A Type 1 Diabetes Genetic Risk Score Can Aid Discrimination Between Type 1 and Type 2 Diabetes in Young Adults

Abstract

Objective: With rising obesity, it is becoming increasingly difficult to distinguish between type 1 diabetes (T1D) and type 2 diabetes (T2D) in young adults. There has been substantial recent progress in identifying the contribution of common genetic variants to T1D and T2D. We aimed to determine whether a score generated from common genetic variants could be used to discriminate between T1D and T2D and also to predict severe insulin deficiency in young adults with diabetes.

Research design and methods: We developed genetic risk scores (GRSs) from published T1D- and T2D-associated variants. We first tested whether the scores could distinguish clinically defined T1D and T2D from the Wellcome Trust Case Control Consortium (WTCCC) (n = 3,887). We then assessed whether the T1D GRS correctly classified young adults (diagnosed at 20-40 years of age, the age-group with the most diagnostic difficulty in clinical practice; n = 223) who progressed to severe insulin deficiency <3 years from diagnosis.

Results: In the WTCCC, the T1D GRS, based on 30 T1D-associated risk variants, was highly discriminative of T1D and T2D (area under the curve [AUC] 0.88 [95% CI 0.87-0.89]; P < 0.0001), and the T2D GRS added little discrimination (AUC 0.89). A T1D GRS >0.280 (>50th centile in those with T1D) is indicative of T1D (50% sensitivity, 95% specificity). A low T1D GRS (<0.234, <5th centile T1D) is indicative of T2D (53% sensitivity, 95% specificity). Most discriminative ability was obtained from just nine single nucleotide polymorphisms (AUC 0.87). In young adults with diabetes, T1D GRS alone predicted progression to insulin deficiency (AUC 0.87 [95% CI 0.82-0.92]; P < 0.0001). T1D GRS, autoantibody status, and clinical features were independent and additive predictors of severe insulin deficiency (combined AUC 0.96 [95% CI 0.94-0.99]; P < 0.0001).

Conclusions: A T1D GRS can accurately identify young adults with diabetes who will require insulin treatment. This will be an important addition to correctly classifying individuals with diabetes when clinical features and autoimmune markers are equivocal.

Conflict of interest statement

The authors declare no conflict of interests.

© 2016 by the American Diabetes Association. Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered.

Figures

Figure 1. The ability of a T1D risk score to discriminate between clinically defined T1D and T2D diabetes in the WTCCC study.

A: A dot plot of T1D GRS by T1D and T2D status. The width of the blue bars represented frequency and the red line is the median. B: An ROC curve of the T1D, T2D and combined genetic risk scores for discriminating between T1D and T2D. The red line represented the T2D GRS, the blue line the T1D GRS and the green line the T1D+T2D GRS. The respective area under the ROC curves are: 0.64, 0.88 and 0.89.

Figure 2. The ability of a T1D genetic risk score to discriminate between severe insulin deficiency and non-insulin deficiency in young adults with diabetes.

A dot plot of T1D GRS by insulin deficiency status. The width of the blue bars represent frequency and the red line is the median.

Figure 3. A series of ROC analyses demonstrating that the T1D-GRS is an additive and independent predictor of insulin deficiency in young-adults with diabetes when compared with known biomarker and clinical discriminators.

T1D-GRS = Type 1 diabetes genetic risk score; AAD = Age at Diagnosis; ABS = autoantibody status for GAD and IA2; ALL includes T1D-GRS, BMI, age at diagnosis and auto-antibodies as predictors. The combination of T1D-GRS, auto-antibodies, BMI and age at diagnosis provides a highly discriminative test.

Figure 4. The incremental increase in discriminatory power of the T1D-GRS as more SNPs are added in.

After ordering by published effect size, this graph shows the increasing AUC for increasing numbers of SNPs. After nine SNPs there is no significantly increased discriminatory power to adding the next SNP. The X-axis starts at 2 SNPs which define the DR3/DR4-DQ8 high-risk haplotypes. The grey bars represent the upper and lower 95% confidence intervals for the AUC estimate.