Polygenic scores in biomedical research

Optimization of multi-ancestry polygenic risk score disease prediction models

Polygenic risk scores (PRS) have ushered in a new era in genetic epidemiology, offering insights into individual predispositions to a wide range of diseases. However, despite recent marked enhancements in predictive power, PRS-based models still need to overcome several hurdles before they can be broadly applied in the clinic. Chiefly, they need to achieve sufficient accuracy, easy interpretability and portability across diverse populations. Leveraging trans-ancestry genome-wide association study (GWAS) meta-analysis, we generated novel, diverse summary statistics for 30 medically-related traits and benchmarked the performance of six existing PRS algorithms using UK Biobank. We built an ensemble model using logistic regression to combine outputs of top-performing algorithms and validated it on the diverse eMERGE and PAGE MEC cohorts. It surpassed current state-of-the-art PRS models, with minimal performance drops in external cohorts, indicating good calibration. To enhance predictive accuracy for clinical application, we incorporated easily-accessible clinical characteristics such as age, gender, ancestry and risk factors, creating disease prediction models intended as prospective diagnostic tests, with easily interpretable positive or negative outcomes. After adding clinical characteristics, 12 out of 30 models surpassed 80% AUC. Further, 25 traits exceeded the diagnostic odds ratio (DOR) of five, and 19 traits exceeded DOR of 10 for all ancestry groups, indicating high predictive value. Our PRS model for coronary artery disease identified 55-80 times more true coronary events than rare pathogenic variant models, reinforcing its clinical potential. The polygenic component modulated the effect of high-risk rare variants, stressing the need to consider all genetic components in clinical settings. These findings show that newly developed PRS-based disease prediction models have sufficient accuracy and portability to warrant consideration of being used in the clinic.

Methodological opportunities in genomic data analysis to advance health equity

The causes and consequences of inequities in genomic research and medicine are complex and widespread. However, it is widely acknowledged that underrepresentation of diverse populations in human genetics research risks exacerbating existing health disparities. Efforts to improve diversity are ongoing, but an often-overlooked source of inequity is the choice of analytical methods used to process, analyse and interpret genomic data. This choice can influence all areas of genomic research, from genome-wide association studies and polygenic score development to variant prioritization and functional genomics. New statistical and machine learning techniques to understand, quantify and correct for the impact of biases in genomic data are emerging within the wider genomic research and genomic medicine ecosystems. At this crucial time point, it is important to clarify where improvements in methods and practices can, or cannot, have a role in improving equity in genomics. Here, we review existing approaches to promote equity and fairness in statistical analysis for genomics, and propose future methodological developments that are likely to yield the most impact for equity.

Overt and covert genetic causes of pediatric acute lymphoblastic leukemia

Pediatric acute lymphoblastic leukemia (pALL) is the most common childhood malignancy, yet its etiology remains incompletely understood. However, over the course of three waves of germline genetic research, several non-environmental causes have been identified. Beginning with trisomy 21, seven overt cancer predisposition syndromes (CPSs)-characterized by broad clinical phenotypes that include an elevated risk of pALL-were first described. More recently, newly described CPSs conferring high risk of pALL are increasingly covert, with six exhibiting only minimal or no non-cancer features. These 13 CPSs now represent the principal known hereditary causes of pALL, and human pangenomic data indicates a strong negative selection against mutations in the genes associated with these conditions. Collectively they affect approximately 1 in 450 newborns, of which just a minority will develop the disease. As evidenced by tailored leukemia care protocols for children with trisomy 21, there is growing recognition that CPSs warrant specialized diagnostic, therapeutic, and long-term management strategies. In this review, we investigate the evidence that the 12 other CPSs associated with high risk of pALL may also see benefits from specialized care - even if these needs are often incompletely mapped or addressed in the clinic. Given the rarity of each syndrome, collaborative international research and shared data initiatives will be crucial for advancing knowledge and improving outcomes for these patients.

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.

Effect of Disclosing a Polygenic Risk Score for Coronary Heart Disease on Adverse Cardiovascular Events

BACKGROUND

In the Myocardial Infarction Genes clinical trial (URL: https://www.clinicaltrials.gov; Unique identifier: NCT01936675), participants at intermediate risk of coronary heart disease (CHD) were randomized to receive a Framingham risk score (Framingham risk score group, n=103) or an integrated risk score (integrated risk score group [IRSg], n=104) that additionally included a polygenic risk score. After 6 months, IRSg participants had higher statin initiation and lower low-density lipoprotein cholesterol. We conducted a post hoc 10-year follow-up analysis to investigate whether disclosure of a polygenic risk score for CHD was associated with a reduction in major adverse cardiovascular events (MACE).

METHODS

Participants were followed from randomization in October 2013 to September 2023 to ascertain MACE, testing for CHD, and changes in risk factors. The primary outcome was time to first MACE, defined as cardiovascular death, nonfatal myocardial infarction, coronary revascularization, and nonfatal stroke. Statistical analyses included Cox proportional hazards regression and linear mixed-effects models.

RESULTS

We followed all participants who completed the trial, 100 in Framingham risk score group and 103 in IRSg (mean age at the end of follow-up, 68.2±5.2; 48% male). During a median follow-up of 9.5 years, 9 MACEs occurred in Framingham risk score group and 2 in IRSg (hazard ratio, 0.20 [95% CI, 0.04-0.94]; P=0.042). In Framingham risk score group, 47 (47%) underwent at least 1 diagnostic test for CHD, compared with 30 (29%) in IRSg (hazard ratio, 0.51 [95% CI, 0.32-0.81]; P=0.004). A higher proportion of IRSg participants were on statin therapy during the first 4 years postrandomization and had a greater reduction in low-density lipoprotein cholesterol for up to 3 years postrandomization. No significant differences were observed between 2 groups in other traditional cardiovascular risk factors during follow-up.

CONCLUSIONS

Disclosure of an integrated risk score that included a polygenic risk score to individuals at intermediate risk for CHD was associated with lower MACE incidence after 10 years, likely due to higher statin initiation, leading to lower low-density lipoprotein cholesterol levels.

Unsupervised Ensemble Learning for Efficient Integration of Pre-trained Polygenic Risk Scores

The growing availability of pre-trained polygenic risk score (PRS) models has enabled their integration into real-world applications, reducing the need for extensive data labeling, training, and calibration. However, selecting the most suitable PRS model for a specific target population remains challenging, due to issues such as limited transferability, data heterogeneity, and the scarcity of observed phenotype in real-world settings. Ensemble learning offers a promising avenue to enhance the predictive accuracy of genetic risk assessments, but most existing methods often rely on observed phenotype data or additional genome-wide association studies (GWAS) from the target population to optimize ensemble weights, limiting their utility in real-time implementation. Here, we present the UNSupervised enSemble PRS (UNSemblePRS), an unsupervised ensemble learning framework, that combines pre-trained PRS models without requiring phenotype data or summaries from the target population. Unlike traditional supervised approaches, UNSemblePRS aggregates models based on prediction concordance across a curated subset of candidate PRS models. We evaluated UNSemblePRS using both continuous and binary traits in the All of Us database, demonstrating its scalability and robust performance across diverse populations. These results underscore UNSemblePRS as an accessible tool for integrating PRS models into real-world contexts, offering broad applicability as the availability of PRS models continues to expand.

Unsupervised Ensemble Learning for Efficient Integration of Pre-trained Polygenic Risk Scores

The growing availability of pre-trained polygenic risk score (PRS) models has enabled their integration into real-world applications, reducing the need for extensive data labeling, training, and calibration. However, selecting the most suitable PRS model for a specific target population remains challenging, due to issues such as limited transferability, data heterogeneity, and the scarcity of observed phenotype in real-world settings. Ensemble learning offers a promising avenue to enhance the predictive accuracy of genetic risk assessments, but most existing methods often rely on observed phenotype data or additional genome-wide association studies (GWAS) from the target population to optimize ensemble weights, limiting their utility in real-time implementation. Here, we present the UN supervised en Semble PRS ( UNSemblePRS ), an unsupervised ensemble learning framework, that combines pre-trained PRS models without requiring phenotype data or summaries from the target population. Unlike traditional supervised approaches, UNSemblePRS aggregates models based on prediction concordance across a curated subset of candidate PRS models. We evaluated UNSemblePRS using both continuous and binary traits in the All of Us database, demonstrating its scalability and robust performance across diverse populations. These results underscore UNSemblePRS as an accessible tool for integrating PRS models into real-world contexts, offering broad applicability as the availability of PRS models continues to expand.

Differential performance of polygenic prediction across traits and populations depending on genotype discovery approach

Polygenic scores (PGS) are widely used for estimating genetic predisposition to complex traits by aggregating the effects of common variants into a single measure. They hold promise in identifying individuals at increased risk for diseases, allowing earlier screening and interventions. Genotyping arrays, commonly used for PGS computation, are affordable and computationally efficient, while whole-genome sequencing (WGS) offers a comprehensive view of genetic variation. Using the same set of individuals, we compared PGS derived from arrays and WGS across multiple traits to evaluate differences in predictive performance, portability across populations, and computational efficiency. We computed PGS for 10 traits across the spectrum of heritability and polygenicity in the three largest genetic ancestry groups in All of Us (European, African American, Admixed American), trained on the multi-ancestry meta-analyses from the Pan-UK Biobank. Using the clumping and thresholding (C+T) method, we found that WGS-based PGS outperformed array-based PRS for highly polygenic traits but showed differentially reduced accuracy for sparse traits in certain populations. This may be attributable to the lower allele frequency observed in clumped variants from WGS compared to arrays. Using the LD-informed PRS-CS method, we observed overall improved prediction performance compared to C+T, with WGS outperforming arrays across most non-cancer traits. In conclusion, while PGS computed using WGS generally provide superior predictive power with PRS-CS, the advantage over arrays is context-dependent, varying by trait, population, and the PGS method. This study provides insights into the complexities and potential advantages of using different genotype discovery approach for polygenic predictions in diverse populations.

Graphical abstract

Influence and role of polygenic risk score in the development of 32 complex diseases

Background

The polygenic risk score (PRS) has been perceived as advantageous in predicting the risk of complex diseases compared to other measures. We aimed to systematically evaluate the influence of PRS on disease outcome and to explore its predictive value.

Methods

We comprehensively assessed the relationship between PRS and 32 complex diseases in the UK Biobank. We used Cox models to estimate the effects of PRS on the incidence risk. Then, we constructed prediction models to assess the clinical utility of PRS in risk prediction. For 16 diseases, we further compared the disease risk and prediction capability of PRS across early and late-onset cases.

Results

Higher PRS led to greater incident risk, with hazard ratio (HR) ranging from 1.07 (95% confidence interval (CI) = 1.06-1.08) for panic/anxiety disorder to 4.17 (95% CI = 4.03-4.31) for acute pancreatitis. This effect was more pronounced in early-onset cases for 12 diseases, increasing by 52.8% on average. Particularly, the early-onset risk of heart failure associated with PRS (HR = 3.02; 95% CI = 2.53-3.59) was roughly twice compared to the late-onset risk (HR = 1.48; 95% CI = 1.46-1.51). Compared to average PRS (20-80%), individuals positioned within the top 2.5% of the PRS distribution exhibited varying degrees of elevated risk, corresponding to a more than five times greater risk on average. PRS showed additional value in clinical risk prediction, causing an average improvement of 6.1% in prediction accuracy. Further, PRS demonstrated higher predictive accuracy for early-onset cases of 11 diseases, with heart failure displaying the most significant (37.5%) improvement when incorporating PRS into the prediction model (concordance index (C-index) = 0.546; standard error (SE) = 0.011 vs. C-index = 0.751; SE = 0.010, P = 2.47 × 10-12).

Conclusions

As a valuable complement to traditional clinical risk tools, PRS is closely related to disease risk and can further enhance prediction accuracy, especially for early-onset cases, underscoring its potential role in targeted prevention for high-risk groups.

Polygenic Risk Score Added to Conventional Case Finding to Identify Undiagnosed Chronic Obstructive Pulmonary Disease

Importance

Chronic obstructive pulmonary disease (COPD) is often undiagnosed. Although genetic risk plays a significant role in COPD susceptibility, its utility in guiding spirometry testing and identifying undiagnosed cases is unclear.

Objective

To determine whether a COPD polygenic risk score (PRS) enhances the identification of undiagnosed COPD beyond a case-finding questionnaire (eg, the Lung Function Questionnaire) using conventional risk factors and respiratory symptoms.

Design, Setting, and Participants

This cross-sectional analysis of participants 35 years or older who reported no history of physician-diagnosed COPD was conducted using data from 2 observational studies: the community-based Framingham Heart Study (FHS) and the COPD-enriched Genetic Epidemiology of COPD (COPDGene) study.

Exposures

Modified Lung Function Questionnaire (mLFQ) scores and COPD PRS.

Main Outcomes and Measures

The primary outcome was spirometry-defined moderate to severe COPD (forced expiratory volume in the first second of expiration/forced vital capacity [FEV1/FVC] <0.7 and FEV1 [percent predicted] <80%). The performance of logistic models was assessed using the PRS, mLFQ score, and PRS plus mLFQ score for predicting spirometry-defined COPD.

Results

Among 3385 FHS participants (median age, 52.0 years; 45.9% male) and 4095 COPDGene participants (median age, 56.8 years; 55.5% male) who reported no history of COPD, 160 (4.7%) FHS and 775 (18.9%) COPDGene participants had spirometry-defined COPD. Adding the PRS to the mLFQ score significantly improved the area under the curve from 0.78 to 0.84 (P < .001) in FHS, 0.69 to 0.72 (P = .04) in COPDGene non-Hispanic African American, and 0.75 to 0.78 (P < .001) in COPDGene non-Hispanic White participants. At a risk threshold for spirometry referral of 10%, the addition of the PRS to the mLFQ score correctly reclassified 13.8% (95% CI, 6.6%-21.0%) of COPD cases in FHS, but not in COPDGene.

Conclusions and Relevance

A COPD PRS enhances the identification of undiagnosed COPD beyond a conventional case-finding approach in the general population. Further research is needed to assess its impact on COPD diagnosis and outcomes.

PennPRS: a centralized cloud computing platform for efficient polygenic risk score training in precision medicine

Polygenic risk scores (PRS) are becoming increasingly vital for risk prediction and stratification in precision medicine. However, PRS model training presents significant challenges for broader adoption of PRS, including limited access to computational resources, difficulties in implementing advanced PRS methods, and availability and privacy concerns over individual-level genetic data. Cloud computing provides a promising solution with centralized computing and data resources. Here we introduce PennPRS (https://pennprs.org), a scalable cloud computing platform for online PRS model training in precision medicine. We developed novel pseudo-training algorithms for multiple PRS methods and ensemble approaches, enabling model training without requiring individual-level data. These methods were rigorously validated through extensive simulations and large-scale real data analyses involving over 6,000 phenotypes across various data sources. PennPRS supports online single- and multi-ancestry PRS training with seven methods, allowing users to upload their own data or query from more than 27,000 datasets in the GWAS Catalog, submit jobs, and download trained PRS models. Additionally, we applied our pseudo-training pipeline to train PRS models for over 8,000 phenotypes and made their PRS weights publicly accessible. In summary, PennPRS provides a novel cloud computing solution to improve the accessibility of PRS applications and reduce disparities in computational resources for the global PRS research community.

Characterizing substructure via mixture modeling in large-scale genetic summary statistics

Genetic summary data are broadly accessible and highly useful, including for risk prediction, causal inference, fine mapping, and incorporation of external controls. However, collapsing individual-level data into summary data, such as allele frequencies, masks intra- and inter-sample heterogeneity, leading to confounding, reduced power, and bias. Ultimately, unaccounted-for substructure limits summary data usability, especially for understudied or admixed populations. There is a need for methods to enable the harmonization of summary data where the underlying substructure is matched between datasets. Here, we present Summix2, a comprehensive set of methods and software based on a computationally efficient mixture model to enable the harmonization of genetic summary data by estimating and adjusting for substructure. In extensive simulations and application to public data, we show that Summix2 characterizes finer-scale population structure, identifies ascertainment bias, and scans for potential regions of selection due to local substructure deviation. Summix2 increases the robust use of diverse, publicly available summary data, resulting in improved and more equitable research.

Understanding and applying gene-environment interactions: a guide for nutrition professionals with an emphasis on integration in African research settings

Noncommunicable diseases (NCDs) are influenced by the interplay between genetics and environmental exposures, particularly diet. However, many healthcare professionals, including nutritionists and dietitians, have limited genetic background and, therefore, they may lack understanding of gene-environment interactions (GxEs) studies. Even researchers deeply involved in nutrition studies, but with a focus elsewhere, can struggle to interpret, evaluate, and conduct GxE studies. There is an urgent need to study African populations that bear a heavy burden of NCDs, demonstrate unique genetic variability, and have cultural practices resulting in distinctive environmental exposures compared with Europeans or Americans, who are studied more. Although diverse and rapidly changing environments, as well as the high genetic variability of Africans and difference in linkage disequilibrium (ie, certain gene variants are inherited together more often than expected by chance), provide unparalleled potential to investigate the omics fields, only a small percentage of studies come from Africa. Furthermore, research evidence lags behind the practices of companies offering genetic testing for personalized medicine and nutrition. We need to generate more evidence on GxEs that also considers continental African populations to be able to prevent unethical practices and enable tailored treatments. This review aims to introduce nutrition professionals to genetics terms and valid methods to investigate GxEs and their challenges, and proposes ways to improve quality and reproducibility. The review also provides insight into the potential contributions of nutrigenetics and nutrigenomics to the healthcare sphere, addresses direct-to-consumer genetic testing, and concludes by offering insights into the field's future, including advanced technologies like artificial intelligence and machine learning.

Development and Evaluation of a Comprehensive Prediction Model for Incident Coronary Heart Disease Using Genetic, Social, and Lifestyle-Psychological Factors: A Prospective Analysis of the UK Biobank

BACKGROUND

Clinical risk calculators for coronary heart disease (CHD) do not include genetic, social, and lifestyle-psychological risk factors.

OBJECTIVE

To improve CHD risk prediction by developing and evaluating a prediction model that incorporated a polygenic risk score (PRS) and a polysocial score (PSS), the latter including social determinants of health and lifestyle-psychological factors.

DESIGN

Cohort study.

SETTING

United Kingdom.

PARTICIPANTS

UK Biobank participants recruited between 2006 and 2010.

MEASUREMENTS

Incident CHD (myocardial infarction and/or coronary revascularization); 10-year clinical risk based on pooled cohort equations (PCE), Predicting Risk of cardiovascular disease EVENTs (PREVENT), and QRISK3; PRS (Polygenic Score Catalog identification: PGS000018) for CHD (PRSCHD); and PSSCHD from 100 related covariates. Machine-learning and time-to-event analyses and model performance indices.

RESULTS

In 388 224 participants (age, 55.5 [SD, 8.1] years; 42.5% men; 94.9% White), the hazard ratio for 1 SD increase in PSSCHD for incident CHD was 1.43 (95% CI, 1.38 to 1.49; P < 0.001) and for 1 SD increase in PRSCHD was 1.59 (CI, 1.53 to 1.66, P < 0.001). Non-White persons had higher PSSCHD than White persons. The effects of PSSCHD and PRSCHD on CHD were independent and additive. At a 10-year CHD risk threshold of 7.5%, adding PSSCHD and PRSCHD to PCE reclassified 12% of participants, with 1.86 times higher CHD risk in the up- versus down-reclassified persons and showed superior performance compared with PCE as reflected by improved net benefit while maintaining good calibration relative to the clinical risk calculators. Similar results were seen when incorporating PSSCHD and PRSCHD into PREVENT and QRISK3.

LIMITATION

A predominantly White cohort; possible healthy participant effect and ecological fallacy.

CONCLUSION

A PSSCHD was associated with incident CHD and its joint modeling with PRSCHD improved the performance of clinical risk calculators.

PRIMARY FUNDING SOURCE

National Human Genome Research Institute.

The PRIMED Consortium: Reducing disparities in polygenic risk assessment

By improving disease risk prediction, polygenic risk scores (PRSs) could have a significant impact on health promotion and disease prevention. Due to the historical oversampling of populations with European ancestry for genome-wide association studies, PRSs perform less well in other, understudied populations, leading to concerns that clinical use in their current forms could widen health care disparities. The PRIMED Consortium was established to develop methods to improve the performance of PRSs in global populations and individuals of diverse genetic ancestry. To this end, PRIMED is aggregating and harmonizing multiple phenotype and genotype datasets on AnVIL, an interoperable secure cloud-based platform, to perform individual- and summary-level analyses using population and statistical genetics approaches. Study sites, the coordinating center, and representatives from the NIH work alongside other NHGRI and global consortia to achieve these goals. PRIMED is also evaluating ethical and social implications of PRS implementation and investigating the joint modeling of social determinants of health and PRS in computing disease risk. The phenotypes of interest are primarily cardiometabolic diseases and cancer, the leading causes of death and disability worldwide. Early deliverables of the consortium include methods for data sharing on AnVIL, development of a common data model to harmonize phenotype and genotype data from cohort studies as well as electronic health records, adaptation of recent guidelines for population descriptors to global cohorts, and sharing of PRS methods/tools. As a multisite collaboration, PRIMED aims to foster equity in the development and use of polygenic risk assessment.

Advancements and limitations in polygenic risk score methods for genomic prediction: a scoping review

This scoping review aims to identify and evaluate the landscape of Polygenic Risk Score (PRS)-based methods for genomic prediction from 2013 to 2023, highlighting their advancements, key concepts, and existing gaps in knowledge, research, and technology. Over the past decade, various PRS-based methods have emerged, each employing different statistical frameworks aimed at enhancing prediction accuracy, processing speed and memory efficiency. Despite notable advancements, challenges persist, including unrealistic assumptions regarding sample sizes and the polygenicity of traits necessary for accurate predictions, as well as limitations in exploring hyper-parameter spaces and considering environmental interactions. We included studies focusing on PRS-based methods for risk prediction that underwent methodological evaluations using valid approaches and released computational tools/software. Additionally, we restricted our selection to studies involving human participants that were published in English language. This review followed the standard protocol recommended by Joanna Briggs Institute Reviewer's Manual, systematically searching Ovid MEDLINE, Ovid Embase, Scopus and Web of Science databases. Additionally, searches included grey literature sources like pre-print servers such as bioRxiv, and articles recommended by experts to ensure comprehensive and diverse coverage of relevant records. This study identified 34 studies detailing 37 genomic prediction methods, the majority of which rely on linkage disequilibrium (LD) information and necessitate hyper-parameter tuning. Nine methods integrate functional/gene annotation, while 12 are suitable for cross-ancestry genomic prediction, with only one considering gene-environment (GxE) interaction. While some methods require individual-level data, most leverage summary statistics, offering flexibility. Despite progress, challenges remain. These include computational complexity and the need for large sample sizes for high prediction accuracy. Furthermore, recent methods exhibit varying effectiveness across traits, with absolute accuracies often falling short of clinical utility. Transferability across ancestries varies, influenced by trait heritability and diversity of training data, while handling admixed populations remains challenging. Additionally, the absence of standard error measurements for individual PRSs, crucial in clinical settings, underscores a critical gap. Another issue is the lack of customizable graphical visualization tools among current software packages. While genomic prediction methods have advanced significantly, there is still room for improvement. Addressing current challenges and embracing future research directions will lead to the development of more universally applicable, robust, and clinically relevant genomic prediction tools.

Associations of Self-Reported Race, Social Determinants of Health, and Polygenic Risk With Coronary Heart Disease

BACKGROUND

Social determinants of health (SDOH) influence the risk of common diseases such as coronary heart disease (CHD).

OBJECTIVES

This study sought to test the associations of self-reported race/ethnicity, SDOH, and a polygenic risk score (PRS), with CHD in a large and diverse U.S.

COHORT

METHODS

In 67,256 All of Us (AoU) participants with available SDOH and whole-genome sequencing data, we ascertained self-reported race/ethnicity and 22 SDOH measures across 5 SDOH domains, and we calculated a PRS for CHD (PRSCHD, PGS004696). We developed an SDOH score for CHD (SDOHCHD). We tested the associations of SDOH and PRSCHD with CHD in regression models that included clinical risk factors.

RESULTS

SDOH across 5 domains, including food insecurity, income, educational attainment, health literacy, neighborhood disorder, and loneliness, were associated with CHD. SDOHCHD was highest in self-reported Black and Hispanic people. Self-reporting as Blacks had higher odds of having CHD than Whites but not after adjustment for SDOHCHD. SDOHCHD and PRSCHD were weakly correlated. In the test set (n = 33,628), 1-SD increases in SDOHCHD and PRSCHD were associated with CHD in models that adjusted for clinical risk factors (OR: 1.32; 95% CI: 1.23-1.41 and OR: 1.36; 95% CI: 1.28-1.44, respectively). SDOHCHD and PRSCHD were associated with incident CHD events (n = 52) over a median follow-up of 214 days (Q1-Q3: 88 days).

CONCLUSIONS

Increased odds of CHD in people who self-report as Black are likely due to a higher SDOH burden. SDOH and PRS were independently associated with CHD. Our findings suggest that including both PRS and SDOH in CHD risk models could improve their accuracy.

Polygenic prediction for underrepresented populations through transfer learning by utilizing genetic similarity shared with European populations

Because current genome-wide association studies are primarily conducted in individuals of European ancestry and information disparities exist among different populations, the polygenic score derived from Europeans thus exhibits poor transferability. Borrowing the idea of transfer learning, which enables the utilization of knowledge acquired from auxiliary samples to enhance learning capability in target samples, we propose transPGS, a novel polygenic score method, for genetic prediction in underrepresented populations by leveraging genetic similarity shared between the European and non-European populations while explaining the trans-ethnic difference in linkage disequilibrium (LD) and effect sizes. We demonstrate the usefulness and robustness of transPGS in elevated prediction accuracy via individual-level and summary-level simulations and apply it to seven continuous phenotypes and three diseases in the African, Chinese, and East Asian populations of the UK Biobank and Genetic Epidemiology Research Study on Adult Health and Aging cohorts. We further reveal that distinct LD and minor allele frequency patterns across ancestral groups are responsible for the dissatisfactory portability of PGS.

Nightingale Health Biobank Collaborative Group, Barrett JC, Esko T, Fischer K, Jostins-Dean L, Jousilahti P, Julkunen H, Jääskeläinen T, Kangas A, Kerimov N, Kerminen S, Kolde A, Koskela H, Kronberg J, Lundgren SN, Lundqvist A, Mäkelä V, Nybo K, Perola M, Salomaa V, Schut K, Soikkeli M, Soininen P, Tiainen M, Tillmann T, Würtz P. Metabolomic and genomic prediction of common diseases in 700,217 participants in three national biobanks

Identifying individuals at high risk of chronic diseases via easily measured biomarkers could enhance efforts to prevent avoidable illness and death. Using 'omic data can stratify risk for many diseases simultaneously from a single measurement that captures multiple molecular predictors of risk. Here we present nuclear magnetic resonance metabolomics in blood samples from 700,217 participants in three national biobanks. We built metabolomic scores that identify high-risk groups for diseases that cause the most morbidity in high-income countries and show consistent cross-biobank replication of the relative risk of disease for these groups. We show that these metabolomic scores are more strongly associated with disease onset than polygenic scores for most of these diseases. In a subset of 18,709 individuals with metabolomic biomarkers measured at two time points we show that people whose scores change have different risk of disease, suggesting that repeat measurements capture changes both to health status and disease risk possibly due to treatment, lifestyle changes or other factors. Lastly, we assessed the incremental predictive value of metabolomic scores over existing clinical risk scores for multiple diseases and found modest improvements in discrimination for several diseases whose clinical utility, while promising, remains to be determined.

Estimating the Number of Polygenic Diseases Among Six Mutually Exclusive Entities of Non-Tumors and Cancer

In the era of precision medicine with increasing amounts of sequenced cancer and non-cancer genomes of different ancestries, we here enumerate the resulting polygenic disease entities. Based on the cell number status, we first identified six fundamental types of polygenic illnesses, five of which are non-cancerous. Like complex, non-tumor disorders, neoplasms normally carry alterations in multiple genes, including in 'Drivers' and 'Passengers'. However, tumors also lack certain genetic alterations/epigenetic changes, recently named 'Goners', which are toxic for the neoplasm and potentially constitute therapeutic targets. Drivers are considered essential for malignant transformation, whereas environmental influences vary considerably among both types of polygenic diseases. For each form, hyper-rare disorders, defined as affecting <1/108 individuals, likely represent the largest number of disease entities. Loss of redundant tumor-suppressor genes exemplifies such a profoundly rare mutational event. For non-tumor, polygenic diseases, pathway-centered taxonomies seem preferable. This classification is not readily feasible in cancer, but the inclusion of Drivers and possibly also of epigenetic changes to the existing nomenclature might serve as initial steps in this direction. Based on the detailed genetic alterations, the number of polygenic diseases is essentially countless, but different forms of nosologies may be used to restrict the number.

Stratifying Lung Adenocarcinoma Risk with Multi-ancestry Polygenic Risk Scores in East Asian Never-Smokers

Polygenic risk scores (PRSs) are promising for risk stratification but have mainly been developed in European populations. This study developed single- and multi-ancestry PRSs for lung adenocarcinoma (LUAD) in East Asian (EAS) never-smokers using genome-wide association study summary statistics from EAS (8,002 cases; 20,782 controls) and European (2,058 cases; 5,575 controls) populations. A multi-ancestry PRS, developed using CT-SLEB, was strongly associated with LUAD risk (odds ratio=1.71, 95% confidence interval (CI):1.61,1.82), with an area under the receiver operating curve value of 0.640 (95% CI:0.629,0.653). Individuals in the highest 20% of the PRS had nearly four times the risk compared to the lowest 20%. Individuals in the 95 th percentile of the PRS had an estimated 6.69% lifetime absolute risk. Notably, this group reached the average population 10-year LUAD risk at age 50 (0.42%) by age 41. Our study underscores the potential of multi-ancestry PRS approaches to enhance LUAD risk stratification in EAS never-smokers.

Breastfeeding, genetic susceptibility, and the risk of asthma and allergic diseases in children and adolescents: a retrospective national population-based cohort study

BACKGROUND

Asthma and allergic diseases (such as allergic rhinitis) are multifactorial chronic respiratory diseases, and have many common pathogenic mechanisms. This study aimed to assess the joint effects of breastfeeding and genetic susceptibility on asthma, allergic disease in children and adolescents and sought to examine whether the effect of breastfeeding was consistent under distinct levels of genetic risk.

METHODS

A total of 351,931 UK Biobank participants were analyzed. Firstly, Cox proportional hazards model was used to evaluate the relation between breastfeeding and asthma, allergic disease and their comorbidity. Next, we incorporated the polygenic risk score as an additional covariate into the model. Then, we explored the role of breastfeeding at each stage of asthma and allergic disease through a multi-state model. Meanwhile, several sensitivity analyses were conducted to evaluate the robustness of our results. Finally, we calculated the attributable protection and population attributable protection of breastfeeding.

RESULTS

Breastfeeding was related to a reduced risk of occurring asthma (adjusted hazard ratio [HR] = 0.89, 95% confidence interval [CI] 0.86 ~ 0.93), allergic disease (HR = 0.89, 95%CI 0.87 ~ 0.91) and comorbidity (HR = 0.89, 95%CI 0.83 ~ 0.94). The effect of breastfeeding was almost unchanged after considering PRS and did not substantially differ across distinct genetic risk levels. Breastfeeding showed a stronger risk-decreased impact on individuals who developed from allergic rhinitis to comorbidity (HR = 0.83, 95%CI 0.73 ~ 0.93). Further, the influence of breastfeeding was robust against covariates considered and the confounding influence of adolescent smoking. Finally, due to breastfeeding, 12.0%, 13.0% or 13.0% of the exposed population would not suffer from asthma, allergic diseases and the comorbidity, while 7.1%, 7.6% or 7.6% of the general population would not suffer from these diseases.

CONCLUSIONS

This study provided supportive evidence for the risk-reduced effect of breastfeeding on asthma, allergic diseases, and the comorbidity in children and adolescents, and further revealed that such an influence was consistent across distinct genetic risk levels.

CCAFE: Estimating Case and Control Allele Frequencies from GWAS Summary Statistics

Methods involving summary statistics in genetics can be quite powerful but can be limited in utility. For instance, many post-hoc analyses of disease studies require case and control allele frequencies (AFs), which are not always published. We present two frameworks to derive case and control AFs from GWAS summary statistics using the odds ratio, case and control sample sizes, and either the total (case and control aggregated) AF or standard error (SE). In simulations and real data, derivations of case and controls AFs using total AF is highly accurate across all settings (e.g., minor AF, condition prevalence). Conversely, derivations using SE underestimate common variant AFs (e.g. minor allele frequency >0.3) in the presence of covariates. We develop an adjustment using gnomAD AFs as a proxy for true AFs, which reduces the bias when using SE. While estimating case and control AFs using the total AF is preferred due to its high accuracy, estimating from the SE can be used more broadly since SE can be derived from p-values and beta estimates, which are commonly provided. The methods provided here expand the utility of publicly available genetic summary statistics and promote the reusability of genomic data. The R package CCAFE, with implementations of both methods, is freely available on Bioconductor and GitHub.

Joint modeling of gene-environment correlations and interactions using polygenic risk scores in case-control studies

Polygenic risk scores (PRSs) are rapidly emerging as a way to measure disease risk by aggregating multiple genetic variants. Understanding the interplay of the PRS with environmental factors is critical for interpreting and applying PRSs in a wide variety of settings. We develop an efficient method for simultaneously modeling gene-environment correlations and interactions using the PRS in case-control studies. We use a logistic-normal regression modeling framework to specify the disease risk and PRS distribution in the underlying population and propose joint inference across the 2 models using the retrospective likelihood of the case-control data. Extensive simulation studies demonstrate the flexibility of the method in trading-off bias and efficiency for the estimation of various model parameters compared with standard logistic regression or a case-only analysis for gene-environment interactions, or a control-only analysis, for gene-environment correlations. Finally, using simulated case-control data sets within the UK Biobank study, we demonstrate the power of our method for its ability to recover results from the full prospective cohort for the detection of an interaction between long-term oral contraceptive use and the PRS on the risk of breast cancer. This method is computationally efficient and implemented in a user-friendly R package.

Genetically Predicted Body Mass Index and Mortality in Chronic Obstructive Pulmonary Disease

Rationale: Body mass index (BMI) is associated with chronic obstructive pulmonary disease (COPD) mortality, but the underlying mechanisms are unclear. The effect of genetic variants aggregated into a polygenic score may elucidate the causal mechanisms and predict risk. Objectives: To examine the associations of genetically predicted BMI with all-cause and cause-specific mortality in COPD. Methods: We developed a polygenic score (PGS) for BMI (PGSBMI) and tested for associations of the PGSBMI with all-cause, respiratory, and cardiovascular mortality in participants with COPD from the COPDGene (Genetic Epidemiology of COPD), ECLIPSE (Evaluation of COPD Longitudinally to Identify Predictive Surrogate End-points), and Framingham Heart studies. We calculated the difference between measured BMI and PGS-predicted BMI (BMIdiff) and categorized participants into groups of discordantly low (BMIdiff <20th percentile), concordant (BMIdiff between the 20th and 80th percentiles), and discordantly high (BMIdiff >80th percentile) BMI. We applied Cox models, examined potential nonlinear associations of the PGSBMI and BMIdiff with mortality, and summarized results with meta-analysis. Measurements and Main Results: We observed significant nonlinear associations of measured BMI and BMIdiff, but not PGSBMI, with all-cause mortality. In meta-analyses, a one-standard deviation increase in the PGSBMI was associated with an increased hazard for cardiovascular mortality (hazard ratio [HR], 1.29; 95% confidence interval [CI], 1.12-1.49), but not for respiratory or all-cause mortality. Compared with participants with concordant measured and genetically predicted BMI, those with discordantly low BMI had higher risks for all-cause mortality (HR, 1.57; 95% CI, 1.41-1.74) and respiratory death (HR, 2.01; 95% CI, 1.61-2.51). Conclusions: In people with COPD, a higher genetically predicted BMI is associated with higher cardiovascular mortality but not respiratory mortality. Individuals with a discordantly low BMI have higher all-cause and respiratory mortality rates than those with a concordant BMI.

The Gut Microbiota and Diabetes: Research, Translation, and Clinical Applications-2023 Diabetes, Diabetes Care, and Diabetologia Expert Forum

This article summarizes the state of the science on the role of the gut microbiota (GM) in diabetes from a recent international expert forum organized by Diabetes, Diabetes Care, and Diabetologia, which was held at the European Association for the Study of Diabetes 2023 Annual Meeting in Hamburg, Germany. Forum participants included clinicians and basic scientists who are leading investigators in the field of the intestinal microbiome and metabolism. Their conclusions were as follows: 1) the GM may be involved in the pathophysiology of type 2 diabetes, as microbially produced metabolites associate both positively and negatively with the disease, and mechanistic links of GM functions (e.g., genes for butyrate production) with glucose metabolism have recently emerged through the use of Mendelian randomization in humans; 2) the highly individualized nature of the GM poses a major research obstacle, and large cohorts and a deep-sequencing metagenomic approach are required for robust assessments of associations and causation; 3) because single-time point sampling misses intraindividual GM dynamics, future studies with repeated measures within individuals are needed; and 4) much future research will be required to determine the applicability of this expanding knowledge to diabetes diagnosis and treatment, and novel technologies and improved computational tools will be important to achieve this goal.

Pleiotropy, epistasis and the genetic architecture of quantitative traits

Pleiotropy (whereby one genetic polymorphism affects multiple traits) and epistasis (whereby non-linear interactions between genetic polymorphisms affect the same trait) are fundamental aspects of the genetic architecture of quantitative traits. Recent advances in the ability to characterize the effects of polymorphic variants on molecular and organismal phenotypes in human and model organism populations have revealed the prevalence of pleiotropy and unexpected shared molecular genetic bases among quantitative traits, including diseases. By contrast, epistasis is common between polymorphic loci associated with quantitative traits in model organisms, such that alleles at one locus have different effects in different genetic backgrounds, but is rarely observed for human quantitative traits and common diseases. Here, we review the concepts and recent inferences about pleiotropy and epistasis, and discuss factors that contribute to similarities and differences between the genetic architecture of quantitative traits in model organisms and humans.

The gut microbiota and diabetes: research, translation, and clinical applications - 2023 Diabetes, Diabetes Care, and Diabetologia Expert Forum

This article summarises the state of the science on the role of the gut microbiota (GM) in diabetes from a recent international expert forum organised by Diabetes, Diabetes Care, and Diabetologia, which was held at the European Association for the Study of Diabetes 2023 Annual Meeting in Hamburg, Germany. Forum participants included clinicians and basic scientists who are leading investigators in the field of the intestinal microbiome and metabolism. Their conclusions were as follows: (1) the GM may be involved in the pathophysiology of type 2 diabetes, as microbially produced metabolites associate both positively and negatively with the disease, and mechanistic links of GM functions (e.g. genes for butyrate production) with glucose metabolism have recently emerged through the use of Mendelian randomisation in humans; (2) the highly individualised nature of the GM poses a major research obstacle, and large cohorts and a deep-sequencing metagenomic approach are required for robust assessments of associations and causation; (3) because single time point sampling misses intraindividual GM dynamics, future studies with repeated measures within individuals are needed; and (4) much future research will be required to determine the applicability of this expanding knowledge to diabetes diagnosis and treatment, and novel technologies and improved computational tools will be important to achieve this goal.

Promoting equity in polygenic risk assessment through global collaboration

The long delay before genomic technologies become available in low- and middle-income countries is a concern from both scientific and ethical standpoints. Polygenic risk scores (PRSs), a relatively recent advance in genomics, could have a substantial impact on promoting health by improving disease risk prediction and guiding preventive strategies. However, clinical use of PRSs in their current forms might widen global health disparities, as their portability to diverse groups is limited. This Perspective highlights the need for global collaboration to develop and implement PRSs that perform equitably across the world. Such collaboration requires capacity building and the generation of new data in low-resource settings, the sharing of harmonized genotype and phenotype data securely across borders, novel population genetics and statistical methods to improve PRS performance, and thoughtful clinical implementation in diverse settings. All this needs to occur while considering the ethical, legal and social implications, with support from regulatory and funding agencies and policymakers.

The Gut Microbiota and Diabetes: Research, Translation, and Clinical Applications-2023 Diabetes, Diabetes Care, and Diabetologia Expert Forum

This article summarizes the state of the science on the role of the gut microbiota (GM) in diabetes from a recent international expert forum organized by Diabetes, Diabetes Care, and Diabetologia, which was held at the European Association for the Study of Diabetes 2023 Annual Meeting in Hamburg, Germany. Forum participants included clinicians and basic scientists who are leading investigators in the field of the intestinal microbiome and metabolism. Their conclusions were as follows: 1) the GM may be involved in the pathophysiology of type 2 diabetes, as microbially produced metabolites associate both positively and negatively with the disease, and mechanistic links of GM functions (e.g., genes for butyrate production) with glucose metabolism have recently emerged through the use of Mendelian randomization in humans; 2) the highly individualized nature of the GM poses a major research obstacle, and large cohorts and a deep-sequencing metagenomic approach are required for robust assessments of associations and causation; 3) because single-time point sampling misses intraindividual GM dynamics, future studies with repeated measures within individuals are needed; and 4) much future research will be required to determine the applicability of this expanding knowledge to diabetes diagnosis and treatment, and novel technologies and improved computational tools will be important to achieve this goal.

Ancestry-aligned polygenic scores combined with conventional risk factors improve prediction of cardiometabolic outcomes in African populations

BACKGROUND

Cardiovascular diseases (CVD) are a major health concern in Africa. Improved identification and treatment of high-risk individuals can reduce adverse health outcomes. Current CVD risk calculators are largely unvalidated in African populations and overlook genetic factors. Polygenic scores (PGS) can enhance risk prediction by measuring genetic susceptibility to CVD, but their effectiveness in genetically diverse populations is limited by a European-ancestry bias. To address this, we developed models integrating genetic data and conventional risk factors to assess the risk of developing cardiometabolic outcomes in African populations.

METHODS

We used summary statistics from a genome-wide association meta-analysis (n = 14,126) in African populations to derive novel genome-wide PGS for 14 cardiometabolic traits in an independent African target sample (Africa Wits-INDEPTH Partnership for Genomic Research (AWI-Gen), n = 10,603). Regression analyses assessed relationships between each PGS and corresponding cardiometabolic trait, and seven CVD outcomes (CVD, heart attack, stroke, diabetes mellitus, dyslipidaemia, hypertension, and obesity). The predictive utility of the genetic data was evaluated using elastic net models containing multiple PGS (MultiPGS) and reference-projected principal components of ancestry (PPCs). An integrated risk prediction model incorporating genetic and conventional risk factors was developed. Nested cross-validation was used when deriving elastic net models to enhance generalisability.

RESULTS

Our African-specific PGS displayed significant but variable within- and cross- trait prediction (max.R2 = 6.8%, p = 1.86 × 10-173). Significantly associated PGS with dyslipidaemia included the PGS for total cholesterol (logOR = 0.210, SE = 0.022, p = 2.18 × 10-21) and low-density lipoprotein (logOR =  - 0.141, SE = 0.022, p = 1.30 × 10-20); with hypertension, the systolic blood pressure PGS (logOR = 0.150, SE = 0.045, p = 8.34 × 10-4); and multiple PGS associated with obesity: body mass index (max. logOR = 0.131, SE = 0.031, p = 2.22 × 10-5), hip circumference (logOR = 0.122, SE = 0.029, p = 2.28 × 10-5), waist circumference (logOR = 0.013, SE = 0.098, p = 8.13 × 10-4) and weight (logOR = 0.103, SE = 0.029, p = 4.89 × 10-5). Elastic net models incorporating MultiPGS and PPCs significantly improved prediction over MultiPGS alone. Models including genetic data and conventional risk factors were more predictive than conventional risk models alone (dyslipidaemia: R2 increase = 2.6%, p = 4.45 × 10-12; hypertension: R2 increase = 2.6%, p = 2.37 × 10-13; obesity: R2 increase = 5.5%, 1.33 × 10-34).

CONCLUSIONS

In African populations, CVD and associated cardiometabolic trait prediction models can be improved by incorporating ancestry-aligned PGS and accounting for ancestry. Combining PGS with conventional risk factors further enhances prediction over traditional models based on conventional factors. Incorporating data from target populations can improve the generalisability of international predictive models for CVD and associated traits in African populations.

Cardiovascular Pharmacogenetics: From Discovery of Genetic Association to Clinical Adoption of Derived Test

Recent breakthroughs in human genetics and in information technologies have markedly expanded our understanding at the molecular level of the response to drugs, i.e., pharmacogenetics (PGx), across therapy areas. This review is restricted to PGx for cardiovascular (CV) drugs. First, we examined the PGx information in the labels approved by regulatory agencies in Europe, Japan, and North America and related recommendations from expert panels. Out of 221 marketed CV drugs, 36 had PGx information in their labels approved by one or more agencies. The level of annotations and recommendations varied markedly between agencies and expert panels. Clopidogrel is the only CV drug with consistent PGx recommendation (i.e., "actionable"). This situation prompted us to dissect the steps from discovery of a PGx association to clinical translation. We found 101 genome-wide association studies that investigated the response to CV drugs or drug classes. These studies reported significant associations for 48 PGx traits mapping to 306 genes. Six of these 306 genes are mentioned in the corresponding PGx labels or recommendations for CV drugs. Genomic analyses also highlighted the wide between-population differences in risk allele frequencies and the individual load of actionable PGx variants. Given the high attrition rate and the long road to clinical translation, additional work is warranted to identify and validate PGx variants for more CV drugs across diverse populations and to demonstrate the utility of PGx testing. To that end, pre-emptive PGx combining genomic profiling with electronic medical records opens unprecedented opportunities to improve healthcare, for CV diseases and beyond. SIGNIFICANCE STATEMENT: Despite spectacular breakthroughs in human molecular genetics and information technologies, consistent evidence supporting PGx testing in the cardiovascular area is limited to a few drugs. Additional work is warranted to discover and validate new PGx markers and demonstrate their utility. Pre-emptive PGx combining genomic profiling with electronic medical records opens unprecedented opportunities to improve healthcare, for CV diseases and beyond.

Polygenic risk scores as a marker for epilepsy risk across lifetime and after unspecified seizure events

A diagnosis of epilepsy has significant consequences for an individual but is often challenging in clinical practice. Novel biomarkers are thus greatly needed. Here, we investigated how common genetic factors (epilepsy polygenic risk scores, [PRSs]) influence epilepsy risk in detailed longitudinal electronic health records (EHRs) of > 700k Finns and Estonians. We found that a high genetic generalized epilepsy PRS (PRSGGE) increased risk for genetic generalized epilepsy (GGE) (hazard ratio [HR] 1.73 per PRSGGE standard deviation [SD]) across lifetime and within 10 years after an unspecified seizure event. The effect of PRSGGE was significantly larger on idiopathic generalized epilepsies, in females and for earlier epilepsy onset. Analogously, we found significant but more modest focal epilepsy PRS burden associated with non-acquired focal epilepsy (NAFE). Here, we outline the potential of epilepsy specific PRSs to serve as biomarkers after a first seizure event.

Effect of disclosing a polygenic risk score for coronary heart disease on adverse cardiovascular events: 10-year follow-up of the MI-GENES randomized clinical trial

Background

The MI-GENES clinical trial (NCT01936675), in which participants at intermediate risk of coronary heart disease (CHD) were randomized to receive a Framingham risk score (FRSg, n=103), or an integrated risk score (IRSg, n=104) that additionally included a polygenic risk score (PRS), demonstrated that after 6 months, participants randomized to IRSg had higher statin initiation and lower low-density lipoprotein cholesterol (LDL-C).

Objectives

In a post hoc 10-year follow-up analysis of the MI-GENES trial, we investigated whether disclosure of a PRS for CHD was associated with a reduction in adverse cardiovascular events.

Methods

Participants were followed from randomization beginning in October 2013 until September 2023 to ascertain adverse cardiovascular events, testing for CHD, and changes in risk factors, by blinded review of electronic health records. The primary outcome was the time from randomization to the occurrence of the first major adverse cardiovascular event (MACE), defined as cardiovascular death, non-fatal myocardial infarction, coronary revascularization, and non-fatal stroke. Statistical analyses were conducted using Cox proportional hazards regression and linear mixed-effects models.

Results

We followed all 203 participants who completed the MI-GENES trial, 100 in FRSg and 103 in IRSg (mean age at the end of follow-up: 68.2±5.2, 48% male). During a median follow-up of 9.5 years, 9 MACEs occurred in FRSg and 2 in IRSg (hazard ratio (HR), 0.20; 95% confidence interval (CI), 0.04 to 0.94; P=0.042). In FRSg, 47 (47%) underwent at least one test for CHD, compared to 30 (29%) in IRSg (HR, 0.51; 95% CI, 0.32 to 0.81; P=0.004). IRSg participants had a longer duration of statin therapy during the first four years post-randomization and a greater reduction in LDL-C for up to 3 years post-randomization. No significant differences between the two groups were observed for hemoglobin A1C, systolic and diastolic blood pressures, weight, and smoking cessation rate during follow-up.

Conclusions

The disclosure of an IRS that included a PRS to individuals at intermediate risk for CHD was associated with a lower incidence of MACE after a decade of follow-up, likely due to a higher rate of initiation and longer duration of statin therapy, leading to lower LDL-C levels.

Real-time dynamic polygenic prediction for streaming data

Polygenic risk scores (PRSs) are promising tools for advancing precision medicine. However, existing PRS construction methods rely on static summary statistics derived from genome-wide association studies (GWASs), which are often updated at lengthy intervals. As genetic data and health outcomes are continuously being generated at an ever-increasing pace, the current PRS training and deployment paradigm is suboptimal in maximizing the prediction accuracy of PRSs for incoming patients in healthcare settings. Here, we introduce real-time PRS-CS (rtPRS-CS), which enables online, dynamic refinement and calibration of PRS as each new sample is collected, without the need to perform intermediate GWASs. Through extensive simulation studies, we evaluate the performance of rtPRS-CS across various genetic architectures and training sample sizes. Leveraging quantitative traits from the Mass General Brigham Biobank and UK Biobank, we show that rtPRS-CS can integrate massive streaming data to enhance PRS prediction over time. We further apply rtPRS-CS to 22 schizophrenia cohorts in 7 Asian regions, demonstrating the clinical utility of rtPRS-CS in dynamically predicting and stratifying disease risk across diverse genetic ancestries.

Interplay of polygenic liability with birth-related, somatic, and psychosocial factors in anorexia nervosa risk: a nationwide study

BACKGROUND

Although several types of risk factors for anorexia nervosa (AN) have been identified, including birth-related factors, somatic, and psychosocial risk factors, their interplay with genetic susceptibility remains unclear. Genetic and epidemiological interplay in AN risk were examined using data from Danish nationwide registers. AN polygenic risk score (PRS) and risk factor associations, confounding from AN PRS and/or parental psychiatric history on the association between the risk factors and AN risk, and interactions between AN PRS and each level of target risk factor on AN risk were estimated.

METHODS

Participants were individuals born in Denmark between 1981 and 2008 including nationwide-representative data from the iPSYCH2015, and Danish AN cases from the Anorexia Nervosa Genetics Initiative and Eating Disorder Genetics Initiative cohorts. A total of 7003 individuals with AN and 45 229 individuals without a registered AN diagnosis were included. We included 22 AN risk factors from Danish registers.

RESULTS

Risk factors showing association with PRS for AN included urbanicity, parental ages, genitourinary tract infection, and parental socioeconomic factors. Risk factors showed the expected association to AN risk, and this association was only slightly attenuated when adjusted for parental history of psychiatric disorders or/and for the AN PRS. The interaction analyses revealed a differential effect of AN PRS according to the level of the following risk factors: sex, maternal age, genitourinary tract infection, C-section, parental socioeconomic factors and psychiatric history.

CONCLUSIONS

Our findings provide evidence for interactions between AN PRS and certain risk-factors, illustrating potential diverse risk pathways to AN diagnosis.

Calibrated prediction intervals for polygenic scores across diverse contexts

Polygenic scores (PGS) have emerged as the tool of choice for genomic prediction in a wide range of fields. We show that PGS performance varies broadly across contexts and biobanks. Contexts such as age, sex and income can impact PGS accuracy with similar magnitudes as genetic ancestry. Here we introduce an approach (CalPred) that models all contexts jointly to produce prediction intervals that vary across contexts to achieve calibration (include the trait with 90% probability), whereas existing methods are miscalibrated. In analyses of 72 traits across large and diverse biobanks (All of Us and UK Biobank), we find that prediction intervals required adjustment by up to 80% for quantitative traits. For disease traits, PGS-based predictions were miscalibrated across socioeconomic contexts such as annual household income levels, further highlighting the need of accounting for context information in PGS-based prediction across diverse populations.

Optimizing clinico-genomic disease prediction across ancestries: a machine learning strategy with Pareto improvement

BACKGROUND

Accurate prediction of an individual's predisposition to diseases is vital for preventive medicine and early intervention. Various statistical and machine learning models have been developed for disease prediction using clinico-genomic data. However, the accuracy of clinico-genomic prediction of diseases may vary significantly across ancestry groups due to their unequal representation in clinical genomic datasets.

METHODS

We introduced a deep transfer learning approach to improve the performance of clinico-genomic prediction models for data-disadvantaged ancestry groups. We conducted machine learning experiments on multi-ancestral genomic datasets of lung cancer, prostate cancer, and Alzheimer's disease, as well as on synthetic datasets with built-in data inequality and distribution shifts across ancestry groups.

RESULTS

Deep transfer learning significantly improved disease prediction accuracy for data-disadvantaged populations in our multi-ancestral machine learning experiments. In contrast, transfer learning based on linear frameworks did not achieve comparable improvements for these data-disadvantaged populations.

CONCLUSIONS

This study shows that deep transfer learning can enhance fairness in multi-ancestral machine learning by improving prediction accuracy for data-disadvantaged populations without compromising prediction accuracy for other populations, thus providing a Pareto improvement towards equitable clinico-genomic prediction of diseases.

Multi-organ imaging-derived polygenic indexes for brain and body health

The UK Biobank (UKB) imaging project is a crucial resource for biomedical research, but is limited to 100,000 participants due to cost and accessibility barriers. Here we used genetic data to predict heritable imaging-derived phenotypes (IDPs) for a larger cohort. We developed and evaluated 4,375 IDP genetic scores (IGS) derived from UKB brain and body images. When applied to UKB participants who were not imaged, IGS revealed links to numerous phenotypes and stratified participants at increased risk for both brain and somatic diseases. For example, IGS identified individuals at higher risk for Alzheimer's disease and multiple sclerosis, offering additional insights beyond traditional polygenic risk scores of these diseases. When applied to independent external cohorts, IGS also stratified those at high disease risk in the All of Us Research Program and the Alzheimer's Disease Neuroimaging Initiative study. Our results demonstrate that, while the UKB imaging cohort is largely healthy and may not be the most enriched for disease risk management, it holds immense potential for stratifying the risk of various brain and body diseases in broader external genetic cohorts.

A Genotype-Phenotype Model for Predicting Resistance Training Effects on Leg Press Performance

This study develops a comprehensive genotype-phenotype model for predicting the effects of resistance training on leg press performance. A cohort of physically inactive adults (N=193) underwent 12 weeks of resistance training, and measurements of maximum isokinetic leg press peak force, muscle mass, and thickness were taken before and after the intervention. Whole-genome genotyping was performed, and genome-wide association analysis identified 85 novel SNPs significantly associated with changes in leg press strength after training. A prediction model was constructed using stepwise linear regression, incorporating seven lead SNPs that explained 40.4% of the training effect variance. The polygenic score showed a significant positive correlation with changes in leg press strength. By integrating genomic markers and phenotypic indicators, the comprehensive prediction model explained 75.4% of the variance in the training effect. Additionally, five SNPs were found to potentially impact muscle contraction, metabolism, growth, and development through their association with REACTOME pathways. Individual responses to resistance training varied, with changes in leg press strength ranging from -55.83% to 151.20%. The study highlights the importance of genetic factors in predicting training outcomes and provides insights into the potential biological functions underlying resistance training effects. The comprehensive model offers valuable guidance for personalized fitness programs based on individual genetic profiles and phenotypic characteristics.

Multi-Ancestry Polygenic Risk Score for Coronary Heart Disease Based on an Ancestrally Diverse Genome-Wide Association Study and Population-Specific Optimization

BACKGROUND

Predictive performance of polygenic risk scores (PRS) varies across populations. To facilitate equitable clinical use, we developed PRS for coronary heart disease (CHD; PRSCHD) for 5 genetic ancestry groups.

METHODS

We derived ancestry-specific and multi-ancestry PRSCHD based on pruning and thresholding (PRSPT) and ancestry-based continuous shrinkage priors (PRSCSx) applied to summary statistics from the largest multi-ancestry genome-wide association study meta-analysis for CHD to date, including 1.1 million participants from 5 major genetic ancestry groups. Following training and optimization in the Million Veteran Program, we evaluated the best-performing PRSCHD in 176,988 individuals across 9 diverse cohorts.

RESULTS

Multi-ancestry PRSPT and PRSCSx outperformed ancestry-specific PRSPT and PRSCSx across a range of tuning values. Two best-performing multi-ancestry PRSCHD (ie, PRSPTmult and PRSCSxmult) and 1 ancestry-specific (PRSCSxEUR) were taken forward for validation. PRSPTmult demonstrated the strongest association with CHD in individuals of South Asian ancestry and European ancestry (odds ratio per 1 SD [95% CI, 2.75 [2.41-3.14], 1.65 [1.59-1.72]), followed by East Asian ancestry (1.56 [1.50-1.61]), Hispanic/Latino ancestry (1.38 [1.24-1.54]), and African ancestry (1.16 [1.11-1.21]). PRSCSxmult showed the strongest associations in South Asian ancestry (2.67 [2.38-3.00]) and European ancestry (1.65 [1.59-1.71]), lower in East Asian ancestry (1.59 [1.54-1.64]), Hispanic/Latino ancestry (1.51 [1.35-1.69]), and the lowest in African ancestry (1.20 [1.15-1.26]).

CONCLUSIONS

The use of summary statistics from a large multi-ancestry genome-wide meta-analysis improved the performance of PRSCHD in most ancestry groups compared with single-ancestry methods. Despite the use of one of the largest and most diverse sets of training and validation cohorts to date, improvement of predictive performance was limited in African ancestry. This highlights the need for larger genome-wide association study datasets of underrepresented populations to enhance the performance of PRSCHD.

Visual detection of seizures in mice using supervised machine learning

Seizures are caused by abnormally synchronous brain activity that can result in changes in muscle tone, such as twitching, stiffness, limpness, or rhythmic jerking. These behavioral manifestations are clear on visual inspection and the most widely used seizure scoring systems in preclinical models, such as the Racine scale in rodents, use these behavioral patterns in semiquantitative seizure intensity scores. However, visual inspection is time-consuming, low-throughput, and partially subjective, and there is a need for rigorously quantitative approaches that are scalable. In this study, we used supervised machine learning approaches to develop automated classifiers to predict seizure severity directly from noninvasive video data. Using the PTZ-induced seizure model in mice, we trained video-only classifiers to predict ictal events, combined these events to predict an univariate seizure intensity for a recording session, as well as time-varying seizure intensity scores. Our results show, for the first time, that seizure events and overall intensity can be rigorously quantified directly from overhead video of mice in a standard open field using supervised approaches. These results enable high-throughput, noninvasive, and standardized seizure scoring for downstream applications such as neurogenetics and therapeutic discovery.

Polygenic architecture of brain structure and function, behaviors, and psychopathologies in children

The human brain undergoes structural and functional changes during childhood, a critical period in cognitive and behavioral development. Understanding the genetic architecture of the brain development in children can offer valuable insights into the development of the brain, cognition, and behaviors. Here, we integrated brain imaging-genetic-phenotype data from over 8,600 preadolescent children of diverse ethnic backgrounds using multivariate statistical techniques. We found a low-to-moderate level of SNP-based heritability in most IDPs, which is lower compared to the adult brain. Using sparse generalized canonical correlation analysis (SGCCA), we identified several covariation patterns among genome-wide polygenic scores (GPSs) of 29 traits, 7 different modalities of brain imaging-derived phenotypes (IDPs), and 266 cognitive and psychological phenotype data. In structural MRI, significant positive associations were observed between total grey matter volume, left ventral diencephalon volume, surface area of right accumbens and the GPSs of cognition-related traits. Conversely, negative associations were found with the GPSs of ADHD, depression and neuroticism. Additionally, we identified a significant positive association between educational attainment GPS and regional brain activation during the N-back task. The BMI GPS showed a positive association with fractional anisotropy (FA) of connectivity between the cerebellum cortex and amygdala in diffusion MRI, while the GPSs for educational attainment and cannabis use were negatively associated with the same IDPs. Our GPS-based prediction models revealed substantial genetic contributions to cognitive variability, while the genetic basis for many mental and behavioral phenotypes remained elusive. This study delivers a comprehensive map of the relationships between genetic profiles, neuroanatomical diversity, and the spectrum of cognitive and behavioral traits in preadolescence.

Generalizability of polygenic prediction models: how is the R2 defined on test data?

BACKGROUND

Polygenic risk scores (PRS) quantify an individual's genetic predisposition for different traits and are expected to play an increasingly important role in personalized medicine. A crucial challenge in clinical practice is the generalizability and transferability of PRS models to populations with different ancestries. When assessing the generalizability of PRS models for continuous traits, the R 2 is a commonly used measure to evaluate prediction accuracy. While the R 2 is a well-defined goodness-of-fit measure for statistical linear models, there exist different definitions for its application on test data, which complicates interpretation and comparison of results.

METHODS

Based on large-scale genotype data from the UK Biobank, we compare three definitions of the R 2 on test data for evaluating the generalizability of PRS models to different populations. Polygenic models for several phenotypes, including height, BMI and lipoprotein A, are derived based on training data with European ancestry using state-of-the-art regression methods and are evaluated on various test populations with different ancestries.

RESULTS

Our analysis shows that the choice of the R 2  definition can lead to considerably different results on test data, making the comparison of R 2  values from the literature problematic. While the definition as the squared correlation between predicted and observed phenotypes solely addresses the discriminative performance and always yields values between 0 and 1, definitions of the R 2 based on the mean squared prediction error (MSPE) with reference to intercept-only models assess both discrimination and calibration. These MSPE-based definitions can yield negative values indicating miscalibrated predictions for out-of-target populations. We argue that the choice of the most appropriate definition depends on the aim of PRS analysis - whether it primarily serves for risk stratification or also for individual phenotype prediction. Moreover, both correlation-based and MSPE-based definitions of R 2 can provide valuable complementary information.

CONCLUSIONS

Awareness of the different definitions of the R 2 on test data is necessary to facilitate the reporting and interpretation of results on PRS generalizability. It is recommended to explicitly state which definition was used when reporting R 2 values on test data. Further research is warranted to develop and evaluate well-calibrated polygenic models for diverse populations.

Characterizing substructure via mixture modeling in large-scale genetic summary statistics

Genetic summary data are broadly accessible and highly useful including for risk prediction, causal inference, fine mapping, and incorporation of external controls. However, collapsing individual-level data into groups masks intra- and inter-sample heterogeneity, leading to confounding, reduced power, and bias. Ultimately, unaccounted substructure limits summary data usability, especially for understudied or admixed populations. Here, we present Summix2, a comprehensive set of methods and software based on a computationally efficient mixture model to estimate and adjust for substructure in genetic summary data. In extensive simulations and application to public data, Summix2 characterizes finer-scale population structure, identifies ascertainment bias, and identifies potential regions of selection due to local substructure deviation. Summix2 increases the robust use of diverse publicly available summary data resulting in improved and more equitable research.

Association between a polygenic lipodystrophy genetic risk score and diabetes risk in the high prevalence Maltese population

BACKGROUND

Type 2 diabetes (T2DM) is genetically heterogenous, driven by beta cell dysfunction and insulin resistance. Insulin resistance drives the development of cardiometabolic complications and is typically associated with obesity. A group of common variants at eleven loci are associated with insulin resistance and risk of both type 2 diabetes and coronary artery disease. These variants describe a polygenic correlate of lipodystrophy, with a high metabolic disease risk despite a low BMI.

OBJECTIVES

In this cross-sectional study, we sought to investigate the association of a polygenic risk score composed of eleven lipodystrophy variants with anthropometric, glycaemic and metabolic traits in an island population characterised by a high prevalence of both obesity and type 2 diabetes.

METHODS

814 unrelated adults (n = 477 controls and n = 337 T2DM cases) of Maltese-Caucasian ethnicity were genotyped and associations with phenotypes explored.

RESULTS

A higher polygenic lipodystrophy risk score was correlated with lower adiposity indices (lower waist circumference and body mass index measurements) and higher HOMA-IR, atherogenic dyslipidaemia and visceral fat dysfunction as assessed by the visceral adiposity index in the DM group. In crude and covariate-adjusted models, individuals in the top quartile of polygenic risk had a higher T2DM risk relative to individuals in the first quartile of the risk score distribution.

CONCLUSION

This study consolidates the association between polygenic lipodystrophy risk alleles, metabolic syndrome parameters and T2DM risk particularly in normal-weight individuals. Our findings demonstrate that polygenic lipodystrophy risk alleles drive insulin resistance and diabetes risk independent of an increased BMI.

Examining intergenerational risk factors for conduct problems using polygenic scores in the Norwegian Mother, Father and Child Cohort Study

The aetiology of conduct problems involves a combination of genetic and environmental factors, many of which are inherently linked to parental characteristics given parents' central role in children's lives across development. It is important to disentangle to what extent links between parental heritable characteristics and children's behaviour are due to transmission of genetic risk or due to parental indirect genetic influences via the environment (i.e., genetic nurture). We used 31,290 genotyped mother-father-child trios from the Norwegian Mother, Father and Child Cohort Study (MoBa), testing genetic transmission and genetic nurture effects on conduct problems using 13 polygenic scores (PGS) spanning psychiatric conditions, substance use, education-related factors, and other risk factors. Maternal or self-reports of conduct problems at ages 8 and 14 years were available for up to 15,477 children. We found significant genetic transmission effects on conduct problems for 12 out of 13 PGS at age 8 years (strongest association: PGS for smoking, β = 0.07, 95% confidence interval = [0.05, 0.08]) and for 4 out of 13 PGS at age 14 years (strongest association: PGS for externalising problems, β = 0.08, 95% confidence interval = [0.05, 0.11]). Conversely, we did not find genetic nurture effects for conduct problems using our selection of PGS. Our findings provide evidence for genetic transmission in the association between parental characteristics and child conduct problems. Our results may also indicate that genetic nurture via traits indexed by our polygenic scores is of limited aetiological importance for conduct problems-though effects of small magnitude or effects via parental traits not captured by the included PGS remain a possibility.

Taking the risk. A systematic review of ethical reasons and moral arguments in the clinical use of polygenic risk scores

Debates about the prospective clinical use of polygenic risk scores (PRS) have grown considerably in the last years. The potential benefits of PRS to improve patient care at individual and population levels have been extensively underlined. Nonetheless, the use of PRS in clinical contexts presents a number of unresolved ethical challenges and consequent normative gaps that hinder their optimal implementation. Here, we conducted a systematic review of reasons of the normative literature discussing ethical issues and moral arguments related to the use of PRS for the prevention and treatment of common complex diseases. In total, we have included and analyzed 34 records, spanning from 2013 to 2023. The findings have been organized in three major themes: in the first theme, we consider the potential harms of PRS to individuals and their kin. In the theme "Threats to health equity," we consider ethical concerns of social relevance, with a focus on justice issues. Finally, the theme "Towards best practices" collects a series of research priorities and provisional recommendations to be considered for an optimal clinical translation of PRS. We conclude that the use of PRS in clinical care reinvigorates old debates in matters of health justice; however, open questions, regarding best practices in clinical counseling, suggest that the ethical considerations applicable in monogenic settings will not be sufficient to face PRS emerging challenges.

Lessons and Applications of Omics Research in Diabetes Epidemiology

PURPOSE OF REVIEW

Recent advances in genomic technology and molecular techniques have greatly facilitated the identification of disease biomarkers, advanced understanding of pathogenesis of different common diseases, and heralded the dawn of precision medicine. Much of these advances in the area of diabetes have been made possible through deep phenotyping of epidemiological cohorts, and analysis of the different omics data in relation to detailed clinical information. In this review, we aim to provide an overview on how omics research could be incorporated into the design of current and future epidemiological studies.

RECENT FINDINGS

We provide an up-to-date review of the current understanding in the area of genetic, epigenetic, proteomic and metabolomic markers for diabetes and related outcomes, including polygenic risk scores. We have drawn on key examples from the literature, as well as our own experience of conducting omics research using the Hong Kong Diabetes Register and Hong Kong Diabetes Biobank, as well as other cohorts, to illustrate the potential of omics research in diabetes. Recent studies highlight the opportunity, as well as potential benefit, to incorporate molecular profiling in the design and set-up of diabetes epidemiology studies, which can also advance understanding on the heterogeneity of diabetes. Learnings from these examples should facilitate other researchers to consider incorporating research on omics technologies into their work to advance the field and our understanding of diabetes and its related co-morbidities. Insights from these studies would be important for future development of precision medicine in diabetes.

Evaluating causal influence of maternal educational attainment on offspring birthweight via observational study and Mendelian randomization analyses

Background

Although extensive discussions on the influence of maternal educational attainment on offspring birthweight, the conclusion remains controversial, and it is challenging to comprehensively assess the causal association between them.

Methods

To estimate effect of maternal educational attainment on the birthweight of first child, we first conducted an individual-level analysis with UK Biobank participants of white ancestry (n = 208,162). We then implemented Mendelian randomization (MR) methods including inverse variance weighted (IVW) MR and multivariable MR to assess the causal relation between maternal education and maternal-specific birthweight. Finally, using the UK Biobank parent-offspring trio data (n = 618), we performed a polygenic score based MR to simultaneously adjust for confounding effects of fetal-specific birthweight and paternal educational attainment. We also conducted simulations for power evaluation and sensitivity analyses for horizontal pleiotropy of instruments.

Results

We observed that birthweight of first child was positively influenced by maternal education, with 7 years of maternal education as the reference, adjusted effect = 44.8 (95%CIs 38.0-51.6, P = 6.15 × 10-38), 54.9 (95%CIs 47.6-62.2, P = 4.21 × 10-128), and 89.4 (95%CIs 82.1-96.7, P = 4.28 × 10-34) for 10, 15 and 20 years of maternal educational attainment, respectively. A causal relation between maternal education and offspring birthweight was revealed by IVW MR (estimated effect = 0.074 for one standard deviation increase in maternal education years, 95%CIs 0.054-0.093, P = 2.56 × 10-13) and by complementary MR methods. This connection was not substantially affected by paternal education or horizontal pleiotropy. Further, we found a positive but insignificant causal association (adjusted effect = 24.0, 95%CIs -150.1-198.1, P = 0.787) between maternal education and offspring birthweight after simultaneously controlling for fetal genome and paternal education; this null causality was largely due to limited power of small sample sizes of parent-offspring trios.

Conclusion

This study offers supportive evidence for a causal association between maternal education and offspring birthweight, highlighting the significance of enhancing maternal education to prevent low birthweight.