Aims/hypothesis: Insulin autoantibodies (IAA) are important in type 1 diabetes risk assessment. However, their determination varies more between laboratories than other diabetes autoantibodies. The Diabetes Antibody Standardization Program (DASP) aims to improve and standardise measurement of autoantibodies associated with type 1 diabetes. We report the results of measurement of IAA from DASP workshops in 2002, 2003 and 2005.
Methods: Up to 32 laboratories in 14 countries participated in each workshop. Aliquots of coded sera from 50 patients with newly diagnosed type 1 diabetes and 100 blood donor controls were circulated to participating laboratories. Reported results were analysed using receiver operator characteristic (ROC) curves. We compared concordance of antibody levels by ranking, IAA and insulin antibody (IA) indices and units derived from an IA standard curve.
Results: In all three workshops IAA assay performance had improved compared with DASP 2000. The median area under the ROC curve was 0.73 in DASP 2002, 0.78 in 2003 and 0.80 in 2005 (p = 0.0012), and median laboratory-assigned sensitivity was 26% in 2002, 36% in 2003 and 45% in 2005 (p < 0.0001). There was, however, marked variation between assays. The range of AUC was 0.36-0.91 and that of laboratory-assigned sensitivity was 22-57%. Concordance of ranking of patient serum samples was related to AUC (p < 0.001). Using an index related to common IAA and IA-positive or -negative control sera improved the concordance between assays (p < 0.0001).
Conclusions/interpretation: The overall performance of IAA assays has improved but there is still wide variation between laboratories. Concordance between assays would be improved by the use of a common reference reagent.