Predictive Value of Autoantibodies to IA-2 for Insulin Requirements in Japanese Subjects With Type 1 Diabetes

GAD autoantibodies (GADAs) are frequently detected before the onset of diabetes and indicate the development of insulin dependency (1). Phosphatase-like protein IA-2 antibodies (IA-2As) in combination with GADA have been shown to improve the positive predictive value for type 1 diabetes (2). However, the presence of IA-2A and its clinical usefulness in Japanese type 1 diabetic patients have not been fully determined. Here we report the significance of serum antibodies to IA-2A and GADA to predict the insulin requirement in Japanese patients with type 1 diabetes.

We studied 101 Japanese patients with type 1 diabetes who fulfilled the classification of the American Diabetes Association (3). We included 47 men and 54 women, aged 41.3 ± 15.3 (mean ± SD) years, who had diabetes for 10.4 ± 9.6 years. Serum GADA and IA-2A levels were determined using a commercially available radioimmunoassay kit (RSR, Cardiff, U.K.) and ¹²⁵I-labeled human recombinant GAD65 and ICA512 on the basis of the first proficiency test of the Diabetes Autoantibody Standardization Program (4). The cutoffs were >0.4 units/ml for IA-2A and >1.3 units/ml for GADA.

Of the 101 Japanese type 1 diabetic patients assayed, GADA and IA-2A were detected in 60 (59%) and 37 (37%) patients, respectively. The mean dosage of daily insulin in all 101 subjects was 24 units. We classified subjects into two groups: high insulin dosing (≥24 units/day of insulin, n = 77) and low insulin dosing (<24 units/day of insulin, n = 24). Mean BMI did not differ significantly between these two groups (20.8 ± 2.9 vs. 21.9 ± 4.0 kg/m², P = 0.126). We observed no significant differences in GADA levels between these two groups (Fig. 1). In contrast, the IA-2A levels in the high-insulin dosing group were significantly higher than in the low-insulin dosing group (1.5 ± 2.8 vs. 0.3 ± 0.8 units/ml, P < 0.05) (Fig. 1). Moreover, by setting the cutoff point at 1.5 units/ml, the mean level of IA-2A in the high-insulin dosing group, all of the IA-2A-positive patients would require treatment with ≥24 units/day of insulin (Fig. 1). No cutoff point for GADA level could predict insulin dependency.

We found a significant difference in IA-2A levels, but not in GADA levels, between high- and low-insulin dosing groups. Our observations show that the presence of high-level IA-2As could be useful in predicting insulin requirements in Japanese type 1 diabetic subjects. A large-scale prospective study is required to verify this hypothesis.