Issue 135 Item 15 Combinations Of Antibodies Predict Diabetes Risk

Measuring levels of antibodies against beta-cell components can replace islet cell antibody (ICA) measurements as a screening procedure in relatives of people with type 1 diabetes. Researchers from the Hopital Edouard Herriot and INSERM 449, Lyon, France, enrolled 3,951 first-degree relatives of people with type 1 diabetes. The researchers examined expression of antibodies to insulin, glutamic acid decarboxylase (GAD) and thyrosine phosphatase (IA-2) as well as ICA levels. The authors assessed the first-degree relatives, who did not initially show diabetes, on a median 2.2 occasions over a maximum 16-year follow up.

On the first examination, 3.6 percent of the relatives expressed GAD antibodies. Furthermore, 4.9 and 2.2 percent expressed IA-2 and insulin antibodies respectively, while 1.1 percent showed high ICA titers. These antibodies were commoner in the ICA-positive relatives than the ICA-negative subjects.

Combining antibodies resulted in a specificity of more than 90 percent. The positive predictive value for developing diabetes depended on the number of positive antibodies. However, combining antibodies against either GAD and IA-2 or GAD and insulin showed higher predictive values and were more sensitive than using high ICA titers. Including insulin autoantibodies increased the predictive value, although sensitivity declined. The combination of GAD and IA-2 showed a sensitivity and predictive value of 87.8 and 13.8 percent respectively.

The authors concluded that measuring combinations of antibodies to beta cell constituents can replace ICA as a screening procedure in relatives of people with type 1 diabetes. However, antibody-positive relatives vary markedly in the time that they take to develop diabetes. Finally, the researchers suggested that genetic marker importance needs further investigation.
Diabetes Metab 2002;28:279-85.