|Home | About | Journals | Submit | Contact Us | Français|
A series of genes and loci influencing the genetic risk of type 1A (immune-mediated) diabetes are now well characterized. These include genes of the major histocompatibility complex (MHC), polymorphisms 5′ of the insulin gene, and PTPN22, as well as more recently defined loci from genome-wide association studies. By far the major determinants of risk for type 1A diabetes are genes within or linked to the MHC and in particular alleles of class II genes (HLA-DR, DQ, and DP). There is evidence that MHC class I alleles contribute and there are additional MHC-linked influences such that for a major subset of relatives of patients there is a risk as high as 80% for siblings, and for the general population a risk as high as 20% can be defined at birth just by analyzing the MHC. We believe the search for additional MHC loci will require analysis of the remarkable long-range identity (up to 9 million base pairs) of extended MHC haplotypes. Current prediction algorithms will likely be greatly improved for the general population when the additional contributing loci of the MHC are defined.
Multiple genetic syndromes are associated with the development of immune mediated diabetes of man. In particular, there are two rare “monogenic disorders” that provide important insight into pathways leading to autoimmune destruction of islet beta cells: IPEX (Immune dysfunction, Polyendocrinopathy, Enteropathy, X-linked) and APSI (Autoimmune Polyendocrine Syndrome Type 1) syndromes. IPEX syndrome results from mutations of the FOXP3 gene leading to a lack of a major population of regulatory T lymphocytes with resulting overwhelming autoimmunity and development of diabetes in approximately 80% of the children with this disorder. Diabetes can occur as early as two days of age. For the APS-I syndrome, and the related mouse model with mutations of the AIRE (Autoimmune Regulator) gene, it is hypothesized that abnormalities in expression of peripheral antigens within the thymus and/or abnormalities of negative selection result in widespread autoimmunity. Approximately 18% of children with this syndrome develop type 1A diabetes.
The more common type 1A diabetes is a “polygenic” disorder primarily determined by genes within or linked to the major histocompatibility complex (MHC).
Type 1A diabetes has a strong genetic component. First degree relatives have a higher risk for type 1A diabetes than the general population, and siblings have a higher risk than offspring. The sibling relative risk for type 1A diabetes is 15 (λS, a measure of familial clustering) .
The class II loci HLA-DRB1 and HLA-DQB1 in the MHC are known to be associated with type 1A diabetes risk based on functional, structural and genetic evidence . DR3-DQA1*0501-DQB1*0201 (DR3) and DR4-DQA1*0301-DQB1*0302 (DR4) are the highest risk DR/DQ haplotypes for type 1A diabetes, and 30–40% of type 1A diabetes patients have the heterozygous genotype DR3/4 [2,3]. Of the general population children in Denver, 2.4% carry this genotype, and these individuals have an absolute risk of approximately 1/15 versus a risk of 1/300 in the general population . In contrast to lower risk homozygous DR4/4 or DR3/3 genotypes, the increased type 1A diabetes risk of DR3/4 individuals is probably a result of DQ alpha and beta heterodimers that form in DR3/4 individuals. In particular, it is hypothesized that the DQA1*0501 on DR3 haplotypes combining with the DQB1*0302 allele on DR4 haplotypes is most important, as that rare African DRB1*0405 haplotypes with DQA1*0301-DQB1*0201 alleles in cis (same as the alternate trans heterodimer of DR3/4 individuals) are lower risk than DQA1*0301-DQB1*0302 haplotypes .
Risk of diabetes is influenced by both DRB1*04 variants and DQ alleles on DR4 haplotypes (Table 1) . Thus there is a hierarchy of DRB1*04 haplotypes, even with the same DQA1*0301-DQB1*0302 alleles, with higher risk from DRB1*0405 (OR=11.4), DRB1*0401 (OR=8.4), DRB1*0402 (OR=3.6), and DRB1*0404 (OR=1.6), while DRB1*0403 is protective (OR=0.27). Similarly, for DRB1*0401, variation of DQB1 influences risk, as haplotypes with DQB1*0302 (OR=8.4) are highly susceptible, while those with DQB1*0301 (OR=0.35) are modestly protective. The influence of DRB1*04-DQ haplotypes on diabetes risk is complex and influenced by genotype, as the proportion of DQB1*0301 is greater in DR1/4 patients (13.1%) than in DR3/4 (0.7%) patients and, additionally within these genotypes, there is a different distribution of DRB1*0401 alleles . There are several DQ risk alleles with aspartic acid at position 57 and these alleles are particularly prominent in the Asian patients, but are also determinants of risk in Caucasian populations .
There are dramatically protective DR-DQ haplotypes [e.g. DRB1*1501-DQA1*0102-DQB1*0602 (OR=0.03), DRB1*1401-DQA1*0101-DQB1*0503 (OR=0.02) and DRB1*0701-DQA1*0201-DQB1*0303 (OR=0.02)]. The most common is the DR2 (DRB1*1501) bearing haplotype. The DR2 haplotype (DRB1*1501-DQA1*0102-DQB1*0602) is dominantly protective (20% of general population individuals have DQB1*0602 but it is present in only 1% of type 1A diabetes patients) . At present, it is not settled as to why a small subset of children with DQB1*0602 still develop type 1A diabetes, but there are reports of genetic heterogeneity at other MHC loci [6,7]
HLA-DPB1 alleles have been reported to be associated with type 1A diabetes, but are not generally recognized as major contributors to type 1A diabetes . Several groups have reported a negative association of DPB1*0402 with type 1A diabetes [9–15]. These studies have found that the frequency of the DPB1*0402 allele is lower in patients than in control groups and that transmission of DPB1*0402 to type 1A diabetes patients is decreased. On average, the relative risk for DPB1*0402 overall from these studies is 0.56. DPB1*0301 and DPB1*0202 have been reported to be predisposing to type 1A diabetes.
HLA-A is associated with type 1A diabetes independent of DR/DQ . A24 is associated with a younger age of onset, A30 (typically on the DR3-B18-A30 haplotype) is higher risk, and A1 (typically on the DR3-B8-A1 haplotype) is lower risk than other DR3 haplotypes . HLA-B and HLA-C alleles are also associated with type 1A diabetes independent of DR/DQ . HLA-B18, B39, B44, C3, C8 and C16 are associated with type 1A diabetes and with age of onset after adjustment for linkage disequilibrium with DR/DQ . Nejentsev et al. used a recursive partitioning, a risk-categorization method of grouping that doesn’t require the risk to be known a priori, to correct for HLA alleles . They concluded that both HLA-A and HLA-B survive correction for HLA-DR and DQ, and that in particular HLA-B39 is important (confers high risk for type 1A diabetes in three different populations), makes up the majority of the signal from HLA-B, and is associated with a lower age of onset of the disease . MICA is reported to be associated with type 1A diabetes when the DR/DQ genotype is fixed (within DR3/4 siblings and offspring) .
The MHC region exhibits strong and extended linkage disequilibrium, and it has been known for years that these haplotypes can be identical for up to 3.2 Mb, from HLA-DRB1 to HLA-A, with one report describing conservation for 6–8 Mb telomeric from HLA-DRB1 . Such extended HLA haplotypes make it difficult to define specific loci that contribute to diabetes risk beyond HLA alleles. Detection of this conservation has been based on microsatellite data, complement gene typing, and HLA-DR/DQ, HLA-B, HLA-C and HLA-A alleles [20–23]. One recent article reports that slightly more than half of Caucasian MHC haplotypes (based on complement alleles, HLA-B and HLA-DR/DQ typing) are fixed from HLA-DR/DQ to HLA-B . Extreme variability in the MHC class III region has been identified using SNP typing .
The DR3-B8-A1 haplotype is extremely conserved [≥99% identity of single nucleotide polymorphisms (SNPs)] . This extreme conservation can extend for up to 9 Mb, from HLA-DRB1 through the 3 Mb of the MHC region and to the telomeric end of the extended MHC (23 Mb from the chromosome 6 telomere) . The DR3-B8-A1 haplotype is the most common extended haplotype, as it is present in 9% of Caucasian MHC control haplotypes and 18% of case MHC haplotypes from individuals with type 1A diabetes .
In addition to the DR3-B8-A1 haplotype, other extended HLA haplotypes have been identified [22,24,25,28]. These haplotypes are common, conserved (based on complement alleles and HLA typing) and extend from DRB1 to HLA-A [24,29].
A VNTR (variable number tandem repeats) microsatellite at the 5′ end of the insulin gene (INS) on chromosome 11 has alleles that are functionally associated with both type 1A diabetes risk and protection, with an allelic odds ratio (from a genome-wide linkage scan) of 1.9 [30–32]. Two groups provided evidence that the allelic variant associated with protection from type 1A diabetes was associated with greater expression of insulin messenger RNA within the thymus [31,33]. Studies in the NOD mouse model of type 1A diabetes implicate insulin/proinsulin as the primary autoantigen where decreased insulin expression in the thymus also correlates with risk , induction of tolerance to proinsulin prevents both diabetes and development of T cells reacting with the islet target molecule IGRP , and mutating a key amino acid of insulin peptide B:9–23 prevents diabetes [36,37].
A gene at 1p13, PTPN22, encodes for LYP (a protein tyrosine phosphatase). Multiple groups have reported an association with type 1A diabetes of a nonsynonymous SNP in PTPN22 at position 1858, with an odds ratio of 3.4 for the homozygous TT genotype [38,39] in many different populations  and the same polymorphism is a major determinant of rheumatoid arthritis [41,42] and other autoimmune disorders . The polymorphism changes an arginine at position 620 to a tryptophan and has been associated with a gain of function . It is of interest that the risk allele of PTPN22 apparently decreases T cell receptor signaling, suggesting that the risk allele may influence autoimmunity by decreasing negative T cell selection within the thymus that is dependent upon strong T cell receptor activation.
A genome-wide association study for type 1A diabetes was completed in 2007 by the Wellcome Trust Case Control Consortium, reporting signals at known loci [6p21 (MHC region), 11q15 (INS), 2q33 (CTLA4), 1p13 (PTPN22), 10p15 (CD25/IL2RA) and 2q24 (IFIH1)] in addition to SNP associations at several novel loci (12q13, 12q24, 16p13, 4q27, 12p13 and 18p11) . The associations at 12q24, 12q13, 16p13 and 18p11 were confirmed in an independent study (Figure 1) .
CTLA4, at 2q33, has an odds ratio of 1.1–1.2, but the association has been replicated in multiple studies [30,47]. The CTLA4 locus is of interest given the role of CTLA4 in T cell signaling. The locus is associated primarily with patients with both thyroid autoimmunity and type 1A diabetes, concordant with its stronger association with thyroid autoimmunity .
Prior to the large whole genome association studies recently reported, other associations have been reported for CD25/IL2RA (10p15), SUMO4 (6q25), and IFIH1 (2q24), in addition to other loci [49–52]. Of these three genes, the clearest confirmation is for CD25/IL2RA (10p15), with the finding of more than one CD25 SNP associated with type 1A diabetes (Figure 1). The CD25/IL2RA locus is implicated in a number of autoimmune disorders including multiple sclerosis, but in that case a different SNP is associated with increased risk. The CD25 SNPs that are associated with risk of type 1A diabetes are associated with differences in circulating levels of CD25, which despite extensive overlap between levels, is highly significant when a large number of individuals are studied . The mechanism by which the SNPs contribute to diabetes risk is currently unknown, and may be complex given two associated SNPs in the same locus, neither of which alters the coding sequence.
In a study of 44 monozygotic and 183 dizygotic twin pairs from the Finnish Twin Cohort the 10-year progression rate to diabetes for monozygotic (MZ) twin-mates was 33%, whereas dizygotic (DZ) twin-mates only had a risk of 3% . The study reported probandwise concordance rates of 42.9% for MZ twins and 7.4% for DZ twins . In addition, they reported that an additive model was the best fit for type 1A diabetes risk, suggesting that 88% of phenotypic variance in type 1A diabetes is due to additive genetic effects and 12% is due to environmental factors . Furthermore, a study describing MZ twins in the United States and Great Britain confirmed this high risk, concluding that the risk for type 1A diabetes, by survival analysis, was 39% for the initially discordant twin at 40 years of discordance .
The Diabetes Autoimmunity Study of the Young (DAISY), with principal investigator Marian Rewers, has HLA-typed more than 30,000 newborns and followed more than 2,000 increased risk children from birth (both general population children and relatives of patients with type 1A diabetes). The cumulative incidence of developing expression of persistent anti-islet autoantibodies by age 7 for DAISY DR3/4 siblings of patient with type 1A diabetes is 41%, 16% for DR3/4 offspring of patients with type 1A diabetes, and general population DR3/4 individuals have a 5% cumulative incidence .
DAISY DR3/4 siblings can share 0, 1 or 2 haplotypes identical by descent with their proband sibling . DR3/4 siblings that share both haplotypes identical by descent with the proband had a risk of 63% for developing anti-islet autoantibodies by age 7 and 85% by age 15, whereas siblings that share 0 or 1 haplotypes with the proband had a risk of 20% of developing anti-islet autoantibodies  (Figure 2) and a concordant risk of progression to type 1A diabetes. The observation that there is a large difference in risk between DR3/4 siblings sharing 2 haplotypes identical by descent and those sharing 0 or 1 indicates that there are additional non-DR/DQ loci in or linked to the MHC that contributes to type 1A diabetes risk . “Fixing” only MHC loci can determine extreme genetic risk.
Ziegler and coworkers have reported that DR3/4 individuals with two first degree relatives with type 1A diabetes also have an extreme risk of developing islet autoimmunity and diabetes . For example, if both parents of a child have type 1A diabetes and the child has the DR3/4 genotype, then both the DR3 and DR4 haplotypes by “necessity” are derived from an affected parent. Both haplotypes are present in a proband, and thus is similar to the sibling algorithm of the DAISY study (see above).
The vast majority of individuals who develop type 1A diabetes do not have a first degree relative with the disorder (>85%). Extending the ability to identify extreme risk to the general population without a relative with type 1A diabetes is thus an important goal. In DAISY DR3/4 general population children there is a protective effect of the HLA-DPB1*0402 allele (Figure 3). DR3/4 children without the protective DPB1*0402 allele have a risk of anti-islet autoimmunity of almost 20% .
There is compelling evidence that the incidence of type 1A diabetes is doubling approximately every 20 years in developed countries [58,59], and thus environmental factors are of importance. Environmental determinants of type 1A diabetes are likely to be common, given the ability to define high genetic risk (perhaps similar to celiac disease with almost universal exposure to a dietary inciting factor, gliadin) . A recent report implicates decreased omega-3 polyunsaturated fatty acid ingestion with type 1A diabetes risk , and a large trial is evaluating exclusion of bovine milk from infant diets following completion of a pilot study . Though diet may influence development of anti-islet autoimmunity, the leading hypothesis for the increasing incidence of type 1A diabetes is similar to that for many immune mediated disorders, namely the “hygiene hypothesis”, postulating decreased pathogen exposure contributes to increased autoimmunity .
With the advent of whole-genome SNP genotyping studies in the past several years, many additional non-MHC loci have been identified that contribute to type 1A diabetes risk. In addition, non-DR/DQ loci within the MHC have been identified that contribute to prediction of type 1A diabetes in the general population (i.e. HLA-DP). These will allow for the potential to identify individuals at extreme risk for type 1A diabetes at birth, and currently clinical trials are forming to prevent disease prior to the onset of autoimmunity in genetically susceptible individuals (i.e. Pre-Point). Intervention prior to the onset of autoimmunity may be key, as it has proven difficult to turn off the autoimmune process once it begins.
This work is supported by grants from the National Institutes of Health (DK32083, DK55969, DK62718, AI50864, DK32493, DK064605), the Diabetes Endocrine Research Center grant from the National Institute of Diabetes and Digestive and Kidney Diseases (P30 DK57516), the American Diabetes Association, the Juvenile Diabetes Foundation (JDRF1-2006-16), and the Children’s Diabetes Foundation.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.