Comparison of Insulin Autoantibodies in Diabetes-Related and Healthy Populations by Precise Displacement ELISA

Betty M Dean; Jessica M McNally; Ezio Bonifacio; Adrian M Jennings; David B Dunger; Edwin A M Gale; Gian Franco Bottazzo

doi:10.2337/diab.38.10.1275

Original Articles

Comparison of Insulin Autoantibodies in Diabetes-Related and Healthy Populations by Precise Displacement ELISA

Betty M Dean,
Jessica M McNally,
Ezio Bonifacio,
Adrian M Jennings,
David B Dunger,
Edwin A M Gale and
Gian Franco Bottazzo

Department of Immunology, University College and Middlesex School of Medicine, and the Department of Diabetes and Immunogenetics, St. Bartholomew's Hospital London Diabetes Department, Royal Hallamshire Hospital Sheffield Department of Pediatrics, John Radcliffe Hospital Headington, Oxford, United Kingdom

Address correspondence and reprint requests to Dr. B.M. Dean, Department of Immunology, University College and Middlesex School of Medicine, Arthur Stanley House, 40-50 Tottenham Street, London W1P 9PG, UK.

Diabetes 1989 Oct; 38(10): 1275-1281. https://doi.org/10.2337/diab.38.10.1275

Abstract

The second international workshop on insulin autoantibodies (IAAs) demonstrated improved concordance among laboratories with both radioimmunoassay (RIA) and enzyme-linked immunosorbent assay (ELISA) when measurements were based on signals displaceable by preincubation with excess insulin. This feature has been incorporated into our original ELISA for IAAs, and the assay was used to compare IAAs in diabetes-related and healthy populations. The serum from a healthy islet cell antibody (ICA)⁺ and IAA⁺ subject was used to construct standard curves with and without preincubation with 1 U/ml human insulin. Reporting results in arbitrary displacement (Δ ± IAA) units derived from the standard curve improved precision and increased the specificity of the assay. Frequency analysis of the results from 200 control adults did not show a normal distribution, and cumulative frequency analysis demonstrated two populations: 10 of 200 (5%) control adults, 8 of 241 (3.3%) healthy schoolchildren, and 18 of 229 (7.9%) non-insulin-dependent diabetes mellitus (NIDDM) patients had IAAs >12 Δ ± IAA units. On the same basis, we tested samples from 89 individuals in the prospective Bart's Windsor Family Study for insulin-dependent diabetes mellitus (IDDM), 31 of whom had ICAs >5 Juvenile Diabetes Foundation (JDF) units. Of those with ICAs >5 JDF units, 12 developed IDDM and 1 developed NIDDM in 10 yr of study. IAAs >12 Δ ± IAA units were found in 6 of 12 (50%) who became diabetic and in 4 of 18 (22%) who remain healthy. A significant association of IgG IAAs with ICAs >5 JDF units is confirmed in this IDDM-susceptible cohort (P < .0005) but was not observed in the other groups studied. The predictive value of a positive test with IgG IAAs alone is poor (20%), but the combined positive predictive value of both markers conjointly for IDDM is 60%, and the predictive value for health is 98.7%. We suggest that screening for IAAs by ELISA should be restricted to those with ICAs >5 JDF units.