DISCORDANT RESULTS IN THE ASSAY OF CYTOPLASMIC ISLET-CELL AUTOANTIBODIES WITH ELISA AND IMMUNOHISTOCHEMICAL TECHNIQUES

We evaluated 6 batches of a solid phase enzyme-linked immunosorbent as say (ELISA) Isletest(TM)-ICA kit commercially available for the determ ination of autoantibodies to pancreatic islet cells, and compared the results with those obtained by a standardized immunohistochemical meth od. Following the immunohistochemical determination of autoantibodies to pancreatic islet cells, sera from patients with insulin-dependent d iabetes mellitus, both positive and negative for autoantibodies to pan creatic islet cells, were randomly selected and analysed by ELISA. Ser a from healthy control subjects, as well as standards recommended by t he International Diabetes Workshop (IDW) ICA (Autoantibodies to Pancre atic Islet Cells) Proficiency Program, were included. Of the sera test ing positive for autoantibodies to pancreatic islet cells in the immun ohistochemical assay, only 14 +/- 5% were found to give a positive rea ction in the ELISA. Among the sera from healthy control subjects and p ancreatic islet cell autoantibody-negative insulin-dependent diabetes mellitus patients, 25 +/- 7% and 1 +/- 1%, respectively, yielded false -positive readings for autoantibodies to pancreatic islet cells. These results clearly show that the ELISA test presently available does not reliably detect autoantibodies to pancreatic islet cells, even qualit atively. Thus, it cannot be used for screening subjects at risk of dev eloping diabetes.