The UK Biobank resource with deep phenotyping and genomic data

Newly identified single-nucleotide polymorphism associated with the transition from nonalcoholic fatty liver disease to liver fibrosis: results from a nested case-control study in the UK biobank

BACKGROUND

Genetic factors may have a significant influence on the likelihood of liver fibrosis in individuals with nonalcoholic fatty liver disease (NAFLD). The present study was conducted to explore how single-nucleotide polymorphism (SNP) impacts the development of fibrosis in those suffering from NAFLD.

MATERIALS AND METHODS

Utilizing the UK Biobank dataset, we conducted a nested case-control analysis among NAFLD participants, defining the case group as those with liver fibrosis and cirrhosis during follow-up. For our in vitro investigations, we employed the LX-2 human hepatic stellate cell line. Our procedures included cultivating these cells, employing SAMM50-rs2073080 plasmid techniques to enhance the expression of recently discovered SNPs, and conducting biochemical assays. To quantify gene expression, we used real-time PCR with fluorescence detection.

RESULTS

The study analyzed data from 5467 participants (1094 cases and 4373 controls). Genome-wide association analysis identified nine significant loci, including the novel rs2073080 variant, strongly associated with NAFLD-associated hepatic fibrosis. In vitro TGF-β modeling revealed significant upregulation of α-SMA and COL1A1, confirming model effectiveness. Oxidative stress markers like elevated malondialdehyde (MDA) and reduced catalase (CAT) and superoxide dismutase (SOD) levels indicated liver damage in the TGF-β group. SAMM50-rs2073080 was upregulated in the NAFLD-associated fibrosis model. In vitro experiments on LX-2 cells showed that SAMM50-rs2073080 overexpression led to increased fibrosis, as indicated by higher cellular MDA levels and lower CAT and SOD levels, compared to the vector group.

CONCLUSION

Our research highlights a significant association of SAMM50-rs2073080 with the progression of NAFLD to hepatic fibrosis, and the in vitro experiments further corroborated these findings.

Data-driven discovery of midlife cardiometabolic profile associated with incident early-onset and late-onset dementia

BACKGROUND

Cardiometabolic risk factors have been associated with the risk of late-onset dementia. However, evidence regarding early-onset dementia was inconsistent, and the impact of clustered cardiometabolic risk factors was unclear. We aimed to investigate the associations of cardiometabolic profiles with incident early-onset and late-onset dementia.

METHODS

Among 289 494 UK Biobank participants, cluster analysis was built on 12 common cardiometabolic markers. Analyses were performed on those aged <65 years at baseline (n = 249 870) for early-onset dementia and those ≥65 at the end of follow-up (n = 191 213) for late-onset dementia.

RESULTS

During a median follow-up of 14.1 years, 279 early-onset dementia cases and 3167 late-onset dementia cases were documented. Among the five clusters of cardiometabolic profiles identified (cluster 1 [obesity-dyslipidemia pattern], cluster 2 [high blood pressure pattern], cluster 3 [high liver enzymes pattern], cluster 4 [inflammation pattern] and cluster 5 [relatively healthy pattern]), cluster 3 was significantly associated with higher risks of both early-onset and late-onset dementia; however, the risk estimate for early-onset dementia (hazard ratio 2.58, 95% CI 1.61-4.14) was larger than that for late-onset dementia (1.36, 1.09-1.71). Cluster 4 was associated with a higher risk of late-onset dementia (hazard ratio 1.39, 95% CI 1.13-1.72). No significant interactions were observed between cardiometabolic clusters and apolipoprotein E ε4 genotype.

CONCLUSIONS

Cardiometabolic patterns characterised by relatively high liver enzyme levels or systemic inflammation were associated with increased risks of early-onset and late-onset dementia. Identification of high-risk subgroups according to distinct cardiometabolic patterns might help develop more precise strategies for dementia prevention regardless of apolipoprotein E (APOE) ε4 status.

Identifying a Genetic Link Between Lung Function and Psoriasis

INTRODUCTION

The common genetic underpinnings of psoriasis and pulmonary comorbidities have yet to be explored.

MATERIAL AND METHODS

In this cross-sectional study, we investigated the single-nucleotide polymorphisms (SNPs) associated with psoriasis and their relationship with pulmonary function using data from the UK Biobank (UKBB) and the Vanderbilt University Medical Center Biobank (BioVU).

RESULTS

Out of the 63 psoriasis-associated SNPs identified in previous genome-wide association studies within the European population, we successfully identified 53 SNPs, including proxy SNPs in UKBB database. Following adjustments using age and sex, 31 SNPs displayed statistically significant associations with psoriasis. Among these, 16 SNPs exhibited significant associations with forced expiratory volume in 1 s (FEV1), 14 with forced vital capacity (FVC), and 5 with the FEV1/FVC ratio in the UKBB. In the validation analysis using the BioVU database, 27 of the 31 psoriasis-associated SNPs were available for examination. Notably, the minor allele of SNP rs8016947 was confirmed to be significant, indicating a reduced risk for psoriasis and improved FEV1. Similarly, the minor alleles of SNPs rs17716942 and rs8016947 were associated with a reduced risk of psoriasis and enhanced FVC. However, none of the 5 SNPs significantly associated with the FEV1/FVC ratio in the UKBB displayed significance in the BioVU dataset.

CONCLUSION

This study has unveiled genetic variants that bridge the realms of psoriasis and lung function. The genes associated with these variants, including IFIH1, Grancalcin gene (GCA), and NFKB inhibitor alpha gene (NFKBIA), regulate innate immune responses, which suggests that immunodysregulation, a central element in psoriasis pathogenesis, may also impact lung function, alluding to a "skin-lung axis."

Genetic effects on gestational diabetes mellitus and their interactions with environmental factors among Japanese women

Gestational diabetes mellitus (GDM) is common in Japanese women, posing serious risks to mothers and offspring. This study investigated the influence of maternal genotypes on the risk of GDM and examined how these genotypes modify the effects of psychological and dietary factors during pregnancy. We analyzed data from 20,399 women in the Tohoku Medical Megabank Project Birth and Three-Generation Cohort. Utilizing two customized SNP arrays for the Japanese population (Affymetrix Axiom Japonica Array v2 and NEO), we performed a meta-analysis to combine the datasets. Gene-environment interactions were assessed by modeling interaction terms between genome-wide significant single nucleotide polymorphisms (SNPs) and psychological and dietary factors. Our analysis identified two SNP variants, rs7643571 (p = 9.14 × 10-9) and rs140353742 (p = 1.24 × 10-8), located in an intron of the MDFIC2 gene, as being associated with an increased risk of GDM. Additionally, although there were suggestive patterns for interactions between these SNPs and both dietary factors (e.g., carbohydrate and fruit intake) and psychological distress, none of the interaction terms remained significant after Bonferroni correction (p < 0.05/8). While nominal significance was observed in some models (e.g., psychological distress, p = 0.04), the data did not provide robust evidence of effect modification on GDM risk once adjusted for multiple comparisons. These findings reveal novel genetic associations with GDM in Japanese women and highlight the importance of gene-environment interactions in its etiology. Given that previous genome-wide association studies (GWAS) on GDM have primarily focused on Western populations, our study provides new insights by examining an Asian population using a population-specific array.

Exposome-Wide Gene-By-Environment Interaction Study of Psychotic Experiences in the UK Biobank

Background

A previous study successfully identified 148 of 23,098 exposures associated with any psychotic experiences (PEs) in the UK Biobank using an exposome-wide association study (XWAS). Furthermore, research has shown that the polygenic risk score for schizophrenia (PRS-SCZ) is associated with PEs. However, the interaction of these exposures with PRS-SCZ remains unknown.

Method

To systematically investigate possible gene-by-environment interactions underlying PEs through data-driven agnostic analyses, we conducted 1) conditional XWAS adjusting for PRS-SCZ to estimate the main effects of the exposures and of PRS-SCZ, 2) exposome-wide interaction study (XWIS) to estimate multiplicative and additive interactions between PRS-SCZ and exposures, and 3) correlation analyses between PRS-SCZ and exposures. The study included 148,502 participants from the UK Biobank.

Results

In the conditional XWAS models, significant effects of PRS-SCZ and 148 exposures on PEs remained statistically significant. In the XWIS model, we found significant multiplicative (multiplicative scale, 1.23; 95% CI, 1.10-1.37; p = 4.0 × 10-4) and additive (relative excess risk due to interaction, 0.55; 95% CI, 0.32-0.77; synergy index, 0.22; 95% CI, 0.14-0.30; and attributable proportion, 1.59; 95% CI, 1.30-1.91; all ps < .05/148) interactions of PRS-SCZ and the variable serious medical conditions/disability with PEs. We additionally identified 6 additive gene-by-environment interactions for mental distress, help-/treatment-seeking behaviors (3 variables), sadness, and sleep problems. In the correlation test focused on 7 exposures that exhibited significant interactions with PRS-SCZ, nonsignificant or small (r < 0.04) gene-by-environment correlations were observed.

Conclusions

These findings reveal evidence for gene-by-environment interactions underlying PEs and suggest that intertwined pathways of genetic vulnerability and exposures may contribute to psychosis risk.

Observational and genetic evidence highlight the association of modifiable risk factors with the incidence and severity of neuroimmunological disorders

Background

Myasthenia gravis (MG), multiple sclerosis (MS), and neuromyelitis optica spectrum disorders (NMOSD) are a heterogeneous group of rare neuroimmunological disorders whose incidence rates have increased in recent years. The relationships between modifiable risk factors and neuroimmunological disorders are not fully understood.

Methods

We utilized multiple logistic regression to estimate the relationships between 38 modifiable risk factors and two neuroimmunological diseases using data from nearly 500,000 individuals in the UK Biobank. Additionally, we applied two-sample Mendelian Randomization (MR) analyses using genetic variants as instrumental variables to investigate the causal relationships of 32 modifiable lifestyle factors with 8 outcomes, representing risk and severity across three neuroimmunological diseases. To further explore the underlying mechanisms, mediation analysis was conducted to elucidate how significant associations might be mediated by intermediate variables.

Results

Our observational and MR analyses consistently found significant associations (P < 0.05) indicating the number of cigarettes smoked daily, television watching, waist circumference, and BMI are all positively associated with the risk of developing MG. In contrast, moderate-to-vigorous physical activity and higher vitamin D levels are associated with a reduced risk of MS. Moreover, we discovered that the impact of television watching on the risk of MG was mediated by BMI (observational mediation analysis: 26.22%; MR mediation analysis: 9.90%).

Conclusions

These findings underscore the importance of modifiable risk factors in the development of neuroimmune diseases and support the identification of personalized intervention and prevention strategies. Notably, BMI significantly mediates the relationship between television watching and MG, indicating potential for targeted interventions to mitigate the risk of MG.

Large-scale benchmarking and boosting transfer learning for medical image analysis

Transfer learning, particularly fine-tuning models pretrained on photographic images to medical images, has proven indispensable for medical image analysis. There are numerous models with distinct architectures pretrained on various datasets using different strategies. But, there is a lack of up-to-date large-scale evaluations of their transferability to medical imaging, posing a challenge for practitioners in selecting the most proper pretrained models for their tasks at hand. To fill this gap, we conduct a comprehensive systematic study, focusing on (i) benchmarking numerous conventional and modern convolutional neural network (ConvNet) and vision transformer architectures across various medical tasks; (ii) investigating the impact of fine-tuning data size on the performance of ConvNets compared with vision transformers in medical imaging; (iii) examining the impact of pretraining data granularity on transfer learning performance; (iv) evaluating transferability of a wide range of recent self-supervised methods with diverse training objectives to a variety of medical tasks across different modalities; and (v) delving into the efficacy of domain-adaptive pretraining on both photographic and medical datasets to develop high-performance models for medical tasks. Our large-scale study (∼5,000 experiments) yields impactful insights: (1) ConvNets demonstrate higher transferability than vision transformers when fine-tuning for medical tasks; (2) ConvNets prove to be more annotation efficient than vision transformers when fine-tuning for medical tasks; (3) Fine-grained representations, rather than high-level semantic features, prove pivotal for fine-grained medical tasks; (4) Self-supervised models excel in learning holistic features compared with supervised models; and (5) Domain-adaptive pretraining leads to performant models via harnessing knowledge acquired from ImageNet and enhancing it through the utilization of readily accessible expert annotations associated with medical datasets. As open science, all codes and pretrained models are available at GitHub.com/JLiangLab/BenchmarkTransferLearning (Version 2).

Sleep-Associated Traits and Hearing Difficulties in Noise: A Bidirectional Mendelian Randomization Study

OBJECTIVES

The aim of this study was to investigate the causal relationships between sleep-associated traits and hearing difficulties in noise (HDinN) by Mendelian randomization (MR) analysis.

DESIGN

Single nucleotide polymorphisms associated with chronotype, insomnia, sleep duration, daytime dozing or sleeping, and ease of getting up in the morning were extracted from European population genome-wide association study pooled data for bidirectional MR analysis. The MR-Egger regression, the inverse variance weighted technique, and the weighted median method were used for data analysis. The study was then expanded to include South Asian, East Asian, African, and Greater Middle Eastern populations.

RESULTS

MR analysis indicated that in European populations, ease of getting up in the morning is a protective factor for HDinN (odds ratio [OR] = 0.932, p = 4.22 × 10 -5 , pFDR = 5.62 × 10 -4 ), while shorter sleep duration was a risk factor (undersleepers: OR = 1.164, p = 0.002, pFDR = 0.014). In addition, there was an indicative causal association between daytime dozing and HDinN (OR = 1.089, p = 0.046, pFDR = 0.123). The conclusions were consistent in African populations (ease of getting up: OR = 0.696, p = 0.012, pFDR = 0.041, sleep duration: OR = 0.677, p = 0.032 pFDR = 0.091, daytime dozing: OR = 1.164, p = 0.002, pFDR = 0.014). In the reverse direction, there was a significant causal association between HDinN and both chronotype (OR = 1.413, p = 0.011, pFDR = 0.042) and ease of getting up in the morning (OR = 0.668, p = 1.75 × 10 -5 , pFDR = 3.49 × 10 -4 ) in European populations, with similar conclusions respectively reached in East Asian (OR = 1.085, p = 0.010, pFDR = 0.045) and African populations (OR = 0.936, p = 0.002, pFDR = 0.012). Furthermore, although not observed in European populations, exploratory studies in non-European populations suggested a potential association between insomnia and HDinN (East Asian: OR = 1.920, p = 0.011, pFDR = 0.043, African: OR = 2.080, p = 0.004, pFDR = 0.019, South Asian: OR = 1.981, p = 1.59 × 10 -4 , PFDR = 0.002, Greater Middle Eastern: OR = 2.394, p = 0.002, pFDR = 0.012), and vice versa (Greater Middle Eastern: OR = 1.056, p = 0.014, pFDR = 0.044).

CONCLUSIONS

This study identified a potential bidirectional causal relationship between sleep-associated traits and HDinN. However, the underlying mechanisms of the causal relationships reported here have yet to be elucidated.

A Bayesian approach to correcting the attenuation bias of regression using polygenic risk score

Polygenic risk score has become increasingly popular for predicting the value of complex traits. In many settings, polygenic risk score is used as a covariate in regression analysis to study the association between different phenotypes. However, measurement error in polygenic risk score causes attenuation bias in the estimation of regression coefficients. In this paper, we employ a Bayesian approach to accounting for the measurement error of polygenic risk score and correcting the attenuation bias in linear and logistic regression. Through simulation, we show that our approach is able to obtain approximately unbiased estimation of coefficients and credible intervals with correct coverage probability. We also empirically compare our Bayesian measurement error model with the conventional regression model by analyzing real traits in the UK Biobank. The results demonstrate the effectiveness of our approach as it significantly reduces the error in coefficient estimates.

Correcting for volunteer bias in GWAS increases SNP effect sizes and heritability estimates

Selection bias in genome-wide association studies (GWASs) due to volunteer-based sampling (volunteer bias) is poorly understood. The UK Biobank (UKB), one of the largest and most widely used cohorts, is highly selected. Using inverse probability (IP) weights we estimate inverse probability weighted GWAS (WGWAS) to correct GWAS summary statistics in the UKB for volunteer bias. Our IP weights were estimated using UK Census data - the largest source of population-representative data - made representative of the UKB's sampling population. These weights have a substantial SNP-based heritability of 4.8% (s.e. 0.8%), suggesting they capture volunteer bias in GWAS. Across ten phenotypes, WGWAS yields larger SNP effect sizes, larger heritability estimates, and altered gene-set tissue expression, despite decreasing the effective sample size by 62% on average, compared to GWAS. The impact of volunteer bias on GWAS results varies by phenotype. Traits related to disease, health behaviors, and socioeconomic status are most affected. We recommend that GWAS consortia provide population weights for their data sets, or use population-representative samples.

Frailty and social contact with dementia risk: A prospective cohort study

BACKGROUND

Frailty and social contact are significant factors influencing dementia risk. While previous studies have separately examined these factors, their combined impact on dementia remains underexplored.

METHODS

This study included 338,567 UK biobank participants from 2006 to 2010, with follow-up until December 2022. Additionally, 30,408 participants with brain magnetic resonance imaging data were analyzed for hippocampal volume. Cox proportional hazards regression and linear regression models were used to assess associations.

RESULTS

The study followed 338,567 participants (mean [SD] age, 60.4 [5.2] years; 54.1 % men) for a median of 13.7 years, documenting 7362 cases of all-cause dementia. Both frailty and lower social contact independently increased the risk of all-cause dementia, Alzheimer's disease (AD), and vascular dementia (VaD). Compared to individuals with non-frailty and high social contact, those with lower social contact and higher frailty had a significantly increased risk of all-cause dementia, with the highest risk observed in individuals with frailty and low social contact (HR = 2.65, 95 % CI: 2.27-3.11). Similar patterns were found for AD and VaD. Furthermore, hippocampal volume was significantly reduced in individuals with frailty and low social contact (β = -0.24, 95 % CI: -0.43 to -0.06) compared to those with non-frailty and high social contact.

LIMITATIONS

The study predominantly included European descent individuals, with most frailty and social contact data based on baseline self-reports.

CONCLUSIONS

The combination of frailty and low social contact is associated with the highest risk of dementia. These findings suggest that both physiological and social factors should be simultaneously considered in dementia prevention strategies.

Combining genetic and non-genetic factors to predict the risk of pancreatic cancer in patients with new-onset diabetes mellitus

BACKGROUND

Recent research suggests that new-onset diabetes mellitus (NODM) often results from pancreatic cancer (PC) rather than causing it. Determining if NODM is type 2 diabetes mellitus (T2DM) or PC-related NODM (NODM-PC) aids in the early diagnosis of PC. We developed a NODM-PC risk prediction model to stratify PC risk in patients with NODM.

METHODS

This study utilized data from the UK Biobank, including 238 NODM-PC cases and 14,825 cancer-free T2DM controls. Polygenic risk scores (PRSs) for PC and T2DM were constructed using previously reported single nucleotide polymorphisms (SNPs) separately, while the NODM-PC PRS was developed by combining SNPs from both. Non-genetic factors were selected as candidate predictors based on prior NODM-PC prediction models. We developed three Cox models to estimate the risk of PC diagnosis within 3 years in the NODM population and evaluated them by internal-external cross-validation.

RESULTS

Elevated NODM-PC PRS and PC PRS scores positively correlated with NODM-PC risk, while T2DM PRS showed an inverse correlation. The NODM-PC PRS achieved the highest AUC at 0.719. Three Cox models were developed: Model 1 included age at T2DM diagnosis, smoking status, HbA1c, PC PRS, and T2DM PRS; Model 2 replaced PC and T2DM PRS with NODM-PC PRS; Model 3 included only non-genetic factors. Model 2 had the highest discrimination (Harrell's C-index 0.823 (95% CI: 0.806-0.840)), demonstrated the best clinical utility with good calibration, and showed significant classification improvement (continuous net reclassification index: 26.89% and 31.93% for cases, 28.51% and 30.90% for controls, compared to Models 1 and 3). The positive predictive value for the top 1% predicted risk in Model 2 was 13.25%.

CONCLUSIONS

This NODM-PC PRS enhances NODM-PC risk prediction, efficiently identifies individuals at high risk for PC screening, and improves PC screening efficiency at the population level among NODM individuals.

Life's Essential 8 scores, socioeconomic deprivation, genetic susceptibility, and new-onset chronic kidney diseases

BACKGROUND

The American Heart Association recently released a new cardiovascular health (CVH) metric, Life's Essential 8 (LE8), for health promotion. However, the association between LE8 scores and the risk of chronic kidney disease (CKD) remains uncertain. We aimed to explore the association of LE8 scores with new-onset CKD and examine whether socioeconomic deprivation and genetic risk modify this association.

METHODS

A total of 286,908 participants from UK Biobank and without prior CKD were included between 2006 and 2010. CVH was categorized using LE8 scores: low (LE8 scores <50), moderate (LE8 scores ≥50 but <80), and high (LE8 scores ≥80). The study outcome was new-onset CKD, ascertained by data linkage with primary care, hospital inpatient, and death data. Cox proportional hazard regression models were used to investigate the association between CVH categories and new-onset CKD.

RESULTS

During a median follow-up of 12.5 years, 8857 (3.1%) participants developed new-onset CKD. Compared to the low CVH group, the moderate (adjusted hazards ratio [HR], 0.50; 95% confidence interval [CI]: 0.47-0.53) and high CVH (adjusted HR, 0.31; 95% CI: 0.27-0.34) groups had a significantly lower risk of developing new-onset CKD. The population-attributable risk associated with high vs. intermediate or low CVH scores was 40.3%. Participants who were least deprived (vs. most deprived; adjusted HR, 0.75; 95% CI: 0.71-0.79) and with low genetic risk of CKD (vs. high genetic risk; adjusted HR, 0.89; 95% CI: 0.85-0.94) had a significantly lower risk of developing new-onset CKD. However, socioeconomic deprivation and genetic risks of CKD did not significantly modify the relationship between LE8 scores and new-onset CKD (both P-interaction >0.05).

CONCLUSION

Achieving a higher LE8 score was associated with a lower risk of developing new-onset CKD, regardless of socioeconomic deprivation and genetic risks of CKD.

A genetically informed brain atlas for enhancing brain imaging genomics

Brain imaging genomics has manifested considerable potential in illuminating the genetic determinants of human brain structure and function. This has propelled us to develop the GIANT (Genetically Informed brAiN aTlas) that accounts for genetic and neuroanatomical variations simultaneously. Integrating voxel-wise heritability and spatial proximity, GIANT clusters brain voxels into genetically informed regions, while retaining fundamental anatomical knowledge. Compared to conventional (non-genetics) brain atlases, GIANT exhibits smaller intra-region variations and larger inter-region variations in terms of voxel-wise heritability. As a result, GIANT yields increased regional SNP heritability, enhanced polygenicity, and its polygenic risk score explains more brain volumetric variation than traditional neuroanatomical brain atlases. We provide extensive validation to GIANT and demonstrate its neuroanatomical validity, confirming its generalizability across populations with diverse genetic ancestries and various brain conditions. Furthermore, we present a comprehensive genetic architecture of the GIANT regions, covering their functional annotation at the molecular levels, their associations with other complex traits/diseases, and the genetic and phenotypic correlations among GIANT-defined imaging endophenotypes. In summary, GIANT constitutes a brain atlas that captures the complexity of genetic and neuroanatomical heterogeneity, thereby enhancing the discovery power and applicability of imaging genomics investigations in biomedical science.

Polygenic overlap between subjective well-being and psychiatric disorders and cross-ancestry validation

Subjective well-being (SWB) is important for understanding human behaviour and health. Although the connection between SWB and psychiatric disorders has been studied, common genetic mechanisms remain unclear. This study aimed to explore the genetic relationship between SWB and psychiatric disorders. Bivariate causal mixture modelling (MiXeR), polygenic risk score (PRS) and Mendelian randomization (MR) analyses showed substantial polygenic overlap and associations between SWB and the psychiatric disorders. Subsequent replication studies in East Asian populations confirmed the polygenic overlap between schizophrenia and SWB. The conditional and conjunctional false discovery rate analyses identified additional or shared genetic loci associated with SWB or psychiatric disorders. Functional annotation revealed enrichment of specific brain tissues and genes associated with SWB. The identified genetic loci showed cross-ancestry transferability between the European and Korean populations. Our findings provide valuable insights into the common genetic mechanisms underlying SWB and psychiatric disorders.

The genetic architecture of and evolutionary constraints on the human pelvic form

Human pelvic evolution following the human-chimpanzee divergence is thought to result in an obstetrical dilemma, a mismatch between large infant brains and narrowed female birth canals, but empirical evidence has been equivocal. By using deep learning on 31,115 dual-energy x-ray absorptiometry scans from UK Biobank, we identified 180 loci associated with seven highly heritable pelvic phenotypes. Birth canal phenotypes showed sex-specific genetic architecture, aligning with reproductive function. Larger birth canals were linked to slower walking pace and reduced back pain but increased hip osteoarthritis risk, whereas narrower birth canals were associated with reduced pelvic floor disorder risk but increased obstructed labor risk. Lastly, genetic correlation between birth canal and head widths provides evidence of coevolution between the human pelvis and brain, partially mitigating the dilemma.

Trans-ancestry GWAS identifies 59 loci and improves risk prediction and fine-mapping for kidney stone disease

Kidney stone disease is a multifactorial disease with increasing incidence worldwide. Trans-ancestry GWAS has become a popular strategy to dissect genetic structure of complex traits. Here, we conduct a large trans-ancestry GWAS meta-analysis on kidney stone disease with 31,715 cases and 943,655 controls in European and East Asian populations. We identify 59 kidney stone disease susceptibility loci, including 13 novel loci and show similar effects across populations. Using fine-mapping, we detect 1612 variants at these loci, and pinpoint 25 causal signals with a posterior inclusion probability >0.5 among them. At a novel locus, we pinpoint TRIOBP gene and discuss its potential link to kidney stone disease. We show that a cross-population polygenic risk score, PRS-CSxEAS&EUR, exhibits superior predictive performance for kidney stone disease than other polygenic risk scores constructed in our study. Relative to individuals in the third quintile of PRS-CSxEAS&EUR, those in the lowest and highest quintiles exhibit distinct kidney stone disease risks with odds ratios of 0.57 (0.51-0.63) and 1.83 (1.68-1.98), respectively. Our results suggest that kidney stone disease patients with higher polygenic risk scores are younger at onset. In summary, our study advances the understanding of kidney stone disease genetic architecture and improves its genetic predictability.

Whole genome sequencing analysis of body mass index identifies novel African ancestry-specific risk allele

Obesity is a major public health crisis associated with high mortality rates. Previous genome-wide association studies (GWAS) investigating body mass index (BMI) have largely relied on imputed data from European individuals. This study leveraged whole-genome sequencing (WGS) data from 88,873 participants from the Trans-Omics for Precision Medicine (TOPMed) Program, of which 51% were of non-European population groups. We discovered 18 BMI-associated signals (P < 5 × 10-9), including two secondary signals. Notably, we identified and replicated a novel low-frequency single nucleotide polymorphism (SNP) in MTMR3 that was common in individuals of African descent. Using a diverse study population, we further identified two novel secondary signals in known BMI loci and pinpointed two likely causal variants in the POC5 and DMD loci. Our work demonstrates the benefits of combining WGS and diverse cohorts in expanding current catalog of variants and genes confer risk for obesity, bringing us one step closer to personalized medicine.

Trade-offs in modeling context dependency in complex trait genetics

Genetic effects on complex traits may depend on context, such as age, sex, environmental exposures, or social settings. However, it remains often unclear if the extent of context dependency, or gene-by-environment interaction (GxE), merits more involved models than the additive model typically used to analyze data from genome-wide association studies (GWAS). Here, we suggest considering the utility of GxE models in GWAS as a trade-off between bias and variance parameters. In particular, we derive a decision rule for choosing between competing models for the estimation of allelic effects. The rule weighs the increased estimation noise when context is considered against the potential bias when context dependency is ignored. In the empirical example of GxSex in human physiology, the increased noise of context-specific estimation often outweighs the bias reduction, rendering GxE models less useful when variants are considered independently. However, for complex traits, we argue that the joint consideration of context dependency across many variants mitigates both noise and bias. As a result, polygenic GxE models can improve both estimation and trait prediction. Finally, we exemplify (using GxDiet effects on longevity in fruit flies) how analyses based on independently ascertained 'top hits' alone can be misleading, and that considering polygenic patterns of GxE can improve interpretation.

The genetics of low and high birthweight and their relationship with cardiometabolic disease

AIMS/HYPOTHESIS

Low birthweight infants are at increased risk not only of mortality, but also of type 2 diabetes mellitus and CVD in later life. At the opposite end of the spectrum, high birthweight infants have increased risk of birth complications, such as shoulder dystocia, neonatal hypoglycaemia and obesity, and similarly increased risk of type 2 diabetes mellitus and CVD. However, previous genome-wide association studies (GWAS) of birthweight in the UK Biobank have primarily focused on individuals within the 'normal' range and have excluded individuals with high and low birthweight (<2.5 kg or >4.5 kg). The aim of this study was to investigate genetic variation associated within the tail ends of the birthweight distribution, to: (1) see whether the genetic factors operating in these regions were different from those that explained variation in birthweight within the normal range; (2) explore the genetic correlation between extremes of birthweight and cardiometabolic disease; and (3) investigate whether analysing the full distribution of birthweight values, including the extremes, improved the ability to detect genuine loci in GWAS.

METHODS

We performed case-control GWAS analysis of low (<2.5 kg) and high (>4.5 kg) birthweight in the UK Biobank using REGENIE software (Nlow=20,947; Nhigh=12,715; Ncontrols=207,506) and conducted three continuous GWAS of birthweight, one including the full range of birthweights, one involving a truncated GWAS including only individuals with birthweights between 2.5 and 4.5 kg and a third GWAS that winsorised birthweight values <2.5 kg and >4.5 kg. Additionally, we performed bivariate linkage disequilibrium (LD) score regression to estimate the genetic correlation between low/normal/high birthweight and cardiometabolic traits.

RESULTS

Bivariate LD score regression analyses suggested that high birthweight had a mostly similar genetic aetiology to birthweight within the normal range (genetic correlation coefficient [rG]=0.91, 95% CI 0.83, 0.99), whereas there was more evidence for a separate set of genes underlying low birthweight (rG=-0.74, 95% CI 0.66, 0.82). Low birthweight was also significantly positively genetically correlated with most cardiometabolic traits and diseases we examined, whereas high birthweight was mostly positively genetically correlated with adiposity and anthropometric-related traits. The winsorisation strategy performed best in terms of locus detection, with the number of independent genome-wide significant associations (p<5×10-8) increasing from 120 genetic variants at 94 loci in the truncated GWAS to 270 genetic variants at 178 loci, including 27 variants at 25 loci that had not been identified in previous birthweight GWAS. This included a novel low-frequency missense variant in the ABCC8 gene, a gene known to be involved in congenital hyperinsulinism, neonatal diabetes mellitus and MODY, that was estimated to be responsible for a 170 g increase in birthweight amongst carriers.

CONCLUSIONS/INTERPRETATION

Our results underscore the importance of genetic factors in the genesis of the phenotypic correlation between birthweight and cardiometabolic traits and diseases.

Associations between multiple ambient air pollutants, genetic risk, and incident mental disorders: An interaction study in the UK population

Mental disorders can be triggered by genetic and environmental risk factors. Limited studies have explored the effects of long-term exposure to air pollution on mental disorders, and most of the studies have focused on individual air pollutants. This study aimed to examine the relationship between long-term exposure to multiple air pollutants and incident mental disorders, including depression, anxiety, and schizophrenia, and whether the associations were affected by genetic susceptibility. Participants in the UK Biobank with no history of mental disorders were followed from baseline (2006 to 2010) to October 31st, 2022. Cox regression was applied to evaluate the correlation between PM2.5 absorbance, PM2.5, PM2.5-10, PM10, NO2, and NOx and any or specific mental disorders. Additive and multiplicative scales were used to measure the interaction between air pollution and schizophrenia polygenic risk score (PRS), depression PRS, or anxiety PRS on specific mental diseases. After a median of 13.36 years of follow-up on 252,376 participants, we observed per interquartile increase of PM2.5 absorbance (0.32 per meter), PM2.5 (1.28 μg/m3), NO2 (10.08 μg/m3), and NOx (16.78 μg/m3) were related to a 2-6 % higher risk of incident mental disorders. The HR (95 % CI) of incident mental disorder for the 2nd, 3rd, and 4th quartile of the air pollution score were 1.05 (1.01-1.18), 1.13 (1.09-1.18), and 1.14 (1.09-1.19), respectively, in comparison to the lowest level of the score. Per interquartile increase in the air pollution score was associated with a 6 %, 24 %, 4 %, and 6 % higher risk of incident mental disorders, schizophrenia, depression, and anxiety, respectively. No interaction between air pollution and genetic risk of schizophrenia, depression or anxiety on corresponding incident disorders was observed. These findings emphasize the importance of implementing air pollution control standards to decrease the burden of mental disorders.

Inherited predisposition to pneumothorax: estimating the frequency of Birt-Hogg-Dubé syndrome from genomics and population cohorts

Birt-Hogg-Dubé syndrome (BHDS) is the most common monogenic cause of pneumothorax. Most affected families have pathogenic variants in the FLCN gene. Using large genomic registries (UK Biobank (UKB), 100,000 Genomes Project and East London Genes & Health) including >550 000 individuals, we demonstrate that the frequency of clinically validated loss-of-function FLCN variants is 1 in 2710 to 4190. While the lifetime risk of pneumothorax in FLCN mutation carriers in the UKB and a BHDS clinical cohort was substantial (28.4% and 37.3%, respectively, to age 65 years), the lifetime risk of renal cancer was significantly lower in UKB than in BHDS patients (1% vs 32.1%). These findings highlight the importance of clinical context in managing individuals with FLCN mutations.

Pitfalls in performing genome-wide association studies on ratio traits

Genome-wide association studies (GWASs) are often performed on ratios composed of a numerator trait divided by a denominator trait. Examples include body mass index (BMI) and the waist-to-hip ratio, among many others. Explicitly or implicitly, the goal of forming the ratio is typically to adjust for an association between the numerator and denominator. While forming ratios may be clinically expedient, there are several important issues with performing GWAS on ratios. Forming a ratio does not "adjust" for the denominator in the sense of conditioning on it, and it is unclear whether associations with ratios are attributable to the numerator, the denominator, or both. Here we demonstrate that associations arising in ratio GWAS can be entirely denominator driven, implying that at least some associations uncovered by ratio GWAS may be due solely to a putative adjustment variable. In a survey of 10 common ratio traits, we find that the ratio model disagrees with the adjusted model (performing GWAS on the numerator while conditioning on the denominator) at around 1/3 of loci. Using BMI as an example, we show that variants detected by only the ratio model are more strongly associated with the denominator (height), while variants detected by only the adjusted model are more strongly associated with the numerator (weight). Although the adjusted model provides effect sizes with a clearer interpretation, it is susceptible to collider bias. We propose and validate a simple method of correcting for the genetic component of collider bias via leave-one-chromosome-out polygenic scoring.

Associations Between Chronotype, Genetic Susceptibility and Risk of Colorectal Cancer in UK Biobank

BACKGROUND

Sleep problems are common in the general population, with evidence suggesting a link between circadian rhythm disruptions and various health outcomes. However, the role of chronotype in influencing colorectal cancer (CRC) risk, particularly in conjunction with genetic predisposition, remains unclear and warrants further investigation.

METHODS

We analyzed data from 295,729 UK Biobank participants, among whom 4305 developed colorectal cancer. Chronotype was self-reported as morning or evening type, and a polygenic risk score for chronotype was generated from 316 genome-wide significant SNPs using 23andMe effect sizes to reduce overlap bias. Colorectal cancer risk was estimated using Cox proportional hazards models adjusted for age, sex, smoking, alcohol consumption, and the Townsend index.

RESULTS

Late chronotype and high polygenic risk were independently associated with an increased risk of CRC. Compared to participants with an early chronotype, those with a late chronotype exhibited a 6.5% increased risk of CRC [HR 1.065, P = 0.046]. Similarly, individuals in the high genetic risk group had a 11.0% increased risk compared with those in the low genetic risk group [HR, 1.110, P = 0.032]. Stratified analyses revealed that individuals with an intermediate genetic risk who had a late chronotype showed a 17.6% higher risk of CRC [OR, 1.176, P = 0.004], whereas those with a high genetic risk had a 25.3% increase [OR, 1.253, P = 0.001]. Through analyzing the combined effects of chronotype and PRS, we found that among individuals with an early chronotype, those with intermediate PRS had a 15.4% increased risk of CRC [HR, 1.154, P = 0.005], and those with high PRS had a 14.7% increased risk [HR, 1.147, P = 0.027].

CONCLUSIONS

Our findings highlight the importance of considering circadian rhythm patterns and genetic predispositions when assessing CRC risk, suggesting that chronotype may be associated with CRC risk, but further studies are needed to integrate objective circadian measurements.

Integration of functional genomics and statistical fine-mapping systematically characterizes adult-onset and childhood-onset asthma genetic associations

BACKGROUND

Genome-wide association studies (GWAS) have identified hundreds of loci underlying adult-onset asthma (AOA) and childhood-onset asthma (COA). However, the causal variants, regulatory elements, and effector genes at these loci are largely unknown.

METHODS

We performed heritability enrichment analysis to determine relevant cell types for AOA and COA, respectively. Next, we fine-mapped putative causal variants at AOA and COA loci. To improve the resolution of fine-mapping, we integrated ATAC-seq data in blood and lung cell types to annotate variants in candidate cis-regulatory elements (CREs). We then computationally prioritized candidate CREs underlying asthma risk, experimentally assessed their enhancer activity by massively parallel reporter assay (MPRA) in bronchial epithelial cells (BECs) and further validated a subset by luciferase assays. Combining chromatin interaction data and expression quantitative trait loci, we nominated genes targeted by candidate CREs and prioritized effector genes for AOA and COA.

RESULTS

Heritability enrichment analysis suggested a shared role of immune cells in the development of both AOA and COA while highlighting the distinct contribution of lung structural cells in COA. Functional fine-mapping uncovered 21 and 67 credible sets for AOA and COA, respectively, with only 16% shared between the two. Notably, one-third of the loci contained multiple credible sets. Our CRE prioritization strategy nominated 62 and 169 candidate CREs for AOA and COA, respectively. Over 60% of these candidate CREs showed open chromatin in multiple cell lineages, suggesting their potential pleiotropic effects in different cell types. Furthermore, COA candidate CREs were enriched for enhancers experimentally validated by MPRA in BECs. The prioritized effector genes included many genes involved in immune and inflammatory responses. Notably, multiple genes, including TNFSF4, a drug target undergoing clinical trials, were supported by two independent GWAS signals, indicating widespread allelic heterogeneity. Four out of six selected candidate CREs demonstrated allele-specific regulatory properties in luciferase assays in BECs.

CONCLUSIONS

We present a comprehensive characterization of causal variants, regulatory elements, and effector genes underlying AOA and COA genetics. Our results supported a distinct genetic basis between AOA and COA and highlighted regulatory complexity at many GWAS loci marked by both extensive pleiotropy and allelic heterogeneity.

Unraveling the genetic landscape of susceptibility to multiple primary cancers

With advances in cancer screening and treatment, there is a growing population of cancer survivors who may develop subsequent primary cancers. While hereditary cancer syndromes account for only a portion of multiple cancer cases, we sought to explore the role of common genetic variation in susceptibility to multiple primary tumors. We conducted a cross-ancestry genome-wide association study (GWAS) and transcriptome-wide association study (TWAS) of 10,983 individuals with multiple primary cancers, 84,475 individuals with single cancer, and 420,944 cancer-free controls from two large-scale studies. Our GWAS identified six lead variants across five genomic regions that were significantly associated (p < 5 × 10-8) with the risk of developing multiple primary tumors (overall and invasive) relative to cancer-free controls (at 3q26, 8q24, 10q24, 11q13.3, and 17p13). We also found one variant significantly associated with multiple cancers when compared with single cancer cases (at 22q13.1). Multi-tissue TWAS detected associations with genes involved in telomere maintenance in two of these regions (ACTRT3 in 3q26 and SLK and STN1 in 10q24) and the development of multiple cancers. Additionally, the TWAS also identified several novel genes associated with multiple cancers, including two immune-related genes, IRF4 and TNFRSF6B. Telomere maintenance and immune dysregulation emerge as central, common pathways influencing susceptibility to multiple cancers. These findings underscore the importance of exploring shared mechanisms in carcinogenesis, offering insights for targeted prevention and intervention strategies.

Cross ethnic Mendelian randomization analysis reveals causal relationship between air pollution and risk of kidney stones

Accumulating evidence has indicated that exposures to air pollution increase the odds of kidney stones. However, the previous research methods were limited. To address this gap, we employed genome-wide association studies (GWAS) datasets and Mendelian randomization (MR) to verify the causation. Applying publicly accessible summary datasets from UK Biobank, FinnGen consortium and Biobank Japan, a two-sample MR, and further multivariate MR were carried out to calculate the causality between air pollution [particulate matter 2.5 (PM2.5), PM2.5 absorbance, PM2.5-10, PM10, nitrogen dioxide, and nitrogen oxides] and kidney stone risk in three different populations (European, East Asian, and South Asian). The inverse variance weighted (IVW) was utilized for its first-step assessment, supplemented with MR-Egger, weighted median, Cochran's Q test, MR-Egger intercept and leave-one-out analysis to ensure the robustness. Employing IVW, we discovered in the European population that PM2.5 absorbance was statistically correlated with kidney stone risk (odds ratio (OR) = 1.40; 95% confidence interval (CI), 1.01-1.94; P = 0.04), with no heterogeneity, pleiotropy, or sensitivity observed. Additionally, the MVMR result revealed the directly causative connection between a single PM2.5 absorbance and the increase in kidney stone risk (OR = 1.77, 95%CI: 1.06-2.98, p = 0.03). Our investigation proposed the correlation between PM2.5 absorbance and an increased risk of kidney stones in European populations. The control of air pollution, especially PM2.5, may have crucial implications for the prevention of kidney stones.

Cross-ancestry analyses of Chinese and European populations reveal insights into the genetic architecture and disease implication of metabolites

Differential susceptibilities to various diseases and corresponding metabolite variations have been documented across diverse ethnic populations, but the genetic determinants of these disparities remain unclear. Here, we performed large-scale genome-wide association studies of 171 directly quantifiable metabolites from a nuclear magnetic resonance-based metabolomics platform in 10,792 Han Chinese individuals. We identified 15 variant-metabolite associations, eight of which were successfully replicated in an independent Chinese population (n = 4,480). By cross-ancestry meta-analysis integrating 213,397 European individuals from the UK Biobank, we identified 228 additional variant-metabolite associations and improved fine-mapping precision. Moreover, two-sample Mendelian randomization analyses revealed evidence that genetically predicted levels of triglycerides in high-density lipoprotein were associated with a higher risk of coronary artery disease and that of glycine with a lower risk of heart failure in both ancestries. These findings enhance our understanding of the shared and specific genetic architecture of metabolites as well as their roles in complex diseases across populations.

Benchmark of computational methods to detect digenism in sequencing data

Digenic inheritance is characterized by the combined alteration of two different genes leading to a disease. It could explain the etiology of many currently undiagnosed rare diseases. With the advent of next-generation sequencing technologies, the identification of digenic inheritance patterns has become more technically feasible, yet still poses significant challenges without any gold standard method. Here, we present a comprehensive overview of the existing methods developed to detect digenic inheritance in sequencing data and provide a classification in cohort-based and individual-based methods. The latter category of methods appeared the most applicable to rare diseases, especially the ones not needing patient phenotypic description as input. We discuss the availability of the different methods, their output and scalability to inform potential users. Focusing on methods to detect digenic inheritance in the case of very rare or heterogeneous diseases, we propose a benchmark using different real-life scenarios involving known digenic and putative neutral pairs of genes. Among these different methods, DiGePred stood out as the one giving the least number of false positives, ARBOCK as giving the greatest number of true positives, and DIEP as having the best balance between both. By synthesizing the state-of-the-art techniques and providing insights into their practical utility, this benchmark serves as a valuable resource for researchers and clinicians in selecting suitable methodologies for detecting digenic inheritance in a wide range of disorders using sequencing data.

Collagen V regulates renal function after kidney injury and can be pharmacologically targeted to enhance kidney repair in mice

Kidney fibrosis determines clinical outcomes in individuals with chronic kidney disease (CKD). The stoichiometric ratio of collagens in renal scar differs from that of healthy kidney extracellular matrix (ECM), but the functional importance of altered collagen types in injured kidneys remains unclear. Using human population studies, we show that circulating protein and renal mRNA amounts of collagen V A1 (COL5A1) exhibited associations with kidney disease and incident CKD risk. We show that Col5a1 regulates the degree of postinjury fibrosis and renal function. Mice with conditionally knocked out Col5a1 (Col5a1 CKO) exhibited decreased renal function and greater renal fibrosis after dietary adenine- or ureteric obstruction-mediated kidney injury. Renal fibroblasts in Col5a1 CKO animals up-regulated the profibrotic αvβ3 integrin. Inhibition of αvβ3 signaling with a small molecule, cilengitide, rescued postinjury renal function in Col5a1 CKO animals. Using the hybrid mouse diversity panel that comprises 100 diverse inbred strains of mice, we observed that gene expression of Col5a1 after injury exhibited genetic variation across 100 strains. Strains with low Col5a1 expression after injury exhibited worse renal function compared with animals that had higher degrees of expression. We next measured Col5a1 expression in peripheral blood mononuclear cells in mice to identify nonresponder strains that did not have increased Col5a1 expression after kidney injury. We observed that administration of cilengitide in nonresponder strains significantly rescued postinjury renal fibrosis and function. These studies point to the feasibility of precision medicine approaches to target Col5a1 for enhancing renal repair.

Hearing impairment, psychological distress, and incident heart failure: a prospective cohort study

BACKGROUND

The relationship between objectively measured hearing ability and the risk of incident heart failure (HF) remains unclear. This study aimed to assess this association, explore potential modifying factors, and examine whether psychological factors mediate this relationship.

METHODS

We included 164 431 participants from the UK Biobank without HF at baseline. Speech-in-noise hearing ability was measured using the Digit Triplets Test and quantified by the speech-reception-threshold (SRT). Incident HF was identified through hospital admission and death records. Mediation analyses assessed the role of social isolation, psychological distress, and neuroticism.

RESULTS

Over a median follow-up of 11.7 years, 4449 (2.7%) participants developed incident HF. Higher SRT levels were associated with an increased risk of HF (adjusted HR per SD increment 1.05, 95% CI 1.02 to 1.08). Compared with those with normal hearing, participants with insufficient hearing, poor hearing, or hearing aid use had higher HF risks (adjusted HRs 1.15, 1.28, and 1.26, respectively). Psychological distress mediated 16.9% of the association between SRT levels and HF, while social isolation and neuroticism mediated 3.0% and 3.1%, respectively. The association was stronger in participants without coronary heart disease or stroke at baseline.

CONCLUSIONS

Poor hearing ability is associated with an increased risk of incident HF, with psychological distress playing a notable mediating role. These findings suggest that hearing health and psychological well-being should be considered in cardiovascular risk assessment and prevention strategies.

Impact of anti-VZV IgG levels on Parkinson's disease risk and progression: a Mendelian randomization analysis

Research suggests a potential link between varicella zoster virus (VZV) and Parkinson's disease (PD), but the causal relationship between anti-VZV IgG levels and PD is not well understood. Using two-sample Mendelian Randomization (MR), we assessed the causal impact of anti-VZV IgG levels on PD risk and progression. Our study found a significant association between higher anti-VZV IgG levels and an increased risk of PD. For PD progression, higher anti-VZV IgG levels were linked to a greater risk of constipation, insomnia, and Restless Legs. These findings remained consistent after sensitivity analyses. In conclusion, our study suggests that elevated anti-VZV IgG levels may contribute to an increased risk and progression of PD, supporting a potential causal link that warrants further mechanistic investigation.

The comprehensive clinical benefits of digital phenotyping: from broad adoption to full impact

Digital phenotyping collects health data digitally, supporting early disease diagnosis and health management. This paper systematically reviews the diversity of research methods in digital phenotyping and its clinical benefits, while also focusing on its importance within the P4 medicine paradigm and its core role in advancing its application in biobanks. Furthermore, the paper envisions the continued clinical benefits of digital phenotyping, driven by technological innovation, global collaboration, and policy support.

The Estonian Biobank's journey from biobanking to personalized medicine

Large biobanks have set a new standard for research and innovation in human genomics and implementation of personalized medicine. The Estonian Biobank was founded a quarter of a century ago, and its biological specimens, clinical, health, omics, and lifestyle data have been included in over 800 publications to date. What makes the biobank unique internationally is its translational focus, with active efforts to conduct clinical studies based on genetic findings, and to explore the effects of return of results on participants. In this review, we provide an overview of the Estonian Biobank, highlight its strengths for studying the effects of genetic variation and quantitative phenotypes on health-related traits, development of methods and frameworks for bringing genomics into the clinic, and its role as a driving force for implementing personalized medicine on a national level and beyond.

Polygenic dissection of treatment-resistant depression with proxy phenotypes in the UK Biobank

BACKGROUND

Treatment-resistant depression (TRD) affects one-third of major depressive disorder (MDD) patients. Previous pharmacogenetic studies suggest genetic variation may influence medication response but findings are heterogeneous. We conducted a comprehensive genetic investigation using proxy TRD phenotypes (TRDp) that mirror the treatment options of MDD from UK Biobank primary care records.

METHODS

Among 15,125 White British MDD patients, we identified TRDp with medication changes (switching or receiving multiple antidepressants [AD]); augmentation therapy (antipsychotics; mood stabilizers; valproate; lithium); or electroconvulsive therapy (ECT). Hospitalized TRDp patients (HOSP-TRDp) were also identified. We conducted genome-wide association analysis, estimated SNP-heritability (hg2), and assessed the genetic burden for nine psychiatric diseases using polygenic risk scores (PRS).

RESULTS

TRDp patients were more often female, unemployed, less educated, and had higher BMI, with hospitalization rates twice as high as non-TRDp. While no credible risk variants emerged, heritability analysis showed significant genetic influence on TRDp (liability hg2 21-24 %), particularly for HOSP-TRDp (28-31 %). TRDp classified by AD changes and augmentation carried an elevated yet varied polygenic burden for MDD, ADHD, BD, and SCZ. Higher BD PRS increased the likelihood of receiving ECT, lithium, and valproate by 1.27-1.80 fold. Patients in the top 10 % PRS relative to the average had a 12-36 % and 24-51 % higher risk of TRDp and HOSP-TRDp, respectively.

CONCLUSIONS

Our findings support a significant polygenic basis for TRD, highlighting genetic and phenotypic distinctions from non-TRD. We demonstrate that different TRDp endpoints are enriched with various spectra of psychiatric genetic liability, offering insights into pharmacogenomics and TRD's complex genetic architecture.

Multiple origins and phenotypic implications of an extended human pseudoautosomal region shown by analysis of the UK Biobank

The 2.7-Mb major pseudoautosomal region (PAR1) on the short arms of the human X and Y chromosomes plays a critical role in meiotic sex chromosome segregation and male fertility and has been regarded as evolutionarily stable. However, some European Y chromosomes belonging to Y haplogroups (Y-Hgs) R1b and I2a carry an ∼115-kb extension (ePAR [extended PAR]) arising from X-Y non-allelic homologous recombination (NAHR). To investigate the diversity, history, and dynamics of ePAR formation, we screened for its presence, and that of the predicted reciprocal X chromosome deletion, among ∼218,300 46,XY males of the UK Biobank (UKB), a cohort associated with longitudinal clinical data. The UKB incidence of ePAR is ∼0.77%, and that of the deletion is ∼0.02%. We found that Y-Hg I2a sub-lineages accounted for nearly 90% of ePAR cases but, by Y haplotyping and breakpoint sequencing, determined that, in total, there have been at least 18 independent ePAR origins, associated with nine different Y-Hgs. We found examples of ePAR linked to Y-Hg K among men of self-declared Pakistani ancestry and Y-Hg E1, typical of men with African ancestry, showing that ePAR is not restricted to Europeans. ePAR formation is likely random, with high frequencies in some Y-Hgs arising through drift and male-mediated expansions. Sequencing recombination junction fragments identified likely reciprocal events, and the heterogeneity of ePAR and X-deletion junctions highlighted the recurrent nature of the NAHR events. A phenome-wide association study revealed an association between ePAR and elevated levels of circulating IGF-1 as well as musculoskeletal phenotypes.

Developing and validating a nomogram prediction model for osteoporosis risk in the UK biobank: a national prospective cohort

BACKGROUND

Osteoporosis is a prevalent bone disease that increases frailty. Developing a nomogram prediction model to predict osteoporosis risk at multiple time points using bone mineral densities, behavioral habits, and clinical risk factors would be essential to identify individual risk and guide prevention.

METHODS

The study population from the UK Biobank was followed from 2014 to December 31st, 2022. The study outcome was identified as the first occurrence of osteoporosis in the UK Biobank during the follow-up period. After rebalancing with the synthetic minority over-sampling technique, a nomogram prediction model was developed using a LASSO Cox regression. Model discrimination between different risk levels was visualised with Kaplan-Meier curves, and model performance was evaluated with integrated c-index, time-dependent AUC, calibration curves and decision curve analysis (DCA).

RESULTS

The model identified several risk factors for osteoporosis, including higher age, underweight, and various clinical risk factors (such as menopause, lower hand grip strength, lower bone mineral density, fracture history within 5 years, and a history of chronic disease including hypercholesterolemia, cardiovascular disease, bone disease, arthritis, and cancer). Kaplan-Meier curves showed that risk levels predicted by the nomogram model were significantly distinct. The c-indexes were 0.844 and 0.823 for training and validation datasets, respectively. Time-dependent AUC, calibration curves and DCA indicated good discrimination, model fit and clinical utility, respectively.

CONCLUSIONS

The nomogram model could properly quantify the five-year risk of osteoporosis and identify high-risk individuals. This might effectively reduce the burden of osteoporosis on the population.

Polygenic score analysis identifies distinct genetic risk profiles in Alzheimer's disease comorbidities

Alzheimer's disease (AD) is usually accompanied by comorbidities such as type 2 diabetes (T2D), epilepsy, major depressive disorder (MDD), and migraine headaches (MH) that can significantly affect patient management and progression. As AD, these comorbidities have their own cumulative common genetic risk component that can be explored in a single individual through polygenic scores. Utilizing data from the UK Biobank, we investigated the correlation between polygenic scores (PGS) for these comorbidities and their actual presentation in AD patients. We show that individuals with higher PGS values showed an elevated risk of developing T2D (OR 2.1, p = 1.07 × 10-11) and epilepsy (OR 1.5, p = 0.0176). High T2D-PGS is also associated with an earlier AD onset in individuals at high genetic risk for AD (AD-PGS). In contrast, no significant genetic associations were found for MDD and MH. Our findings show distinct common genetic risk factors for T2D and epilepsy carried by AD patients that are associated with increased prevalence and earlier disease onset. These results highlight the contribution of common genetic variation to the broader clinical landscape of AD and will contribute to future tailored patient management strategies for individuals at high genetic risk.

Opportunities and challenges of local ancestry in genetic association analyses

Recently, admixed populations make up an increasing percentage of the US and global populations, and the admixture is not uniform over space or time or across genomes. Therefore, it becomes indispensable to evaluate local ancestry in addition to global ancestry to improve genetic epidemiological studies. Recent advances in representing human genome diversity, coupled with large-scale whole-genome sequencing initiatives and improved tools for local ancestry inference, have enabled studies to demonstrate that incorporating local ancestry information enhances both genetic association analyses and polygenic risk predictions. Along with the opportunities that local ancestry provides, there exist challenges preventing its full usage in genetic analyses. In this review, we first summarize methods for local ancestry inference and illustrate how local ancestry can be utilized in various analyses, including admixture mapping, association testing, and polygenic risk score construction. In addition, we discuss current challenges in research involving local ancestry, both in terms of the inference itself and its role in genetic association studies. We further pinpoint some future study directions and methodology development opportunities to help more effectively incorporate local ancestry in genetic analyses. It is worth the effort to pursue those future directions and address these analytical challenges because the appropriate use of local ancestry estimates could help mitigate inequality in genomic medicine and improve our understanding of health and disease outcomes.

Machine learning reveals connections between preclinical type 2 diabetes subtypes and brain health

Previous research has established type 2 diabetes mellitus as a significant risk factor for various disorders, adversely impacting human health. While evidence increasingly links type 2 diabetes to cognitive impairment and brain disorders, understanding the causal effects of its preclinical stage on brain health is yet to be fully known. This knowledge gap hinders advancements in screening and preventing neurological and psychiatric diseases. To address this gap, we employed a robust machine learning algorithm (Subtype and Stage Inference, SuStaIn) with cross-sectional clinical data from the UK Biobank (20 277 preclinical type 2 diabetes participants and 20 277 controls) to identify underlying subtypes and stages for preclinical type 2 diabetes. Our analysis revealed one subtype distinguished by elevated circulating leptin levels and decreased leptin receptor levels, coupled with increased body mass index, diminished lipid metabolism, and heightened susceptibility to psychiatric conditions such as anxiety disorder, depression disorder, and bipolar disorder. Conversely, individuals in the second subtype manifested typical abnormalities in glucose metabolism, including rising glucose and haemoglobin A1c levels, with observed correlations with neurodegenerative disorders. A >10-year follow-up of these individuals revealed differential declines in brain health and significant clinical outcome disparities between subtypes. The first subtype exhibited faster progression and higher risk for psychiatric conditions, while the second subtype was associated with more severe progression of Alzheimer's disease and Parkinson's disease and faster progression to type 2 diabetes. Our findings highlight that monitoring and addressing the brain health needs of individuals in the preclinical stage of type 2 diabetes is imperative.

EndoPRS: Incorporating endophenotype information to improve polygenic risk scores for clinical endpoints-A study in asthma

Polygenic risk score (PRS) prediction of complex diseases can be improved by leveraging related phenotypes. This has motivated the development of several multi-trait PRS methods that jointly model genetically correlated traits. However, these methods do not account for vertical pleiotropy, where one trait acts as a mediator for another. Here, we introduce endoPRS, a weighted lasso model that incorporates information from relevant endophenotypes to improve disease risk prediction without making assumptions about the genetic architecture underlying the endophenotype-disease relationship. Through extensive simulation analysis, we demonstrate the robustness of endoPRS in a variety of complex genetic frameworks. We also apply endoPRS to predict the risk of childhood-onset asthma in UK Biobank and All of Us by leveraging a paired genome-wide association study of eosinophil count, a relevant endophenotype. We find that endoPRS significantly improves prediction and transferability compared to many existing PRS methods, including multi-trait PRS methods MTAG and wMT-BLUP, which suggests advantages of endoPRS in real-life clinical settings.

Metabolic reaction fluxes as amplifiers and buffers of risk alleles for coronary artery disease

Genome-wide association studies have identified thousands of variants associated with disease risk but the mechanism by which such variants contribute to disease remains largely unknown. Indeed, a major challenge is that variants do not act in isolation but rather in the framework of highly complex biological networks, such as the human metabolic network, which can amplify or buffer the effect of specific risk alleles on disease susceptibility. Here we use genetically predicted reaction fluxes to perform a systematic search for metabolic fluxes acting as buffers or amplifiers of coronary artery disease (CAD) risk alleles. Our analysis identifies 30 risk locus-reaction flux pairs with significant interaction on CAD susceptibility involving 18 individual reaction fluxes and 8 independent risk loci. Notably, many of these reactions are linked to processes with putative roles in the disease such as the metabolism of inflammatory mediators. In summary, this work establishes proof of concept that biochemical reaction fluxes can have non-additive effects with risk alleles and provides novel insights into the interplay between metabolism and genetic variation on disease susceptibility.

Genomic determinants of biological age estimated by deep learning applied to retinal images

With the development of deep learning (DL) techniques, there has been a successful application of this approach to determine biological age from latent information contained in retinal images. Retinal age gap (RAG) defined as the difference between chronological age and predicted retinal age has been established previously to predict the age-related disease. In this study, we performed discovery genome-wide association analysis (GWAS) on the RAG using the 31,271 UK Biobank participants and replicated our findings in 8034 GoDARTS participants. The genetic correlation between RAGs predicted from the two cohorts was 0.67 (P = 0.021). After meta-analysis, we found 13 RAG loci which might be related to retinal vessel density and other aging processes. The SNP-wide heritability (h2) of RAG was 0.15. Meanwhile, by performing Mendelian randomization analysis, we found that glycated hemoglobin, inflammation hemocytes, and anemia might be associated with accelerated retinal aging. Our study explored the biological implications and molecular-level mechanism of RAG, which might enable causal inference of the aging process as well as provide potential pharmaceutical intervention targets for further treatment.

Alzheimer's Disease polygenic risk, the plasma proteome, and dementia incidence among UK older adults

Alzheimer's Disease (AD) is a complex polygenic neurodegenerative disorder. Its genetic risk's relationship with all-cause dementia may be influenced by the plasma proteome. Up to 40,139 UK Biobank participants aged ≥ 50y at baseline assessment (2006-2010) were followed-up for ≤ 15 y for dementia incidence. Plasma proteomics were performed on a sub-sample of UK Biobank participants (k = 1,463 plasma proteins). AD polygenic risk scores (PRS) were used as the primary exposure and Cox proportional hazards models were conducted to examine the AD PRS-dementia relationship. A four-way decomposition model then partitioned the total effect (TE) of AD PRS on dementia into an effect due to mediation only, an effect due to interaction only, neither or both. The study found that AD PRS tertiles significantly increased the risk for all-cause dementia, particularly among women. The study specifically found that AD PRS was associated with a 79% higher risk for all-cause dementia for each unit increase (HR = 1.79, 95% CI: 1.70-1.87, P < 0.001). Eighty-six plasma proteins were significantly predicted by AD PRS, including a positive association with PLA2G7, BRK1, the glial acidic fibrillary protein (GFAP), neurofilament light chain (NfL), and negative with TREM2. Both GFAP and NfL significantly interacted synergistically with AD PRS to increase all-dementia risk (> 10% of TE is pure interaction), while GFAP was also an important consistent mediator in the AD PRS-dementia relationship. In summary, we detected significant interactions of NfL and GFAP with AD PRS, in relation to dementia incidence, suggesting potential for personalized dementia prevention and management.

Relatively independent and complementary roles of family history and polygenic risk score in age at onset and incident cases of 12 common diseases

Few studies have systematically compared the overlap and complementarity of family history (FH) and polygenic risk score (PRS) in terms of disease risk. We here investigated the impacts of FH and PRS on the risk of incident diseases or age at disease onset, as well as their clinical value in risk prediction. We analyzed 12 diseases in the prospective cohort study of UK Biobank (N = 461,220). First, restricted mean survival time analysis was performed to evaluate the influences of FH and PRS on age at onset. Then, Cox proportional hazards model was employed to estimate the effects of FH and PRS on the incident risk. Finally, prediction models were constructed to examine the clinical value of FH and PRS in the incident disease risk. Compared to negative FH, positive FH led to an earlier onset, with an average of 2.29 years earlier between the top and bottom 2.5% PRSs and high blood pressure showing the greatest difference of 6.01 years earlier. Both FH and PRS were related to higher incident risk; but they only exhibited weak interactions on high blood pressure and Alzheimer's disease/dementia, and provided relatively independent and partially complementary information on disease susceptibility, with PRS explaining 7.0% of the FH effect but FH accounting for only 1.1% of the PRS effect for incident cases. Further, FH and PRS showed additional predictive value in risk evaluation, with breast cancer showing the greatest improvement (31.3%). FH and PRS significantly affect a variety of diseases, and they are not interchangeable measures of genetic susceptibility, but instead offer largely independent and partially complementary information. Incorporating FH, PRS, and clinical risk factors simultaneously leads to the greatest predictive value for disease risk assessment.

Exploring the Link Between C-Reactive Protein Change and Stroke Risk: Insights From a Prospective Cohort Study and Genetic Evidence

BACKGROUND

Previous research on how changes in CRP (C-reactive protein) levels predict stroke risk is limited. This study aimed to examine the association between CRP change and the risk of stroke and its subtypes.

METHODS AND RESULTS

Based on the UK Biobank data, we investigated the association between CRP change and the risk of stroke and its subtypes with Cox proportional hazards regression analysis. We further performed genetic analyses including genetic correlation, pairwise genome-wide association study, and polygenic risk score. Our study involved 14 754 participants with a median follow-up time of 10.4 years. After categorizing participants by CRP percentage change and making adjustments for potential confounders, it was observed that those with an elevated percentage of CRP change had a higher risk of any stroke (hazard ratio [HR], 1.44 [95% CI, 1.12-1.85]) and ischemic stroke (HR, 1.65 [95% CI, 1.24-2.18]). After categorization by CRP change types and adjustment for confounders, the group that became high level had a higher any-stroke risk (HR, 1.45 [95% CI, 1.04-2.02]), with the group that remained at a high level facing the greatest risk (HR, 1.74 [95% CI, 1.30-2.33]). Similar trends were observed for ischemic stroke. The group that remained at a high level also had a heightened hemorrhagic stroke risk (HR, 1.91 [95% CI, 1.07-3.44]). Genetic analysis showed a significant genetic correlation between CRP and stroke (rg, 0.257; rg_P=2.39E-07). Pairwise genome-wide association study analysis identified 5 shared genomic regions between CRP and stroke. Polygenic risk score analysis showed that participants with high stroke polygenic risk score and elevated or remaining high CRP levels have the highest risk of stroke.

CONCLUSIONS

Both any stroke and ischemic stroke are related to elevated and remaining high CRP levels, while hemorrhagic stroke is only related to remaining high CRP levels.

Plasma proteomics-based brain aging signature and incident dementia risk

Investigating brain-enriched proteins with machine learning methods may enable a brain-specific understanding of brain aging and provide insights into the molecular mechanisms and pathological pathways of dementia. The study aims to analyze associations of brain-specific plasma proteomic aging signature with risks of incident dementia. In 45,429 dementia-free UK Biobank participants at baseline, we generated a brain-specific biological age using 63 brain-enriched plasma proteins with machine learning methods. The brain age gap was estimated, and Cox proportional hazards models were used to study the association with incident all-cause dementia, Alzheimer's disease (AD), and vascular dementia. Per-unit increment in the brain age gap z-score was associated with significantly higher risks of all-cause dementia (hazard ratio [95% confidence interval], 1.67 [1.56-1.79], P < 0.001), AD (1.85 [1.66-2.08], P < 0.001), and vascular dementia (1.86 [1.55-2.24], P < 0.001), respectively. Notably, 2.1% of the study population exhibited extreme old brain aging defined as brain age gap z-score > 2, correlating with over threefold increased risks of all-cause dementia and vascular dementia (3.42 [2.25-5.20], P < 0.001, and 3.41 [1.05-11.13], P = 0.042, respectively), and fourfold increased risk of AD (4.45 [2.32-8.54], P < 0.001). The associations were stronger among participants with healthier lifestyle factors (all P-interaction < 0.05). These findings were corroborated by magnetic resonance imaging assessments showing that a higher brain age gap aligns global pathophysiology of dementia, including global and regional atrophy in gray matter, and white matter lesions (P < 0.001). The brain-specific proteomic age gap is a powerful biomarker of brain aging, indicative of dementia risk and neurodegeneration.

Measuring the associations between brain morphometry and polygenic risk scores for substance use disorders in drug-naive adolescents

Substance use has been associated with differences in adult brain morphology; however, it is unclear whether these differences precede or are a result of substance use substance use. We investigated the impact of polygenic risk scores (PRSs) for cannabis use disorder (CUD) and general substance use and substance use disorder liability (SU/SUD) on brain morphology in drug-naïve adolescents. Baseline data were used from 1,874 European-descent participants (ages 9-11) comprising 222, 328 and 387 pairs of MZ twins, DZ twins, and Non-Twin Siblings, respectively, in the Adolescent Brain Cognitive Development Study. We fitted multivariate twin models to estimate the putative effects of CUD, SU/SUD, and brain region-specific PRSs. These models assessed their influence on six subcortical and two cortical phenotypes. PRS for CUD and SU/SUD were created based on GWAS conducted by Johnson et al. (2020) and Hatoum et al. (2023), respectively. When decomposing variance in each brain region of interest (ROI), we used the corresponding ROI-specific PRS. Brain morphometry in drug-naive subjects was unrelated to CUD PRS. The variance explained in each ROI by its corresponding PRS ranged from 0.8-4.4%. The SU/SUD PRS showed marginally significant effects (0.2-0.4%) on cortical surface area and nucleus accumbens volume, but overall effect sizes were small. Our findings indicate that differences in brain morphometry among baseline drug-naive individuals are not associated with the genetic risk for CUD but show a weak association with general addiction and substance use risk (SU/SUD), particularly in nucleus accumbens volume and total cortical surface area.

Personalization of primary prevention: Exploring the role of coronary artery calcium and polygenic risk score in cardiovascular diseases

Personalized healthcare is becoming increasingly popular given the vast heterogeneity in disease manifestation between individuals. Many commonly encountered diseases within cardiology are multifactorial in nature and disease progression and response is often variable due to environmental and genetic factors influencing disease states. This makes accurate early identification and primary prevention difficult in certain populations, especially young patients with limited Atherosclerotic Cardiovascular Disease (ASCVD) risk factors. Newer strategies, such as coronary artery calcium (CAC) scans and polygenic risk scores (PRS), are being implemented to aid in the detection of subclinical disease and heritable risk, respectively. Data surrounding CAC scans have shown promising results in their ability to detect subclinical atherosclerosis and predict the risk of future coronary events, especially at the extremes; however, predictive variability exists among different patient populations, limiting the test's specificity. Furthermore, relying only on CAC scores and ASCVD risk scores may fail to identify a large group of patients needing primary prevention who lack subclinical disease and traditional risk factors, but harbor genetic variabilities strongly associated with certain cardiovascular diseases. PRS can overcome these limitations. These scores can be measured in individuals as early as birth to identify genetic variants placing them at elevated risk for developing cardiovascular disease, irrespective of their current cardiovascular health status. By applying PRS alongside CAC scores, previously overlooked patient populations can be identified and begin primary prevention strategies early to achieve optimal outcomes. In this review, we expand on the current knowledge surrounding CAC scores and PRS and highlight the future possibilities of these technologies for preventive cardiology.