Diabetes Care, Vol 21, Issue 9 1445-1450, Copyright © 1998 by American Diabetes Association
Two-step islet autoantibody screening for risk assessment of type 1 diabetes in relatives
MR Pastore, E Bazzigaluppi, R Bonfanti, N Dozio, A Sergi, A Balini, C Belloni, F Meschi, E Bonifacio and E Bosi
Department of Internal Medicine, Istituto Scientifico Ospedale San Raffaele, University of Milan, Italy.
OBJECTIVE: To examine the performance of islet cell antibodies (ICAs) and
antibodies to glutamate decarboxylase (GADA), IA-2 (IA-2 antibody [IA-2A]),
and insulin (insulin autoantibody [IAA]), alone and in combination, in
assessing type 1 diabetes risk within type 1 diabetic families to identify
a practical and effective screening strategy for predicting type 1 diabetes
in relatives. RESEARCH DESIGN AND METHODS: ICA, GADA, IA-2A, and IAA were
determined in 806 first-degree relatives participating in a prospective
type 1 diabetes family study (median follow-up 6.17 years, range 0.6-8.3).
The conferred risk of developing type 1 diabetes within 6 years was
evaluated by Kaplan-Meier for each antibody marker, used alone or in
combination. RESULTS: ICAs were detected in 3%, GADA in 5.1%, IA-2A in
2.5%, and IAA in 3.7% of relatives; > or =1 antibody markers were
detected in 10.7% of relatives and > or =2 were detected in 1.9% of
relatives. The risk of type 1 diabetes at 6 years was 1.5% in relatives
with only 1 marker and 24.8% in relatives with > or =2 markers. As a
practical and effective strategy for type 1 diabetes risk assessment in
relatives, this study indicates a first-step screening based on GADA and
IA-2A measurement--which identified 6.5% of relatives, including all who
developed the disease, with a 6-year type 1 diabetes risk of 9.0%--followed
by a second step based on ICA and IAA measurement in relatives with either
GADA or IA-2A, which identified a total of 1.9% of all relatives as having
> or =2 markers, and a 6-year risk of 24.8%, including 6 of 7 who
developed type 1 diabetes. CONCLUSIONS: A two-step antibody screening,
based first on GADA and IA-2A and then on ICA and IAA measurements in
identified individuals, is likely to be a practical, sensitive, and
effective strategy for predicting type 1 diabetes in first-degree
relatives.