast cancer risk prediction appeared first on Permanente Medicine.