The stage- and subgroup-specific impact of non-HLA polymorphisms on preclinical type 1 diabetes progression

Abstract

Besides variation within the HLA gene complex determining a major part of genetic susceptibility to Type 1 diabetes, genome-wide association studies have identified over 60 non-HLA loci also contributing to disease risk. While individual single nucleotide polymorphisms (SNPs) have limited predictive power, genetic risk scores (GRS) can identify at-risk individuals. However, current models do not fully capture the heterogeneous progression of asymptomatic islet autoimmunity, especially in autoantibody-positive subjects. In this study, we investigated the additional stage-specific impact of 17 non-HLA loci on previously established prediction models in 448 persistently autoantibody-positive first-degree relatives. Cox regression and Kaplan Meier survival analysis were used to assess their influence on progression from single to multiple autoantibody-positivity, and from there to clinical onset. FUT2 and CTSH significantly accelerated progression of single to multiple autoAb-positivity, but only in presence of insulin autoantibodies and HLA-DQ2/DQ8, respectively. At the stage of multiple autoantibody-positivity, progression to clinical onset was impacted by various non-HLA SNPs either as independent predictors (GLIS3, CENPW, IL2, GSDM, MEG3A, and NRP-1) or through interaction with HLA class I alleles (CLEC16A, NRP-1, TCF7L2), maternal diabetes status (CTSH), or a high-risk autoantibody-profile (CD226). Our data indicate that, unlike for GRS, the weight of distinct non-HLA polymorphisms varies significantly among individuals at risk, depending on disease stage and other stage-specific risk factors. They refine our previous stage-specific prediction models including age, autoantibody-profile, HLA genotype, and other non-HLA SNPs, and emphasize the importance of stratifying accordingly to personalize time-to-event prediction in risk groups, or for preparing or interpreting prevention trials.

Keywords: Autoimmunity; Non-HLA SNPs; Prediabetes; Prediction; Type 1 diabetes.

Fig. 1

Interaction effects of non-HLA SNPs on progression of single to multiple autoAb-positivity. Kaplan-Meier survival analysis of progression from single to multiple autoAb-positivity for FDRs with (red) or without (blue) FUT2 CC according to the presence (A) or absence (B) of IAA as first autoAb; CTSH CC according to the presence (C) or absence (D) of HLA-DQ2/DQ8; and the absence of genotypes CTLA4 AG and CTSH CC (green), the presence of only one of these genotypes (blue), and the presence of both genotypes (red), in HLA-DQ2/DQ8+ (E) and HLA-DQ2/DQ8- (F) relatives. P-values (p) are from the log-rank test on differences in survival. The numbers of individuals at risk are indicated below the time axis. For each arm, the genotype and number (cases/events) are shown in the legend.

Fig. 2

Independent effects of non-HLA SNPs on progression from multiple autoAb-positivity to clinical onset. Kaplan-Meier survival analysis for FDRs with (red) or without (blue) CENPW CC (A), IL2 AG (B), GSDM CT (C), GLIS3 CC (D), NRP-1 AA (E) or MEG3A AA (F) genotype. P-values (p) are from the log-rank test on differences in survival. The numbers of individuals at risk are indicated below the time axis. For each arm, the genotype and number (cases/events) are shown above the graph.

Fig. 3

Interaction effects of non-HLA SNPs on progression from multiple autoAb-positivity to clinical onset. Kaplan-Meier survival analysis for HLA-A∗24 positive (A,C) or negative (B,D) FDRs with (red) or without (blue) NRP-1 AA (A,B) or TCF7L2 CC (C,D) genotype, and for FDRs with (red) or without (blue) CLEC16A GG, according to the presence (E) or absence (F) of for HLA-B∗18. P-values (p) are from the log-rank test on differences in survival. The numbers of individuals at risk are indicated below the time axis. For each arm, the genotype and number (cases/events) are shown above the graph.