UK Biobank Research Articles

The associations of n-3 fatty acid intake with handgrip strength and muscle mass indices in older adults: A cross-sectional study from UK Biobank

ObjectivesTo investigate the associations of n-3 fatty acid intake with handgrip strength and muscle mass indices in older adults. A secondary aim was to investigate whether these associations differed by physical activity status. Research methods & proceduresA cross-sectional study included 53,170 participants aged 60 years and over from the UK biobank (25,324 men and 27,846 women). The primary outcomes were grip strength index and muscle mass index, the predictor variable was n-3 fatty acid intake and the covariates were age, ethnicity, Townsend deprivation index, physical activity, multimorbidity count, total energy intake, body fat percentage and the month of assessment. Multivariable linear regression analyses were performed across 5 models. Model 1 was unadjusted; model 2 adjusted for age, ethnicity, deprivation index and month of assessment; model 3 adjusted as in model 2 plus total energy intake; model 4 as in model 2 plus multimorbidity count; and model 5 as in model 4 plus body fat percentage. ResultsIn model 5, higher n-3 fatty acid intake was positively associated with the grip strength index in women. For each additional gram of n-3 fatty acid consumed per day, there was an increase of 0.03 kg/m2 (95% CI: 0.00 to 0.06 kg/m2) in active women and 0.04 kg/m2 (95% CI: 0.00 to 0.07 kg/m2) in inactive women. However, no significant associations were observed in men, whether active (p = 0.405) or inactive (p = 0.323). Additionally, no significant associations were found between n-3 fatty acid intake and muscle mass index in either active (p = 0.858) or inactive (p = 0.250) men, or in active (p = 0.909) or inactive (p = 0.187) women. ConclusionAlthough n-3 fatty acid intake was associated with grip strength index in older women, regardless of their activity status, the magnitude of this association was very small and unlikely to be clinically relevant. Additionally, n-3 fatty acid was not associated with muscle mass index.

Association of glycaemic control with intraocular pressure in a large general population: Results from the UK Biobank.

To evaluate the association of glycated haemoglobin (HbA1c) and serum glucose with intraocular pressure (IOP) in a large UK general population. Participants were selected from the UK Biobank, excluding those with eye conditions that may affect IOP. IOP was measured using an ocular response analyser. Goldmann-correlated IOP (IOPg) and corneal-compensated IOP (IOPcc) were outcomes of interest, and ocular hypertension was defined as left-eye IOPg or IOPcc > 21 mmHg. HbA1c and random (non-fasting) serum glucose were the exposures of interest. Multivariate restricted cubic spline models, as well as linear regression, were applied to explore the associations of interest. Among 68 806 participants (46.5% male), the mean age was 56.7 years. The mean (standard deviation) for IOPg was 15.7 (3.6) mmHg and 15.9 (3.6) mmHg for IOPcc. Occular hypertension was prevalent in 8055 participants (11.7%) and 4178 participants (6.1%) had diabetes. Those with diabetes had higher IOP and a higher prevalence of ocular hypertension. After adjustment for demographic and clinical variables, HbA1c was positively associated with IOP in participants with diabetes, but not in those without diabetes. For every 10-mmol/mol increase in HbA1c, IOPg increased by 0.20 mmHg (95% confidence interval [CI] 0.12, 0.28) and IOPcc by 0.15 mmHg (95% CI 0.07, 0.23); the odds of ocular hypertension was increased by 6% (95% CI 1.00, 1.13) in participants with diabetes. A borderline positive association between serum glucose and IOP was found only in participants without diabetes. Impaired glycaemic control was associated with elevated IOP and a possible risk of ocular hypertension among participants with diabetes but of normal ocular health.

Artificial intelligence-enabled electrocardiogram for mortality and cardiovascular risk estimation: a model development and validation study

Artificial intelligence (AI)-enabled electrocardiography (ECG) can be used to predict risk of future disease and mortality but has not yet been adopted into clinical practice. Existing model predictions do not have actionability at an individual patient level, explainability, or biological plausibi. We sought to address these limitations of previous AI-ECG approaches by developing the AI-ECG risk estimator (AIRE) platform. The AIRE platform was developed in a secondary care dataset (Beth Israel Deaconess Medical Center [BIDMC]) of 1 163 401 ECGs from 189 539 patients with deep learning and a discrete-time survival model to create a patient-specific survival curve with a single ECG. Therefore, AIRE predicts not only risk of mortality, but also time-to-mortality. AIRE was validated in five diverse, transnational cohorts from the USA, Brazil, and the UK (UK Biobank [UKB]), including volunteers, primary care patients, and secondary care patients. AIRE accurately predicts risk of all-cause mortality (BIDMC C-index 0·775, 95% CI 0·773-0·776; C-indices on external validation datasets 0·638-0·773), future ventricular arrhythmia (BIDMC C-index 0·760, 95% CI 0·756-0·763; UKB C-index 0·719, 95% CI 0·635-0·803), future atherosclerotic cardiovascular disease (0·696, 0·694-0·698; 0·643, 0·624-0·662), and future heart failure (0·787, 0·785-0·789; 0·768, 0·733-0·802). Through phenome-wide and genome-wide association studies, we identified candidate biological pathways for the prediction of increased risk, including changes in cardiac structure and function, and genes associated with cardiac structure, biological ageing, and metabolic syndrome. AIRE is an actionable, explainable, and biologically plausible AI-ECG risk estimation platform that has the potential for use worldwide across a wide range of clinical contexts for short-term and long-term risk estimation. British Heart Foundation, National Institute for Health and Care Research, and Medical Research Council.

Genetically Predicted Sleep Traits and Sensorineural Hearing Loss: A Mendelian Randomization Study.

Observational studies suggest a potential association between sleep characteristics, sensorineural hearing loss (SNHL), and sudden SNHL (SSNHL), but causal evidence is scarce. We sought to clarify this issue using two-sample Mendelian randomization analysis. The inverse-variance weighted (IVW) method was performed as primary analysis to assess bidirectional causal associations between sleep traits (chronotype, sleep duration, insomnia, daytime sleepiness, and snoring) and SNHL/SSNHL using publicly available Genome-Wide Association Studies summary data from two large consortia (UK Biobank and FinnGen). Sensitivity analyses, including Mendelian randomization (MR)-Egger, Mendelian randomization pleiotropy residual sum and outlier, weight median, Cochran's Q test, leave-one-out analysis, and potential pleiotropy analysis, were conducted to ensure robustness. IVW analysis found suggestive associations of morning chronotype (odds ratio [OR] = 1.08, 95% confidence interval [CI] = 1.01-1.16, p = 0.031) and daytime sleepiness (OR = 1.88, 95% CI = 1.24-2.87, p = 0.003) with SNHL onset. Additionally, morning chronotype was nominally associated with SSNHL onset using IVW method (OR = 1.37, 95% CI = 1.10-1.71, p = 0.006). However, there was no evidence for the causal effect of SNHL and SSNHL on different sleep traits (all p > 0.05). Sensitivity analysis showed that the results were stable. Within the MR limitations, morning chronotype and daytime sleepiness were underlying causal contributors to the burden of SNHL, indicating that optimal sleep might facilitate the prevention and development of SNHL. 3 Laryngoscope, 134:4723-4729, 2024.

Degree of joint risk factor control and incident chronic kidney disease among individuals with obesity.

To investigate the extent to which joint risk factor control might attenuate the excess risk of chronic kidney disease (CKD) in participants with obesity. We included a total of 97 538 participants who were obese at baseline and matched 97 538 normal weight control participants from the UK Biobank in the analysis. The degree of joint risk factor control was assessed based on six major CKD risk factors, including blood pressure, glycated haemoglobin, low-density lipoprotein cholesterol, albuminuria, smoking and physical activity. The Cox proportional hazards models were used to estimate associations between the degree of risk factor control and risk of CKD, following participants from their baseline assessment until the occurrence of CKD, death, or the end of the follow-up period. Among participants with obesity, joint risk factor control showed an association with a stepwise reduction of incident CKD risk. Each additional risk factor control corresponded to an 11% (hazard ratio: 0.89; 95% confidence interval: 0.86-0.91) reduced risk of CKD among participants with obesity, with the optimal controlling of all six risk factors associated with a 49% (hazard ratio: 0.51; 95% confidence interval: 0.43-0.61) decrease in risk of CKD. Furthermore, in individuals with obesity who jointly controlled all six risk factors, the excess risk of CKD associated with obesity was effectively neutralized compared with normal weight control subjects. Notably, the protective correlations between the degree of joint risk factor control and the incidence of CKD were more pronounced in men compared with women, in individuals with a lower healthy food score versus a higher score, and among diabetes medication users as opposed to non-users (pinteraction = 0.017, 0.033 and 0.014, respectively). The joint risk factor control is associated with an inverse association of CKD risk in an accumulative manner among individuals with obesity. Achieving ideal control over risk factors may effectively counterbalance the excessive risk of CKD typically associated with obesity.

Gestational diabetes mellitus, body mass index, and cardiometabolic multimorbidity: a prospective cohort study

PurposeGestational diabetes mellitus (GDM) could increase the risks of type 2 diabetes mellitus (T2DM) and cardiovascular disease. However, evidence on its association with cardiometabolic multimorbidity (CMM) was limited. This study aimed to evaluate the association between GDM and the prevalence, incidence, patterns, and progression of CMM; and the role of body mass index (BMI) in such association. MethodsThis study included 203,372 women who have given birth in UK Biobank. The diagnoses of GDM and cardiometabolic diseases (including stroke, coronary heart disease [CHD], and T2DM) were reported by participants or obtained through linkage to inpatient hospital data until 31st December 2020. BMI was assessed at the baseline assessment. CMM was defined as having two or more of included cardiometabolic diseases. Logistic regression models and Cox proportional hazard models were used to assess the association between GDM and CMM, and the modifications on both additive and multiplicative scales were assessed to evaluate the effect of BMI on such association. ResultsA total of 1,217 women had a history of GDM, 2,351 participants had CMM at the end of follow-up and 1,601 was newly diagnosed during follow-up. GDM was associated with higher prevalence (odds ratio [OR]=4.64, 95% confidence interval [95% CI]=3.54-6.08) and incidence (hazard ratio [HR]=3.62, 95% CI=2.62-5.00) of CMM. In particular, GDM was associated with higher odds of T2DM, coexisting T2DM and vascular disease, and T2DM followed by vascular disease. Formal testing for effect modification suggested multiplicative modification by BMI for the association between GDM and incident CMM. ConclusionsGDM was associated with CMM in women’s late life, with multiplicative modification effects of BMI. Our results suggest that maternal and lifestyle interventions (e.g., weight management) are warranted for the primary and secondary prevention of CMM, particularly in women with a history of GDM.

The impact of anxiety on the risk of kidney stone disease: Insights into eGFR-mediated effects.

Previous studies have linked kidney stone disease (KSD) with depression, but there are no reports on the relationship between anxiety and KSD, and the mechanism underlying the potential relationship remains unclear. Associations of anxiety and incident KSD were assessed in the National Health and Nutrition Examination Survey (NHENES) using multivariate logistic regression. Two-sample bidirectional Mendelian randomization studies and a two-step two-sample MR was used to estimate the mediating factors that influence KSD risk. Examinations of NHANES data revealed that a rise in the frequency and intensity of anxiety were independently associated with incident KSD. In MR analysis, anxiety (uk-a-51 and uk-b-6519) were from the UK Biobank, with sample sizes of 328,717 and 450,765 respectively. KSD data were from the FinnGen, including 8597 cases and 333,128 controls. In the IVW analysis, genetically predicted anxieties (ukb-a-51 and ukb-b-6519) were found to be causally associated with a higher risk of KSD, with odds ratios of 6.18 (95% CI 2.54-15.04) and 3.44 (95% CI 1.67-7.08), respectively. There were no reverse causal effects. Further mediation analysis indicated that anxiety increases the risk of KSD by raising eGFR, through which 11.8% of the effect of anxiety on KSD risk was mediated. The research was confined to individuals of European heritage, and there could be specific genetic variances among diverse ethnicities. The current study suggests anxiety as an independent causal risk factor for KSD and unveils a new pathogenic mechanism, showing that anxiety raises eGFR, thereby increasing the risk of KSD.

Independent and joint relationships of cardiorespiratory fitness and body mass index with liver fat content.

To investigate the relationship between cardiorespiratory fitness (CRF) and liver fat content (LFC) in community-based participants and highlight their relationship in people with different body mass indices (BMIs). Using UK Biobank data, CRF was estimated with bicycle ergometer fitness testing and was evaluated based on physical work capacity at 75% maximum heart rate (PWC75%). LFC was quantified through liver proton density fat fraction (PDFF) on magnetic resonance imaging. Multivariate linear regression models were used to analyse the associations of CRF and BMI with absolute reduction and percentage change in PDFF (%). In total, 5765 participants with a mean age of 55.57 years and a median (range) follow-up of 10.7 (4.0-17.7) years were included. Compared with the lowest PWC75% tertile, the absolute reduction and percentage change in PDFF in the highest PWC75% tertile were -0.450 (95% confidence interval [CI] -0.699 to -0.192) and -4.152 (95% CI -6.044 to -2.104), respectively. These associations were independent of BMI, and individuals with obesity and normal weight had the largest absolute reduction and percentage change in LFC, respectively (p for interaction <0.001). Joint analysis showed that PWC75% and BMI had a negative dose-response relationship with PDFF. These associations were consistent in different sex and age subgroups (p for interaction >0.05). There was a significant negative association between CRF and LFC, and this association was independent of BMI. The results of this study strongly recommend improving CRF to mitigate LFC.

The effect of depression status on osteoarthritis: A powerful two-step Mendelian randomization study.

There is no conclusive epidemiological evidence regarding the relationship between OA, depression, and whole-body fat mass. In this study, we conducted a two-step Mendelian randomization analysis to determine the causal relationships between them. The published summary-level data are from genome-wide association studies (GWAS). Our study included 357,957 samples and 10,828,862 SNPs. Finally, the outcome GWAS data for OA came from a GWAS on the genetic architecture of OA using UK Biobank data. This study included 50,508 samples and 15,845,511 SNPs. We used five different modes of analysis, including inverse variance weighted meta-analysis (IVW), MR-Egger regression, weighted median, simple mode, and weighted mode, to explore causal relationships. We found a positive correlation between depression and body fat mass, with depression leading to body fat mass an increase in (IVW result: p=3.39E-07, OR (95% CI) =2.16 (1.61, 2.90)). We also found a positive correlation between body fat mass and OA, with body fat mass increasing the risk of OA (IVW result: p=1.65E-33, OR (95% CI)=1.98 (1.77, 2.21). Body fat mass played an important role as a mediator in the causal relationship between depression and OA, with approximately 14% of the risk of OA caused by depression being mediated by body fat mass. Our study offers reliable evidence that depression has a detrimental impact on the risk of OA. Future research can support these associations from improving depressed effect, including social, biological, and behavioral factors, to reduce the risk of chronic diseases such as osteoarthritis. And we identified high-risk variation of alleles which associated with OA and depression can be used to predict disease and provide a basis for clinical intervention and treatment of OA.

Vitamin D Deficiency Increases Mortality Risk in the UK Biobank

Vitamin D Deficiency Increases Mortality Risk in the UK Biobank

Vitamin D Deficiency Increases Mortality Risk in the UK Biobank

Vitamin D Deficiency Increases Mortality Risk in the UK Biobank

Associations of Serum Lipid Traits With Fracture and Osteoporosis: A Prospective Cohort Study From the UK Biobank.

Previous studies reveal inconsistent associations between serum lipid traits and the risks of fractures and osteoporosis in the general population. This prospective cohort study analysed data from 414 302 UK Biobank participants (223 060 women and 191 242 men, aged 37-73 years) with serum lipid measurements: apolipoprotein A (Apo A), apolipoprotein B (Apo B), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), triglycerides (TG) and lipoprotein A (Lp(a)). Multivariable Cox proportional hazard models with penalized cubic splines were used to explore potential nonlinear associations of each lipid trait with the risks of fractures and osteoporosis. Subgroup analyses by age, sex, BMI categories and pre-existing cardiovascular disease were conducted. Mediation analyses using the g-formula were performed to quantify to which extent bone mineral density (BMD) may mediate the association between serum lipids and fracture risk. Over a median follow-up period of 13.8 years, 25 918 (6.8%) of the 383 530 participants without prior fracture had incident fracture cases, and 7591 (4.1%) of the 184 919 participants with primary care data and without baseline osteoporosis were diagnosed with osteoporosis. TG had nonlinear associations with fractures and osteoporosis, whereas Apo B, TC and LDL-C had linear associations. There were also nonlinear associations of Apo A and HDL-C with fractures. Individuals in the highest quintiles for Apo A (fracture: HR 1.15 [95% CI 1.10, 1.21]; osteoporosis: HR 1.13 [1.02, 1.25]) and HDL-C (fracture: HR 1.27 [1.20, 1.34]; osteoporosis: HR 1.31 [1.18, 1.46]) were associated with higher risks of fractures and osteoporosis. Conversely, those in the highest quintile for Apo B (fracture: HR 0.85 [0.81, 0.89]; osteoporosis: HR 0.86 [0.79, 0.94]), LDL-C (fracture: HR 0.89 [0.85, 0.93]; osteoporosis: HR 0.91 [0.83, 1.00]) and TG (fracture: HR 0.78 [0.74, 0.82]; osteoporosis: HR 0.75 [0.68, 0.82]) were associated with lower risks. The associations of Apo A (ratio of HR [RHR] 1.05 [1.02, 1.09]) and HDL-C (RHR 1.06 [1.03, 1.09]) with fracture risk were more pronounced in men compared to women. Except for TG and Lp(a), the associations between serum lipids and fractures appear to be partially mediated through BMD (mediation proportions: 5.30% to 40.30%), assuming causality. Our study reveals a complex interplay between different lipid markers and skeletal health, potentially partially mediated through BMD. Routine lipid profile assessments, including HDL-C and Apo A among other lipid traits, may be integrated into the strategies for fracture risk stratification.

Causal role of immune cells in muscle atrophy: mendelian randomization study

Immune system and inflammation had a great influence on the progression of muscle atrophy. However, the causal relationship with specific immune cell traits remained uncertain. The aim of this study was to elucidate the genetic influences on these associations, providing insights into the underlying mechanisms of muscle atrophy. A bidirectional two-sample Mendelian randomization (MR) analysis was conducted to investigate the causal relationship between immune cell phenotypes and muscle atrophy. Data on immune cell phenotypes were obtained from a research cohort containing data on 731 immune cell phenotypes and data on muscle atrophy were sourced from a Finnish database. MR analysis was performed using the MR-Egger method, weighted median, inverse variance weighting, heterogeneity testing, sensitivity analysis, and multiplicity analysis, with results subjected to false discovery rate(FDR) correction. Additionally, the UK Biobank cohort was utilized as an external validation. A total of 31 immune phenotypes with causal relationships with muscle atrophy were identified, including various phenotypes of conventional dendritic cells, myeloid cells, T cells/B cells/natural killer cells, regulatory cells, and T cell maturation stages. Among them, 12 immune phenotypes were identified as exhibiting a positive causal relationship with muscle atrophy, while 19 immune phenotypes were demonstrated to have a negative causal association, highlighting the complex interactions between immune cells and muscle health. The results of the reverse MR analysis indicated that a negative correlation between muscle atrophy and CD28 on secreting Treg (OR = 0.9038, 95%CI:0.8308 ~ 0.9832, P = 0.0186). A significant positive correlation was revealed by external datasets between the CD25 on IgD + CD38- immune phenotype and the risk of muscle atrophy, which was consistent with the trend observed in the training group (OR = 1.1041, 95% CI: 1.1005–1.1076, P = 0.0263). No evidence of pleiotropy was observed, and the reliability of these findings was demonstrated by the leave-one-out analysis. The findings highlight significant correlations between certain immune cell features and muscle atrophy, providing potential targets for further investigation of immunological mechanisms and therapeutic interventions for this condition.

Prevalence and implications of cardiomyopathy-associated genetic variants in early-onset atrial fibrillation

Abstract Background Atrial fibrillation (AF) is a common and morbid arrhythmia. Despite a substantial heritable component, genetics does not currently play a role in personalized medicine for AF. Rare genetic variants in genes associated with inherited cardiomyopathies (CMP) are enriched in AF patients. However, prior studies on rare variants in AF mainly relied on hospital-based populations, and knowledge on clinical outcomes remains limited. Purpose We aimed to evaluate the potential diagnostic yield and prognostic implication of pathogenic or likely pathogenic (PLP) genetic variants for CMP in persons with AF, using large population-based cohorts. Methods This study utilized sequencing data and electronic health records (EHR) from the All of Us Researh Program (AllofUs) and UK Biobank (UKB). Using ClinVar, we curated a set of PLP variants in 36 CMP-associated genes and assessed the carrier prevalence overall, in participants with AF, and in participants with early-onset AF (below 65, 55 and 45 years). Next, we used Cox proportional hazards models to evaluate the association between variant carrier status and incident risk of heart failure or CMP, after AF diagnosis. In all time-to-event analyses, individuals with prevalent heart failure or CMP at AF diagnosis were excluded and models were adjusted for age, sex, and principal components of ancestry. Sensitivity analyses were performed requiring a minimum period from AF to HF or CMP, of 1, 3 and 6 months. Results After excluding related individuals, we identified 32,281 (8.19%) and 9,666 (6.06%) AF cases in UKB and AllofUs, respectively. Compared to the overall populations (UKB: 1.15%; AllofUs: 1.36%), PLP variants for CMP were about ~5 times more prevalent in AF with onset before 45 (UKB: 4.98%; AllofUs: 5.26%), ~3 times more prevalent in AF with onset before 55 (UKB: 3.41%; AllofUs: 4.34%) and ~2 times more prevalent in all AF patients (UKB: 2.03%; AllofUs: 2.67%). In turn, AF patients carrying PLP variants had increased risk of developing heart failure or CMP, after AF diagnosis (AllofUs HR 1.81, 1.44-2.29, p&amp;lt;1e10-6, UKB:HR=1.39 1.08-1.78, p&amp;lt;0.01), which remained present when requiring a minimum of 6 months after AF diagnosis (AllofUs:HR 1.70, 1.29-2.25, p&amp;lt;0.001; UKB:HR=1.59, 1.21-2.07, p&amp;lt;0.001). Effect sizes were greater when limiting analysis to AF patients with onset before 65 (cases=1021, HR 1.91, 1.35-2.68, p&amp;lt;0.0001) and before 55 years (cases=384, HR 2.53, 1.55-4.1, p &amp;lt;0.001) in AllofUs. UK Biobank did not reflect this increase (cases&amp;lt;65=513, HR&amp;lt;65=1.27, 0.63-2.39, p&amp;lt;65&amp;gt;0.05). Considering only HF events in AllofUs, the effect remained present (HR 1.70, 1.32-2.19, p&amp;lt;0.0001). Conclusion The yield of CMP-associated PLP variants is substantial among persons with early-onset AF, even when considering a 55 year cutoff. AF patients with PLP variants have an increased risk of future heart failure or CMP. Our results further support the clinical relevance of genetic testing in early-onset AF.Prevalence of CMP variants in AF by age

Improving cardiovascular risk stratification with ECG-predicted risk factors in primary prevention

Abstract Background Cardiovascular risk- and screening scores guide clinical practice. Built upon simple risk factors, they could benefit from information on actual, subclinical cardiac changes. The ECG provides proxies for various cardiac states, which can be detected using machine learning. We systematically assess the added value of ECG-based cardiovascular risk factor estimates for six outcomes in two population-based cohorts: Study of Health in Pomerania (SHIP-Start) and UK Biobank (UKB). Methods We employed convolutional neural networks to analyze ECGs from the Hamburg City Health Study (HCHS) for the assessment of multiple risk factors, including age, weight, diabetes, atrial fibrillation (AF) and heart failure. Cox models were fitted on UKB and SHIP to predict 5 and 15 year time-to-event. Endpoints were death, cardiovascular death, AF, stroke, heart failure and myocardial infarction. Covariates were sex, age, body mass index, diabetes, hypertension and smoking (Cox-RF), ECG-predicted risk factors (Cox-AI), and a combination (Cox-RF/AI). Stratified Monte Carlo cross-validation was used as validation and groups of C-indices were compared using Mann-Whitney U tests. Results The study included 9103 HCHS participants (mean age 62.0±8.4 years, 51% women), 3546 SHIP participants (age 49.7±16.4, 51% women), and 32828 UKB participants (age 64±7.6 years, 52% women). Cox-AI performed comparable or better than Cox-RF for AF (SHIP: 0.88 vs. 0.84, UKB: 0.71 vs. 0.72) and heart failure (SHIP: 0.8 vs. 0.79, UKB: 0.76 vs. 0.75), but was less strong for the other outcomes. The combination of information on cardiovascular risk factors and ECG information (Cox-RF/AI) significantly improved the predictive ability compared to risk factors alone (Cox-RF) for AF (SHIP: 0.89 vs. 0.84, UKB: 0.76 vs. 0.72), cardiovascular death (SHIP: 0.90 vs. 0.89, UKB: 0.75 vs. 0.74) and heart failure (SHIP: 0.81 vs. 0.79, UKB: 0.80 vs. 0.75). Discussion ECG-based risk factors complement risk stratification by incorporating nuanced information on cardiac function and structure, improving risk prediction and thus possibly clinical decision making and patient outcomes.

Rare genetic variants in LDLR, APOB, and PCSK9 are associated with aortic stenosis

Abstract Background Aortic stenosis (AS) is the most common cardiac valvular disease in the developed world. Although AS is associated with substantial morbidity and mortality, current treatment options are limited to invasive procedures and reserved for severe symptomatic disease. Genetic studies have linked common variants near lipid genes to AS and cohort studies have linked diagnoses of familial hypercholesterolemia to AS. In contrast, randomized controlled trials have shown no benefit of lipid-lowering therapy in AS. Purpose We aimed to assess the impact on AS conferred by lifelong decreased or increased LDL levels, by studying rare genetic variation in three important lipid handling genes (LDLR, APOB, PCSK9) from a genome-first perspective. Methods We utilized sequencing data and electronic health records from the UK Biobank (UKB) and the All of Us Research Program (AllofUs). We used LOFTEE to identify carriers of truncating variants in LDLR (LDLRtv), APOB (APOBtv) and PCSK9 (PCSK9tv); we used AlphaMissense to identify predicted-damaging missense variants (LDLR-AM, APOB-AM, PCSK9-AM). Linear regression analyses were used to assess the impact of the variant groupings on LDL levels in the UK Biobank, adjusting for sex, age, age^2, and principal components of ancestry. We then conducted Firth’s logistic regression analyses in UK Biobank and AllofUs modelling AS as outcome and variant status as predictor, adjusting for the same covariates; we used an inverse-variance-weighted meta-analysis to combine results. Results We identified 421,058 unrelated participants (N=5,322 AS cases) in the UK Biobank, and 232,170 unrelated participants (N=1,087 AS cases) in AllofUs. Across both cohorts, we identified 198 LDLRtv carriers, 1095 LDLR-AM carriers, 662 APOBtv carriers, 1422 APOB-AM carriers, 712 PCSK9tv carriers and 794 PCSK9-AM carriers. In the UK Biobank LDL values were substantially elevated in LDLRtv (P=1.9x10-48) and LDLR-AM carriers (P=7.5e-200), while LDL levels were substantially decreased in APOBtv (P&amp;lt;1x10-200), APOB-AM (P=6.3x10-19), PCSK9tv (P=1.2x10-114) and PCSK9-AM carriers (P=9.9x10-73; Figure 1). Across UK Biobank and AllofUs, the risk of being diagnosed with AS was strongly increased in LDLRtv (odds ratio [OR]=9.2 [4.32–19.50], P=8e-9) and LDLR-AM carriers (OR=2.76 [1.75–4.35], P=1e-5; Figure 2). Directionally consistent with these effects, we observed a trend towards decreased AS risk among carriers of either APOBtv or PCSK9tv (OR=0.57, [0.30–1.08], P=0.08) and a significantly reduced risk among carriers of APOB-AM or PCSK9-AM (OR=0.46, [0.26-0.82], P=0.009). Conclusion Rare genetic variants that confer lifelong increased LDL levels are associated with substantially increased risk of AS, and vice versa. Our results support a causal role of LDL in AS pathogenesis. In the context of trials demonstrating no benefit in AS, our findings suggest that lipid-lowering may need to be initiated earlier to affect AS development.LDL levels in UK BiobankAS risk in UK Biobank and AllofUs

Predicting progression from coronary artery disease to heart failure in community cohorts

Abstract Background Approximately 10% of people develop heart failure (HF) within 5 years after an acute coronary event. Once ischaemic damage has occurred, the heart undergoes remodeling, which can lead to subclinical dysfunction and HF. However, remodeling varies considerably between patients and is difficult to predict. Purpose To assess whether adding genetic information to clinical factors improves the accuracy of predicting incident non-fatal/fatal HF within 5 years in people with coronary artery disease (CAD). Methods In the UK Biobank (UKB) [1], we developed a clinical risk model (predictors age, sex, ethnicity, body mass index, social deprivation, smoking, personal medical history, time since most recent CAD admission, lipids, HbA1C, creatinine, medications). Separately, we performed a genome-wide association meta-analysis of 13,360 (6,315 + 7,045) cases and 58,126 (37,245 + 20,881) non-cases from the UKB and deCODE [2] cohorts, relating common genetic variants to non-fatal/fatal HF or cardiomyopathy in people with CAD. For the current analysis of the UKB cohort, we derived a polygenic risk score using summary statistics from deCODE alone. We then compared risk estimates and assessed reclassification of 5-year risk using the clinical risk model with and without polygenic risk score, in UKB. Risk groups were &amp;lt;5%, 5-9.9%, 10-14.9%, ≥15%. Results Among people with CAD in the 5 years prior to assessment in UKB (n=8,880, mean age 61y, 27% women) or deCODE (n=10,113, mean age 66y, 34% women), 377 (4%) and 915 (9%) developed non-fatal/fatal HF over 5 years follow-up, respectively. In the UKB cohort, median 5-year risk with the clinical model was 3.2% (IQR 2-5.4%, C-statistic 0.72, 95% CI 0.70–0.74, Table 1) and plots of predicted versus observed risk showed very good calibration across deciles. The polygenic risk score was predictive of non-fatal/fatal HF in the UKB (hazard ratio 1.03, 95% CI 1.00 to 1.05 per standard deviation) with significance decreasing after adjusting for the prognostic index of the clinical risk model (hazard ratio 1.02, 95% CI 0.99 to 1.05 per standard deviation). Median 5-year risk by the clinical+polygenic model was 3.0% (IQR 1.9-5.0%). Net reclassification with the addition of the polygenic score improved accuracy of risk prediction for 3.6% of non-cases but deteriorated accuracy of risk prediction for 6% of cases. Overall net reclassification was deterioration of -0.025. Conclusion Our data suggest that common genetic variants may be associated with progression from CAD to HF, but that adding genetic data did not improve the accuracy of predicting progression from CAD to HF beyond established clinical predictors in the UKB cohort. Confirmation of these findings in deCODE and other large cohorts is warranted.

Using deep learning to detect atherosclerotic plaques on carotid ultrasound images in the UK Biobank

Abstract Background Atherosclerosis is the main underlying cause of cardiovascular disease (CVD). Existing CVD risk assessment tools do not consider the burden of subclinical atherosclerosis. The presence of carotid plaques on carotid ultrasound is a well-known marker of subclinical atherosclerosis. The accumulation of population-scale data on the presence of atherosclerotic plaques, along with deep phenotyping, can allow not only to address the effectiveness of carotid ultrasound in routine clinical practice, but to shed light on the biology of atherosclerosis development. Purpose To develop an effective deep learning model for plaque detection in carotid ultrasound images in the UK Biobank. Methods We used 680 carotid ultrasound images with manually annotated plaques to train a deep learning model employing the YOLOv8 architecture. Different augmentation techniques were used to increase the generalizability of the model. The developed model was applied to automatically detect plaques in raw ultrasound images from 19,507 UK Biobank participants. Logistic and Cox regression were used to explore the associations of plaque presence and number as predicted by the model with conventional CVD risk factors and the risk of future CVD events over follow-up. To explore the genetic architecture of subclinical atherosclerosis, we conducted a genome-wide association study (GWAS) on plaque presence, followed by meta-analysis with data from the CHARGE Consortium. Results Our plaque detection model achieved high classification metrics of accuracy, sensitivity, and specificity (89.3%, 89.5%, and 89.2%, respectively) and detected atherosclerotic plaques in 44% of UK Biobank participants. As expected, plaques were more common among men than women and their prevalence increased linearly with age. Both plaque presence and number of plaques were correlated with conventional CVD risk factors including diabetes, hypertension, and hyperlipidemia, and showed strong associations with future risk of incident CVD events (Hazard Ratio for plaque presence: 1.48 [95%CI: 1.21-1.82], for 2 plaques or more: 1.65, [95% CI: 1.28-2.13]). Incorporating plaque-derived phenotypes minimally altered the C-index of the time-to-event model. GWAS meta-analysis of carotid plaque presence revealed 5 previously known loci, as well as a significant locus including the LPA gene that had not previously been associated with carotid plaque. Conclusion We have developed and implemented an efficient plaque detection model to data from the UK Biobank, which holds significant promise for studying atherosclerosis at a population-wide scale through integration with multiomics data and electronic health records.

Electrocardiogram-based artificial intelligence to identify prevalent coronary artery disease

Abstract Background Coronary artery disease (CAD) often goes undetected for years before it manifests and results in substantial morbidity and mortality. Some patients have neither typical risk factors nor symptomatic angina despite progressive disease. Purpose Develop a deep-learning model to detect prevalent CAD and identify people at risk for adverse events using electrocardiograms (ECG) in a primary care setting. Methods We developed a convolutional neural network using 12-lead ECG waveforms to discriminate the presence of CAD defined using diagnostic codes ("ECG2CAD"). ECG2CAD was trained on 764,670 ECGs from 137,199 individuals at the Massachusetts General Hospital (MGH). Model performance for discrimination of prevalent CAD was measured using AUROC and AUPRC, and compared against a model comprising age and sex, as well as the Pooled Cohort Equations (PCE), in three test sets independent of model training: MGH, Brigham and Women’s Hospital (BWH) and UK Biobank. ECG2CAD was assessed across subgroups of age, sex and self-reported ethnicity and for incident CAD-related events in the BWH primary care subset. Results ECG2CAD was evaluated in MGH (N=18,706 [N=6,051 cases], age 57±16 years), BWH (N=88,270 [N=27,898 cases], age 57±16 years), and UK Biobank (N=42,147 [N=1,509 cases], age 65±8 years). ECG2CAD consistently discriminated prevalent CAD (MGH: AUROC 0.782, AUPRC 0.639; BWH: AUROC 0.747, AUPRC 0.588; UK Biobank: AUROC 0.760, AUPRC 0.155) and incrementally improved upon both a model based on age and sex (p&amp;lt;0.01) and the PCE (p&amp;lt;0.01) in MGH and BWH. In the BWH primary care subset (N=51,808), model performance was consistent in subgroups defined by age, sex and self-reported ethnicity/race. The incorporation of ECG2CAD in addition to age and sex consistently improved discrimination (AUROC delta of 0.011-0.086) in all subgroups. In the BWH primary care subset with over 10 years of follow-up, we observed that either the presence of ECG2CAD-predicted CAD (HR 2.18, 95% CI 2.00-2.38, P&amp;lt;0.001) and ICD-based known CAD (HR 2.13, 95% CI 1.86-2.45, P&amp;lt;0.001) were associated with increased risk for incident myocardial infarction. The presence of both definitions was associated with an almost five-fold increase in risk (HR 4.97, 95% CI 4.51-5.48, P&amp;lt;0.001). The risk of incident adverse events was highest among people in the top quintile of ECG-based predicted risk (high risk) compared with people from quintile 4-2 (intermediate risk) or the lowest quintile (low risk), even when excluding those with known prevalent CAD at baseline. Conclusions Artificial intelligence-enabled analysis of the 12-lead ECG may facilitate efficient identification of individuals with possible undiagnosed CAD and inform downstream testing and preventive measures.

Association between GDF-15 levels and cardiac morphology and function in cardiac magnetic resonance imaging: insights from UK Biobank data analysis

Abstract Background/Introduction Growth Differentiation Factor-15 (GDF-15) has emerged as a biomarker associated with cardiovascular disease and adverse cardiac outcomes. Cardiac morphology and function, assessed through cardiac magnetic resonance imaging (CMR), are critical indicators of cardiovascular pathology. Understanding the relationship between GDF-15 levels and LV characteristics could provide valuable insights into cardiac pathophysiology and disease progression. Purpose This study aims to investigate the association between GDF-15 levels and cardiac morphology and function, utilizing data from the UK Biobank and CMR. By analyzing various CMR-derived parameters, including LV strain, myocardial mass, LV mass-to-volume ratio, ventricular and atrial volumes, we seek to elucidate the impact of GDF-15 on cardiac structure and function. Methods Data from the UK Biobank, including 2,264 participants, were analyzed. The median age of the analyzed participants was 55 years, with a median BMI of 26.0. Among the participants, 53% were females and 47% were males. CMR-derived variables related to cardiac morphology and function, along with GDF-15 levels, were examined. Multiple linear regression models were constructed to assess the association between GDF-15 levels and cardiac parameters, adjusting for age, sex, and other relevant covariates. Confidence intervals (CI) and p-values were calculated to determine the significance of observed associations. Results Analysis of the UK Biobank data revealed significant associations between GDF-15 levels and specific cardiac parameters (table1). Notably, GDF-15 levels were negatively associated with LV myocardial mass (LVM)(coefficient = -1.146, p = 0.0071), LV end-diastolic volume LVEDV)(coefficient = -3.321, p &amp;lt; 0.0001), LV end-systolic volume (coefficient = -1.876, p &amp;lt; 0.0001), as well as right atrial maximum volume (coefficient = -3.025, p &amp;lt; 0.0001) and minimum volume (coefficient = -1.858, p = 0.0003). In addition, LVM/LVEDV was positively associated with GDF-15 level (p&amp;lt;0.03). However, no significant associations were observed between GDF-15 levels and LV strain parameters, ejection fraction, or left atrial volumes. Conclusion This study provides evidence of a significant association between GDF-15 levels and selected parameters of cardiac morphology and function, as assessed by CMR in the UK Biobank population. These findings suggest a potential role of GDF-15 in modulating specific aspects of cardiac structure, function and left ventricle concentric remodeling warranting further investigation.GDF 15