Objective: To design a dynamic prediction model for estimating the time of progression from a single glutamic acid decarboxylase autoantibody (GADA) to multiple islet autoantibodies and type 1 diabetes in children, exploring different longitudinally measured risk variables.

Research Design and Methods: GADA-positive children (n = 379) participating in The Environmental Determinants of Diabetes in the Young (TEDDY) study were followed for the appearance of additional autoantibodies against either insulin autoantibody (IAA), insulinoma-like 2 autoantibody (IA-2A), or zinc transporter 8 antibody (ZnT8A) and type 1 diabetes. A dynamic prediction model was designed, including trajectories of longitudinal risk variables, autoantibody titers, and metabolic variables (C-peptide, glucose, and HbA1c) together with time-invariant variables (gender, age at GADA positivity, and high-risk HLA genotypes).

Results: Transition risk from GADA to multiple autoantibodies was increased by lower age (p < 0.001) and by increased GADA titers during follow-up (p < 0.001), and was less likely in children with HLA DQ2/X but not DQ2/8 (p = 0.004). The transition risk from multiple autoantibodies without IA-2A to IA-2A positivity was associated with increased levels of 2 h glucose following oral glucose tolerance test (OGTT) (p < 0.001) and increased ZnT8A titers (p < 0.001). Increasing HbA1c (p < 0.001) and GADA titers (p < 0.001) were associated with an increased risk of transition from GADA only to type 1 diabetes; while increasing HbA1c (p < 0.001) was associated with the transition from multiple autoantibodies to type 1 diabetes. Risk of transition from multiple autoantibodies, including IA-2A to type 1 diabetes was also associated with 2 h glucose level (p < 0.001).

Conclusion: The dynamic prediction model presented an individual time-specific risk of transition from a single GADA to multiple autoantibodies and type 1 diabetes.