Objective
Type 1 Diabetes genetic risk scores (T1D GRSs) have been developed using data from the current wealth of knowledge in populations of European ancestry and, to a lesser extent, smaller datasets from other populations to aid in T1D risk assessment. The initial T1D GRS was based on single nucleotide polymorphisms (SNPs) in both HLA and non-HLA regions, primarily identified in cohorts of European ancestry. SNPs in the HLA region “tagged” known HLA alleles and haplotypes associated with T1D risk. Although several forms of the T1D GRS have been proposed, the T1D GRS1 had good performance in European ancestry groups, classifying individuals at increased risk of T1D, distinguishing risk of T1D from type 2 diabetes (T2D), and identifying those with T2D who require insulin for glucose control. However, as the T1D GRSs are based on data primarily from European ancestry populations, there are gaps in genes and variants associated with T1D from other, under-studied populations. Importantly, incorporating genetic risk variants that differ from European ancestry populations is essential to improve and develop a truly “global” T1D GRS.
Genetic risk scores in T1D, as with other complex human diseases, has shown that the T1D GRS works best in European ancestry populations and performance is degraded in other tested populations. Further, there are key populations that constitute the majority of the world’s population (East Asian and South Asian) for which limited (or no) data on genetic risk of T1D is known. In our initial research with a small number of T1D cases and controls from East and South Asia, the current T1D GRS has the worst performance in determining risk when compared with European, African and Admixed American ancestry groups. Thus, there is a need to perform genetic studies in these under-represented populations to identify genetic variants associated with T1D risk, incorporate these variants into a “global” T1D GRS, and evaluate the transferability of a “cosmopolitan” T1D GRS globally across diverse ancestry groups to ensure inclusivity in risk assessment and avoid barriers to access.
Our objective is to focus on East Asian populations (e.g., Chinese, Korean) that have low T1D incidence rates but represent ~20% of the global population. Development of T1D GRSs for EAS is critical for risk prediction, prevention, and reducing misclassification in young-onset diabetes. In this project, four sources of clinically confirmed T1D (and controls) will greatly enhance our knowledge of genetic risk in the EAS population: Taichung Veterans General Hospital (Taichung VGH), Taiwan Precision Medicine Initiative (TPMI), the Seoul National University Hospital (SNUH) Type 1 Diabetes Study, and the Type 1 Diabetes Genetics Consortium (T1DGC).
Analysis of EAS data will identify novel SNPs and candidate genes associated with T1D risk. These results, along with HLA alleles and haplotypes, will be used to develop an optimized T1D GRS for EAS population. Using enhanced GRS methods, we will then integrate these EAS specific risk variants (both HLA and non-HLA) with existing T1D GRS to construct a more accurate and globally applicable T1D GRS. This effort is especially critical for general population where 95% of those who develop T1D do not have a family history.
Background Rationale
Type 1 diabetes (T1D) is a chronic autoimmune disease in which an unknown event triggers the body’s immune system to attack and destroy the insulin-producing beta cells in the pancreas. As only 5% of those who develop T1D have a family history of the disease, the symptoms are often unrecognized and, with increasing destruction of the beta cells, there is insufficient insulin to convert glucose into energy and the body starts breaking down tissues, producing toxic ketones. As many as 40% of children who develop T1D present with diabetic ketoacidosis (DKA), whose presence is associated with increased risk of complications later in life. The treatment for T1D is lifetime delivery of insulin, although recent FDA-approved immune intervention (teplizumab) has been shown to delay onset of T1D by 2-3 years after a single 2-week infusion. Currently, eligibility for immune intervention is restricted to those individuals who have at least two (of four) islet autoantibodies and evidence of dysglycemia (increased glucose), suggesting that the destruction of the beta cells is in an advanced stage.
Twin and family studies have shown that genetic factors account for ~50% of the risk of T1D. Variation in HLA class I (HLA-A, HLA-B, HLA-C) and HLA class II (HLA-DRB1, HLA-DQA1, HLA-DQB1) genes in the human Major Histocompatibility Complex accounts for 30% - 50% of the genetic risk of T1D. We led the Type 1 Diabetes Genetics Consortium (T1DGC) and conducted the first large genome-wide association study (GWAS) and fine-mapping (with the ImmunoChip) to identify T1D susceptibility loci. Our work has led to the discovery of over 100 loci in the human genome that contain single nucleotide variants (SNVs) associated with T1D risk. These studies demonstrated that the T1D SNVs were enriched in DNA regulatory regions (not in protein-coding elements in genes) and were more likely to be in enhancers (not promoters) in immune-relevant cell types (CD4+ and CD8+ T cells and CD19+ B cells). These SNVs have been assembled into genetic risk scores (GRS) that provide a single measure of genetic risk for T1D.
The majority of genetic studies in complex human diseases, including T1D, have focused on populations of Northern European genetic ancestry. As a result, the majority of published T1D GRS models performs well at predicting risk in in individuals of European ancestry, but shows reduced accuracy in those of African- and Admixed ancestry, and performs much worse in East Asian and South Asian groups that historically have been poorly represented in genetic studies. Recent findings from multiple consortia demonstrated that including large, diverse cohorts improves GRS performance, both by inclusion of risk variants identified in novel loci and by uncovering ancestry-specific risk variants within known loci. Thus, increasing genetic diversity and sample sizes in under-represented populations is critical for global detection of genetic risk using T1D GRS. Individuals of East Asian (EAS) ancestry (e.g., China, South Korea, Japan) currently have relatively low T1D incidence yet these countries are among the fastest growing in population and T1D incidence in these populations is rising rapidly.
There is a need to evaluate the transferability of a T1D GRS globally across diverse ancestry groups to ensure inclusivity in risk assessment and avoid barriers to access. This proposed research will focus on populations of East Asian ancestry, specifically Taiwan and South Korea, with the assembly of a cohort of ~20,000 individuals with genome-wide data in ~3,000 T1D cases and ~17,000 controls. We anticipate that this effort will result in an improved GRS for East Asians but also allow the development of multi-ancestry GRSs that will improve T1D prediction globally.
Description of Project
Genetic factors account for nearly half a child’s risk of developing type 1 diabetes (T1D). A significant portion of this genetic risk, between one third and one-half, can be attributed to genetic variations in the human Major Histocompatibility Complex which includes HLA class I (A, B, C) and HLA class II (DRB1, DQA1, DQB1) genes. Genome-wide studies conducted by our group and others have identified T1D risk variants at over 100 sites in the genome, collectively accounting for 80%-90% of known genetic risk for T1D. However, the vast majority of these discoveries have been made in individuals with Northern European genetic ancestry, limiting their generalizability to other populations. Using these data from European ancestry populations, we and others have developed T1D Genetic Risk Scores (T1D GRS) that assigns a "genetic risk" to each individual of developing T1D. This single score can help identify who might benefit from subsequent islet autoantibody screening, and eligibility for immune intervention to delay or stop the disease process in T1D. We and others have shown that for T1D, like other complex human diseases, the T1D GRS works best for people of Northern European ancestry and not as well for other ancestries (African, Hispanic/Latino, Asia). In order to establish a T1D GRS that can be used globally, across all populations, there needs to be genetic studies to establish the identity and impact of genetic variants associated with T1D risk in genetically diverse populations. In particular, for populations with a low incidence of T1D (e.g., East Asia) but very large population size, there is a lack of studies with sufficient statistical power to provide data that could improve the T1D GRS. In this research project, we will use genome-wide data already collected on individuals in Taiwan and South Korea (~3,000 with T1D, ~17,000 without T1D) to form the largest study of T1D in East Asia. These results will provide novel insights (likely novel genes and new genetic variants in known T1D regions) that will support the development of a globally applicable T1D GRS.
Anticipated Outcome
A type 1 diabetes (T1D) genetic risk score (GRS) has been shown to be useful in identifying children (mainly of European ancestry) who are at genetic risk of developing T1D. As 95% of children who develop T1D has no family history of the disease and are not likely to be tested for T1D-specific autoantibodies that indicate an immune attack on the insulin-producing beta cells is happening, a genetic test could be a valuable screening tool. In order to optimize the The current versions of a T1D GRS have been shown to have promise in non-European populations but at less precision than in European ancestry groups from which the T1D GRS was developed. In order to improve the T1D GRS for many diverse ancestry populations, there needs to increase the knowledge of genetic variants associated with T1D in non-European populations.
The ability to identify novel genes and variants associated with risk of T1D is effectively absent in populations of East Asia (20% of the world’s population, including China, Japan and South Korea) and South Asia (25% of the world’s population including India, Pakistan, and Sri Lanka). In order to determine the genetic variants associated with T1D, there needs to be samples in those with T1D and controls, large population sizes, and genetic data. In this research, we focus on collaborators with these resources in East Asian populations. The Taichung Veterans General Hospital (VGH), located in central Taiwan, has 324 individuals with T1D with genome-wide data as well as 1,620 sex- and year of birth-matched genotyped controls. An additional 181 cases with T1D with whole exome sequencing (WES) with 905 matched controls available to detect protein-coding variants. The Taiwan Precision Medicine Initiative (TPMI) Consortium, through Academia Sinica, has conducted a comprehensive retrieval across 16 hospitals using the same ICD-based criteria as in the Taichung VGH. From a total of ~76,000 subjects, 2,097 individuals with T1D and 10,435 matched controls have been genotyped using the same Axiom genotyping array. Thus, from Taiwan, the sample size of ~2,500 T1D cases and ~12,000 controls are available to identify genetic variants associated with T1D with small-modest effects.
The Seoul National University Hospital (SNUH) Type 1 Diabetes Study has genotyped 400 of 568 confirmed cases with T1D with a population control group of 4,000 (out of 11,500) individuals. Genetic data was obtained through use of the Korean Biobank Array, low-pass sequencing (Genotyping-by-Sequencing) through the Twist platform, with an additional subset with WES using the SPIKE-IN WES method. Additional 200-300 T1D cases are available for genotyping. In total, we will analyze a statistically robust sample of EAS ancestry in over 3,000 cases and 17,000 controls. In addition to our genome-wide association analysis, we will obtain HLA alleles and haplotypes that, in European ancestry populations, account for ~40% of the genetic risk of T1D). Combined, these data will create an EAS-specific T1D GRS and incorporate these data in an improved T1D GRS that can assess T1D risk across the world.
Our team has extensive experience in development of diabetes-focused methods for GRS that incorporate increased genetic diversity. Ultimately, we plan to better inform T1D risk across age by incorporating islet autoantibody status and other T1D risk factors with genetic variation into our T1D risk prediction models. We anticipate that incorporation of East Asian T1D-associated variants is a critical step into development of a T1D GRS model that will improve prediction accuracy not only in East Asians but also in other diverse populations.
Relevance to T1D
Type 1 diabetes (T1D) is an autoimmune disease in which an individual’s immune system targets and destroys the insulin-producing beta cells in the pancreas. The incidence of T1D in children and adolescents is increasing by ~3% per year. The natural history of T1D is complex, as it occurs in all ethnic groups (although most common in those of European ancestry, yet the highest prevalence currently is in Qatar) and nearly one-half of new-onset cases occur in adults (although milder than in children and adolescents). The initiation of the autoimmune attack that leads to T1D is unknown but is characterized by the presence of pancreatic islet autoantibodies in the blood, requiring specific laboratory testing over time for evaluation of risk. Only a portion of those with a single islet autoantibody progress to multiple islet autoantibodies (initially with normal blood sugar levels); however, the presence of multiple islet autoantibodies and decline in glucose tolerance will lead to clinical T1D.
Early detection of islet autoimmunity that leads to T1D provides the opportunity for education on symptom awareness, monitoring of disease progression, prevention of diabetic ketoacidosis (DKA), improved management of glycemia at T1D onset, and potential immune intervention to delay disease onset. Teplizumab (a humanized mouse anti-CD3 monoclonal antibody) received FDA approval in November 2022 for use in children with multiple islet autoantibodies and decreased glucose tolerance, as it delays progression to clinical T1D by nearly 3 years. However, identification of who to screen for early detection is complex, as only 5% of those developing T1D have a family history of the disease.
T1D risk in children and adolescents is partly driven by genetic factors. We and others have discovered over 100 sites in the human genome that account for 80%-90% of the genetic risk in European ancestry populations, with additional sites identified in populations of African ancestry. While most individual genetic variants contribute relatively little to risk (with the exception of HLA genes), combining them into a genetic risk score (T1D GRS) effectively captures overall risk and has shown clinical utility in identifying children and adolescents at “high” genetic risk and who may benefit from islet autoantibody screening.
Developing a T1D GRS in the European ancestry populations benefitted from large sample size (our initial genome-wide association included 7,500 subjects with T1D and 9,000 controls); however, our discovery of African ancestry genetic variants that accounted for T1D genetic risk in both European and African populations only had 1,500 T1D subjects and 3,000 controls. With increased sample size, new sites in the human genome and new variants in known sites can be used to improve the prediction of the T1D GRS. Our goal of general population screening for genetic risk of T1D focuses on a population not previously included in T1D genetic studies (East Asian) with expectation of discovery of new genetic variants that would enhance global use of the T1D GRS. We have identified sources of genetic data in over 3,000 subjects with T1D and 17,000 controls from Taiwan, South Korea, and in the Type 1 Diabetes Genetics Consortium (T1DGC) data that we led. Use of these novel data in an improved T1D GRS, not only for East Asians, but also globally, will aid in identifying those at high genetic risk for islet autoantibody testing (focused on those without a family history of T1D) and ultimately increasing availability of immune interventions, improving education and pre-onset care, and reducing the occurrence of DKA.