Using the manufacturer's suggested cutoff values for anti-glutamic acid decarboxylase antibodies (GADA) results in intermediate levels (1–2 U/mL) in which the patient's status is unclear. Our aim was to re-evaluate these thresholds by characterizing patient phenotype according to GADA level.

All patients under 25 years of age who underwent GADA assessment by CentAK® radiobinding assay (RBA) in a context of type-1 diabetes (T1D) between January 2020 and May 2023 were retrospectively included. Clinical phenotype, age at evaluation, number and nature of positive islet autoantibodies, and C-peptide and GADA levels were analyzed.

The study involved 219 subjects: 110 with recent onset of T1D and 109 asymptomatic at-risk relatives from T1D families. Most patients with T1D had positive GADA levels (> 2 U/mL) whereas most at-risk relatives had negative levels ( < 1 U/mL). Seventeen subjects (7.8%) had intermediate GADA levels. On ROC analysis, 2 U/mL and 1 U/mL were the best cutoffs, yielding diagnostic sensitivity of 80% and 90% and specificity of 91.7% and 88.1%, respectively. Inter-assay reproducibility was higher at 2 U/mL than at 1 U/mL. GADA-positive T1D patients significantly more often showed multiple islet autoantibodies than those with negative or intermediate levels.

We propose a single > 2 U/mL threshold for the CentAK® RBA to identify GADA-positive children and young adults, as it yields higher specificity and identifies patients prone to a wider anti-islet autoimmune reaction.