Poster Presentation The 13th International Congress of the Immunology of Diabetes Society 2013

Higher prediction values of ECL-IAA and -GADA assays for type 1 diabetes (#133)

Liping Yu 1 , Andrea Steck 1 , Dongmei Miao 1 , LI ZHANG 1 , K. Michelle Guyer 1 , Ling Jiang 1 , Jeffrey Krischer 2 , Marian Rewers 1 , The TrialNet Study Group
  1. Barbara Davis Daibetes Center, Aurora, CO, United States
  2. University of South Florida, Tampa, FL, United States

Islet autoantibodies (iAbs) precede clinical type 1 diabetes (T1D) by years and are an essential tool for T1D prediction. We recently developed new electrochemiluminescence (ECL) insulin autoantibody (IAA) and GAD65 autoantibody (GADA) assays which discriminated high affinity, high risk diabetes specific autoantibodies from “low risk”, low-affinity positives by radioassay. This is particular value in subjects positive for a single iAb who are generally at low risk for T1D. Here, we report on a pilot study of ECL-IAA and -GADA assays in TrialNet study participants. ECL assay sensitivity, diabetes specificity, and disease prediction values were compare with those current NIDDK standard IAA (mIAA) and GADA radioassays. Of 3484 TrialNet subjects, 2682 were positive for mIAA, GADA, IA-2A and/or ZnT8A radioassays and 384 progressed to T1D. The sensitivity of ECL assay in pre-diabetics were similar to radioassay, 55% (210/384) for ECL-IAA vs 48% (186/384) for mIAA and 79% (304/384) for ECL-GADA vs 77% (295/384) GADA radioassay. Among 2682 Ab+ subjects, 1814 were single iAb+ and majority of them were either mIAA (441) or GADA (1228) only. Only 24% (107/441) of single mIAA+ and 46% (65/1228) of single GADA+ were found positive on their respective ECL-IAA and -GADA assays. In contrast, ECL-IAA were positive for 91% (421/462; p<0.0001) and ECL-GADA positive for 94% (702/743; p<0.0001) for samples from subjects with multiple iAbs. The positive prediction values in terms of progression to T1D for both ECL-IAA (31%, 210/675) and ECL-GADA (23%, 304/1330) were significantly higher than mIAA (21%, 186/903; p<0.0001) and GADA radioassay (15%, 300/1971; p<0.0001). The negative predictive values for both ECL-IAA (94%, 2635/2809) and ECL-GADA (96%, 2074/2154) were also significantly higher than mIAA (92%, 2383/2581; p<0.05) and GADA radioassay (94%, 1429/1513; p=0.01). In conclusion, both ECL-IAA and -GADA assays are more disease specific and are better predictors for development of T1D.