Aims/hypothesis: Islet autoantibodies are important in diabetes classification and risk assessment, and as endpoints in observational studies. The Diabetes Autoantibody Standardization Program (DASP) aims to improve and standardise measurement of autoantibodies associated with type 1 diabetes. We report results for glutamic acid decarboxylase autoantibodies (GADA) and islet antigen-2 autoantibodies (IA-2A) from three DASP workshops (2002--2005).
Methods: Up to 60 laboratories in 18 countries participated in each workshop. Participants received coded serum aliquots from 50 patients with newly diagnosed type 1 diabetes (median age 18 years, range 9-35 years) and 100 blood donor controls. Results were analysed using receiver operator characteristic (ROC) curves with sensitivity adjusted to 95% specificity in workshop controls.
Results: GADA assays performed well in all three workshops (median area under the ROC curve [AUC] 0.94; interquartile range 0.91-0.95) and performance was similar to DASP 2000. Performance of IA-2A assays improved over the workshop programme. Median AUC was 0.81 (interquartile range 0.79-0.83) in DASP 2002, 0.82 (interquartile range 0.78-0.84) in 2003, and 0.85 (interquartile range 0.82-0.87) in 2005 (p < 0.0001). Performance of GADA ELISA improved between 2002 and 2005, and, in DASP 2005, achieved higher median AUC and adjusted sensitivity than RIA. IA-2A ELISA improved and, in DASP 2005, achieved AUCs equivalent to in-house RIA. Assays using IA-2ic or full length IA-2 clones were more sensitive than those using IA-2bdc, with higher AUC (p = 0.004).
Conclusions/interpretation: GADA and IA-2A assays perform well in discriminating health and disease. The workshop format highlights systematic differences related to assay method and allows full evaluation of novel methods. The programme of autoantibody workshops in type 1 diabetes provides a model for other autoimmune diseases.