Deep Learning Analyzes Genetics For Aortic Aneurysm Risk

To identify patients at high risk of aortic aneurysm, Massachusetts General Hospital (MGH) researchers used deep learning to analyze how genetics could impact aorta size. According to researchers, the findings could assist in developing preventive and targeted therapies.

An aortic aneurysm is when the aorta is abnormally enlarged and can tear or rupture, causing sudden cardiac death. The issue the condition presents to clinicians is that patients often show no signs or symptoms before the aorta fails.

The research team implemented deep learning using data collected from a UK Biobank study that performed multiple magnetic resonance imaging tests of the heart and aorta in more than 40,000 individuals.

“There were no aortic measurements provided by the UK Biobank, and we wanted to read the aortic diameter in all of the images collected,” lead author and cardiologist at MGH, James Pirruccello, MD, said in a press release.

“That is very hard for a human to do because it would take a long time, which motivated our use of deep learning models to do this process at a large scale.”

The research team trained deep learning models to analyze the dimensions of the ascending and descending sections of the aorta in 4.6 million cardiac images. The team then examined the study participants’ genes to identify variations in 82 genetic regions, or loci, linked to the diameter of the ascending aorta and 47 linked to the diameter of the descending aorta.

“When we added up the genetic variants into what’s called a polygenic score, people with a higher score were more likely to be diagnosed with aortic aneurysm by a doctor,” Pirruccello said.

“This suggests that, after further development and testing, such a score might one day be useful to help us identify people at high risk of an aneurysm. The genetic loci that we discovered also offer a useful starting point for trying to identify new drug targets for aortic enlargement.”

According to Pirruccello, the research supports evidence that deep learning and other machine learning methods could assist in accelerating scientific analyses of complex biomedical data, including imaging results.

The research was supported by Leducq, the National Institutes of Health, the American Heart Association, the John S. LaDue Memorial Fellowship, a Sarnoff Cardiovascular Research Foundation Scholar Award, the Burroughs Wellcome Fund, the Fredman Fellowship for Aortic Disease, the Toomey Fund for Aortic Dissection Research, Bayer AG, and the Susan Eid Tumor Heterogeneity Initiative.