A combined analysis of whether islet cell autoantibodies (ICAs) are cross-reactive with mouse pancreas, with glutamate decarboxylase (GAD) antibodies, and with 64K antibodies was performed in a large sample of recently diagnosed type I diabetic patients. The disappearance rates of these different autoantibodies were compared in some patients after onset of the disease. The aims were to determine patterns in GAD/64K antibodies with regard to cross-species reaction of ICA and to assess whether GAD could contribute to ICA positivity in mouse and human pancreases and whether the simultaneous search for all the antibody specificities enhances the detection of autoimmune stigma.
ICA detected by immunofluorescence in human and mouse pancreases, antibodies immunoprecipitating the 64K rat islet antigen, and antibodies immunotrapping brain GAD activity were quantified at diagnosis of diabetes in 95 patients and in sequential samples during 1 year after diagnosis in 13 patients. The contribution of GAD to ICA positivity in mouse and human pancreases was evaluated by the analysis of correlations between tests and by the ability of brain homogenate to block ICA reactivity in pancreases from both species.
ICAs were detected in human pancreases in sera from 63 (66%) patients, among which 61% bound also to a mouse pancreas. GAD and 64K antibodies were strongly correlated (P < 0.0001) and were detected in 69 and 73% of the patients, respectively. All but two patients with ICA in human pancreas also displayed either ICA in mouse pancreas or GAD/64K antibodies. Among 32 patients without ICA in human pancreas, 54% displayed either GAD/64K antibodies or ICA in mouse pancreas. Only 16% of the patients displayed neither ICA nor GAD/64K antibodies. A correlation (P < 0.005) was found between ICA in human and mouse pancreases. GAD or 64K antibodies were strongly correlated with ICA in human pancreas (P < 0.0001), but not with ICA in mouse pancreas. After preincubation of six sera with GAD-containing brain homogenate, ICA titers were unaffected in mouse pancreas but reduced in human pancreas. ICA titers in mouse pancreas were decreased after 3 months (P < 0.01) in diabetic patients, contrasting with the stability of ICA in human pancreas and GAD antibodies by 1 year after diagnosis.
According to cross-species reaction, we confirm the heterogeneity of ICA in a large series of type I diabetic patients, ICAs that cross-reacted with mouse pancreas being more frequent than ICAs without cross-species reactivity. GAD and 64K antibodies were also present in a majority of patients. The simultaneous search for all the antibody specificities enhances the detection of autoimmune stigma so that only a few patients did not display any autoantibody at diagnosis. GAD is not the target of ICAs in mouse pancreas, whereas GAD accounts for ICA positivity in human pancreas. The conclusion that ICAs in mouse pancreas are not GAD-reactive is reinforced by the fact that they are more transient after onset of diabetes than are GAD antibodies or the complex mixture of ICAs in human pancreas.