Heart-brain connections: phenotypic and genetic insights from 40,000 cardiac and brain magnetic resonance images

Abstract

Cardiovascular health interacts with cognitive and psychological health in complex ways. Yet, little is known about the phenotypic and genetic links of heart-brain systems. Using cardiac and brain magnetic resonance imaging (CMR and brain MRI) data from over 40,000 UK Biobank subjects, we developed detailed analyses of the structural and functional connections between the heart and the brain. CMR measures of the cardiovascular system were strongly correlated with brain basic morphometry, structural connectivity, and functional connectivity after controlling for body size and body mass index. The effects of cardiovascular risk factors on the brain were partially mediated by cardiac structures and functions. Using 82 CMR traits, genome-wide association study identified 80 CMR-associated genomic loci (P < 6.09 × 10-10), which were colocalized with a wide spectrum of heart and brain diseases. Genetic correlations were observed between CMR traits and brain-related complex traits and disorders, including schizophrenia, bipolar disorder, anorexia nervosa, stroke, cognitive function, and neuroticism. Our results reveal a strong heart-brain connection and the shared genetic influences at play, advancing a multi-organ perspective on human health and clinical outcomes.

The role of sleep in the human brain and body: insights from multi-organ imaging genetics

“…We examined the phenotypic associations between 7 sleep conditions and a wide range of brain and cardiac MRI traits, including 101 regional brain volumes 54 , 63 cortical thickness measures 60 , 110 DTI parameters 56 , 92 parcellation-based network-level traits in resting and task fMRI 58 , respectively, as well as 82 heart imaging traits 60,61 ( Table S1 ). We performed regression analysis with unrelated white British subjects (average n = 29,025, Methods).…”

“…Then we identified those variants (and variants in linkage disequilibrium [LD] with them, r 2 ≥ 0.6) that were also significant in GWAS of brain and cardiac MRI traits (Methods). We found that a total of 39 genomic loci showed shared genetic influences on both sleep and imaging traits, covering regional brain volumes 54 , DTI parameters 56 , whole brain independent component analysis (ICA)-based resting fMRI traits 53,55,84 , parcellation- based resting and task fMRI traits 58 , and cardiac MRI traits 60 ( Table S4 ). We tagged previous GWAS for a wide range of sleep conditions, including insomnia 39 , chronotype 85,86 , daytime nap 39 , depressive symptom (sleep problems) 87 , getting up 39 , hypersomnia 88 , sleep duration 37,89 , and snoring 90 ( Fig.…”

“…In the phenotypic analysis above, UKB subjects who had both imaging and sleep data were considered, whereas more than 80% UKB subjects had only sleep data and thus were excluded. In order to include these non-imaging UKB subjects in our study, we developed polygenic risk scores (PRS) by using GWAS summary statistics of imaging traits from previous studies 54,56,58,60 . These genetically predicted imaging traits allowed us to repeat the above sleep-imaging association analysis in non-imaging UKB unrelated white British subjects (average n = 217,254, removing relatives of imaging subjects, Methods).…”

“…, 110 DTI parameters56 , 92 network-level parcellation-based resting fMRI traits58 (the same traits as in the phenotypic analysis), and 82 cardiac MRI traits60 . GWAS summary statistics of //gwas.mrcieu.ac.uk/).…”

“…To overcome these challenges, here we examined the phenotypic and genetic sleep- brain/heart connections using multi-modal cardiac and brain MRI data from more than 40,000 subjects in the UK Biobank (UKB) study 59 . We included three major brain MRI modalities: 1) brain anatomical and neuropathological structures from structural MRI, including regional brain volumes 54 and cortical thickness traits 60 ; 2) white matter microstructures from diffusion MRI, including multiple diffusion tensor imaging (DTI) parameters 56 ; and 3) intrinsic and extrinsic functional organizations of the cerebral cortex from resting and task fMRI, including functional connectivity and amplitude traits 58 . For heart, we used 82 traits extracted from short-axis, long-axis, and aortic cine cardiac MRI 61 , including global and regional measures of 4 cardiac chambers (the left ventricle, right ventricle, left atrium, and right atrium), as well as 2 aortic sections (the ascending aorta and descending aorta).…”