Genetics of Cardiac Aging Implicate Organ-Specific Variation

Area of Science:

• Cardiology
• Medical Imaging
• Genetics

Background:

• Cardiac aging varies among individuals, prompting interest in estimating cardiac age acceleration.
• Existing methods for cardiac age estimation lack specificity or feature richness, hindering research into genetic contributions.
• Deep learning (DL) on cardiac-masked MRI offers a potential solution for precise cardiac aging assessment.

Purpose of the Study:

• To develop and validate a video-based DL model for estimating cardiac age acceleration using heart-masked cardiac MRI data.
• To investigate the associations of cardiac age acceleration with cardiac function, lifestyle factors, serum proteins, and brain MRI characteristics.
• To explore the heritability and genetic underpinnings of cardiac age acceleration.

Main Methods:

• A video-based DL model was trained on cardiac MRI data from 61,691 UK Biobank participants, excluding non-cardiac pixels.
• Cardiac age acceleration was calculated as the difference between predicted heart age and calendar age.
• Genome-wide association studies (GWAS) and Mendelian randomization were employed to identify genetic loci and protein associations.

Main Results:

• The DL model explained 71.1% of calendar age variance with a mean absolute error of 3.3 years.
• Cardiac age acceleration correlated with adverse cardiac geometry, dysfunction, lifestyle factors, specific serum proteins, and brain MRI findings.
• Cardiac age acceleration was found to be heritable (h2g 26.6%), with GWAS identifying 24 associated loci, 21 novel for cardiac age acceleration.

Conclusions:

• A novel DL approach provides a cardiac-specific measure of aging.
• Cardiac age acceleration is influenced by genetic, lifestyle, and environmental factors, and is associated with increased risk of cardiovascular disease and mortality.
• These findings highlight the importance of heart- and vascular-specific factors in cardiac aging.