OBJECTIVE—Autoantibodies to insulin and GAD are features of preclinical type 1 diabetes in children. For insulin autoantibodies, the antibody affinity and epitope specificity predict which children progress to diabetes. We asked whether autoantibodies to GAD (GADAs) are heterogeneous in affinity and epitope recognition and whether diabetes-related GADA are restricted to high-affinity responses. RESEARCH DESIGN AND METHODS—GADA affinity was measured by competitive binding experiments with [125I]-labeled and -unlabeled recombinant human GAD65 in the first GADA-positive sample from 95 children with a type 1 diabetes family history who were prospectively followed from birth and in follow-up samples from 65 of these children. RESULTS—At first GADA appearance, affinity ranged from 107 to 1010 l/mol. Affinity was higher in multiple islet autoantibody-positive children (P < 0.0001) and in HLA DR3–positive children (P = 0.006). GADA affinities were >109 l/mol in 52 of 53 multiple autoantibody-positive children. In contrast, children who were single GADA positive often had lower affinity GADA and/or GADA with specificities that were restricted to minor NH2-terminal GAD65 epitopes. At follow-up, affinity increased from low to high in 3 of 65 children. All 24 children who developed diabetes had high-affinity GADAs before diabetes onset. CONCLUSIONS—Children develop discrete, heterogeneous antibody responses to GAD that could arise from distinct immunization events, only some of which are diabetes relevant. Subtyping the GADA responses using affinity measurement will improve type 1 diabetes risk assessment.

Among children of patients with type 1 diabetes, the appearance of thyroid autoantibodies is frequent, is not synchronous to the appearance of other autoantibodies and is most common in late childhood and adolescence.