Integrative Observational and Genetic Evidence From the UK Biobank Supports Carotid Intima‐Media Thickness as a Window Into Cardiac Remodeling

Abstract

Background

Carotid intima‐media thickness (cIMT) is a noninvasive marker of subclinical atherosclerosis linked to future cardiovascular disease (CVD). However, its relationship with early cardiac remodeling and shared genetic basis with cardiovascular disease remain poorly understood. We investigated the phenotypic and genetic associations between cIMT, aortic and cardiac traits, and cardiovascular disease in the UK Biobank, a large population‐based cohort with comprehensive imaging and genetics.

Methods

We analyzed 46 102 participants with carotid ultrasound and cardiac magnetic resonance imaging, excluding those with prevalent heart failure, atrial fibrillation, or stroke. Multivariable linear and Cox regression models assessed associations between common cIMT, aortic and cardiac traits, and incident cardiovascular disease, with effect estimates expressed per 1 SD increase in cIMT. Genetic correlations and pathway enrichment analyses were performed using genome‐wide association studies summary statistics.

Results

Higher cIMT was associated with aortic enlargement, left ventricular hypertrophy, and greater chamber volumes, alongside reduced biatrial ejection fractions (all P<0.001). Higher left cIMT was associated with lower dilated cardiomyopathy (hazard ratio [HR], 0.475 [95% CI, 0.23–0.98], P=0.04) risk. Genetically, cIMT correlated positively with aortic and ventricular traits and negatively with left atrial ejection fraction. Seven shared genes (CDH13, HAND2, ITCH, FBXO32, TBX20, FBN1, and CBFA2AT3) were enriched in pathways related to extracellular matrix organization, vascular development, and cardiac morphogenesis.

Conclusion

Increased cIMT is associated with systemic aortic and cardiac remodeling and decreased dilated cardiomyopathy risk. Integrative genetic analyses revealed shared heritable mechanisms, highlighting cIMT as a marker of early cardiovascular structural change, potentially bridging vascular and cardiac disease processes.

Nonstandard Abbreviations and Acronyms

cIMT

carotid intima‐media thickness

DCM

dilated cardiomyopathy

HOCM

hypertrophic obstructive cardiomyopathy

Research Perspective.

What Is New?

• In ∼46 000 UK Biobank participants, increased carotid intima‐media thickness was associated with subclinical alterations in aortic and cardiac structure and function, as well as decreased risk of dilated cardiomyopathy.

• Genetic correlation analyses identified shared genetic architecture between carotid intima‐media thickness and multiple cardiovascular traits, particularly descending aortic diameter, left ventricular mass, left ventricular wall thickness, and left atrial ejection fraction, with overlapping genes enriched in pathways related to cardiovascular remodeling, cardiac morphogenesis, and extracellular matrix organization.

What Question Should Be Addressed Next?

• What are the mechanistic roles of left ventricular wall thickening, altered stroke volume, and abnormal ejection fraction in mediating changes in aortic and cardiac structure and function associated with increased carotid intima‐media thickness?

Atherosclerotic cardiovascular disease (CVD) is a leading global cause of death, with atherosclerosis as the primary underlying pathology. ¹ Atherosclerosis is marked by early lipid accumulation, vascular remodeling, chronic inflammation, and plaque formation. These changes may progress subclinically from adolescence and culminate in major cardiovascular events such as myocardial infarction or stroke. ² , ³ Ultrasound is a widely used noninvasive technique for assessing carotid intima‐media thickness (cIMT), a surrogate marker of subclinical atherosclerosis. ⁴ , ⁵ Large epidemiological cohorts, including the MESA (Multiethnic Study of Atherosclerosis) ⁶ and the ARIC (Atherosclerosis Risk in Communities) study, ⁷ have shown that increased cIMT predicts future cardiovascular events, highlighting its potential utility in early risk stratification.

As a systemic vascular disorder, atherosclerosis also affects major arteries, including the aorta and carotids, with implications for cardiac structure and function. ⁸ To examine this, studies have linked cIMT to cardiac morphology and function, as well as to CVD outcomes such as stroke, myocardial infarction, and coronary heart disease. ⁵ , ⁹ However, prior investigations have often been limited by small sample sizes for cardiac morphological analyses, heterogeneous measurement protocols, and incomplete characterization of cardiac and aortic phenotypes.

At the genetic level, twin and family studies have demonstrated that cIMT and plaque characteristics are heritable, with heritability estimates for cIMT ranging from ∼30% to 69% across populations and varying with degree of cardiovascular risk factor adjustment, and heritability of carotid plaque presence estimated at ∼52%. ¹⁰ , ¹¹ , ¹² , ¹³ Moreover, genome‐wide association studies have identified loci associated with cIMT and related atherosclerotic traits as well as cardiac traits. ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ However, whether cIMT shares a common genetic architecture with cardiac or aortic remodeling traits remains largely unexplored. Understanding such shared heritability could provide insight into the molecular pathways underlying early cardiovascular remodeling and identify potential mechanisms linking subclinical atherosclerosis to overt cardiac disease.

In this study, we performed an integrative epidemiological and genetic analysis using the UK Biobank cohort and publicly available genetic data from the CHARGE (Cohorts for Heart and Aging Research in Genomic Epidemiology). ¹⁸ We first analyzed the phenotypic associations of cIMT and aortic and cardiac structure and function, as well as with CVD outcomes in asymptomatic individuals. We explored the shared genetic architecture of cIMT, aortic, and cardiac traits to gather insights into the early molecular and structural pathways linking subclinical atherosclerosis to CVD.

METHODS

Data Availability Statement

This research was conducted using data from the UK Biobank (Application Number 12113). The data used in this study are available to bona fide researchers upon application. Derived data generated in this study will be returned to the UK Biobank in accordance with their data sharing policy and will be made available to future approved researchers through the UK Biobank resource.

Study Sample

We used participants from the UK Biobank imaging cohort in this study, a large prospective population‐based study that recruited >500 000 individuals aged 40 to 69 years across the United Kingdom between 2006 and 2010. Detailed information regarding the UK Biobank cohort has been previously published. ¹⁹ , ²⁰ Comprehensive whole‐body imaging and genetic data were available for >60 000 of these participants. ²¹ Cardiac magnetic resonance images and B‐mode ultrasound were used for cardiac characterization and cIMT assessment, respectively. ²² , ²³ Based on linked clinical records, individuals with a history of myocardial infarction, angina, heart failure (HF), stroke, cardiomyopathy, or atrial fibrillation (AF) were excluded from the analysis. Therefore, the study sample consisted of participants free from clinically manifestCVDat baseline. Individual‐level data on age, sex, body mass index, plaque status, systolic blood pressure, diastolic blood pressure, low‐density lipoprotein cholesterol, total cholesterol, and smoking status were also extracted from the UK Biobank.

The Northwest Multi‐Centre Research Ethics Committee approved ethical approval for the UK Biobank Study, and all participants provided electronic informed consent before participation. Generic approval for the UK Biobank Study was granted by National Health Service National Research Ethics Service (Ref. 11/NW/0382).

Carotid IMT Imaging and Extraction

Carotid ultrasound examinations were performed using the CardioHealth Station system (Panasonic Biomedical Sales Europe BV, Leicestershire, UK) equipped with a~9 MHz high‐frequency linear array transducer. B‐mode settings were standardized for carotid imaging, including gain, depth, focal zone, and dynamic range adjustments to enhance visualization of the far wall of the common carotid artery. Image acquisition was performed by trained staff following a standardized UK Biobank protocol. ²¹

Common cIMT measurements were acquired at fixed angles, 150° and 120° for the right carotid artery and 210° and 240° for the left, resulting in 4 cIMT values per participant. A 10‐mm region‐of‐interest box was automatically positioned to track the far wall of the common carotid artery at each acquisition angle. ²¹ , ²⁴ As measurements were fully automated, interobserver variability is confined to image acquisition rather than measurement. Published reproducibility data demonstrate interoperator limits of agreement of −0.063 to +0.056 mm (mean bias −0.003 mm), with excellent inter‐ and intraobserver agreement confirmed in the context of the UK Biobank carotid imaging specifically. ¹⁹ , ²⁵

Plaque status, and maximum, minimum, and mean values were recorded at end‐diastole. ²¹ The average of the 4 common cIMT measurements was calculated by integrating the maximum values across multiple carotid angles, incorporating data from both the near and far arterial walls. The final mean cIMT value for each participant was calculated as the average across all measured angles, providing a representative estimate of overall carotid wall thickness (WT). ²⁶ cIMT measurements that did not meet quality control standards, either due to zero values or identified errors, were omitted from the study. ²³

A custom in‐house pipeline was developed to extract cIMT data from raw vascular imaging records. An optical character recognition‐based workflow was implemented to convert image‐embedded text into machine‐readable outputs. ²⁷ The script detected predefined text anchors and numerical patterns corresponding to left and right cIMT mean values. To evaluate accuracy, a manual quality control step was performed on a randomly selected subset of 30 imaging records. Detailed information on the extraction workflow is provided in Figure S1.

Cardiac Magnetic Resonance Protocol and Image Processing

Cardiac magnetic resonance was performed using a 1.5 Tesla MRI scanner (MAGNETOM Aera, Syngo Platform VD13A; Siemens Healthcare, Erlangen, Germany) with an 18‐channel anterior body surface coil used in combination with 12 elements of an integrated 32‐element spine coil, and ECG gating for cardiac synchronization, as previously described in full. ²⁸ , ²⁹ , ³⁰ For cardiac magnetic resonance images, we processed the left ventricle (LV) short‐axis, long‐axis, and aortic images. Each of these images was segmented using a pretrained neural network to identify the aorta and the 4 cardiac chambers: left atrium, right atrium, LV, and right ventricle. Detailed methodology and validation of the segmentation model can be found in Bai et al. ³⁰ Based on the segmentation, we quantified 25 cardiac traits, spanning the aorta (4 traits), atria (4/4 left atrium/right atrium), ventricular (9/4 LV/right ventricle); see Table S1 for details. ³¹

Demographic and Clinical Data

The UK Biobank data dictionary showcase provides hospital inpatient data and national death and cancer registry data in tabular formats (comma‐separated values or short format text), with most categorical variables encoded. ³² Using these resources, we extracted key variables for participants with cIMT measurements, including age, body mass index, sex, systolic blood pressure, diastolic blood pressure, low‐density lipoprotein cholesterol, total cholesterol, smoking status, and CVD outcomes (AF, HF, dilated cardiomyopathy [DCM], and hypertrophic obstructive cardiomyopathy [HOCM]). Clinical outcomes were identified from linked National Health Service electronic health records based on predefined International Classification of Diseases (ICD), and Office of Population Censuses and Surveys code lists. The ukbpheno package was used to process the clinical data and extract the required diagnostic codes. ³³ For the CVD analyses, participants with a prevalent CVD diagnosis before or within 6 months of the imaging visit were classified as controls to minimize reverse causation and ensure focus on subclinical cardiovascular changes.

Statistical Analysis

Characteristics of the participants were explored through stratification by sex. Continuous variables with a normal distribution were summarized as mean±SD, and skewed variables as median (interquartile range). Categorical variables were presented as percentages (%). The Mann–Whitney U test was used to assess the significant differences in measurement values between women and men, and the chi‐square test was used to examine the significance of differences in categorical variables between the 2 groups. Associations between cIMT (independent variable), aortic and cardiac phenotypes, and CVD were examined using multivariable linear and Cox regression models. All primary models were adjusted for age, sex, body mass index, systolic blood pressure, diastolic blood pressure, low‐density lipoprotein cholesterol, total cholesterol, and smoking status.

Separate models were fitted for each dependent variable, including cardiac structural and functional parameters (ventricular volumes, myocardial mass, WT, and atrial volumes and function) and CVD outcomes (AF, HF, DCM, and HOCM). For continuous variables, results are reported as the change in the dependent variable per 1 SD increase in the independent variable. Results from Cox regression analyses are expressed as hazard ratios for incident CVD per 1 SD increase in the independent variable, with corresponding 95% CIs. The statistical significance threshold was set at P<0.05. Statistical analysis was performed using Python 3.13.2 (https://www.python.org).

Genetic Analysis

Genetic Correlation Between cIMT and Cardiac Structure and Function

Linkage disequilibrium score regression (v1.0.1) was used to estimate genetic correlations between cIMT and a range of aortic and cardiac traits. ⁸ Genome‐wide association studies summary statistics for the cIMT mean phenotypes were downloaded from the CHARGE consortium (http://www.chargeconsortium.com/), and summary statistics for aortic and cardiac traits were derived from the UK Biobank cohort, both derived from individuals of European ancestry. ³ , ¹⁸ , ³⁴ Ancestral alignment supports the validity of the linkage disequilibrium score regression estimates, and residual population stratification is partially controlled by the linkage disequilibrium score regression method.

Variant‐Level Annotations and Genetic Analysis

Genome‐wide association results for cIMT‐associated, aortic, and cardiac traits were analyzed using GWASLab. ³⁵ Genome‐wide significant lead variants, defined as surpassing the significance threshold of P<5×10⁻⁸ within a 500 kb window, were used to identify a set of independent single‐nucleotide variants associated with cIMT phenotypes, aortic, and cardiac traits. Variants exhibiting −log₁₀(P) values <2 were omitted to optimize computational efficiency during plotting, and variants with −log₁₀(P) values >20 were rescaled down to this limit to improve the clarity and interpretability of the visualization. The closest protein‐coding gene was annotated for each corresponding lead variant, and the resulting gene set was used to identify genes shared between cIMT, aortic, and cardiac traits.

Gene Set Enrichment and Pathway Analyses

Functional and pathway enrichment analyses were performed using g:GOSt within the g:Profiler suite, which integrates annotations from Gene Ontology, Human Phenotype Ontology, Reactome, Kyoto Encyclopedia of Genes and Genomes, and WikiPathways. Genes mapped to validated loci for cIMT, aortic, and cardiac traits were used as input, and results were summarized to highlight enriched biological pathways and molecular processes. ³⁶

RESULTS

Study Population

This study included 46 102 participants (51.2% female), with details of the population listed in Table 1. Female participants were slightly younger (mean age 64.4 versus 65.7 years) and had a lower body mass index (25.9 versus 26.9 kg/m²). Compared with men, women had a lower prevalence of smoking, with a greater proportion of never smokers (64.0% versus 56.8%) and lower proportions of former smokers (30.4% versus 35.4%) and current smokers (5.3% versus 7.6%). In contrast, men had higher systolic and diastolic blood pressure than women (systolic blood pressure: 140.4 versus 132.9 mm Hg; diastolic blood pressure: 83.5 versus 79.3 mm Hg). Low‐density lipoprotein cholesterol levels were comparable between men and women (3.6 mmol/L), and women exhibited higher total cholesterol levels than men (5.8 versus 5.59 mmol/L). Carotid plaque was more common in men (right 0.05%, left 0.03%) than in women (0.01 each). Men also had a higher prevalence of CVD, including AF (3.9% versus 1.5%), HF (1.4% versus 0.5%), DCM (0.11% versus 0.02%), and HOCM (0.11% versus 0.04%). Mean cIMT values were lower in women than in men, with smaller overall aortic and cardiac dimensions; however, women exhibited slightly higher ejection fraction (EF; P<0.001). Details of statistical analysis can be found in Table S2.

Table 1.

1Characteristics of the UK Biobank Participants Stratified by Sex

Variable	Female sex (N=23 586)	Male se (N=22 516)	P value
Age, y	64.4 (7.6)	65.7 (7.8)	<0.001
Body mass index, kg/m2	25.9 (4.6)	26.9 (3.8)	<0.001
Nonsmoker	15 106 (64)	12 793 (56.8)	<0.001
Previous smoker	7181 (30.4)	7975 (35.4)
Current smoker	1247 (5.3)	1705 (7.6)
Systolic blood pressure, mm Hg	132.9 (19.1)	140.4 (17.5)	<0.001
Diastolic blood pressure, mm Hg	79.3 (10.4)	83.5 (10.2)	<0.001
Low‐density lipoprotein cholesterol, mmol/L	3.6 (0.8)	3.6 (0.8)	<0.001
Total cholesterol, mmol/L	5.8 (1.1)	5.59 (1.1)	<0.001
cIMT measurements
Left mean cIMT	0.68 (0.60, 0.78)	0.74 (0.63, 0.87)	<0.001
Right mean cIMT	0.68 (0.61, 0.77)	0.72 (0.62, 0.83)	<0.001
Right plaque present	3 (0.01)	13 (0.05)	0.05
Left plaque present	3 (0.01)	7 (0.03)	0.31
Cardiovascular disease
Atrial fibrillation	363 (1.5)	869 (3.9)	<0.001
Heart failure	111 (0.5)	309 (1.4)	<0.001
Dilated cardiomyopathy	4 (0.02)	25 (0.11)	<0.001
Hypertrophic cardiomyopathy	9 (0.04)	25 (0.11)	0.006

Data are reported as mean±SD for normally distributed and median (Q1, Q3) for nonnormally distributed continuous variables; categorical variables are summarized as count (%). cIMT indicates carotid intima‐media thickness.

Associations Between cIMT, Aortic, and Cardiac Traits, and CVD

In fully adjusted multivariable regression models, higher left and right mean cIMT were consistently associated with larger aortic cross‐sectional areas, increased chamber size, and greater atrial and ventricular volumes (all P<0.001; Figure 1). Conversely, cIMT demonstrated significant inverse associations with left and right atrial EFs (P<0.001), as well as LVEF (P<0.001), indicating early functional impairment accompanying structural enlargement. No association was observed between left cIMT and right ventricular EF (P=0.145), whereas a weak association was noted for right cIMT (P=0.037).

Figure 1. Associations between left and right mean carotid intima‐media thickness ‐ and cardiac structure and function (N=46 102).

Multivariable linear regression models examined left cIMT (A, C, E) and right cIMT (B, D, F) in relation to (A, B) aortic traits and left ventricular wall thickness, (C, D) left heart morphology and function, and (E, F) right heart morphology and function. Models were adjusted for age, sex, body mass index, systolic blood pressure, diastolic blood pressure, low‐density lipoprotein cholesterol, total cholesterol, and smoking status. Results are shown as beta coefficients (β) with corresponding 95% CIs, representing the change in each cardiac trait per 1 SD increase in cIMT. * denotes statistically significant association (P<0.05). AAo indicates ascending aortic area; cIMT, carotid intima‐media thickness; DAo, descending aortic area; LAEF, left atrial ejection fraction; LASV, left atrial stroke volume; LAV, left atrial volume; LVCO, left ventricular cardiac output; LVEDV, left ventricular end‐diastolic volume; LVEF, left ventricular ejection fraction; LVESV, left ventricular end‐systolic volume; LVM, left ventricular mass; LVSV, left ventricular stroke volume; RAEF, right atrial ejection fraction; RASV, right atrial stroke volume; RAV, right atrial volume; RVEDV, right ventricular end‐diastolic volume; RVEF, right ventricular ejection fraction; RVESV, right ventricular end‐systolic volume; RVSV, right ventricular stroke volume; and WT, wall thickness.

Although higher cIMT correlated positively with left atrial stroke volume, no association was observed with right atrial stroke volume (P>0.05). Moreover, an inverse association was observed between cIMT, and LV cardiac output, borderline significant for right cIMT and slightly stronger for left cIMT (all P<0.05). Further details can be found in Table S3.

The median follow‐up times were 3374 days for AF, 3207 days for HF, 2861 days for DCM, and 2820 days for HOCM. In adjusted Cox proportional hazards models, higher left cIMT was significantly associated with a decreased risk of DCM (hazard ratio [HR], 0.475 [95% CI, 0.23 –0.98], P=0.04), shown in Figure 2C. No significant associations were found with AF, HF, or HOCM (all P>0.05), shown in Figures 2A, 2B, and 2D. Details can be found in Table S4.

Figure 2. Associations between left and right mean carotid intima‐media thickness and cardiovascular disease (atrial fibrillation, heart failure, dilated cardiomyopathy, hypertrophic obstructive cardiomyopathy) estimated using Cox proportional hazards regression (N=46 102).

Cox regression models of left and right cIMT with (A) atrial fibrillationAF, (B) heart failureHF, (C) dilated cardiomyopathy, and (D) hypertrophic obstructive cardiomyopathy. Models were adjusted for age, sex, body mass index, systolic blood pressure, diastolic blood pressure, low‐density lipoprotein cholesterol, total cholesterol, and smoking status. Participants with a diagnosis of CVD before, or within 6 mo following, the imaging visit were excluded from all analyses. Results are shown as HRs with corresponding 95% CIs, representing the risk of incident CVD per 1 SD increase in cIMT. * denotes statistically significant association (P<0.05). AF indicates atrial fibrillation; cIMT, carotid intima‐media thickness; DCM, dilated cardiomyopathy; and HOCM, hypertrophic obstructive cardiomyopathy.

Genetic Correlations of cIMT With Aortic and Cardiac Phenotypes

Linkage disequilibrium score regression analysis revealed significant genetic correlations between mean cIMT and measures of ventricular WT, with the strongest association for LV WT (basal; rg=0.50, P<0.001), and weaker but directionally consistent correlations for WT mid and WT apical (Figure 3A). Positive correlations were also identified with descending aortic area (maximum; rg=0.2, P=0.01), whereas no significant correlations were found with ascending aortic dimensions.

Figure 3. Genetic correlations between carotid intima‐media thickness and quantitative aortic and cardiac phenotypes (N=46 102).

Linkage disequilibrium score regression analysis of cIMT with (A) aortic and global cardiac morphology and function, (B) right heart morphology and function, and (C) left heart morphology and function. * denotes statistically significant genetic correlations (P<0.05). AAo indicates ascending aortic area; cIMT, carotid intima‐media thickness; DAo, descending aortic area; LAEF, left atrial ejection fraction; LASV, left atrial stroke volume; LAV, left atrial volume; LVCO, left ventricular cardiac output; LVEDV, left ventricular end‐diastolic volume; LVEF, left ventricular ejection fraction; LVESV, left ventricular end‐systolic volume; LVM, left ventricular mass; LVSV, left ventricular stroke volume; RAEF, right atrial ejection fraction; RASV, right atrial stroke volume; RAV, right atrial volume; RVEDV, right ventricular end‐diastolic volume; RVEF, right ventricular ejection fraction; RVESV, right ventricular end‐systolic volume; RVSV, right ventricular stroke volume; and WT, wall thickness.

For right heart traits, cIMT showed positive genetic correlations with right ventricular end‐diastolic volume (rg=0.2, P=0.01) and stroke volume (rg=0.3, P<0.001) but not with other right‐sided parameters (Figure 3B). With respect to the left atrium, cIMT was positively correlated with atrial volume, stroke volume, and LV mass (all r≈0.4, P<0.001) but negatively correlated with left atrial EF (P=0.01), and no significant correlation with LVEF or end‐systolic volume (all P>0.05; Figure 3C); see Table S5 for details.

Shared Genes and Molecular Pathways

A total of 33 independent genetic loci associated with cIMT mean, and 193 loci related to aortic and cardiac traits, as well as gene expression, were identified in the genome‐wide association studies. Among the 193 prioritized genes associated with aortic and cardiac traits, 7 genes (CBFA2T3, CDH13, FBN1, FBXO32, HAND2, ITCH, and TBX20), were consistently shared with cIMT mean; see Figure S2.

Pathway enrichment analysis using gProfiler highlighted biological processes central to vascular and cardiac development (Figure 4). In the Gene Ontology Molecular Function category, significant enrichment was observed for extracellular matrix (ECM) structural constituents, proteoglycan binding, and integrin interactions (adjusted P=1.1×10⁻³–2.3×10⁻²). Within biological processes, key pathways included blood vessel and circulatory system development, vasculature morphogenesis, smooth muscle cell differentiation, and regulation of growth factor responses (P=7.1×10⁻⁸); see Figure S3.

Figure 4. Summary of overlapping biological pathways identified for carotid intima‐media thickness, aortic, and cardiac parameters.

Candidate genes associated with cIMT, aortic, and cardiac traits from the primary genome‐wide association studies were analyzed for shared biological pathways and manually categorized into functional groups related to atherosclerosis and cardiac development. cIMT indicates carotid intima‐media thickness.

Across cIMT, aortic, and cardiac traits, 25 pathways were shared (Figure 4), encompassing cardiac morphogenesis, ECM remodeling, vascular and tissue development, and cellular adhesion and migration. These convergent processes reinforce that cIMT reflects integrated vascular–cardiac remodeling mechanisms central to early CVD pathogenesis.

DISCUSSION

To our knowledge, this is the largest population‐based study to systematically investigate the relationship between cIMT, a comprehensive spectrum of aortic and cardiac phenotypes, and clinical CVD using imaging. We found that increased cIMT was associated with subclinical structural remodeling, including increased aortic diameter, chamber enlargement, and myocardial hypertrophy, and with a lower risk of DCM, indicating that cIMT reflects early remodeling rather than overt systolic dysfunction. Genetic correlation analyses further revealed substantial shared heritable influences between cIMT and multiple vascular and myocardial traits, particularly descending aortic dimensions, LV WT, and myocardial mass. Consistent with this, pathway analyses demonstrated convergence on biological processes involved in vascular and cardiac development, ECM organization, and tissue remodeling, suggesting that cIMT, aortic, and cardiac traits are genetically and biologically coupled. Together, these findings position cIMT as a marker of subclinical structural cardiovascular remodeling, rather than a universal predictor of clinical CVD outcomes.

In addition, femoral plaques have demonstrated stronger associations with traditional cardiovascular risk factors and positive coronary artery calcium scores compared with carotid plaques, suggesting that assessment of both vascular beds may improve cardiovascular risk stratification and the detection of subclinical coronary artery disease. ³⁷ , ³⁸ , ³⁹ Although the predictive value of both carotid and femoral plaques increases with overall plaque burden, particularly with progression over time, these findings further highlight the clinical importance of evaluating both plaque presence and progression.

Observational Associations of cIMT, Aortic and Cardiac Traits, and CVD

The mechanisms linking increased carotid WT to downstream aortic and myocardial remodeling remain incompletely understood. ⁵ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ⁴⁴ Moreover, few prior studies have simultaneously examined both left and right cIMT in relation to early‐stage aortic and cardiac dysfunction within large, deeply phenotyped cohorts. Consequently, incorporating data from both carotid arteries provides a more comprehensive and accurate assessment of carotid artery morphology, thereby improving the detection of subclinical vascular disease. Previous evidence suggests that cIMT reflects more than just vascular atherosclerosis. In the MESA study, cIMT independently predicted impaired myocardial strain, a marker of subclinical systolic and diastolic dysfunction, in >500 asymptomatic individuals. ⁴³ Similarly, in a study of 4301 UK Biobank participants without CVD, higher cIMT was strongly associated with concentric remodeling of left and right ventricular chambers, impaired atrial function, increased LV mass, and subclinical LV dysfunction. ⁴⁰

Our findings extend these observations by providing a more comprehensive characterization of atrial structural and functional changes, as well as aortic cross‐sectional dimensions. We observed that higher cIMT values were associated with significantly enlarged ascending and descending aortic areas, a pattern that may reflect increased systemic vascular resistance and altered vascular compliance. Compensatory increases in cardiac output to maintain perfusion could impose chronic hemodynamic load on the heart, supporting the observed associations with LV wall thickening and biatrial and biventricular structural and functional alterations. ⁴⁵ , ⁴⁶ , ⁴⁷

One plausible explanation is that subclinical atherosclerosis induces myocardial ischemia due to a supply–demand mismatch, triggering adverse remodeling, such as cardiomyocyte hypertrophy and reduced ventricular compliance. ⁴⁸ Chronic exposure to atherosclerosis‐related risk factors (eg, hypertension, dyslipidemia, and alcohol use) may further exacerbate this process, progressively impairing myocardial structure and function. ⁴⁹ Over time, these adaptations may evolve into HF with preserved EF or, in the presence of additional stressors such as myocardial infarction, progress to HF with reduced EF. Chronic low‐grade inflammation accompanying increased carotid WT may also initiate early myocardial remodeling, including hypertrophy and fibrosis, that precedes clinically overt HF. ⁵⁰

Importantly, these systemic and myocardial remodeling processes may not be uniform across the carotid vasculature, providing a potential explanation for the differential associations observed between left and right cIMT and cardiac morphology. Previous studies have shown that left cIMT was consistently greater than right cIMT across the studied age range in our cohort. Previous studies also have reported that left cIMT is more strongly associated with metabolic risk factors, such as diabetes and dyslipidemia, whereas right cIMT is more closely associated with hypertension. Anatomically, the left cIMT arises directly from the aortic arch and is therefore more influenced by hydrostatic pressure, whereas the right cIMT originates from the brachiocephalic trunk and is exposed to different pulsatile flow dynamics from the ascending aorta. ⁵¹ , ⁵² Furthermore, age‐related reductions in endothelial shear stress have been shown to promote atherosclerotic plaque development, potentially altering circumferential wall tension and carotid remodeling, which may contribute to downstream aortic and cardiac morphology and function. ⁵³ , ⁵⁴ , ⁵⁵ , ⁵⁶

Taken together, our findings suggest that cIMT may serve not only as a marker of vascular atherosclerosis but also as an early hemodynamic biomarker of maladaptive cardiac remodeling. This dual role supports its potential utility in risk stratification and intervention before the onset of symptomatic CVD. Consistent with prior studies, such as a UK Biobank analysis of 29 292 individuals showing cIMT as an independent predictor of coronary heart disease and myocardial infarction, ⁵ we also observed associations between increased cIMT and subclinical aortic and cardiac remodeling. In contrast, we did not find significant associations with overt AF, HF, or HOCM, likely due to the predominantly healthy composition of the UK Biobank imaging cohort and limited event rates.

Interestingly, higher left cIMT was associated with a lower risk of incident DCM, despite its association with subclinical structural remodeling. This apparent divergence likely reflects the heterogeneity of DCM, which includes genetic, inflammatory (eg, viral myocarditis), toxic (eg, alcohol‐related), and peripartum forms, many of which are not primarily driven by atherosclerotic processes. ⁵⁷ As such, the biological pathways reflected by cIMT may not align with the dominant mechanisms underlying most DCM cases. Notably, ischemic cardiomyopathy represents the subtype most plausibly linked to atherosclerosis, where a positive association with cIMT might be expected. However, subtype stratification was not possible in the present data set. The inverse association observed may reflect differences in underlying remodeling pathways; however, this requires further investigation. The relationship between cIMT and incident cardiovascular outcomes is therefore end point specific, with greater relevance to atherosclerosis‐driven conditions than to primary cardiomyopathies.

Genetic Associations of cIMT, Aortic, and Cardiac Traits

In line with known epidemiology, the direction of genetic correlations between cIMT and traits such as ascending aortic area, right atrial chamber size and function, and LV end stroke volume was opposite to that observed in the phenotypic analysis. Nevertheless, strong positive genetic correlations were identified between cIMT and LV WT, as well as LV chamber size and volume. In contrast, LVEF showed only weak but statistically significant correlations with cIMT traits. These findings indicate that genetically mediated LV structure may play a greater role than right‐sided traits in early atherosclerotic remodeling.

Previous studies have shown shared genetic correlations of cIMT with obesity and glucometabolic traits, suggesting that some variants influence vascular remodeling and metabolic regulation simultaneously. ³ , ¹⁶ , ⁵⁸ Such pathways may alter endothelial function and arterial stiffness, ultimately influencing cardiac structure. Thus, the linkage disequilibrium score regression correlations likely reflect shared heritable metabolic‐vascular‐cardiac mechanisms rather than secondary hemodynamic effects alone.

The divergence between genetic and observational findings likely reflects differences in study design and phenotype derivation rather than biological contradiction. Genetic approaches capture lifelong inherited influences, whereas observational analyses reflect cumulative environmental exposure, lifestyle, and comorbidity effects. ⁵⁹ , ⁶⁰ , ⁶¹ Therefore, a comprehensive understanding of these relationships necessitates the integration of both inherited and modifiable risk factors. Because this is the first study to examine cIMT mean using genetic correlation analysis, independent validation is warranted.

Commonalities and Distinctions in Underlying Biological Pathways

Only a small subset of single‐nucleotide variants associated with cIMT also showed associations with cardiac traits. Shared variants in genes such as CDH13, HAND2, ITCH, and TBX20 align with established mechanisms in vascular remodeling, ECM dynamics, inflammation, and cardiac development, processes central to both atherosclerosis and myocardial structural changes. For example, CDH13, which encodes T‐cadherin, is implicated in activation of the NF‐κB (nuclear factor‐κB) signaling pathway and smooth muscle proliferation, processes linked to carotid atherosclerosis and CVD. ⁶² , ⁶³ , ⁶⁴ Moreover, HAND2, may exert protective effects by suppressing proinflammatory genes, thereby reducing the risk of plaque formation and cardiovascular events. ⁶⁵ ITCH modulates NF‐κB and TGF‐β (transforming growth factor‐β) signaling, thereby influencing proatherogenic gene expression and plaque formation. ⁶⁶ In addition, ITCH has been implicated in cardiac hypertrophy and fibrosis through ubiquitination of key signaling molecules, contributing to both carotid atherosclerosis and pathological cardiac remodeling. Collectively, these findings suggest mechanistic links between early vascular changes and subclinical cardiac remodeling. Hence, incorporating genetic data with imaging phenotypes may enhance risk assessment and facilitate personalized preventative methods, including targeted lifestyle or pharmacological therapies to alleviate vascular and myocardial stress.

Limitations

The UK Biobank cohort comprised predominantly healthy, less deprived, middle‐aged to older UK volunteers, which may limit generalizability to populations with different health profiles, age distributions, and ethnic backgrounds. Furthermore, this study may be subject to selection bias due to missing imaging and covariate data, which resulted in the exclusion of certain participants from the analysis and may therefore influence the representativeness of the study sample. Moreover, the cross‐sectional study design precludes the establishment of causal inferences regarding the observed relationships between carotid atherosclerosis, aortic, and cardiac traits. Although multivariable analyses were adjusted for key potential confounders, the influence of residual or unmeasured confounding factors cannot be entirely ruled out and may contribute to the observed associations. Additionally, we were unable to distinguish between ischemic and nonischemic subtypes of DCM, which may have influenced subtype‐specific associations with cIMT. In the genetic analysis, although shared genes were identified, limitations include polygenicity, measurement variability, and heterogeneity across phenotypes, which may have reduced detectable associations. Future studies should incorporate longitudinal imaging and multiomics to elucidate how variants contribute to cardiovascular remodeling over time.

CONCLUSIONS

We demonstrated that increased cIMT was associated with alterations in cardiac structure and function and with a lower prevalence of DCM. Genetic correlation analyses further revealed that cIMT, aortic dimensions, and cardiac traits share common genetic architecture and biological pathways, including those involved in aortic and cardiac chamber development, ECM organization, and cellular proliferation. These findings suggest that routine cIMT assessment in high‐risk individuals may enable earlier identification of subclinical cardiac remodeling before overt CVD becomes clinically manifest, supporting its potential utility as an early, noninvasive biomarker for risk stratification, surveillance, and preventive intervention. The results reinforce the need for comprehensive management of cardiovascular risk factors and motivate future investigation into the transcriptomic and epigenetic mechanisms driving these associations.

Sources of Funding

Adama Fatima Saccoh is supported by a 4‐year PhD studentship from the British Heart Foundation. Marion van Vugtis supported by a postdoc talent grant from the Amsterdam Cardiovascular Sciences. Amand Floriaan Schmidt is supported by British Heart Foundation grant PG/22/10989, the University College London British Heart Foundation Research Accelerator AA/18/6/34223, the University College London British Heart Foundation Centre of Research Excellence RE/24/130013, MR/V033867/1, the National Institute for Health and Care Research University College London Hospitals Biomedical Research Centre, and the European Union Horizon scheme (AI4HF 101 080 430 and DataTools4Heart 101 057 849). Nikhil Paliwal is supported by the University College London British Heart Foundation Research Accelerator AA/18/6/34223, the Aortic Dissection Charitable Trust and the National Institute for Health and Care Research University College London Hospitals Biomedical Research Centre. University College London also receives support from a British Heart Foundation Centre of Research Excellence award.

Disclosures

The authors have declared no competing interest.

Supporting information

Tables S1–S5

Figures S1–S3

STROBE Checklist