Abstract
OBJECTIVE: Childhood obesity may impact the risk of islet autoimmunity (IA). The trajectory of BMI through childhood resembles the early peak incidence of first-appearing autoantibodies against insulin (IAA-first) but not GAD65 (GADA-first). We studied whether a child's BMI can impact the age-related risk of first-appearing IA phenotypes. RESEARCH DESIGN AND METHODS: We identified 7,724 children at risk for IA with at least three BMI measurements in The Environmental Determinants of Diabetes in the Young (TEDDY) study. We modeled the risk of IAA-first, GADA-first, and IA overall on a child's BMI z score and change in BMI during infancy (age 2 weeks to 1.5 years, n = 7,724), early childhood (age 1.5-8.5 years, n = 6,396), and puberty (age 8.5-15 years, n = 4,732) using joint modeling of longitudinal BMI and time-to-event IA. RESULTS: An infant's BMI z score was not associated with IA risk before 18 months of age (n = 185, hazard ratio [HR] 1.03 [95% CI 0.88, 1.19]). In contrast, a child's BMI correlated with an increased risk of IA from 1.5 to 8.5 years of age (n = 470, HR 1.20 [95% CI 1.04, 1.32]) and from 8.5 to 15 years of age (n = 209, HR 1.27 [95% CI 1.09, 1.49]). No interactions with first-appearing IA phenotypes were observed. However, high BMI z score (SD >0.5) from age 9 months increased the risk of IA in early childhood, specifically for children with HLA-DR4/4 or HLA-DR4/8 and not with HLA-DR3/3 or HLA-DR3/4 (HLA ∗ BMI interaction, P < 0.005). CONCLUSIONS: The contribution of BMI to risk of IA during early childhood is dependent on the HLA-DR-DQ genotype more so than the first-appearing IA phenotype.