Delayed time to stage 3 type 1 diabetes after GAD-alum treatment in HLA-DR3-DQ2-positive children: follow-up of the randomised placebo-controlled Diabetes Prevention – Immune Tolerance trial

Abstract Aims/hypothesis The aim of this study was to assess the long-term treatment effects of alum-formulated human recombinant GAD65 (GAD-alum) after completion of the Diabetes Prevention – Immune Tolerance (DiAPREV-IT) trial in 2017. A second aim was to determine whether the HLA - DR3-DQ2 haplotype affected the treatment efficacy in delaying or preventing stage 3 type 1 diabetes. Methods In the investigator-initiated RCT DiAPREV-IT (EUdraCT number 2008-007484-16/ClinicalTrials.gov registration no. NCT01122446), children 4–17.99 years old with GADA and at least one additional islet autoantibody were randomised 1:1 to receive two subcutaneous injections of 20 µg GAD-alum ( n =25) or placebo ( n =25), 30 days apart, with a 5 year follow-up. Randomisation was stratified according to islet autoantibody status (two vs three or more islet autoantibodies), and participants were assigned randomisation numbers sequentially according to a prespecified randomisation list. All study personnel, participants and caregivers were masked to treatment allocation. The RCT did not reach significant treatment effect in delaying or preventing stage 3 type 1 diabetes in the predefined analyses. In this long-term follow-up, diagnosis and the date of onset of stage 3 type 1 diabetes were obtained from the Swedish National Diabetes Registry, or, for unregistered participants, through telephone interviews after informed consent. The time from treatment start to stage 3 type 1 diabetes diagnosis was analysed using Cox proportional hazards regression, with Kaplan–Meier estimates being used to visualise survival curves and estimate median time to stage 3 type 1 diabetes diagnosis. Incidence rates were analysed using Poisson regression. Both analyses were performed in the full population and in the subgroups with and without the HLA - DR3-DQ2 haplotype. Results After a median 12.9 years follow-up (range 1.0–13.8 years), 35 of 49 participants progressed to stage 3 type 1 diabetes. One of the original 50 participants was excluded from the analysis of the follow-up study due to their glucose values at the baseline OGTT in the original DiAPREV-IT study meeting the threshold for stage 3 type 1 diabetes. Among the 27 participants with the HLA-DR3-DQ2 haplotype, 22 developed the disease. In the overall population, GAD-alum showed no significant effect on progression (HR 1.017; 95% CI 0.523, 1.977; p =0.960; n =49). In HLA-DR3-DQ2 -positive participants, GAD-alum significantly delayed time to stage 3 type 1 diabetes (HR 0.315; 95% CI 0.120, 0.823; p =0.018; n =27), while an increased rate of progression was observed in individuals without the HLA-DR3-DQ2 haplotype ( HR 3.500; 95% CI 1.052, 11.630; p =0.040; n =22). In HLA-DR3-DQ2 -positive participants, Kaplan–Meier estimates showed a median time to stage 3 type 1 diabetes of 9.0 years with GAD-alum compared to 3.4 years with placebo ( p =0.096), and the incidence rate ratio was 0.419 (95% CI 0.176, 0.997; p =0.049; n =27). Conclusions/interpretation In the current study, GAD-alum significantly delayed progression to stage 3 type 1 diabetes in HLA-DR3-DQ2 -positive children with multiple islet autoantibodies, while the opposite was observed in the subgroup without this haplotype, supporting the importance of precision medicine-based study designs and the potential of GAD-alum as a treatment option to delay progression to clinical onset. Trial registration ClinicalTrials.gov NCT01122446