UK Biobank Research Articles

Body mass index and all-cause mortality in HUNT and UK biobank studies: revised non-linear Mendelian randomisation analyses

ObjectivesTo estimate the shape of the causal relationship between body mass index (BMI) and mortality risk in a Mendelian randomisation framework.DesignMendelian randomisation analyses of two prospective population-based cohorts.SettingIndividuals of European...

Proteomic aging clock (PAC) predicts age-related outcomes in middle-aged and older adults.

Beyond mere prognostication, optimal biomarkers of aging provide insights into qualitative and quantitative features of biological aging and might, therefore, offer useful information for the testing and, ultimately, clinical use of gerotherapeutics. We aimed to develop a proteomic aging clock (PAC) for all-cause mortality risk as a proxy of biological age. Data were from the UK Biobank Pharma Proteomics Project, including 53,021 participants aged between 39 and 70 years and 2923 plasma proteins assessed using the Olink Explore 3072 assay®. 10.9% of the participants died during a mean follow-up of 13.3 years, with the mean age at death of 70.1 years. The Spearman correlation between PAC proteomic age and chronological age was 0.77. PAC showed robust age-adjusted associations and predictions for all-cause mortality and the onset of various diseases in general and disease-free participants. The proteins associated with PAC proteomic age deviation were enriched in several processes related to the hallmarks of biological aging. Our results expand previous findings by showing that biological age acceleration, based on PAC, strongly predicts all-cause mortality and several incident disease outcomes. Particularly, it facilitates the evaluation of risk for multiple conditions in a disease-free population, thereby, contributing to the prevention of initial diseases, which vary among individuals and may subsequently lead to additional comorbidities.

The landscape of rare genetic variation associated with inflammatory bowel disease and Parkinson’s disease comorbidity

BackgroundInflammatory bowel disease (IBD) and Parkinson’s disease (PD) are chronic disorders that have been suggested to share common pathophysiological processes. LRRK2 has been implicated as playing a role in both diseases. Exploring the genetic basis of the IBD-PD comorbidity through studying high-impact rare genetic variants can facilitate the identification of the novel shared genetic factors underlying this comorbidity.MethodsWe analyzed whole exomes from the BioMe BioBank and UK Biobank, and whole genomes from a cohort of 67 European patients diagnosed with both IBD and PD to examine the effects of LRRK2 missense variants on IBD, PD and their co-occurrence (IBD-PD). We performed optimized sequence kernel association test (SKAT-O) and network-based heterogeneity clustering (NHC) analyses using high-impact rare variants in the IBD-PD cohort to identify novel candidate genes, which we further prioritized by biological relatedness approaches. We conducted phenome-wide association studies (PheWAS) employing BioMe BioBank and UK Biobank whole exomes to estimate the genetic relevance of the 14 prioritized genes to IBD-PD.ResultsThe analysis of LRRK2 missense variants revealed significant associations of the G2019S and N2081D variants with IBD-PD in addition to several other variants as potential contributors to increased or decreased IBD-PD risk. SKAT-O identified two significant genes, LRRK2 and IL10RA, and NHC identified 6 significant gene clusters that are biologically relevant to IBD-PD. We observed prominent overlaps between the enriched pathways in the known IBD, PD, and candidate IBD-PD gene sets. Additionally, we detected significantly enriched pathways unique to the IBD-PD, including MAPK signaling, LPS/IL-1 mediated inhibition of RXR function, and NAD signaling. Fourteen final candidate IBD-PD genes were prioritized by biological relatedness methods. The biological importance scores estimated by protein–protein interaction networks and pathway and ontology enrichment analyses indicated the involvement of genes related to immunity, inflammation, and autophagy in IBD-PD. Additionally, PheWAS provided support for the associations of candidate genes with IBD and PD.ConclusionsOur study confirms and uncovers new LRRK2 associations in IBD-PD. The identification of novel inflammation and autophagy-related genes supports and expands previous findings related to IBD-PD pathogenesis, and underscores the significance of therapeutic interventions for reducing systemic inflammation.

Skeletal muscle hypertrophy rewires glucose metabolism: An experimental investigation and systematic review.

Proliferating cancer cells shift their metabolism towards glycolysis, even in the presence of oxygen, to especially generate glycolytic intermediates as substrates for anabolic reactions. We hypothesize that a similar metabolic remodelling occurs during skeletal muscle hypertrophy. We used mass spectrometry in hypertrophying C2C12 myotubes in vitro and plantaris mouse muscle in vivo and assessed metabolomic changes and the incorporation of the [U-13C6]glucose tracer. We performed enzyme inhibition of the key serine synthesis pathway enzyme phosphoglycerate dehydrogenase (Phgdh) for further mechanistic analysis and conducted a systematic review to align any changes in metabolomics during muscle growth with published findings. Finally, the UK Biobank was used to link the findings to population level. The metabolomics analysis in myotubes revealed insulin-like growth factor-1 (IGF-1)-induced altered metabolite concentrations in anabolic pathways such as pentose phosphate (ribose-5-phosphate/ribulose-5-phosphate: +40%; P=0.01) and serine synthesis pathway (serine: -36.8%; P=0.009). Like the hypertrophy stimulation with IGF-1 in myotubes in vitro, the concentration of the dipeptide l-carnosine was decreased by 26.6% (P=0.001) during skeletal muscle growth in vivo. However, phosphorylated sugar (glucose-6-phosphate, fructose-6-phosphate or glucose-1-phosphate) decreased by 32.2% (P=0.004) in the overloaded muscle in vivo while increasing in the IGF-1-stimulated myotubes in vitro. The systematic review revealed that 10 metabolites linked to muscle hypertrophy were directly associated with glycolysis and its interconnected anabolic pathways. We demonstrated that labelled carbon from [U-13C6]glucose is increasingly incorporated by ~13% (P=0.001) into the non-essential amino acids in hypertrophying myotubes, which is accompanied by an increased depletion of media serine (P=0.006). The inhibition of Phgdh suppressed muscle protein synthesis in growing myotubes by 58.1% (P<0.001), highlighting the importance of the serine synthesis pathway for maintaining muscle size. Utilizing data from the UK Biobank (n=450243), we then discerned genetic variations linked to the serine synthesis pathway (PHGDH and PSPH) and to its downstream enzyme (SHMT1), revealing their association with appendicular lean mass in humans (P<5.0e-8). Understanding the mechanisms that regulate skeletal muscle mass will help in developing effective treatments for muscle weakness. Our results provide evidence for the metabolic rewiring of glycolytic intermediates into anabolic pathways during muscle growth, such as in serine synthesis.

Analysis of somatic mutations in whole blood from 200,618 individuals identifies pervasive positive selection and novel drivers of clonal hematopoiesis.

Human aging is marked by the emergence of a tapestry of clonal expansions in dividing tissues, particularly evident in blood as clonal hematopoiesis (CH). CH, linked to cancer risk and aging-related phenotypes, often stems from somatic mutations in a set of established genes. However, the majority of clones lack known drivers. Here we infer gene-level positive selection in whole blood exomes from 200,618 individuals in UK Biobank. We identify 17 additional genes, ZBTB33, ZNF318, ZNF234, SPRED2, SH2B3, SRCAP, SIK3, SRSF1, CHEK2, CCDC115, CCL22, BAX, YLPM1, MYD88, MTA2, MAGEC3 and IGLL5, under positive selection at a population level, and validate this selection pattern in 10,837 whole genomes from single-cell-derived hematopoietic colonies. Clones with mutations in these genes grow in frequency and size with age, comparable to classical CH drivers. They correlate with heightened risk of infection, death and hematological malignancy, highlighting the significance of these additional genes in the aging process.

Association between household size and risk of incident dementia in the UK Biobank study

Currently, the relationship between household size and incident dementia, along with the underlying neurobiological mechanisms, remains unclear. This prospective cohort study was based on UK Biobank participants aged ≥ 50 years without a history of dementia. The linear and non-linear longitudinal association was assessed using Cox proportional hazards regression and restricted cubic spline models. Additionally, the potential mechanisms driven by brain structures were investigated by linear regression models. We included 275,629 participants (mean age at baseline 60.45 years [SD 5.39]). Over a mean follow-up of 9.5 years, 6031 individuals developed all-cause dementia. Multivariable analyses revealed that smaller household size was associated with an increased risk of all-cause dementia (HR, 1.06; 95% CI 1.02–1.09), vascular dementia (HR, 1.08; 95% CI 1.01–1.15), and non-Alzheimer’s disease non-vascular dementia (HR, 1.09; 95% CI 1.03–1.14). No significant association was observed for Alzheimer’s disease. Restricted cubic splines demonstrated a reversed J-shaped relationship between household size and all-cause and cause-specific dementia. Additionally, substantial associations existed between household size and brain structures. Our findings suggest that small household size is a risk factor for dementia. Additionally, brain structural differences related to household size support these associations. Household size may thus be a potential modifiable risk factor for dementia.

The association between being breastfed in infancy and risks of cancer in adulthood—a UK Biobank study

BackgroundBeing breastfed has established benefits for infant health, but its long-term effects on adult diseases, including cancer, remain underexplored. We examined associations between being breastfed in infancy and the risks of common cancers.MethodsData from 339,115 participants (191,117 women) enrolled in the UK Biobank with self-reported breastfeeding data were linked to national cancer registries. Cox models estimated sex-specific hazard ratios (HR) and 95% confidence intervals (CI) for the associations between being breastfed (ever/never) and risks of overall cancer as well as common cancer sites.ResultsIn total, 34,392 incident cancers (17,895 in women) were registered. The HR of overall cancer associated with being breastfed was 1.05 (95% CI 1.01–1.09) in women and 1.00 (95% CI 0.96–1-04) in men. In site-specific analysis being breastfed was associated with an increased risk of breast and ovarian cancer in women and a reduced risk of oesophageal cancer in men.DiscussionWe found that having been breastfed was associated with a marginally increased risk of adult cancer in women, but we found no evidence of an association in men. These findings should be viewed within the study limitations, and do not outweigh the many benefits that breastfeeding provides.

To weight or not to weight? The effect of selection bias in 3 large electronic health record-linked biobanks and recommendations for practice.

To develop recommendations regarding the use of weights to reduce selection bias for commonly performed analyses using electronic health record (EHR)-linked biobank data. We mapped diagnosis (ICD code) data to standardized phecodes from 3 EHR-linked biobanks with varying recruitment strategies: All of Us (AOU; n = 244071), Michigan Genomics Initiative (MGI; n = 81243), and UK Biobank (UKB; n = 401167). Using 2019 National Health Interview Survey data, we constructed selection weights for AOU and MGI to represent the US adult population more. We used weights previously developed for UKB to represent the UKB-eligible population. We conducted 4 common analyses comparing unweighted and weighted results. For AOU and MGI, estimated phecode prevalences decreased after weighting (weighted-unweighted median phecode prevalence ratio [MPR]: 0.82 and 0.61), while UKB estimates increased (MPR: 1.06). Weighting minimally impacted latent phenome dimensionality estimation. Comparing weighted versus unweighted phenome-wide association study for colorectal cancer, the strongest associations remained unaltered, with considerable overlap in significant hits. Weighting affected the estimated log-odds ratio for sex and colorectal cancer to align more closely with national registry-based estimates. Weighting had a limited impact on dimensionality estimation and large-scale hypothesis testing but impacted prevalence and association estimation. When interested in estimating effect size, specific signals from untargeted association analyses should be followed up by weighted analysis. EHR-linked biobanks should report recruitment and selection mechanisms and provide selection weights with defined target populations. Researchers should consider their intended estimands, specify source and target populations, and weight EHR-linked biobank analyses accordingly.

Peripheral GFAP and NfL as early biomarkers for dementia: longitudinal insights from the UK Biobank

BackgroundPeripheral glial fibrillary acidic protein (GFAP) and neurofilament light chain (NfL) are sensitive markers of neuroinflammation and neuronal damage. Previous studies with highly selected participants have shown that peripheral GFAP and NfL levels are elevated in the pre-clinical phase of Alzheimer’s disease (AD) and dementia. However, the predictive value of GFAP and NfL for dementia requires more evidence from population-based cohorts.MethodsThis was a prospective cohort study to evaluate UK Biobank participants enrolled from 2006 to 2010 using plasma GFAP and NfL measurements measured by Olink Target Platform and prospectively followed up for dementia diagnosis. Primary outcome was the risk of clinical diagnosed dementia. Secondary outcomes were cognition. Linear regression was used to assess the associations between peripheral GFAP and NfL with cognition. Cox proportional hazard models with cross-validations were used to estimate associations between elevated GFAP and NfL with risk of dementia. All models were adjusted for covariates.ResultsA subsample of 48,542 participants in the UK Biobank with peripheral GFAP and NfL measurements were evaluated. With an average follow-up of 13.18 ± 2.42 years, 1312 new all-cause dementia cases were identified. Peripheral GFAP and NfL increased up to 15 years before dementia diagnosis was made. After strictly adjusting for confounders, increment in NfL was found to be associated with decreased numeric memory and prolonged reaction time. A greater annualized rate of change in GFAP was significantly associated with faster global cognitive decline. Elevation of GFAP (hazard ratio (HR) ranges from 2.25 to 3.15) and NfL (HR ranges from 1.98 to 4.23) increased the risk for several types of dementia. GFAP and NfL significantly improved the predictive values for dementia using previous models (area under the curve (AUC) ranges from 0.80 to 0.89, C-index ranges from 0.86 to 0.91). The AD genetic risk score and number of APOE*E4 alleles strongly correlated with GFAP and NfL levels.ConclusionsThese results suggest that peripheral GFAP and NfL are potential biomarkers for the early diagnosis of dementia. In addition, anti-inflammatory therapies in the initial stages of dementia may have potential benefits.

Association of sarcopenia with the long-term risk of atrial fibrillation: A prospective cohort study.

The relationship between sarcopenia and the long-term risk of atrial fibrillation (AF) remains unclear. This study recruited a large prospective Caucasian cohort from the UK Biobank. Participants were assessed at baseline with handgrip strength and muscle mass and were categorized into groups of non-sarcopenia, probable sarcopenia, and confirmed sarcopenia. Kaplan-Meier method and Cox proportional hazards model were used to explore the association between sarcopenia and the incidence of AF. The genetic predisposition of AF was assessed by polygenic risk score. Sensitivity analyses were performed to validate the results. A total of 384,433 participants with a median age of 58 years and 54.3% women were enrolled in this study. There were 24,007 cases of new-onset AF over a median follow-up of 12.56 years. The groups of non-sarcopenia, probable sarcopenia, and confirmed sarcopenia accounted for 22,290 (6.1%), 1665 (9.2%), and 52 (11.9%) cases, respectively. Compared with the non-sarcopenia group, participants with probable sarcopenia or confirmed sarcopenia had an 8% (95% CI, 1.03-1.14) or 61% (95% CI, 1.23-2.12) higher risk of AF incidence. The findings remained robust in multiple sensitivity analyses, such as subgroup analysis and further adjustment of genetic predisposition. Notably, the association between sarcopenia and a high AF risk was more pronounced in younger participants, women, and those with valvular heart disease. In conclusion, sarcopenia was associated with a high long-term risk of AF in Caucasians, supporting sarcopenia as a new independent risk factor of AF.

ComorbidPGS: an R package assessing shared predisposition between Phenotypes using Polygenic Scores.

Introduction Polygenic Score (PGS) is a valuable method for assessing the estimated genetic liability to a given outcome or genetic variability contributing to a quantitative trait. While PRSs are widely used for complex traits, their application in uncovering shared genetic predisposition between phenotypes, i.e. when genetic variants influence more than one phenotype, remains limited. Methods We developed an R package, comorbidPGS, which facilitates a systematic evaluation of shared genetic effects among (cor)related phenotypes using PGSs. The comorbidPGS package takes as input a set of Single Nucleotide Polymorphisms (SNPs) along with their established effects on the original phenotype (Po), referred to as Po-PGS. It generates a comprehensive summary of effect(s) of Po-PGS on target phenotype(s) (Pt) with customisable graphical features. Results We applied comorbidPGS to investigate the shared genetic predisposition between phenotypes defining elevated blood pressure (Systolic Blood Pressure, SBP; Diastolic Blood Pressure, DBP; Pulse Pressure, PP) and several cancers (Breast Cancer, BrC; Pancreatic Cancer, PanC; Kidney Cancer, KidC; Prostate Cancer, PrC; Colorectal Cancer, CrC) using the European ancestry UK Biobank individuals and GWAS meta-analyses summary statistics from independent set of European ancestry individuals. We report a significant association between elevated DBP and the genetic risk of PrC (β (SE)=0.066 (0.017), P-value=9.64×10^(-5)), as well as between CrC PGS and both, lower SBP (β (SE)=-0.10 [0.029], P-value=3.83×10^(-4))) and lower DBP (β (SE)=-0.055 [0.017], P-value=1.05×10^(-3)). Our analysis highlights two nominally significant relationships for individuals with genetic predisposition to elevated SBP leading to higher risk of KidC (OR [95%CI]=1.04 [1.0039-1.087], P-value=2.82×10^(-2)) and PrC (OR [95%CI]=1.02 [1.003-1.041], P-value=2.22×10^(-2)). Conclusion Using comorbidPGS, we underscore mechanistic relationships between blood pressure regulation and susceptibility to three comorbid malignancies. This package offers valuable means to evaluate shared genetic susceptibility between (cor)related phenotypes through polygenic scores.

Modifiable lifestyle factors and lifetime risk of atrial fibrillation: longitudinal data from the Korea NHIS-HealS and UK Biobank cohorts

BackgroundThe reason for higher incidence of atrial fibrillation (AF) in Europe compared with East Asia is unclear. We aimed to investigate the association between modifiable lifestyle factors and lifetime risk of AF in Europe and East Asia, along with race/ethnic similarities and disparities.Methods1:1 propensity score matched pairs of 242,763 East Asians and 242,763 White Europeans without AF were analyzed. Modifiable lifestyle factors considered were blood pressure, body mass index, cigarette smoking, diabetes, alcohol consumption, and physical activity, categorized as non-adverse or adverse levels. Lifetime risk of AF was estimated from the index age of 45 years to the attained age of 85 years, accounting for the competing risk of death.ResultsThe overall lifetime risk of AF was higher in White Europeans than East Asians (20.9% vs 15.4%, p < 0.001). The lifetime risk of AF was similar between the two races in individuals with non-adverse lifestyle factor profiles (13.4% vs 12.9%, p = 0.575), whereas it was higher in White Europeans with adverse lifestyle factor profiles (22.1% vs 15.8%, p < 0.001). The difference in the lifetime risk of AF between the two races increased as the burden of adverse lifestyle factors worsened (1 adverse lifestyle factor; 4.3% to ≥ 3 adverse lifestyle factors; 11.2%). Compared with East Asians, the relative risk of AF in White Europeans was 23% and 62% higher for one (hazard ratio [HR] 1.23, 95% confidence interval [CI] 1.16–1.29) and ≥ 3 adverse lifestyle factors (HR 1.62, 95% CI 1.51–1.75), respectively.ConclusionsThe overall higher lifetime risk of AF in White Europeans compared with East Asians might be attributable to adverse lifestyle factors. Adherence to healthy lifestyle factors was associated with the lifetime risk of AF of about 1 in 8 regardless of race/ethnicity.

Serum urate levels and neurodegenerative outcomes: a prospective cohort study and mendelian randomization analysis of the UK Biobank

BackgroundPrevious studies on the associations between serum urate levels and neurodegenerative outcomes have yielded inconclusive results, and the causality remains unclear. This study aimed to investigate whether urate levels are associated with the risks of Alzheimer’s disease and related dementias (ADRD), Parkinson’s disease (PD), and neurodegenerative deaths.MethodsThis prospective study included 382,182 participants (45.7% men) from the UK Biobank cohort. Cox proportional hazards models were used to assess the associations between urate levels and risk of neurodegenerative outcomes. In the Mendelian randomization (MR) analysis, urate-related single-nucleotide polymorphisms were identified through a genome-wide association study. Both linear and non-linear MR approaches were utilized to investigate the potential causal associations.ResultsDuring a median follow-up period of 12 years, we documented 5,400 ADRD cases, 2,553 PD cases, and 1,531 neurodegenerative deaths. Observational data revealed that a higher urate level was associated with a decreased risk of ADRD (hazard ratio [HR]: 0.93, 95% confidence interval [CI]: 0.90, 0.96), PD (HR: 0.87, 95% CI: 0.82, 0.91), and neurodegenerative death (HR: 0.88, 95% CI: 0.83, 0.94). Negative linear associations between urate levels and neurodegenerative events were observed (all P-values for overall < 0.001 and all P-values for non-linearity > 0.05). However, MR analyses yielded no evidence of either linear or non-linear associations between genetically predicted urate levels and the risk of the aforementioned neurodegenerative events.ConclusionAlthough the prospective cohort study demonstrated that elevated urate levels were associated with a reduced risk of neurodegenerative outcomes, MR analyses found no evidence of causality.

Transcript and protein signatures derived from shared molecular interactions across cancers are associated with mortality

BackgroundCharacterization of shared cancer mechanisms have been proposed to improve therapy strategies and prognosis. Here, we aimed to identify shared cell–cell interactions (CCIs) within the tumor microenvironment across multiple solid cancers and assess their association with cancer mortality.MethodsCCIs of each cancer were identified by NicheNet analysis of single-cell RNA sequencing data from breast, colon, liver, lung, and ovarian cancers. These CCIs were used to construct a shared multi-cellular tumor model (shared-MCTM) representing common CCIs across cancers. A gene signature was identified from the shared-MCTM and tested on the mRNA and protein level in two large independent cohorts: The Cancer Genome Atlas (TCGA, 9185 tumor samples and 727 controls across 22 cancers) and UK biobank (UKBB, 10,384 cancer patients and 5063 controls with proteomics data across 17 cancers). Cox proportional hazards models were used to evaluate the association of the signature with 10-year all-cause mortality, including sex-specific analysis.ResultsA shared-MCTM was derived from five individual cancers. A shared gene signature was extracted from this shared-MCTM and the most prominent regulatory cell type, matrix cancer-associated fibroblast (mCAF). The signature exhibited significant expression changes in multiple cancers compared to controls at both mRNA and protein levels in two independent cohorts. Importantly, it was significantly associated with mortality in cancer patients in both cohorts. The highest hazard ratios were observed for brain cancer in TCGA (HR [95%CI] = 6.90[4.64–10.25]) and ovarian cancer in UKBB (5.53[2.08–8.80]). Sex-specific analysis revealed distinct risks, with a higher mortality risk associated with the protein signature score in males (2.41[1.97–2.96]) compared to females (1.84[1.44–2.37]).ConclusionWe identified a gene signature from a comprehensive shared-MCTM representing common CCIs across different cancers and revealed the regulatory role of mCAF in the tumor microenvironment. The pathogenic relevance of the gene signature was supported by differential expression and association with mortality on both mRNA and protein levels in two independent cohorts.

Phenome-wide identification of therapeutic genetic targets, leveraging knowledge graphs, graph neural networks, and UK Biobank data.

The ongoing expansion of human genomic datasets propels therapeutic target identification; however, extracting gene-disease associations from gene annotations remains challenging. Here, we introduce Mantis-ML 2.0, a framework integrating AstraZeneca's Biological Insights Knowledge Graph and numerous tabular datasets, to assess gene-disease probabilities throughout the phenome. We use graph neural networks, capturing the graph's holistic structure, and train them on hundreds of balanced datasets via a robust semi-supervised learning framework to provide gene-disease probabilities across the human exome. Mantis-ML 2.0 incorporates natural language processing to automate disease-relevant feature selection for thousands of diseases. The enhanced models demonstrate a 6.9% average classification power boost, achieving a median receiver operating characteristic (ROC) area under curve (AUC) score of 0.90 across 5220 diseases from Human Phenotype Ontology, OpenTargets, and Genomics England. Notably, Mantis-ML 2.0 prioritizes associations from an independent UK Biobank phenome-wide association study (PheWAS), providing a stronger form of triaging and mitigating against underpowered PheWAS associations. Results are exposed through an interactive web resource.

Tea Consumption, Milk or Sweeteners Addition, Genetic Variation in Caffeine Metabolism, and Incident Venous Thromboembolism.

The association between tea consumption and venous thromboembolism (VTE) remains unknown. We aimed to evaluate the association between tea consumption with different additives (milk and/or sweeteners) and incident VTE, and the modifying effects of genetic variation in caffeine metabolism on the association. A total of 190,189 participants with complete dietary information and free of VTE at baseline in the UK Biobank were included. The primary outcome was incident VTE, including incident deep vein thrombosis and pulmonary embolism. During a median follow-up of 12.1 years, 4,485 (2.4%) participants developed incident VTE. Compared with non-tea drinkers, tea drinkers who added neither milk nor sweeteners (hazard ratio [HR]: 0.85; 95% confidence interval [95% CI]: 0.76-0.94), only milk (HR: 0.86; 95% CI: 0.80-0.93), and both milk and sweeteners to their tea (HR: 0.90; 95% CI: 0.81-0.99) had a lower risk of VTE, while those who added only sweeteners to their tea did not (HR: 0.94; 95% CI: 0.75-1.17). Moreover, there was an L-shaped relationship between tea consumption and incident VTE among tea drinkers who added neither milk nor sweeteners, only milk, and both milk and sweeteners to their tea, respectively. However, a nonsignificant association was found among tea drinkers who added only sweeteners to their tea. Genetic variation in caffeine metabolism did not significantly modify the association (p-interaction = 0.659). Drinking unsweetened tea, with or without added milk, was associated with a lower risk of VTE. However, there was no significant association between drinking tea with sweeteners and incident VTE.

IGWAS: Image-based genome-wide association of self-supervised deep phenotyping of retina fundus images.

Existing imaging genetics studies have been mostly limited in scope by using imaging-derived phenotypes defined by human experts. Here, leveraging new breakthroughs in self-supervised deep representation learning, we propose a new approach, image-based genome-wide association study (iGWAS), for identifying genetic factors associated with phenotypes discovered from medical images using contrastive learning. Using retinal fundus photos, our model extracts a 128-dimensional vector representing features of the retina as phenotypes. After training the model on 40,000 images from the EyePACS dataset, we generated phenotypes from 130,329 images of 65,629 British White participants in the UK Biobank. We conducted GWAS on these phenotypes and identified 14 loci with genome-wide significance (p<5×10-8 and intersection of hits from left and right eyes). We also did GWAS on the retina color, the average color of the center region of the retinal fundus photos. The GWAS of retina colors identified 34 loci, 7 are overlapping with GWAS of raw image phenotype. Our results establish the feasibility of this new framework of genomic study based on self-supervised phenotyping of medical images.

Associations between dietary habits and bipolar disorder: a diet-wide mendelian randomization study

BackgroundDiet/nutrition is critically important in the pathogenesis, progression, and treatment outcomes of various mental disorders. Current research predominantly focuses on the role of diet in the development and treatment of depression, with less attention given to the relationship between diet and Bipolar Disorder (BD).MethodWe employed Mendelian Randomization (MR) to investigate the relationship between 28 dietary habits and BD. An analysis was conducted using publicly available genome-wide association study data from the UK Biobank dataset. Various dietary habits were analyzed as exposures with BD as the outcome, mainly using the Inverse Variance Weighted (IVW) method.ResultsIntake of non-oily fish and sponge pudding both have a positive association with BD. Oily fish, dried fruit, apples, salt, and cooked vegetables intake also appeared potentially risky for BD, although the possibility of false positives cannot be ruled out. Sensitivity analysis further confirmed the robustness of these findings.ConclusionOur research provides evidence of a relationship between various dietary habits and BD. It underscores the need for careful dietary management and balance to reduce the risk of BD, suggesting caution with dietary preferences for fish and sponge pudding. Furthermore, more detailed studies are needed to further understand the potential impacts of high-sugar and high-protein diets on BD development.

A novel method for multiple phenotype association studies based on genotype and phenotype network.

Joint analysis of multiple correlated phenotypes for genome-wide association studies (GWAS) can identify and interpret pleiotropic loci which are essential to understand pleiotropy in diseases and complex traits. Meanwhile, constructing a network based on associations between phenotypes and genotypes provides a new insight to analyze multiple phenotypes, which can explore whether phenotypes and genotypes might be related to each other at a higher level of cellular and organismal organization. In this paper, we first develop a bipartite signed network by linking phenotypes and genotypes into a Genotype and Phenotype Network (GPN). The GPN can be constructed by a mixture of quantitative and qualitative phenotypes and is applicable to binary phenotypes with extremely unbalanced case-control ratios in large-scale biobank datasets. We then apply a powerful community detection method to partition phenotypes into disjoint network modules based on GPN. Finally, we jointly test the association between multiple phenotypes in a network module and a single nucleotide polymorphism (SNP). Simulations and analyses of 72 complex traits in the UK Biobank show that multiple phenotype association tests based on network modules detected by GPN are much more powerful than those without considering network modules. The newly proposed GPN provides a new insight to investigate the genetic architecture among different types of phenotypes. Multiple phenotypes association studies based on GPN are improved by incorporating the genetic information into the phenotype clustering. Notably, it might broaden the understanding of genetic architecture that exists between diagnoses, genes, and pleiotropy.

Evaluation of Deep Learning Estimation of Whole Heart Anatomy from Automated Cardiovascular Magnetic Resonance Short- and Long-Axis Analyses in UK Biobank.

Standard methods of heart chamber volume estimation in cardiovascular magnetic resonance (CMR) typically utilize simple geometric formulae based on a limited number of slices. We aimed to evaluate whether an automated deep learning neural network prediction of 3D anatomy of all four chambers would show stronger associations with cardiovascular risk factors and disease than standard volume estimation methods in the UK Biobank. A deep learning network was adapted to predict 3D segmentations of left and right ventricles (LV, RV) and atria (LA, RA) at ∼1mm isotropic resolution from CMR short and long axis 2D segmentations obtained from a fully automated machine learning pipeline in 4723 individuals with cardiovascular disease (CVD) and 5733 without in the UK Biobank. Relationships between volumes at end-diastole (ED) and end-systole (ES) and risk/disease factors were quantified using univariate, multivariate and logistic regression analyses. Strength of association between deep learning volumes and standard volumes was compared using the area under the receiving operator characteristic curve (AUC). Univariate and multivariate associations between deep learning volumes and most risk and disease factors were stronger than for standard volumes (higher R2 and more significant P values), particularly for sex, age, and body mass index. AUC for all logistic regressions were higher for deep learning volumes than standard volumes (p<0.001 for all four chambers at ED and ES). Neural network reconstructions of whole heart volumes had significantly stronger associations with cardiovascular disease and risk factors than standard volume estimation methods in an automatic processing pipeline.