Genetics and AI Improve Diagnosis of Aortic Stenosis

Aortic stenosis is a progressive narrowing of the aortic valve that restricts blood flow from the heart and can be fatal if left untreated. There are currently no medical therapies that can prevent or slow its progression, leaving surgery or catheter-based valve replacement as the only effective options once the disease becomes severe. Although genetic variants have been linked to diagnosed cases, the earliest biological mechanisms driving disease onset remain unclear. New research now suggests that subtle differences in normal aortic valve function may share genetic roots with an increased risk of aortic stenosis.

Researchers at the University of California, San Francisco (San Francisco, CA, USA), in collaboration with the Broad Institute of MIT and Harvard (Cambridge, MA, USA), have developed a deep learning model capable of estimating detailed aortic valve measurements directly from cardiac MRI scans. This approach allowed them to assess valve function in large numbers of otherwise healthy individuals without relying on rare clinical disease diagnoses.

Instead of focusing only on advanced disease, the researchers analyzed continuous MRI-derived measures of aortic valve function. These included peak blood flow velocity, mean pressure gradient, and aortic valve area. By linking these measurements with genome-wide association studies, the team aimed to identify common genetic variants that influence both normal valve physiology and future disease susceptibility.

Using MRI data from nearly 60,000 healthy participants in the UK Biobank, the analysis identified 61 genetic loci associated with normal aortic valve function. A separate meta-analysis of more than 40,000 aortic stenosis cases and 1.5 million controls across multiple biobanks revealed 91 disease-associated loci. Combining both datasets in a multi-trait analysis uncovered 166 genetic loci linked to valve function and aortic stenosis.

The findings, published in Nature Genetics, showed strong genetic overlap between normal valve measurements and diagnosed aortic stenosis, suggesting disease risk develops along the same biological pathways that shape everyday valve function. Associations were also found with coronary artery disease, lipid metabolism, and phosphate regulation, pointing to possible preventive targets. Researchers emphasize that clinical validation is required before translating these insights into therapies. Future work will explore whether modifying these pathways could delay or prevent disease progression.

“Our findings suggest that risk for aortic stenosis is conferred at least in part through the same genetic mechanisms that drive normal variation in aortic valve function in the healthy population,” said James Pirruccello, senior author of the study.

Related Links:
University of California, San Francisco
Broad Institute of MIT and Harvard