Impact of diabetes susceptibility loci on progression from pre-diabetes to diabetes in at-risk individuals of the diabetes prevention trial-type 1 (DPT-1)

Abstract

Objective: The unfolding of type 1 diabetes involves a number of steps: defective immunological tolerance, priming of anti-islet autoimmunity, and destruction of insulin-producing beta-cells. A number of genetic loci contribute to susceptibility to type 1 diabetes, but it is unclear which stages of the disease are influenced by the different loci. Here, we analyzed the frequency of type 1 diabetes-risk alleles among individuals from the Diabetes Prevention Trial-Type 1 (DPT-1) clinical trial, which tested a preventive effect of insulin in at-risk relatives of diabetic individuals, all of which presented with autoimmune manifestations but only one-third of which eventually progressed to diabetes.

Research design and methods: In this study, 708 individuals randomized into DPT-1 were genotyped for 37 single nucleotide polymorphisms in diabetes susceptibility loci.

Results: Susceptibility alleles at loci expected to influence immunoregulation (PTPN22, CTLA4, and IL2RA) did not differ between progressors and nonprogressors but were elevated in both groups relative to general population frequencies, as was the INS promoter variant. In contrast, HLA DQB1*0302 and DQB1*0301 differed significantly in progressors versus nonprogressors (DQB*0302, 42.6 vs. 34.7%, P = 0.0047; DQB*0301, 8.6 vs. 14.3%, P = 0.0026). Multivariate analysis of the factors contributing to progression demonstrated that initial titers of anti-insulin autoantibodies (IAAs) could account for some (P = 0.0016) but not all of this effect on progression (P = 0.00038 for the independent effect of the number of DQB*0302 alleles). The INS-23 genotype was most strongly associated with anti-IAAs (median IAA levels in TT individuals, 60 nU/ml; AT, 121; and AA, 192; P = 0.000037) and only suggestively to the outcome of oral insulin administration.

Conclusions: With the exception of HLA, most susceptibility loci tested condition the risk of autoimmunity rather than the risk of failed immunoregulation that results in islet destruction. Future clinical trials might consider genotyping INS-23 in addition to HLA alleles as disease/treatment response modifier.

FIG. 1.

CTLA4 haplotype representation in DPT-1 individuals. A: Allelic composition of major CTLA4 haplotypes computationally reconstructed in DPT-1 individuals. For reference, corresponding frequencies in various population groups are indicated (CEPH-HGDP DNA panel [28]). B: Skewing (OR) of CTLA4 SNPs/haplotype frequency in DPT-1 individuals when compared with ethnicity-matched control cohorts. C: Frequencies of major CTLA4 haplotypes in DPT-1 and control cohorts.

FIG. 2.

Diabetes-free survival and HLA-DQB1 genotypes. Diabetes-free survival in all (n = 638) (A) or high-risk (n = 295) (B) individuals screened at age <10,000 days according to their HLA-DQB1*0302 or HLA-DQB1*0301 genotypes. Significance of differences in survival is evaluated by log-rank test. The number of diabetes-free individuals in each category and at distinct time point is shown on the bottom of the figure. C: Disease-free survival in individuals screened at age <5,000 or between 5,000 and 10,000 days, stratified based on their HLA-DQB*0302 or *0301 genotypes.

FIG. 3.

Response to treatment according to HLA-DQB1 and INS-23 genotype. A: All 638 individuals were stratified according to the number of HLA-DQB1*0302 alleles and treatment group. Significance of differences in survival is evaluated by log-rank test. I, intervention group; P, placebo/observation. B: Individuals were stratified based on the number of INS-23A risk alleles.

FIG. 4.

A: Cox proportional hazard modeling was applied to evaluate the individual contributions of DQB1 status and initial IAA levels to progression toward type 1 diabetes. B: Diabetes-free survival according to DQB1*0302 status fitted on averaged IAA levels in the Cox model.