- In this study, machine learning was used to assess the size of the ascending and descending thoracic aorta from MRI data on 43,243 UK Biobank participants (4.4 million images)
- 75 previously unreported genetic loci were identified in the ascending aorta and 43 in the descending aorta
- Several potential gene targets were pinpointed, including SVIL, which was strongly associated with descending aortic diameter
- A polygenic score for ascending aortic size was an independent risk factor for thoracic aortic aneurysm or dissection
- A model incorporating a polygenic score and clinical risk factors might identify asymptomatic individuals at high risk of ascending aortic aneurysm who would benefit from thoracic imaging
Thoracic aortic aneurysm is typically asymptomatic until the time of dissection or rupture, and there are no guidelines for screening. There's keen interest in identifying the genetic basis for pathologic aortic enlargement to aid in the development of pharmacologic interventions and identify individuals at risk.
Subscribe to the latest updates from Cardiovascular Advances in Motion
James P. Pirruccello, MD, a cardiologist at the Corrigan Minehan Heart Center at Massachusetts General Hospital, Patrick T. Ellinor, MD, PhD, director of the Telemachus & Irene Demoulas Family Foundation Center for Cardiac Arrhythmias,cardiologist Mark Lindsay, MD, PhD, and colleagues used a form of artificial intelligence to measure aortic diameters in more than 40,000 individuals, then developed a polygenic risk score that predicted future risk of aortic disease. They published their findings in Nature Genetics.
Training the Model
As the first step, the lead researcher annotated 116 cardiovascular magnetic resonance images (MRI) from the UK Biobank, manually identifying and labeling all pixels. The team then used those annotations to train a deep learning model to perform the same task.
The ascending and descending thoracic aorta were quantified separately because they have different biologic origins and independent risk factors for aneurysm formation. In the validation set of 24 MRIs, the model was 97% accurate in categorizing pixels for the ascending aorta and 94% accurate for the descending aorta.
Heritability of Aortic Diameter
The model was applied to 4,374,900 images from 43,243 UK Biobank participants. 38,694 participants had data that passed quality control and contributed to genetic analyses of the ascending aortic diameter, and 39,688 contributed to analyses of the descending aortic diameter.
Both traits were highly heritable. The single nucleotide polymorphism heritability of the size of the ascending aorta was 63% and that of the descending aorta was 50%.
Genome-wide association studies (GWAS) of the two traits identified the genetic basis for variation of aortic diameter:
- 82 loci were associated with the diameter of the ascending aorta, including 75 not previously reported in GWAS of aortic dimension or disease
- 47 loci were associated with the diameter of the descending aorta, of which 43 were previously unreported and one was located on the X chromosome
- Only 14 loci were associated with both traits, emphasizing their distinct biology
Other Association Studies
The genetic loci are a starting point for identifying new drug targets. By integrating GWAS with transcriptome-wide association studies and rare-variant testing, it was possible to prioritize several potential gene targets, including SVIL, which was strongly associated with descending aortic diameter. Using single nucleus RNA sequencing, the researchers identified likely cell types of relevance.
Polygenic Risk Score
The researchers built a polygenic score for ascending aortic size and applied it to 385,621 UK Biobank participants who did not have a diagnosis of aortic disease at enrollment. Over a median follow-up of 11 years:
- The score was strongly associated with the 685 incident cases of thoracic aortic aneurysm or dissection (HR, 1.43 per standard deviation in score; P = 3.3 × 10−20)
- Participants in the top 10% of the score had a 2.1-fold higher HR than the other 90% (P = 7.3 × 10−15)
Hope for the Future
This study suggests that a model incorporating a polygenic score and clinical risk factors might identify asymptomatic individuals at high risk of ascending aortic aneurysm who would benefit from thoracic imaging.
More broadly, the results highlight the value of studying quantitative traits with deep learning, which is broadly applicable to various types of biomedical imaging, to gain a greater understanding of disease processes.
Learn more about the Telemachus & Irene Demoulas Family Foundation Center for Cardiac Arrhythmias
Refer a patient to the Corrigan Minehan Heart Center