Genetic Risk Score Modelling for Disease Progression in New-Onset Type 1 Diabetes Patients: Increased Genetic Load of Islet-Expressed and Cytokine-Regulated Candidate Genes Predicts Poorer Glycemic Control

Abstract

Genome-wide association studies (GWAS) have identified over 40 type 1 diabetes risk loci. The clinical impact of these loci on β-cell function during disease progression is unknown. We aimed at testing whether a genetic risk score could predict glycemic control and residual β-cell function in type 1 diabetes (T1D). As gene expression may represent an intermediate phenotype between genetic variation and disease, we hypothesized that genes within T1D loci which are expressed in islets and transcriptionally regulated by proinflammatory cytokines would be the best predictors of disease progression. Two-thirds of 46 GWAS candidate genes examined were expressed in human islets, and 11 of these significantly changed expression levels following exposure to proinflammatory cytokines (IL-1β + IFNγ + TNFα) for 48 h. Using the GWAS single nucleotide polymorphisms (SNPs) from each locus, we constructed a genetic risk score based on the cumulative number of risk alleles carried in children with newly diagnosed T1D. With each additional risk allele carried, HbA1c levels increased significantly within first year after diagnosis. Network and gene ontology (GO) analyses revealed that several of the 11 candidate genes have overlapping biological functions and interact in a common network. Our results may help predict disease progression in newly diagnosed children with T1D which can be exploited for optimizing treatment.

Figure 1

Cytokine-regulated candidate genes in human islets. Isolated human islets were left untreated or exposed to cytokines (IL-1β + IFNγ + TNFα) for 48 h. Gene expression of candidate genes was determined by real-time PCR. Target gene expression was normalized to the geometric mean of three housekeeping genes. (a) Genes upregulated in response to cytokine treatment. (b) Genes downregulated in response to cytokine treatment. Data are means ± SEM of n = 8-9, except for IL10 (n = 3). ^∗ p < 0.05, ^∗∗ p < 0.01, and ^∗∗∗ p < 0.001.

Figure 2

Correlation between HbA1c and IDAA1c levels and risk allele numbers. HbA1c (a) and IDAA1c (b) in carriers with <25% (n = 65), 25–75% (n = 96), or >75% (n = 21) risk alleles at 1, 3, 6, 9, and 12 months following disease onset. Data are means ± SEM, ^∗ p < 0.05, ^∗∗∗ p < 0.001.

Figure 3

Protein interaction network of the 11 genes. The network was constructed using the STRING tool (http://string-db.org) and the 11 candidate genes as input. The width of the interactions depends on the confidence score to each association in STRING.