Progression likelihood score identifies substages of presymptomatic type 1 diabetes in childhood public health screening

Abstract

Aims/hypothesis: The aim of this study was to develop strategies that identify children from the general population who have late-stage presymptomatic type 1 diabetes and may, therefore, benefit from immune intervention.

Methods: We tested children from Bavaria, Germany, aged 1.75-10 years, enrolled in the Fr1da public health screening programme for islet autoantibodies (n=154,462). OGTT and HbA_1c were assessed in children with multiple islet autoantibodies for diagnosis of presymptomatic stage 1 (normoglycaemia) or stage 2 (dysglycaemia) type 1 diabetes. Cox proportional hazards and penalised logistic regression of autoantibody, genetic, metabolic and demographic information were used to develop a progression likelihood score to identify children with stage 1 type 1 diabetes who progressed to stage 3 (clinical) type 1 diabetes within 2 years.

Results: Of 447 children with multiple islet autoantibodies, 364 (81.4%) were staged. Undiagnosed stage 3 type 1 diabetes, presymptomatic stage 2, and stage 1 type 1 diabetes were detected in 41 (0.027% of screened children), 30 (0.019%) and 293 (0.19%) children, respectively. The 2 year risk for progression to stage 3 type 1 diabetes was 48% (95% CI 34, 58) in children with stage 2 type 1 diabetes (annualised risk, 28%). HbA_1c, islet antigen-2 autoantibody positivity and titre, and the 90 min OGTT value were predictors of progression in children with stage 1 type 1 diabetes. The derived progression likelihood score identified substages corresponding to ≤90th centile (stage 1a, n=258) and >90th centile (stage 1b, n=29; 0.019%) of stage 1 children with a 4.1% (95% CI 1.4, 6.7) and 46% (95% CI 21, 63) 2 year risk of progressing to stage 3 type 1 diabetes, respectively.

Conclusions/interpretation: Public health screening for islet autoantibodies found 0.027% of children to have undiagnosed clinical type 1 diabetes and 0.038% to have undiagnosed presymptomatic stage 2 or stage 1b type 1 diabetes, with 50% risk to develop clinical type 1 diabetes within 2 years.

Keywords: Clinical trial modelling; Glucose tolerance; Immunotherapy; Islet autoantibodies; Population screening; Progression score; Public health screening; Type 1 diabetes.

Fig. 1

Flow of participants in the Fr1da study. DKA, diabetic ketoacidosis

Fig. 2

Progression from stage 1 or stage 2 type 1 diabetes to stage 3 type 1 diabetes. Cumulative risks (lines with 95% CIs indicated with shading) of developing stage 3 type 1 diabetes in children with stage 1 (red) or stage 2 (blue) type 1 diabetes. The follow-up starts at the initial staging using OGTTs. The numbers of children at risk are indicated below each time point

Fig. 3

Multivariable analysis of risk of developing stage 3 type 1 diabetes. The HRs (95% CIs) are shown for variables included in the multivariable CPH analysis that were significantly associated with the development of stage 3 type 1 diabetes in univariable CPH models. IA-2A category as negative (0) and tertiles of IA-2A titres: 3–100 arbitrary units (1), 100–290 arbitrary units (2), >290 arbitrary units (3). Standardised BMI was calculated using the WHO Child Growth Standards based on height and weight and age

Fig. 4

Progression to stage 3 type 1 diabetes in children with stage 1 type 1 diabetes stratified by the progression likelihood score. (a) Sensitivity (in red) and cumulative risk (in black) of developing stage 3 type 1 diabetes within 2 years in children with stage 1 type 1 diabetes, stratified by the tenth centiles of the risk scores calculated using Cox regression (progression likelihood score). Analyses were performed using 287 children with stage 1 type 1 diabetes and complete data for all three variables were included in the risk score (glucose at 90 min in OGTTs, IA-2A titre category and HbA_1c). (b) Cumulative risks (lines with 95% CIs indicated with shading) of developing stage 3 type 1 diabetes in children with stage 1 type 1 diabetes and a risk score >90th centile (in blue) or ≤90th centile (in red). The follow-up starts at the initial staging. The numbers of children at risk are indicated below each time point. T1D, type 1 diabetes

Fig. 5

Progression to stage 3 type 1 diabetes in children with stage 2 or stage 1 type 1 diabetes stratified by the progression likelihood score. The cumulative risks (lines with 95% CIs indicated with shading) of developing stage 3 type 1 diabetes are shown for children with either stage 2 type 1 diabetes or stage 1b, corresponding to stage 1 type 1 diabetes, and a progression likelihood score >90th centile (in blue), children with stage 1a and a progression likelihood score between the 30th and 90th centiles (in green), and children with stage 1a and a progression likelihood score below the 30th centile (in red). The follow-up starts at the initial staging. The numbers of children at risk are indicated below each time point