부산대학교 도서관

BACKGROUND: For cost-effective population-based diabetes prediction and confirmation, islet autoantibody assays must be made more economical. METHODS: We evaluated glutamic acid decarboxylase (GAD)-Ruc (renilla luciferase) and IA2ic (also known as ICA512ic)-Ruc (renilla luciferase) fusion protein constructs in high-throughput islet antibody assay formats. RESULTS: Antigen production via transfection onto COS cells in 100 mm culture dishes yielded sufficient antigen to assay 375 and 535 serum samples for GAD and IA2ic per dish, respectively. Antigen was usably stable after − 80 °C storage for 40–80 days after which luciferase activity decreased. The mean signal-to-noise ratios for luciferase-based immunoprecitation system (LIPS) GAD and LIPS IA2ic were 88 ± 24 and 219 ± 89, respectively, comparing favourably to radio-binding assays (RBA) in the same format. However, the coefficient of variation among triplicate wells was higher for IA2ic than for GAD in LIPS, similar to findings in RBA format. Correlation coefficients between autoantibody indices determined from the RBA and LIPS methods were only R = 0.79 and R = 0.75 for GAD and IA2ic, respectively, raising the possibility that different epitopes were favoured in the two different assay formats. Nevertheless, overall concordance for the two assay types was high, at 228/240 = 95.0% for GAD and 494/521 = 94.8% for IA2ic. Using optimal cutoffs, Diabetes Autoantibody Standardization Program (DASP) 2010 sensitivity/specificity was 80/99% for GAD RBA, 80/99% for GAD LIPS, 70/98% for IA2 RBA, and 72/99% for IA2 LIPS. CONCLUSION: The LIPS assays for islet autoantibodies to GAD and IA2ic performed as well as RBA in DASP 2010. With further refinements in expression and storage, these assays may be more economical than current methods to measure islet autoantibodies in type 1 diabetes. Copyright © 2011 John Wiley & Sons, Ltd.