High‐resolution genotyping indicates that children with type 1 diabetes and celiac disease share three HLA class II loci in DRB3, DRB4 and DRB5 genes

Type 1 diabetes (T1D) and celiac disease (CD) share common genetic loci, mainly within the human leukocyte antigen (HLA) class II complex. Extended genotyping of HLA class II alleles and their potential risk for developing both diseases remains to be studied. The present study compared extended HLA‐class II gene polymorphisms in children with T1D, CD, and a subgroup diagnosed with both diseases (T1D w/CD). Next‐generation targeted sequencing (NGTS) of HLA‐DRB3, DRB4, DRB5, DRB1, DQA1, DQB1, DPA1, and DPB1 alleles from DNA collected from 68 T1D, 219 CD, and seven T1D w/CD patients were compared with 636 HLA‐genotyped Swedish children from the general population selected as controls. In comparison to controls, the DRB4*01:03:01 allele occurred more frequently in T1D w/CD (odds ratio (OR) = 7.84; 95% confidence interval (95% CI) = (2.24, 34.5), P = 0.0002) and T1D (OR = 3.86; 95% CI, (2.69, 5.55), P = 1.07 × 10−14), respectively. The DRB3*01:01:02 allele occurred more frequently in CD as compared to controls (OR = 7.87; 95% CI, (6.17, 10.03), P = 4.24 × 10−71), but less frequently in T1D (OR = 2.59; 95% CI, (1.76, 3.81), P = 7.29 × 10−07) and T1D w/CD (OR = 0.87; 95% CI, (0.09, 3.96), P ≤ 0.999). The frequency of the DRB4*01:03:01‐DRB1*04:01:01‐DQA1*03:01:01‐DQB1*03:02:01 (DR4‐DQ8) haplotype was higher in T1D w/CD (OR = 12.88; 95% CI (4.35, 38.14) P = 3.75 × 10−9), and moderately higher in T1D (OR = 2.13; 95% CI (1.18, 3.83) P = 0.01) compared with controls, but comparable in CD (OR = 1.45; 95% CI (0.94, 2.21), P = 0.08) and controls. Children with T1D and CD are associated with DRB4*01:03:01, DRB3*01:01:02, and DRB3*02:02:01 of which DRB4*01:03:01 confers the strongest risk allele for developing T1D w/CD.


| INTRODUCTION
Type 1 diabetes (T1D) and celiac disease (CD) are two of the most common autoimmune diseases in the western world. 1,2 Estimated incidence rates of T1D continue rising by 1.4% annually, 3 and CD prevalence is estimated at 1.4% worldwide. 4 CD occurs in 3% to 16% of patients with previously diagnosed T1D. 5,6 Conversely, individuals with prior CD are at a 3-fold increased risk for T1D before the age of 20. 7 The risk of developing both diseases is thus significantly higher compared to the general population (GP), which is proposed to be partly explained by shared genetics. 8,9 Albeit both diseases are also associated with major histocompatibility complex (MHC) class I gene variants, 10 the strongest genetic association has been attributed to specific MHC class II HLA-DRB1, -DQA1 and -DQB1 genes that are in linkage disequilibrium. 11 Nearly all patients with CD and T1D carry either the DRB1*03-DQA1*05:01-DQB1*02:01 (DR3-DQ2.5) or DRB1*04-DQA1*03-DQB1*03:02 (DR4-DQ8) haplotypes. 12,13 The distribution of these two haplotypes in different genotypes further confers different risk for T1D and CD. Individuals homozygous for the DR3-DQ2.5 genotype are at the highest risk for developing CD, whereas DR3-DQ2.5/DR4-DQ8 is the high-risk genotype for T1D. 14 The risk of developing both T1D and CD (T1D w/CD) has previously been attributed to being homozygous for DQ2 or heterozygous for DQ2.5/DQ8. 15,16 Although the HLA-DRB1, -DQA1 and -DQB1 genotypes mentioned above are prerequisites for CD and T1D, additional genes likely contribute to the disease risk. 16,17 To date, genome-wide association studies (GWAS) have identified 41 non-HLA loci associated with CD 18 and 50 susceptibility genes associated with T1D. 19 Of these, only three non-HLA loci were identified as related to T1D w/CD; RGS1 on chromosome 1q31, IL18RAP on chromosome 2q12 and TAGAP on chromosome 6q25, 20 respectively.
Since next-generation targeted sequencing (NGTS) for extended genotyping of the HLA gene complex was developed, new associations between subtypes of DRB1, DRB3, DRB4 and DRB5 and T1D have been found. 21 In a previous study applying NGTS, we found that the risk of TID was further modulated by having the DRB3*01:01:02 and DRB3*02:02:01 alleles, or not. 22 When using the same methodology for full-length HLA-genotyping, it was later discovered that the DRB3*01:01:02 and DRB3*02:02:01 alleles also distinguished the risk for CD in DR3-DQ2 carriers. 23 However, studies of HLA-DR and DQ polymorphisms that contribute to common genetic estimation for both diseases remain to be investigated. The aim of the present study was to extend the findings of our two previous investigations and perform high-resolution genotyping using NGTS in children with T1D and CD in the search for shared extended HLA class II loci in children that develop both diseases.

| Study subjects
Included were children prospectively followed in a birth cohort screened for T1D and CD between 2004 and 2010 at the Unit of Diabetes and Celiac disease, Department of Clinical Sciences, Lund University, Malmö, Sweden, as described elsewhere. 24 Two hundred nineteen children were diagnosed with CD (137 females, 82 males) at median age 4.5 (range 1.1-11.0) years according to ESPGHAN criteria, 25 68 children were diagnosed with T1D (39 females, 29 males) at median age 5.5 (range 0.9-11.3) years according to the American Diabetes Association criteria, 26 and seven children (5 females, 2 males) were diagnosed with T1D w/CD (Table 1). Representing the general population (GP), 448 healthy Swedish children 27 and 188 healthy children randomly selected from the LifeGene prospective cohort study were included as controls. 28 Local ethical review board approval and parental informed consent were obtained for all the study participants.
2.2 | HLA class II high-resolution sequencing DNA was extracted from either a dried blood spot punch or a small volume whole blood lysate specimen format from the study subjects. HLA-class II allele sequencing was performed using the ScisGo HLA v4 typing kit (in collaboration with Scisco Genetics Inc., Seattle, Washington). Known haplotypes were used to phase the extended haplotypes and to predict the genotypes. Amplicon-based 2-stage polymerase chain reaction (PCR)-based amplification of HLA-class II alleles and sequencing-by-synthesis approach by using fluorescently labeled reversible terminator nucleotides with MiSeq v2 PE500 (Illumina, San Diego, California) technology was performed as previously described. 29,30 Briefly, the laboratory steps comprise consecutive PCR reactions with barcoding incorporated into the PCRs for individual DNA sample tracking. After assay-specific amplification, samples were tagged with unique indexes and were pooled together and applied to the MiSeq device, where they were amplified as individual clusters and ordinarily sequenced using universal sequencing primers. Subsequently, sequences were analyzed using genetic system software to report unambiguous HLA-class II alleles and haplotypes to patients' samples simultaneously.

| Statistical analysis
Allelic frequency distribution analysis of HLA-class II genes was performed using the relative predispositional effects (RPE) analysis. 30 Crude odds ratios (ORs) and their associated 95% confidence intervals (95% CI) were calculated, and χ 2 tests and Fisher's exact tests (if any cell contained fewer than three observations) were used to test whether the frequencies of a given allele/haplotype differed between cases and noncases. The RPE method was used to identify the disease risk alleles, haplotypes or genotypes with the strongest predisposing or protective effects at each iteration. The selected alleles were then removed from the dataset, and the analysis was repeated until no risk or protective alleles were identified. Comparisons of DR and DQ allele frequencies were performed for both exons 2 and 3 of chromosome 6p21 by performing pairwise comparisons between all study groups and listed in order of increasing P-value, followed by the extended haplotype and genotype frequencies. In the HLA-DR locus, alleles of HLA-DRB1 are in linkage disequilibrium with alleles of either HLA-DRB3 or DRB4 or DRB5. Hence, by treating these highly dependent DR subtypes and their allelic variations as different alleles, analysis of HLA-DRB1 and the secondary DRB3, DRB4 or DRB5 alleles by PRE method produces estimates of frequencies for all haplotypes provided that expectation numbers of corresponding haplotypes are five or more copies in both patients and control subjects. P-values ≤.05 were considered statistically significant, and alleles with a low frequency (≤1%) were not shown in the analysis. The P-values presented are nominal and not adjusted for multiple comparisons. Analyses were performed in R (r-project.org) version 3.6.1 and R package epiDisplay version 3.5.0.1.

| CD children compared with controls
HLA-DRB345, -DRB1, -DQA1 and -DQB1 allele frequencies in children with CD compared with controls are summarized in Table S1. The allelic distributions of the disease associated DRB alleles differed significantly between all the patient groups and the controls ( Table 2  and Table S5

| DISCUSSION
The present study aimed to fill the gap of knowledge of why some children with either T1D or CD are at an increased risk for developing both diseases (T1D w/CD) by analyzing extended HLA class II genes. The main findings were that the DRB4*01:03:01, DRB3*01:01:02, and DRB3*02:02:01 alleles, were found to be associated with T1D w/CD. Of these three alleles DRB4*01:03:01 was associated with T1D only, DRB3*01:01:02 was associated with CD only, but inversely associated with T1D only. The DRB3*02:02:01 allele was associated with a low predisposition in all three groups. The novelty of the study lies in the fact that several alleles in DRB3, DRB4 and DRB5 among T1D w/CD children seem to have an extended HLA polymorphism more similar to that in children with T1D than that in children with CD.
Dissecting the extended DRB1-DRB3-DRB4-DRB5 haplotypes, DRB4*01:03:01-containing haplotypes conferred a positive association with T1D w/CD as well as T1D in comparison to DRB3*01:01:02-containing haplotypes that were positively associated with CD. Around 50% of T1D and 70% of T1D w/CD haplotypes carry the former allele compared with 60% of CD haplotypes which carry the latter allele. Noting that all copies of chromosome 6  On a genotype level, differences between T1D w/CD and T1D children were found of whom T1D w/CD children were more likely to be homozygous for DR4-DQ8/DR4-DQ8 compared with T1D children. In line with other studies, the DR3-DQ2 haplotype occurred in over a third of T1D children, 30 and as previously showed, DR3-DQ2/DR4-DQ8 was the most frequent genotype. 31 Moreover, HLA-DQ2.5 homozygosity was more common among CD children compared with T1D w/CD and T1D children, confirming the HLA dosage effect of DR3-DQ2 on the risk of CD. 32 These results are all in line with a previous Norwegian study, 33 and a study conducted on Dutch patients, 10 which showed that the T1D risk heterozygous genotype (DQ2.5/DQ8) provided a comparable frequency with T1D w/CD. In contrast, Bakker et al 34 reported that HLA-DQ2.5 homozygosity is expected in 30% of the T1D w/CD group, indicating that a double dose of DQ2.5 confers the highest risk for T1D patients to develop CD as shown previously. 35 The suggested hypothesis for the increased susceptibility to CD and T1D coexistence is the putative presence of DQ heterodimers encoded by alleles in trans, in addition to the DQ molecules encoded by alleles in cis, in linkage disequilibrium; DQA1*05, and DQB1*02, encoding the DQ2.5 molecule, and DQA1*03 and DQB1*03, encoding the DQ8 molecule. 36 Consequently, the associations of these two DR-DQ genotypes (DR3-DQ2.5/DR3-DQ2.5 and DR3-DQ2/DR4-DQ8) with CD and T1D indicate that the mechanism of autoimmune susceptibility may partly be overlapping.
The strengths of the present study were the use of high-resolution NGTS for extended HLA genotyping, which enabled examining the disease susceptibility between T1D w/CD and extended HLA-DRB3, DRB4 and DRB5 alleles. The RPE analysis used to estimate the association between HLA alleles (or haplotypes) and each of the outcomes (T1D only, CD only, T1D w/CD) accounts for the fact that a high frequency of a given allele "induces" a lower frequency of all other alleles, as their total must remain constant. 37 A limitation of the study was the small sample size for the comparison of genotype effects both within and between the three disease groups. Secondly, the study included study participants from a single site comparing children at high-risk genotypes constituting a relatively homogeneous population with little HLA diversity. It cannot be ruled out that shared HLA loci may be different in other populations.