Biobank Dataset Research Articles

In search of a marker of altered cerebrovascular function in hypertension: Analysis of the fractional amplitude of low-frequency fluctuations in UK Biobank resting state fMRI data

The selfish brain mechanism proposes that in some patients with impaired cerebral blood flow (CBF) or cerebrovascular function, hypertension may develop as a compensatory mechanism that aims to maintain CBF by increasing systemic blood pressure through an increase in cardiovascular sympathetic tone. The amplitude of low frequency fluctuations (ALFF) in the resting state blood oxygenation level dependent (BOLD) functional MRI signal has been previously posited as an index of cerebrovascular reactivity.We investigated whether regional fractional ALFF (fALFF) differs between 2054 hypertensives and 1724 normotensives using data from the UK Biobank dataset. Our primary hypothesis was that cerebrovascular function in the medulla and other regions involved in sympathetic regulation differs between hypertensives and normotensives, and that this is reflected by regional variations in fALFF.There is a significant regional variation in fALFF (F(14) =1126.17, p < 2 × 10−16, partial η2 = 0.22), but this regional variation does not differ between hypertensives and normotensives (F(14) = 0.23, p = 0.99, partial η2 = 8 × 10−5).Prospective longitudinal studies of cerebral haemodynamics in hypertensives and normotensives are required to further investigate the selfish brain mechanism.

Circulating levels of cytokines and risk of inflammatory bowel disease: evidence from genetic data.

Prior epidemiological studies have established a correlation between inflammatory cytokines and inflammatory bowel disease (IBD). However, the nature of this relationship remains uncertain. Mendelian randomization (MR) study has the advantages of avoiding confounding and reverse causality compared with traditional observational research. We aimed to evaluate whether genetically determined circulating levels of cytokines are associated with the risk of IBD by using the MR approach. We selected genetic variants associated with circulating levels of 28 cytokines at the genome-wide significance level from a genome-wide association study (GWAS) including 8,293 individuals. Summary-level data for IBD (including Crohn's disease and ulcerative colitis) were obtained from the International Inflammatory Bowel Disease Genetics Consortium and UK Biobank. We performed the primary analysis using the inverse-variance weighted method, as well as sensitivity analyses to test the stability of our results. We subsequently replicated the results of IBD in the UK Biobank dataset. A reverse MR analysis was also conducted to evaluate the possibility of reverse causation. Genetically predicted elevated levels of interleukin-17 (IL-17) and monokine induced by interferon-gamma (MIG) were associated with an increased risk of IBD[odds ratio (OR): 1.52, 95% confidence interval (CI):1.10-2.08, P =0.010 for IL-17 and OR: 1.58, 95% CI: 1.24-2.00, P = 1.60×10-4 for MIG]. Moreover, we observed suggestive associations between β-NGF and MIP-1β with the risk of Crohn's disease (OR: 0.71, 95% CI: 0.52-0.98, P = 0.039) and ulcerative colitis (OR: 1.08, 95% CI: 1.01-1.15, P= 0.019). In the reverse MR study, there was no evidence of causal effects of IBD and these cytokines. Our study suggests the potential causal associations of IL-17 and MIG with IBD. Further studies are needed to determine whether IL-17 and MIG or their downstream effectors could be useful in the management of IBD.

Proteome-wide analysis reveals potential therapeutic targets for Colorectal cancer: a two-sample mendelian randomization study

BackgroundColorectal cancer (CRC) is a leading cause of cancer-related mortality, highlighting an unmet clinical need for more effective therapies. This study aims to evaluate the causal relationship between 4,489 plasma proteins and CRC to identify potential therapeutic targets for CRC.MethodsWe conducted two-sample Mendelian randomization (MR) analysis to examine the causal effects of plasma proteins on CRC. Mediation analysis was performed to assess the indirect effects of plasma proteins on CRC through associated risk factors. In addition, we conducted a phenome-wide association study using the UK Biobank dataset to examine associations between these plasma proteins and other phenotypes.ResultsOut of 4,489 plasma proteins, MR analysis revealed causal associations with CRC for 23 proteins, including VIMP, MICB, TNFRSF11B, C5orf38 and SLC5A8. Our findings also confirm the associations between reported risk factors and CRC. Mediation analysis identified mediating effects of proteins on CRC outcomes through risk factors. Furthermore, MR analysis identified 154 plasma proteins are causally linked to at least one CRC risk factor.ConclusionsOur study evaluated the causal relationships between plasma proteins and CRC, providing a more complete understanding of potential therapeutic targets for CRC.

A brain‐wide risk score of Alzheimer’s disease based on multimodal neuroimaging predicts cognition in large N&gt;37000 population data

AbstractBackgroundAlzheimer’s disease (AD) is the most frequent age‐related dementia impairing memory, thinking, and social skills. Much research has been conducted in recent years to investigate the genetic risk of AD, but little effort has been made to find a brain imaging‐based AD risk evaluation. Here, we proposed a multimodal neuroimaging‐based measure of AD risk, called brain‐wide risk score (BRS).MethodFunctional network connectivity (FNC) and voxelwise gray matter (GM) from the Open Access Series of Imaging Studies (OASIS)‐3 (N = 1302 samples) cohort was used as a reference dataset to define mild cognitive impairment (MCI) and control (CN)group to evaluate individuals in the UK Biobank (N = 37,784) dataset. We calculated the mean of FNC and GM of MCI and CN. Next, we removed the mean FNC and GM from each sample FNC and GM of the target dataset. Then, we calculated the distance between the target dataset and CN (i.e., dist_CN) and MCI (i.e., dist_MCI) of the reference dataset using Euclidian distance. Finally, we calculated Δdiff = dist_CN‐dist_MCI as a BRS of AD, where the positive values are associated with higher AD BRS. Using two BRSs estimated based on FNC and GM for target dataset, we identified four AD biotypes. Samples with two negative Δdist represent low AD risk (Biotype1), samples with two positive Δdist represent high AD risk (Biotype4), and samples with one negative and one positive Δdist represent moderate AD risk (Biotype2/3). Next, we compared cognitive scores including fluid intelligence (FI) and reaction time (RT) across four biotypes.ResultFig.1 shows the 2D histogram of two BRS estimated from FNC and GM. We identified 11704 participants as biotype1, 10910 as biotype2, 7804 as biotype3 and 7366 as biotype4. We also found that Biotype1, with lower AD BRS, significantly shows a higher FI score than Biotype3 and Biotype4 (Fig.2, FI panel). Additionally, Biotype 4, with the highest BRS, had the most increased RT among all four AD biotypes (Fig.2, RT panel).ConclusionWe develop a new multimodal neuroimaging‐based BRS based which we identified four AD biotypes. We found that BRS predict cognitive scores in the identified biotypes.

Deep learning-based phenotype imputation on population-scale biobank data increases genetic discoveries

Biobanks that collect deep phenotypic and genomic data across many individuals have emerged as a key resource in human genetics. However, phenotypes in biobanks are often missing across many individuals, limiting their utility. We propose AutoComplete, a deep learning-based imputation method to impute or ‘fill-in’ missing phenotypes in population-scale biobank datasets. When applied to collections of phenotypes measured across ~300,000 individuals from the UK Biobank, AutoComplete substantially improved imputation accuracy over existing methods. On three traits with notable amounts of missingness, we show that AutoComplete yields imputed phenotypes that are genetically similar to the originally observed phenotypes while increasing the effective sample size by about twofold on average. Further, genome-wide association analyses on the resulting imputed phenotypes led to a substantial increase in the number of associated loci. Our results demonstrate the utility of deep learning-based phenotype imputation to increase power for genetic discoveries in existing biobank datasets.

How far do we have to go on the road to calcium supplementation?

Sleep disorder is a growing concern, and calcium supplementation is often recommended as a potential intervention for sleep disorders. However, the causal relationship between calcium levels and the incidence of sleep disorders remains unclear. Mendelian randomization techniques utilizing genetic variants that affect calcium levels, can provide valuable insights into causality. This study aims to examine the association between calcium levels and sleep disorders in a diverse population that includes both adolescents and adults, and investigate the effects of calcium levels on sleep disorders. Mendelian randomization analysis was conducted using data from UK Biobank and FinnGen datasets. The inverse-variance weighting (IVW) was selected as the primary method. In addition, traditional mediation analysis was performed on a subset of the NHANES data spanning from 2007 to 2018. Our findings provide evidence supporting a causal relationship between calcium intake and reduced risk of sleep disorders (beta=-0.079, SE=0.0395, P=0.0457). While not reaching statistical significance, other MR methods such as weighted median and Mr-Egger exhibited similar directional trends. Analysis of the NHANES cohort revealed a negative association between calcium levels and the prevalence of sleep disorders in male, black, and physically active populations. However, this association was not observed in other demographic groups. Our results suggested that there is no significant correlation between calcium levels and sleep disorder in non-exercise populations. This raises concerns about the long-term high-dose calcium supplementation in clinical practice, which requires further investigation.

Large-scale genetic and multi-omics analyses identify molecular pathways underlying dilated cardiomyopathy

Abstract Background Dilated cardiomyopathy (DCM) is a heart muscle disease that represents a major public health burden. While both common and rare genetic variants are known to affect DCM risk, previous genome-wide association studies (GWAS) have yielded only few loci and the molecular pathways underlying DCM remain largely unknown. Purpose We aimed to discover novel genetic loci for DCM by combining data from biobank and case-control studies, and by combining GWAS with multi-trait analyses. We also pursued multi-omics Mendelian randomization to nominate novel molecular pathways underlying DCM. Methods and results We first performed a GWAS of ICD-code defined DCM (N=1472 cases) from two large biobank datasets (UK Biobank and Mass General Brigham Biobank), and found good genetic correlation (rg=0.82, SE=0.22) and similar heritability estimates (h2=0.12, SE=0.03 vs. h2=0.14, SE=0.03) when compared with a previously published GWAS of recruited DCM cases. In a large meta-analysis across case-control and biobank datasets (N=4191 cases), we identified 10 loci significantly associated with DCM. We then integrated this novel meta-analysis into a multi-trait GWAS (MTAG) with MRI-derived left ventricular traits from the UK Biobank (LVEF and LVESV; N=36083), which yielded 44 lead variants at 36 significant loci. Novel loci overlap reported Mendelian cardiomyopathy genes (ACTN2, OBSCN, MYH7B), support previous mechanistic findings (HSPB8; heat shock protein family), and include genomic regions also arising from a recent GWAS of hypertrophic cardiomyopathy (notably, SVIL). We then performed Mendelian randomization analyses on the MTAG GWAS, using genetic instruments from published multi-omics datasets. Mendelian randomisation of the blood metabolome suggests potential causal roles for metabolites, notably those involved in the pentose phosphate pathway (ribitol; β=0.25, SE=0.04) and glycemic excursion (1,5-anhydroglucitol; β=0.08, SE=0.02). Analyses of the blood proteome suggest potential causal roles for several serum proteins, including proteins involved in redox balance (glutathione synthetase; β=-0.34, SE=0.05), mTORC1 signalling (LAMTOR3; β=0.08, SE=0.01), and the pentose phosphate pathway (G6PE; β=0.09, SE=0.02). Conclusion Our analyses link several new genomic regions with DCM, and implicate redox disbalance and alternative usage of (gluco)metabolic pathways as putative causal mechanisms of disease. Future functional studies could determine whether these represent actionable findings.

Abstract 15806: An Age Index Derived From Heart Rate Metrics is Associated With Cardiovascular Risk Biomarkers

Introduction: Conventional cardiovascular (CV) risk scores depend on clinical and laboratory measurements, and therefore have limited accessibility in the general population. Consumer wearables provide heart rate measures which individually correlate with CV risk, but it is not known how well they align collectively with conventional clinical biomarkers. Objective: To demonstrate the alignment of a composite index based on heart rate metrics available from consumer wearables with clinical biomarkers and risk scores. Methods: As chronological age is a strong predictor of cardiovascular health, we trained a linear regression model on the UK Biobank (UKB) dataset to predict chronological age from sex, body mass index, resting heart rate (RHR), and mean HR, maximum HR, HR recovery and estimated VO2max during a cycle ergometry test. Data from 53,670 individuals aged 40-70 years were used: 30,247 for training and 23,423 for testing. A cardiovascular age (CVAge) index was obtained as the difference between chronological and predicted age of the individual: a positive/negative CVAge index indicates the CV health is better/worse than the age would suggest. Heart rate predictors and CVAge index were z-scored using age and sex data on the training set. Results: Individuals in the bottom quintile (“Poor CVAge index”) of the test set had higher systolic blood pressure (+5.0 mmHg), diastolic blood pressure (+3.0 mmHg), Framingham 10-year risk score (+1.8%) and ASCVD risk score (+0.8%), than those in the top quintile (“Very Good CVAge index”) - Figure 1. Conclusions: Continuous monitoring of heart rate metrics may help users track their heart health even in the absence of clinical and laboratory measurements.

Abstract 17610: A Novel Approach Using Deep Neural Networks on Clinical Cardiac MRIs Can Identify Patients With Clonal Hematopoiesis of Indeterminate Potential With High Accuracy

Background: Clonal hematopoiesis of indeterminate potential (CHIP) has been shown to be an independent risk factor for heart disease and heart failure. DNA sequencing is the gold standard to diagnose CHIP. It is unclear if deep learning techniques using routine clinical imaging can aid in identifying those who are likely to have CHIP and thus likely at increased risk for heart disease and heart failure. Hypothesis: We hypothesized that deep learning models trained on routine cardiac MRI (CMR) would be able to help predict who has CHIP Methods: We enrolled patients into an IRB approved study to evaluate for CHIP and associated imaging findings. Whole exome sequencing and our somatic variant pipeline was used to identify those who had CHIP. Various machine learning techniques were employed for training and testing, including SVM, random forest and convolutional neural networks (CNN) to predict who had CHIP from CMR imaging. Results: We enrolled 98 unique patients with CMR and found that 40 (41%) had CHIP. Majority were female 65 (66%) and there was no significant difference in median age between CHIP vs no CHIP (69, IQR 62-78 vs 65, IQR 54-72). There were no significant differences between CHIP vs no CHIP patients in co-morbidities such as CAD, DM, HTN, HF. Compared to SVM and random forest, the best performing model with high sensitivity, specificity and accuracy was derived from CNN (AUC 0.64 vs 0.71 vs 0.85 respectively). We found that multi-view MRI inputs were more accurate than single view MRI inputs for the CNN model. Using an unsupervised approach with CNN and multi-view input, we were able to predict CHIP with 82% accuracy. Conclusions: CNN is a promising deep learning technique that can accurately classify CHIP and thus those at higher cardiovascular disease risk. Ongoing analyses are underway to validate these findings using the UK Biobank and TOPMED datasets and predict who will develop cardiomyopathy with CHIP.

ExplaiNAble BioLogical Age (ENABL Age): an artificial intelligence framework for interpretable biological age

Biological age is a measure of health that offers insights into ageing. The existing age clocks, although valuable, often trade off accuracy and interpretability. We introduce ExplaiNAble BioLogical Age (ENABL Age), a computational framework that combines machine-learning models with explainable artificial intelligence (XAI) methods to accurately estimate biological age with individualised explanations. To construct the ENABL Age clock, we first predicted an age-related outcome (eg, all-cause or cause-specific mortality), and then rescaled these predictions to estimate biological age, using UK Biobank and National Health and Nutrition Examination Survey (NHANES) datasets. We adapted existing XAI methods to decompose individual ENABL Ages into contributing risk factors. For broad accessibility, we developed two versions: ENABL Age-L, based on blood tests, and ENABL Age-Q, based on questionnaire characteristics. Finally, we validated diverse ageing mechanisms captured by each ENABL Age clock through genome-wide association studies (GWAS) association analyses. Our ENABL Age clock was significantly correlated with chronological age (r=0·7867, p<0·0001 for UK Biobank; r=0·7126, p<0·0001 for NHANES). These clocks distinguish individuals who are healthy (ie, their ENABL Age is lower than their chronological age) from those who are unhealthy (ie, their ENABL Age is higher than their chronological age), predicting mortality more effectively than existing clocks. Groups of individuals who were unhealthy showed approximately three to 12 times higher log hazard ratio than healthy groups, as per ENABL Age. The clocks achieved high mortality prediction power with an area under the receiver operating characteristic curve of 0·8179 for 5-year mortality and 0·8115 for 10-year mortality on the UK Biobank dataset, and 0·8935 for 5-year mortality and 0·9107 for 10-year mortality on the NHANES dataset. The individualised explanations that revealed the contribution of specific characteristics to ENABL Age provided insights into the important characteristics for ageing. An association analysis with risk factors and ageing-related morbidities and GWAS results on ENABL Age clocks trained on different mortality causes showed that each clock captures distinct ageing mechanisms. ENABL Age brings an important leap forward in the application of XAI for interpreting biological age clocks. ENABL Age also carries substantial potential in practical settings, assisting medical professionals in untangling the complexity of ageing mechanisms, and potentially becoming a valuable tool in informed clinical decision-making processes. National Science Foundation and National Institutes of Health.

The association between thyroid and breast cancers: a bidirectional mendelian randomization study.

Thyroid and breast cancers are the two most frequent types of endocrine-related tumors among women worldwide, and their incidence is still on the rise. Observational studies have shown a relationship between thyroid and breast cancers. Nevertheless, many confounders predispose the results to interference. Accordingly, we performed a two-sample Mendelian randomization (MR) study to investigate the causal association between thyroid and breast cancers. We acquired breast cancer data from the UK Biobank (13,879 breast cancer cases and 198,523 controls) and the Breast Cancer Association Consortium (BCAC; 122,977 breast cancer cases and 105,974 controls), and thyroid cancer data from FinnGen Biobank (989 thyroid cancer and 217,803 controls). Then, the multiplicative random effects inverse variance weighting (IVW), weight median (WM), and MR Egger methods were executed for MR analysis. Overall, IVW showed a causal effect of breast cancer on thyroid cancer using the BCAC dataset (odds ratio [OR] = 1.17; 95% confidence interval [CI] = 1.036-1.322; P = 0.011), and this relationship was also supported by the UK Biobank dataset (OR = 23.899; 95% CI = 2.331-245.003; P = 0.007), which showed that breast cancer patients were more likely to be diagnosed with thyroid cancer. On the whole, the reverse MR analysis did not show a causal effect of breast cancer on thyroid cancer. However, IVW showed a causal effect of thyroid cancer on estrogen receptor -negative breast cancer using the BCAC dataset (OR = 1.019; 95% CI = 1.001-1.038; P = 0.043), which suggested that people with thyroid cancer were more likely to develop breast cancer. Breast cancer represents a possible risk factor for thyroid cancer and thyroid cancer also represents a possible risk factor for ER-negative breast cancer. Future studies using powerful genetic tools to determine the causal relationship between breast and thyroid cancers are required.

Investigating the Relationship Between Smoking Behavior and Global Brain Volume

BackgroundPrevious studies have shown that brain volume is negatively associated with cigarette smoking, but there is an ongoing debate about whether smoking causes lowered brain volume or a lower brain volume is a risk factor for smoking. We address this debate through multiple methods that evaluate directionality: Bradford Hill’s criteria, which are commonly used to understand a causal relationship in epidemiological studies, and mediation analysis. MethodsIn 32,094 participants of European descent from the UK Biobank dataset, we examined the relationship between a history of daily smoking and brain volumes, as well as an association of genetic risk score to ever smoking with brain volume. ResultsA history of daily smoking was strongly associated with decreased brain volume, and a history of heavier smoking was associated with a greater decrease in brain volume. The strongest association was between total gray matter volume and a history of daily smoking (effect size = −2964 mm3, p = 2.04 × 10−16), and there was a dose-response relationship with more pack years smoked associated with a greater decrease in brain volume. A polygenic risk score for smoking initiation was strongly associated with a history of daily smoking (effect size = 0.05, p = 4.20 × 10−84), but only modestly associated with total gray matter volume (effect size = −424 mm3, p = .01). Mediation analysis indicated that a history of daily smoking mediated the relationship between the smoking initiation polygenic risk score and total gray matter volume. ConclusionsA history of daily smoking is strongly associated with a decreased total brain volume.

Smoking and alcohol consumption with the risk of 11 common otolaryngological diseases: a bidirectional Mendelian randomization.

In this study, a bidirectional mendelian randomization was applied to evaluate the association of smoking and alcohol consumption with 11 otolaryngological diseases. A total of 85,22,34 and 7 single nucleotide polymorphisms were used as instrumental variables for smoking initiation, cigarettes per day, alcoholic drinks per week and alcohol consumption, respectively. Genetic associations with 11 common otolaryngological diseases were obtained from the UK Biobank and FinnGen dataset. IVW, weighted median, MR-Egger, MR-PRESSO and leave-one-out method were used in this analysis. Smoking initiation increased the risk of vocal cord and larynx diseases (OR 1.002; 95% CI 1.001-1.004; P = 4 × 10-4), head and neck cancer (OR 1.001; 95% CI 0.999-1.003; P = 0.027), thyroid cancer (OR 1.538; 95% CI 1.006-2.351; P = 0.047) and sleep apnoea (OR 1.286; 95% CI 1.099-1.506; P = 0.002). Cigarettes per day was associated with chronic sinusitis (OR 1.152; 95% CI 1.002-1.324; P = 0.046), chronic rhinitis and pharyngitis (OR 1.200; 95% CI 1.033-1.393; P = 0.017), vocal cord and larynx diseases (OR 1.001; 95% CI 0.999-1.002; P = 0.021) and head and neck cancer (OR 1.001; 95% CI 0.999-1.003; P = 0.017). Alcoholic drinks per week only was significantly associated with the risk of head and neck cancer (OR 1.003; 95% CI 1.001-1.006; P = 0.014). However, there was no evidence to support that genetically predicted alcohol consumption increased the risk of otolaryngological diseases. Reverse MR also did not find outcomes effect on exposures. This study shows that smoking and heavy alcohol consumption promote the occurrence of some otolaryngological diseases indicating that lifestyle modification might be beneficial in preventing otolaryngological diseases.

Lifestyle factors, serum parameters, metabolic comorbidities, and the risk of kidney stones: a Mendelian randomization study.

The early identification of modifiable risk factors is important for preventing kidney stones but determining causal associations can be difficult with epidemiological data. We aimed to genetically assess the causality between modifiable factors (lifestyle factors, serum parameters, and metabolic comorbidities) and the risk of kidney stones. Additionally, we aimed to explore the causal impact of education on kidney stones and its potential mediating pathways. We conducted a two-sample Mendelian randomization (MR) study to explore the causal association between 44 modifiable risk factors and kidney stones. The FinnGen dataset initially explored the causal relationship of risk factors with kidney stones and the UK Biobank dataset was used as the validation set. Then, a meta-analysis was conducted by combining discovery and validation datasets. We used two-step MR to assess potential mediators and their mediation proportions between education and kidney stones. The combined results indicated that previous exposures may increase the risk of kidney stones, including sedentary behavior, urinary sodium, the urinary sodium/potassium ratio, the urinary sodium/creatinine ratio, serum calcium, 25-hydroxyvitamin D (25OHD), the estimated creatinine-based glomerular filtration rate (eGFRcrea), GFR estimated by serum cystatin C (eGFRcys), body mass index (BMI), waist circumference, type 2 diabetes mellitus (T2DM), fasting insulin, glycated hemoglobin, and hypertension. Coffee intake, plasma caffeine levels, educational attainment, and the urinary potassium/creatinine ratio may decrease the risk of kidney stones. Ranked by mediation proportion, the effect of education on the risk of kidney stones was mediated by five modifiable risk factors, including sedentary behavior (mediation proportion, 25.7%), smoking initiation (10.2%), BMI (8.2%), T2DM (5.8%), and waist circumference (3.2%). This study provides MR evidence supporting causal associations of many modifiable risk factors with kidney stones. Sedentary lifestyles, obesity, smoking, and T2DM are mediating factors in the causal relationship between educational attainment and kidney stones. Our results suggest more attention should be paid to these modifiable factors to prevent kidney stones.

Personalized prediction for multiple chronic diseases by developing the multi-task Cox learning model.

Personalized prediction of chronic diseases is crucial for reducing the disease burden. However, previous studies on chronic diseases have not adequately considered the relationship between chronic diseases. To explore the patient-wise risk of multiple chronic diseases, we developed a multitask learning Cox (MTL-Cox) model for personalized prediction of nine typical chronic diseases on the UK Biobank dataset. MTL-Cox employs a multitask learning framework to train semiparametric multivariable Cox models. To comprehensively estimate the performance of the MTL-Cox model, we measured it via five commonly used survival analysis metrics: concordance index, area under the curve (AUC), specificity, sensitivity, and Youden index. In addition, we verified the validity of the MTL-Cox model framework in the Weihai physical examination dataset, from Shandong province, China. The MTL-Cox model achieved a statistically significant (p<0.05) improvement in results compared with competing methods in the evaluation metrics of the concordance index, AUC, sensitivity, and Youden index using the paired-sample Wilcoxon signed-rank test. In particular, the MTL-Cox model improved prediction accuracy by up to 12% compared to other models. We also applied the MTL-Cox model to rank the absolute risk of nine chronic diseases in patients on the UK Biobank dataset. This was the first known study to use the multitask learning-based Cox model to predict the personalized risk of the nine chronic diseases. The study can contribute to early screening, personalized risk ranking, and diagnosing of chronic diseases.

Proteome-wide mendelian randomization study implicates therapeutic targets in common cancers

BackgroundThe interest in targeted cancer therapies has been growing rapidly. While numerous cancer biomarkers and targeted treatment strategies have been developed and employed, there are still significant limitations and challenges in the early diagnosis and targeted treatment of cancers. Accordingly, there is an urgent need to identify novel targets and develop new targeted drugs.MethodsThe study was conducted using combined cis-Mendelian randomization (cis-MR) and colocalization analysis. We analyzed data from 732 plasma proteins to identify potential drug targets associated with eight site-specific cancers. These findings were further validated using the UK Biobank dataset. Then, a protein–protein interaction network was also constructed to examine the interplay between the identified proteins and the targets of existing cancer medications.ResultsThis MR analysis revealed associations between five plasma proteins and prostate cancer, five with breast cancer, and three with lung cancer. Subsequently, these proteins were classified into four distinct target groups, with a focus on tier 1 and 2 targets due to their higher potential to become drug targets. Our study indicatied that genetically predicted KDELC2 (OR: 0.89, 95% CI 0.86–0.93) and TNFRSF10B (OR: 0.74, 95% CI 0.65–0.83) are inversely associated with prostate cancer. Furthermore, we observed an inverse association between CPNE1 (OR: 0.96, 95% CI 0.94–0.98) and breast cancer, while PDIA3 (OR: 1.19, 95% CI 1.10–1.30) were found to be associated with the risk of breast cancer. In addition, we also propose that SPINT2 (OR: 1.05, 95% CI 1.03–1.06), GSTP1 (OR: 0.82, 95% CI 0.74–0.90), and CTSS (OR: 0.91, 95% CI 0.88–0.95) may serve as potential therapeutic targets in prostate cancer. Similarly, GDI2 (OR: 0.85, 95% CI 0.80–0.91), ISLR2 (OR: 0.87, 95% CI 0.82–0.93), and CTSF (OR: 1.14, 95% CI 1.08–1.21) could potentially be targets for breast cancer. Additionally, we identified SFTPB (OR: 0.93, 95% CI 0.91–0.95), ICAM5 (OR: 0.95, 95% CI 0.93–0.97), and FLRT3 (OR: 1.10, 95% CI 1.05–1.15) as potential targets for lung cancer. Notably, TNFRSF10B, GSTP1, and PDIA3 were found to interact with the target proteins of current medications used in prostate or breast cancer treatment.ConclusionsThis comprehensive analysis has highlighted thirteen plasma proteins with potential roles in three site-specific cancers. Continued research in this area may reveal their therapeutic potential, particularly KDELC2, TNFRSF10B, CPNE1, and PDIA3, paving the way for more effective cancer treatments.

Revolutionizing Cancer Research: The Impact of Artificial Intelligence in Digital Biobanking.

Biobanks are vital research infrastructures aiming to collect, process, store, and distribute biological specimens along with associated data in an organized and governed manner. Exploiting diverse datasets produced by the biobanks and the downstream research from various sources and integrating bioinformatics and "omics" data has proven instrumental in advancing research such as cancer research. Biobanks offer different types of biological samples matched with rich datasets comprising clinicopathologic information. As digital pathology and artificial intelligence (AI) have entered the precision medicine arena, biobanks are progressively transitioning from mere biorepositories to integrated computational databanks. Consequently, the application of AI and machine learning on these biobank datasets holds huge potential to profoundly impact cancer research. In this paper, we explore how AI and machine learning can respond to the digital evolution of biobanks with flexibility, solutions, and effective services. We look at the different data that ranges from specimen-related data, including digital images, patient health records and downstream genetic/genomic data and resulting "Big Data" and the analytic approaches used for analysis. These cutting-edge technologies can address the challenges faced by translational and clinical research, enhancing their capabilities in data management, analysis, and interpretation. By leveraging AI, biobanks can unlock valuable insights from their vast repositories, enabling the identification of novel biomarkers, prediction of treatment responses, and ultimately facilitating the development of personalized cancer therapies. The integration of biobanking with AI has the potential not only to expand the current understanding of cancer biology but also to pave the way for more precise, patient-centric healthcare strategies.

O.4.2-9 Shared genetic factors may partly explain the associations between physical activity and cardiometabolic diseases

Abstract Shared genetic factors may contribute to the associations between higher levels of physical activity (PA) and lower risk for cardiometabolic diseases (CMDs), and may partially explain these associations observed in cohort studies. To explore this, we used novel methodology to calculate PA genotypes (polygenic risk score, PRS) and validated them against measured or reported PA in three independent cohorts. We then investigated the associations between polygenic inheritance of PA and cardiometabolic risk factors and diseases in two large population-based biobank datasets, and examined whether selected associations were independent of self-reported PA. Our study utilized the UK Biobank as a base dataset (N = 400,124) and constructed genomewide PRSs for both self-reported and device-measured PA using single nucleotide polymorphism (SNP)-specific weights and SBayesR methodology. Both PRSs for PA included over one million SNPs. PRSs were constructed in the Finnish Twin cohort (N = 759–11,528), the Northern Finland Birth Cohort 1966 (N = 3,263–4,061), the Trøndelag Health Study cohort (HUNT, N = 47,148), and the FinnGen (N = 218,792). Cardiometabolic risk factors were measured in laboratory conditions, and CMD outcomes were derived from national health registers (ICD codes). We utilized linear, logistic, and cox regression methods for analysis. Our results showed that genotypes predisposing to higher PA were associated with higher levels of PA in independent datasets, but PRSs accounted for only a limited amount of variation (0.13-1.44%). Genotypes supporting higher PA were associated with lower body mass index [B=-0.002 in HUNT and B=-0.025 in FinnGen] and favorable cardiometabolic health in HUNT (waist circumference [B=-0.003] and HDL cholesterol [B = 0.004]). Genotypes supporting higher PA volumes were associated with lower incidence of CMDs in both HUNT and FinnGen. The strongest associations were found in hypertensive diseases and Type 2 Diabetes. In HUNT, the observed associations were not materially changed after accounting for self-reported PA. Higher PRS for PA was also associated with lower risk of mortality in FinnGen. Our findings suggest small pleiotropic effects between PA and CMDs. This means that same genetic variation may explain both physical activity behaviour and risk of diseases. PRSs provide new tools for genetic studies in sport science, but they currently have substantial practical limitations.

Long-term outcomes of offspring from multiple gestations: a two-sample Mendelian randomization study on multi-system diseases using UK Biobank and FinnGen databases

BackgroundAssisted reproductive technologies (ART) have increased the incidence of multiple births, which can have a negative impact on maternal and offspring health. The study aimed to investigate the association between genetically predicted multiple birth and the risk of 42 common diseases of the nervous, psychiatric, cardiovascular, respiratory, digestive, and endocrine systems.MethodsThe study utilized two-sample Mendelian randomization (MR) analysis to explore the potential causal relationship between genetically predicted multiple birth and the genetically predicted risk of diseases. The study used the FinnGen and UK Biobank datasets for analysis.ResultsThe study found no significant causal relationship between multiple birth and psychiatric disorders. However, the lower limits of the 95% confidence intervals for bipolar affective disorder and anxiety disorders were not robust, indicating a need for further investigation. The study found that multiple birth may be a strong risk factor for infantile cerebral palsy, and caution is necessary in both natural and ART multiple births. The study revealed a potential causal relationship between multiple birth and coronary heart disease, ischemic heart disease, and deep vein thrombosis, which may be related to abnormal intrauterine environments in multiple pregnancies. Surprisingly, multiple birth appears to have a protective effect against some respiratory diseases, such as chronic obstructive pulmonary disease and asthma.ConclusionsThe study highlights the need for caution regarding the risk of infantile cerebral palsy, cardiovascular diseases, and psychiatric disorders in multiple birth. Our study can lead to the development of preventive strategies and improved clinical management for affected infants.

Gene-based burden scores identify rare variant associations for 28 blood biomarkers

BackgroundA relevant part of the genetic architecture of complex traits is still unknown; despite the discovery of many disease-associated common variants. Polygenic risk score (PRS) models are based on the evaluation of the additive effects attributable to common variants and have been successfully implemented to assess the genetic susceptibility for many phenotypes. In contrast, burden tests are often used to identify an enrichment of rare deleterious variants in specific genes. Both kinds of genetic contributions are typically analyzed independently. Many studies suggest that complex phenotypes are influenced by both low effect common variants and high effect rare deleterious variants. The aim of this paper is to integrate the effect of both common and rare functional variants for a more comprehensive genetic risk modeling.MethodsWe developed a framework combining gene-based scores based on the enrichment of rare functionally relevant variants with genome-wide PRS based on common variants for association analysis and prediction models. We applied our framework on UK Biobank dataset with genotyping and exome data and considered 28 blood biomarkers levels as target phenotypes. For each biomarker, an association analysis was performed on full cohort using gene-based scores (GBS). The cohort was then split into 3 subsets for PRS construction and feature selection, predictive model training, and independent evaluation, respectively. Prediction models were generated including either PRS, GBS or both (combined).ResultsAssociation analyses of the cohort were able to detect significant genes that were previously known to be associated with different biomarkers. Interestingly, the analyses also revealed heterogeneous effect sizes and directionality highlighting the complexity of the blood biomarkers regulation. However, the combined models for many biomarkers show little or no improvement in prediction accuracy compared to the PRS models.ConclusionThis study shows that rare variants play an important role in the genetic architecture of complex multifactorial traits such as blood biomarkers. However, while rare deleterious variants play a strong role at an individual level, our results indicate that classical common variant based PRS might be more informative to predict the genetic susceptibility at the population level.