Prediction of type 1 diabetes postpartum in patients with gestational diabetes mellitus by combined islet cell autoantibody screening: a prospective multicenter study.

Women with gestational diabetes mellitus (GDM) have a considerable risk of developing diabetes later in life. To determine the predictive value of autoantibody markers in gestational diabetic pregnancy for the development of type 1 diabetes postpartum, we tested 437 patients with GDM (289 women treated with diet only [GDM-A] and 148 requiring insulin treatment during pregnancy [GDM-B]) for antibodies to islet cells (ICAs), GAD (GADAs), and tyrosine phosphatase ICA512/IA-2 (IA2As). We prospectively followed them with repeated oral glucose tolerance tests and antibody determinations for up to 7 years postpartum (mean, 1.6 years; range, 0-7.2 years). The cumulative risk of diabetes up to 5 years postpartum was 17% (95% CI 12-22%). The risk of type 1 diabetes was 3% (2-5%) by 9 months and 7% (4-9%) 2 years after delivery. At delivery, 8.5% of all patients were ICA+, 9.5% were GADA+, 6.2% were IA2A+, and 18.1% were positive for at least one antibody (12.6% for GDM-A vs. 30.4% for GDM-B, P < 0.0001). During follow-up, GADAs persisted in 75%, ICAs in 35%, and IA2As in 30% of the subjects positive for the respective marker at delivery. By 2 years postpartum, 29% (19-39%) of patients positive for at least one antibody developed type 1 diabetes, compared with 2% (1-4%) of antibody-negative patients (P < 0.0001). Thereby, the risk for type 1 diabetes 2 years postpartum increased with the number of antibodies present at delivery from 17% (6-28%) for one antibody, to 61% (30-91%) for two antibodies, and to 84% (55-100%) for 3 antibodies. Risk of progression to type 1 diabetes postpartum was also associated with the status of parity. Women with one or more pregnancies before the index pregnancy had a higher risk for type 1 diabetes 2 years after delivery (14.7% [4.9.-24.5%]) than women having their first (i.e., index) pregnancy (5% [2.9-7.1%]) (P < 0.006). A comparison of different prediction strategies showed that single antibody screening with GADA yielded the highest sensitivity of 63% (45-75%), compared with ICA (48% [31-65%]) and IA2A (34% [13-47%]). Combined screening with two autoantibodies increased sensitivity to 74% (58-90%) and 75% (60-92%) when using GADA plus ICA or GADA plus IA2A, respectively. Screening with all three markers improved sensitivity further to 82% (67-100%). Beta-cell autoantibodies determined at delivery in women with GDM are highly predictive for the development of type 1 diabetes postpartum. Autoantibody screening in pregnant women with GDM from populations at high risk for type 1 diabetes should therefore be considered to allow early diagnosis and appropriate therapy.