Article Text

The effect of the MHC locus on autoantibodies in type 1 diabetes
  1. H-Q Qu,
  2. C Polychronakos
  1. Departments of Pediatrics and Human Genetics, McGill University, Montreal, Québec, Canada
  1. Dr C Polychronakos, Departments of Pediatrics and Human Genetics Director, Pediatric Endocrinology McGill University Health Center (Children’s Hospital), 2300 Tupper, Montréal, QC Canada H3H 1P3; constantin.polychronakos{at}


Objective: To investigate whether the presence of autoantibodies specific for type 1 diabetes (T1D) is determined by the major genetic susceptibility locus for the disease at the HLA genes, using the T1D Genetics Consortium data.

Methods: We analysed anti-IA-2 and anti-GAD 65 autoantibody data from 2282 T1D patients from 1117 multiplex families. HLA genotyping was available for all cases and their parents and association with autoantibody positivity was tested by the transmission disequilibrium test.

Results: Association of anti-IA-2 with the HLA genes was detected at high statistical significance. HLA-DRB1*0401 confers both the strongest type 1 diabetes risk, and positive association of anti-IA-2, whereas the DRB1*03- DQA1*0501-DQB1*0201 haplotype, associated less strongly with T1D, showed a significant negative association with anti-IA-2 positivity. Interestingly, HLA-A*24 is also negatively associated with anti-IA-2, independently of the DRB1*03- DQA1*0501-DQB1*0201 haplotype. No statistically significant association was identified between anti-GAD65 and HLA.

Conclusions: This study highlights that IA-2 as an autoantigen depends on HLA genotype and suggests new insights into the mechanism of loss of immune tolerance.

Statistics from

Type 1 diabetes (T1D) is due to T cell mediated autoimmune destruction of the pancreatic β cells. The role of humoral autoimmunity in this destruction is less clear but antibodies against a number of T1D specific autoantigens are detectable in most cases. The protein tyrosine phosphatase-2 (also known as islet antigen-2, IA-2, encoded by PTPRN) and the 65 kD glutamate decarboxylase 2 (GAD65) are the most highly predictive.1 However, a significant proportion of patients diagnosed with T1D show absence of either or both autoantibodies.2 To which extent this represents heterogeneity in the causative cellular autoantigens is an unanswered question. Such heterogeneity may be due to the effect of genotype at the major T1D susceptibility locus that involves the highly polymorphic antigen presenting proteins of the human leukocyte antigen (HLA) genes.

The association between the HLA genes and autoantibodies in T1D has been reported by previous studies. A consistent finding was the association between anti-IA-2 and the HLA class II genes—for example, the positive association between anti-IA-2 and HLA DR4,35 and the negative association between anti-IA-2 and HLA DQ2.6 In this study, we aimed to examine systematically the association of anti-IA-2 and anti-GAD65 positivity with both HLA class I and class II genes in a large family cohort.



This study was based on the December 2007 data release by the T1D Genetics Consortium (T1DGC) ( The T1DGC genotyped HLA-A, HLA-B, HLA-Cw, HLA-DPA1, HLA-DPB1, HLA-DQA1, HLA-DQB1, and HLA-DRB1, in 10 072 individuals from 2315 families with at least two children diagnosed with diabetes (4951 T1D cases). Criteria of T1D diagnosis were age at diagnosis below 35 years and uninterrupted treatment with insulin within 6 months of diagnosis. For siblings of probands diagnosed under the age of 35, the age-at-diagnosis limit was extended to 45 if they were lean and had positive antibodies and/or low C peptide values at diagnosis. HLA genotypes and results of anti-IA-2 and anti-GAD65 testing were available in 2282 T1D patients from 1117 multiplex families with HLA genotyping data. The patients were recruited worldwide but the majority of the patients (>90%) are of European descent. We compared autoantibody(−) versus autoantibody(+) patients in terms of the transmission ratio of the eight HLA genes by the transmission disequilibrium test (TDT), an approach immune to population stratification.


For the transmission disequilibrium test (TDT) we used the TDTPHASE program in the UNPHASED software (∼fdudbrid/software/unphased/).8 A total of 130 comparisons were performed (the eight HLA genes have a total of 130 alleles) tested for association with positivity for two autoantibodies. We used the Bonferroni corrected significance threshold α = 0.05/260 = 1.92×10−4. This is a conservative approach since, because of linkage disequilibrium (LD), the HLA loci examined are associated with each other.


Among 2282 T1D patients, 696 (30.5%) patients are autoantibody(−), and the remaining 1586 (69.5%) patients have at least one autoantibody(+). The statistics are summarised in table 1. The distribution of anti-IA-2 has no obvious gender difference, whereas anti-GAD65 has lower prevalence in male patients. For either anti-IA-2 or anti-GAD65, the antibody(−) group had younger age-of-onset of T1D and longer disease duration at the time of sampling.

Table 1 General information of the type 1 diabetes patients

Since our study was performed on cases only, autoantibody association with the HLA genes is independent of the effect of HLA on T1D risk. No statistically significant difference was observed in the transmission ratio of each HLA allele in the anti-GAD65 (−) group versus the anti-GAD65(+) group. The lowest p value of 8.72×10−3 is larger than the corrected significant level α = 1.9×10−4. By contrast, differences in transmission on the basis of anti-IA-2 reactivity met the Bonferroni corrected significance threshold (p<1.92×10−4) in five HLA alleles (table 2). As shown by previous studies, autoantibodies may disappear years after the disease onset,9 which can explain the phenomenon that the antibody(−) group had longer disease duration at the time of sampling, as shown in table 1. Therefore, the antibody(−) cases with long disease duration may have been antibody(+) previously. To address this issue, we compared the transmission ratio of the HLA alleles between the antibody(−) cases with disease duration ⩽10 years and the antibody(−) cases with disease duration >10 years. We found no statistical difference of the transmission ratio of the five anti-IA-2-associated alleles in terms of disease duration. The anti-IA-2 associations of these HLA alleles remain valid if we look at short disease duration and long disease duration separately (table 3).

Table 2 The anti-IA-2 association of the HLA genes in type 1 diabetes (T1D) cases
Table 3 The type 1 diabetes (T1D) disease duration and the anti-IA-2 association of the HLA genes

HLA-DRB1*0401 entails both the strongest T1D risk and positive association of anti-IA-2 (table 2). The T1D cases with at least one HLA-DRB1*0401 allele have higher prevalence of anti-IA-2 (57.7%) than the T1D cases without the HLA-DRB1*0401 allele (36.3%). This result confirms a recent report of higher anti-IA-2 prevalence in patients carrying HLA-DRB1*04.5 In addition, we found that HLA class II alleles were negatively associated with anti-IA-2. HLA-DQB1*0201, HLA-DRB1*03, and HLA-DQA1*0501 entail medium T1D risk, and are negatively associated with anti-IA-2 (table 2). DRB1*03, DQA1*0501, and DQB1*0201 are in tight LD and compose a remarkably conserved, T1D predisposing haplotype.10 The T1D association of the DRB1*03-DQA1*0501-DQB1*0201 haplotype has an odds ratio (OR) of 2.06 (95% confidence interval (CI) 1.91 to 2.22), while the association with anti-IA-2 has an OR of 0.63 (95% CI 0.51 to 0.77), p = 6.06×10−6. The T1D cases with at least one DRB1*03-DQA1*0501-DQB1*0201 haplotype have lower prevalence of anti-IA-2 (41.2%), than the T1D cases without the DRB1*03-DQA1*0501-DQB1*0201 haplotype (55.6%). Therefore, in this large cohort of T1D cases, we validated a previous report that anti-IA-2 is less common in patients with DQ2 haplotype.6

Interestingly, we found that HLA class I allele HLA-A*24 was also negatively associated with anti-IA-2. HLA-A*24 is in very low LD with the HLA class II alleles DRB1*03, DQA1*0501, and DQB1*0201 (supplemental fig 1). The negative association cannot be explained by the LD. In the T1D cases without any of the DRB1*03, DQA1*0501, and DQB1*0201 alleles, HLA-A*24 is negatively associated with anti-IA-2 with an OR of 0.57 (95% CI 0.38 to 0.84), p = 0.004.

Key points

  • The type 1 diabetes (T1D) autoantibodies, anti-IA-2 and anti-GAD65A, are reliable markers of the autoimmune destruction of pancreatic β-cells. However, a significant proportion of patients diagnosed with T1D show absence of either or both autoantibodies.

  • The anti-IA-2 has been reported to be associated with the major T1D genetic susceptibility locus HLA by previous studies—that is, positive association with HLA DR4 and negative association with HLA DQ2.

  • By the analysis of 2282 T1D patients from 1,117 multiplex families, this study finds no association between anti-GAD65 and HLA, and validates positive anti-IA-2 association of HLA-DRB1*0401 and negative anti-IA-2 association of the DRB1*03- DQA1*0501-DQB1*0201 haplotype.

  • A novel finding of this study is the negative anti-IA-2 association of HLA-A*24, independent of the DRB1*03- DQA1*0501-DQB1*0201 haplotype.

IA-2 is a major islet cell autoantigen.11 The auto-antigenicity is from the cytoplasmic domain of IA-2.12 IA-2 is involved in glucose stimulated insulin secretion and its targeted disruption can result in alterations of insulin secretion.13 Our findings show that the relative importance of IA-2 as an autoantigen depends on HLA genotype and permit some insight into the mechanism of loss of tolerance. As shown by Sidney et al,14 the IA-2 epitopes GVAGLLVALAV (586–596) and MSSGSFINISV (499–509) can bind with DQB1*0201, but not DRB*0401. Our data are consistent with the conclusion that efficient presentation of IA-2 induces immune tolerance and inhibits anti-IA-2 reactivity. This may, at least partially, explain the somewhat lower predisposition conferred by DQB1*0201 than by DRB1*0401. The role of HLA class I genes in T1D was highlighted by recent genetic study.15 Our study suggests that anti-IA-2 may be involved in the HLA class I genetic effect in T1D.


This research utilises resources provided by the Type 1 Diabetes Genetics Consortium, a collaborative clinical study sponsored by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), National Institute of Allergy and Infectious Diseases (NIAID), National Human Genome Research Institute (NHGRI), National Institute of Child Health and Human Development (NICHD), and Juvenile Diabetes Research Foundation International (JDRF), and is supported by U01 DK062418.


Supplementary materials


  • Additional figure is published online only at

  • Funding: HQQ is supported by a fellowship from the Canadian Institutes of Health Research.

  • Competing interests: None.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.