Simplifying prediction of disease progression in pre-symptomatic type 1 diabetes using a single blood sample

Abstract

Aims/hypothesis: Accurate prediction of disease progression in individuals with pre-symptomatic type 1 diabetes has potential to prevent ketoacidosis and accelerate development of disease-modifying therapies. Current tools for predicting risk require multiple blood samples taken during an OGTT. Our aim was to develop and validate a simpler tool based on a single blood draw.

Methods: Models to predict disease progression using a single OGTT time point (0, 30, 60, 90 or 120 min) were developed using TrialNet data collected from relatives with type 1 diabetes and validated in independent populations at high genetic risk of type 1 diabetes (TrialNet, Diabetes Prevention Trial-Type 1, The Environmental Determinants of Diabetes in the Young [1]) and in a general population of Bavarian children who participated in Fr1da.

Results: Cox proportional hazards models combining plasma glucose, C-peptide, sex, age, BMI, HbA1c and insulinoma antigen-2 autoantibody status predicted disease progression in all populations. In TrialNet, the AUC for receiver operating characteristic curves for models named M60, M90 and M120, based on sampling at 60, 90 and 120 min, was 0.760, 0.761 and 0.745, respectively. These were not significantly different from the AUC of 0.760 for the gold standard Diabetes Prevention Trial Risk Score, which requires five OGTT blood samples. In TEDDY, where only 120 min blood sampling had been performed, the M120 AUC was 0.865. In Fr1da, the M120 AUC of 0.742 was significantly greater than the M60 AUC of 0.615.

Conclusions/interpretation: Prediction models based on a single OGTT blood draw accurately predict disease progression from stage 1 or 2 to stage 3 type 1 diabetes. The operational simplicity of M120, its validity across different at-risk populations and the requirement for 120 min sampling to stage type 1 diabetes suggest M120 could be readily applied to decrease the cost and complexity of risk stratification.