Genetic Variants Predisposing Most Strongly to Type 1 Diabetes Diagnosed Under Age 7 Years Lie Near Candidate Genes That Function in the Immune System and in Pancreatic β-Cells

Abstract

Objective: Immunohistological analyses of pancreata from patients with type 1 diabetes suggest distinct autoimmune islet β-cell pathology between those diagnosed at <7 years (<7 group) and those diagnosed at age ≥13 years (≥13 group), with both B- and T-lymphocyte islet inflammation common in children in the <7 group, whereas B cells are rare in the ≥13 group. Based on these observations, we sought to identify differences in genetic susceptibility between these prespecified age-at-diagnosis groups to inform on the etiology of the most aggressive form of type 1 diabetes that initiates in the first years of life.

Research design and methods: Using multinomial logistic regression models, we tested if known type 1 diabetes loci (17 within the HLA and 55 non-HLA loci) had significantly stronger effect sizes in the <7 group compared with the ≥13 group, using genotype data from 27,071 individuals (18,485 control subjects and 3,121 case subjects diagnosed at <7 years, 3,757 at 7-13 years, and 1,708 at ≥13 years).

Results: Six HLA haplotypes/classical alleles and six non-HLA regions, one of which functions specifically in β-cells (GLIS3) and the other five likely affecting key T-cell (IL2RA, IL10, IKZF3, and THEMIS), thymus (THEMIS), and B-cell development/functions (IKZF3 and IL10) or in both immune and β-cells (CTSH), showed evidence for stronger effects in the <7 group.

Conclusions: A subset of type 1 diabetes-associated variants are more prevalent in children diagnosed under the age of 7 years and are near candidate genes that act in both pancreatic β- and immune cells.

Figure 1

Classical HLA haplotype/alleles association with type 1 diabetes diagnosed at <7 years old (red circle; mean log-odds ratio age at diagnosis ± 95% CI), 7–13 years old (green circle; mean log-odds ratio age at diagnosis 7–13 ± 95% CI), and ≥13 years old (blue circle; mean log-odds ratio age at diagnosis ≥13 ± 95% CI), from a multinomial logistic regression. Left panel shows the log-odds ratios, with a dashed red line showing a log-odds ratio of 0. The right panel shows the association statistics from a likelihood ratio test comparing a multinomial logistic regression constraining the log-odds ratios from the <7 to ≥13 groups to be equal compared with an unconstrained model. The red dotted line shows nominal significance in heterogeneity (P < 0.05), and the red dashed line shows Bonferroni-corrected significance in heterogeneity. T1D, type 1 diabetes.

Figure 2

Non-HLA type 1 diabetes–associated loci showing on the left panel the log-odds ratios for the minor allele for those diagnosed at <7 years old (red circle; log-odds ratio age at diagnosis ± 95% CI), 7–13 years old (green circle; log-odds ratio age at diagnosis 7–13 ± 95% CI), and ≥13 years old (blue circle; log-odds ratio age at diagnosis ≥13 ± 95% CI), from a multinomial logistic regression; the dashed red line shows a log-odds ratio of 0. The right panel shows the association statistics from a likelihood ratio test comparing a multinomial logistic regression constraining the log-odds ratios from the <7 to ≥13 groups to be equal compared with an unconstrained model. The red dotted line shows the threshold for FDR of <0.1, and the red dashed line shows the threshold for Bonferroni-corrected heterogeneity (showing only loci with an FDR of <0.1). T1D, type 1 diabetes.

Figure 3

Results from colocalization and fine mapping in the IKZF3 region (A), CTSH region (B), and THEMIS region (C). Analyses include individuals from the U.K. and Northern Ireland and only control and case subjects diagnosed at <7 years. The first panel shows the association absolute z scores from a whole-blood eQTL study examining variant effects on IKZF3 (A), CTSH (B), and THEMIS (C) mRNA levels, colored by LD r² to the most strongly associated variant with the respective mRNA expression. The second panel shows the association absolute z scores from logistic regression examining variant associations with type 1 diabetes risk at <7 years, colored by LD r² to the most strongly associated variant with IKZF3 (A), CTSH (B), and THEMIS (C) mRNA. The third panel shows the gene positions (genome build 37), with arrows indicating direction of transcription. The fourth panel shows univariable early-diagnosed type 1 diabetes log-odds ratios and 95% CIs for the minor allele for each of the most likely causally associated variants as prioritized by GUESSFM. The fifth panel shows the log_e(absolute eQTL z score) if z score >0 and −log_e(absolute eQTL z score) if z score <0 for the effect of the minor allele, so direction of effect can be compared, including only eQTLs with a P value <5 × 10⁻¹⁵⁰ (A), <5 × 10⁻⁵⁰ (B), and <5 × 10⁻²⁵ (C). The symbols are colored red if contained in the set of most likely causal variants, as produced by GUESSFM, and the shape corresponds to the gene that the variant is affecting transcription of, with the genes shown in the center of the figure. T1D, type 1 diabetes.