Background: The study is aimed at assessing the possibility of type 1 diabetes mellitus (T1DM) prediction on the basis of autoantibody concentrations and their dynamics. Antibodies against glutamic acid decarboxylase (GADA), tyrosine phosphatase (IA-2A), and zinc transporter 8 (ZnT8A) were assayed. Methods: Regression modeling was applied to repeated measured longitudinal data from a total of 517 participants: 314 children with T1DM and 203 healthy siblings. Results: Among healthy siblings, the high risk of T1DM was associated with 1) high baseline concentration of all three antibodies (an average of 57.5-92.0 times compared with the reference values); 2) significant and rapid decrease in GADA and IA-2A by −23.29 and −43.30 IU/mL/month, respectively; 3) insignificant and very slow declining of ZnT8A level by −5.30 U/mL/month. Conclusions: Modeling of longitudinal GADA, IA-2A, and ZnT8A profiles may be a basis for development of more complex and accurate diagnostic systems. Such an approach appears to be promising but it requires further investigations.
Корнева et al. (Fri,) studied this question.