Abstract
OBJECTIVE: To study the predictive value of autoantibodies for type 1 (T1DM) and type 2 (T2DM) diabetes morbidity after gestational diabetes (GDM) in a 23-year follow-up study. DESIGN: Prospective population-based cohort study. METHODS: We studied 391 women with GDM, and 391 age- and parity-matched controls, who delivered in 1984-1994. Four autoantibodies were analysed in first-trimester blood samples: islet cell autoantibodies (ICAs), glutamic acid decarboxylase autoantibodies (GADAs), insulin autoantibodies (IAAs) and insulinoma-associated antigen-2 autoantibodies (IA-2As). Two follow-up questionnaires (1995-1996, 2012-2013) were sent to assess development of T1DM and T2DM. Predictive value of autoantibodies and clinical factors were analysed by conditional linear regression and ROC analyses. RESULTS: Single autoantibody positivity was detected in 12% (41/342) of the GDM cohort and in 2.3% (8/353) of the control cohort. In the GDM cohort, 2.6% (9/342) tested positive for two autoantibodies and 2.3% (8/342) for three autoantibodies, whereas only one subject in the control cohort had two autoantibodies. ICA positivity was found in 12.5% of the cases, followed by GADA (6.0%), IA-2A (4.9%) and IAA (1.2%). In the control cohort, GADA positivity was found in 1.4%, IA-2A in 0.8%, IAA in 0.6%, and ICA in 0.3% of the subjects. Detection of ICA, GADA and/or IA-2A autoantibodies decreased T1DM-free survival time and time to diagnosis. All subjects with three positive autoantibodies developed T1DM within seven years from the GDM pregnancy. Development of T2DM after GDM occurred independent of autoantibody positivity. CONCLUSION: Development of T1DM can be reliably predicted with GADA and ICA autoantibodies during early pregnancy.