Heart health is brain health. We take this adage for granted, and now a massive new study published in Science on June 2 backs it up. Using deep machine learning to decipher and compare heart and brain MRI scans of more than 40,000 people, researchers led by Hongtu Zhu at the University of North Carolina, Chapel Hill, unearthed numerous connections between the fine-grained structural and functional features of the two organs. For example, an enlarged left ventricle of the heart—an indicator of poor cardiac health—correlated with erosion of certain white-matter tracts in the brain. Weaving genome-wide association studies into the canvas, the authors were able to see shared genetic ties underlying traits and diseases that affect each organ. In all, the study points to strong genetic and causal relationships between heart and brain.

  • Heart and brain MRI scans compared for 40,000 U.K. Biobank participants.
  • There were thousands of correlations between features of heart and brain.
  • Genetic variants associated with heart structure also linked to brain traits and disorders.

“This is the most in-depth analysis to date of heart-brain connections in human health and disease,” commented Berislav Zlokovic of the University of Southern California in Los Angeles and Daniel Nation of the University of California, Irvine. “The study findings have substantial relevance to Alzheimer's disease and related disorders given prior work underscoring the importance of vascular contributions to cognitive impairment and dementia” (comment below).

“[The authors] identify multiple genetic links between distinct aspects of cardiovascular function and brain health,” wrote Julia Sacher and Veronica Witte of the University of Leipzig Medical Center in Germany, in an accompanying perspective in Science. “By offering a multidimensional analysis of heart-brain connections, this study could contribute to the development of personalized disease risk prediction.”

Still, Sacher, Witte, and other commentators caution that while bioinformatic and machine-learning techniques enabled such a massive, detailed analysis, the computational methods also make it difficult to draw clinically meaningful conclusions.

Studies resoundingly agree that the functions of the heart and brain are intimately intertwined. For example, ill cardiovascular health in early mid-life foreshadows brain shrinkage, flagging cognition, and neurodegenerative disease later on (Apr 2017 news; Aug 2018 news; Aug 2020 news). However, which structural and functional features tie the two organs together, and how genetics versus lifestyle and environment contribute to these connections, remain unclear.

Heart-Brain-Gene Triangle. The study compared myriad features spied by MRI scans of the hearts and brains of 40,000 U.K. Biobank participants, and correlated genetic variants associated with the structural characteristics of each organ. [Courtesy of Zhao et al., Science, 2023.]

To investigate this requires a massive dataset, and the U.K. Biobank has that. The cohort study logs multiple health characteristics of participants, including MRI scans of the hearts and brains. First author Bingxin Zhao and colleagues used recently developed algorithms to extract different features of these scans, pulling out 82 traits from cardiovascular magnetic resonance (CMR) scans, which included multiple measures of the four heart chambers as well as the two sections of the aorta (Bai et al., 2020). They then hunted for relationships between each of these 82 cardiac traits and thousands of features gleaned from different brain scans, including structural, diffusion, and functional MRI.

What did they find? In a nutshell, 4,000 significant correlations between features of the heart and brain. The 82 traits picked up by CMR related to a wide variety of brain MRI traits, including regional brain volumes, cortical thickness, white-matter integrity, and functional connectivity. Consider a few examples. Thickness of muscle walls of the heart correlated with larger subcortical regions in the brain, particularly the putamen. The larger the area of the distal descending aorta, the heftier the prefrontal cortex and the hippocampus. Conversely, this same aortic measure negatively correlated with connectivity of major white-matter tracts in the brain, as measured by fractional anisotropy.

Cardiac Connections. Heart wall thickness (left) and the descending aorta minimum area (middle) correlated with volume of subcortical regions in the brain. The descending aorta minimum area negatively correlated with the integrity of major white-matter tracts as seen on fractional anisotropy scans (right). [Courtesy of Sacher and Witte, Science, 2023.]

What could these associations mean? Hoping that genetics would offer at least partial explanations, the scientists looked for genetic variants among the U.K. Biobank participants that associated with the 82 CMR traits. They identified 80 genetic loci that associated with 49 of these heart scan features.

They then cross-referenced these newly identified CMR loci with genetic variants tied to complex traits and diseases in other GWAS. Perhaps unsurprisingly, they found that many of the 80 CMR loci also correlated with cardiovascular traits and diseases. For example, 41 of the genetic loci were linked to different aspects of blood pressure, such as diastolic and systolic, pulse, and mean arterial pressures. Also, many of the heart scan loci were linked to disorders of the brain. Some associated with thicker heart walls or correlated with stroke. Others cropped up as GWAS hits for cognitive function, anorexia nervosa, or schizophrenia.

Some of the heart trait variants were found to influence expression levels of genes within the brain. One SNP correlated with brain levels of ALDH2, an enzyme that breaks down toxic aldehydes, and is reportedly more active in the putamina and temporal cortices of people with AD (D’Souza et al., 2015).

“Our findings indicate that cardiovascular conditions share substantial genetic components with brain diseases, mental health traits, and cognitive functions, suggesting a potential genetic basis for heart-brain connections,” the authors wrote.

Using Mendelian randomization to infer inherited causal relationships between the heart and brain disorders, the researchers found that in most cases, shared genetic variants work by setting off cardiovascular problems first, which then lead to brain disorders. For example, a variant that leads to thickening of the heart muscle walls may make the heart pump less efficiently, reducing oxygen levels in the brain, and perhaps setting off mood disorders.

These bioinformatics-based findings open a window to new relationships between the heart and the brain, but they are several steps removed from unlocking the mechanisms involved, noted Costantino Iadecola. “What we need to explore now is the biological relevance of these connections,” he said.

David Knopman of the Mayo Clinic in Rochester, Minnesota, made a related point, noting that many underlying risk factors such as diabetes, hypertension, smoking, and hyperlipidemia are known to affect the health of heart and brain simultaneously. The same is true for genetic variants that confer common susceptibility to the risk factors. “So, while this article highlights numerous structural heart-brain associations, and over 80 genomic loci, there is a lot of work to be done to determine which of the associations are clinically meaningful,” Knopman said.—Jessica Shugart


  1. This is the most in-depth analysis to date of heart-brain connections in human health and disease, utilizing genetic and MRI analysis of both heart and brain traits in over 40,000 individuals from the U.K. Biobank and other data repositories. Numerous robust correlations were observed between heart and brain MRI metrics, genetics, and endophenotypes, several of which were found to have causal (heart-to-brain) significance in Mendelian randomization analysis.

    The study findings have substantial relevance to Alzheimer's disease and related disorders, given prior work underscoring the importance of vascular contributions to cognitive impairment and dementia. Specifically, the authors report important genetic links between heart MRI metrics and brain structural and functional connectivity metrics known to underpin cognitive functions affected by ADRD, as well as links to vascular brain injuries and cerebral small vessel diseases known to contribute to cognitive decline in the majority of cases.

    Detailed analysis of the genes involved in heart-brain axis correlations also revealed that many of these genes are targeted by existing medications that prior studies have suggested may benefit both cardiac and neuropsychiatric conditions, highlighting the potential clinical implications of the study findings. Thus, further study of the heart-brain axis is warranted, and will likely lead to additional insights into novel prevention and treatment strategies for ADRD.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Vascular Disease in 50s Begets Brain Amyloid in 70s
  2. Brain Damage from Cardiovascular Disease Starts Earlier Than You Think
  3. Heart Health Is Brain Health, and It Starts in Your 20s

Paper Citations

  1. . A population-based phenome-wide association study of cardiac and aortic structure and function. Nat Med. 2020 Oct;26(10):1654-1662. Epub 2020 Aug 24 PubMed.
  2. . Characterization of Aldh2 (-/-) mice as an age-related model of cognitive impairment and Alzheimer's disease. Mol Brain. 2015 Apr 25;8:27. PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . Heart-brain connections: Phenotypic and genetic insights from magnetic resonance images. Science. 2023 Jun 2;380(6648):abn6598. PubMed.