BACKGROUND: Type 1 diabetes includes all forms of autoimmune-mediated and idiopathic beta-cell destruction leading to an absolute insulin deficiency. Evidence of an autoimmune pathogenesis was assessed by studying cytoplasmic islet cell antibodies (ICA), antibodies to glutamic acid decarboxylase (GADA), antibodies reacting with an islet tyrosine phosphatase-related molecule referred to as ICA512 (ICA512A), or its homologue phogrin (phogrin-A). In comparison with ICA, the best validation to assess the risk of type 1 diabetes, shows that a combination of antibodies to GADA with ICA512A has the power to detect a majority of ICA and 97~100% of subjects who progressed to overt diabetes. These findings suggest the possibility of replacing the laborious ICA test in the screening programs to identify subjects at risk of progressing to type 1 diabetes or forclassifying the stage of diabetes at the time of diagnosis. Up to now, it is unclear whether these results are applicable to the slowly progressive type 1 diabetes that appears to be more prevalent in Asian than in western countries. The prevalence of combined autoantibody testing (1 of GADA, ICA512A, or phogrin-A) was investigated in the patients with type 1 diabetes (typical and slowly progressive) and type 2 diabetes, and compared with that of ICA which is a more laborious and insensitive test. METHODS: The ICA assay was performed using immunoenzymatic staining of frozen human (blood group O) pancreatic sections with serial dilutions of serum samples with peroxidase-labeled protein A. For the GADA determination, commercially available GADA radioimmunoassay kits utilizing the 125I-labeled recombinant GAD65 (RSR , United Kingdom) as an antigen was used. Either ICA512A or phogrin-A were detected by a radioligand-binding assay after in vitro transcription and translation using the clone ICA512bdc or phogrin cDNA. Serum was obtainedfrom 76 patients with type 1 diabetes (mean age 22.8+/-14.0 years), 22 patients with slowly progressive type 1 diabetes (mean age 37.9+/-13.9 years) and 39 patients with type 2 diabetes (mean age 45.3+/-12.3 years). Typical and slowly progressive type 1 diabetes patients had the disease for between 4.0+/-4.6 and 10.1+/-9.5 years, respectively at the earliest serum sampling. RESULTS: 1) In typical type 1 diabetes, 30% of patients tested positive for ICA and 57% for the combined autoantibody test (1 of GADA, ICA512A, or phogrin-A). In the slowly progressive type 1 diabetes group, 18% of patients tested positive for ICA and 50% for the combined autoantibody test. In type 2 diabetes, 7.7% and 5.1% tested positive, respectively. 2) Ninety-six percent of ICA-positive patients expressed one or more of the 3 auto-antibody specificities in typical type 1 diabetes. Among the 53 ICA-negative patients with typical type 1 diabetes, 40% had one or more of these auto-antibodies. In the slowly progressive type 1 diabetes, 100% of the ICA-positive and 39% of the ICA- negative patients expressed one or more of the 3 autoantibody specificities. 3) Of the 23 patients with ICA-positive typical type 1 diabetes patients, 87% had a positive result for GADA, 48% for ICA512A, 44% for phogrin-A, and 96% for GADA or ICA512A. Of the 4 patients with ICA-positive slowly progressive type 1 diabetes, three had a positive result for GADA, and 1 for ICA512A. 4) When the prevalence of combined autoantibody testing was analyzed according to the duration of diabetes, the prevalence in patients tested within 4 years after the diagnosis and more than 4 years after the diagnosis was 61% and 52%, respectively in typical type 1 diabetes. Furthermore, that for the ICA was 37% and 21%, respectively. In the slowly progressive type 1 diabetes, the prevalence of combined auto-antibody testing was 88% and 25%, respectively (p<0.05), while that of ICA was 25% and 13%, respectively. 5) In typical type 1 diabetes, ICA were detected more frequently in patients younger than 15 years of age (48%) than in older patients (23%) (p<0.05), while the prevalence of combined auto-antibody testing -was not different according to the onset age (65% vs 53%). CONCLUSION: Combined autoantibody testing for GADA and ICA512A is more sensitive that ICA in type 1 diabetes. Therefore, it could replace the laborious ICA measurement and may be useful for discriminating the etiology of adult onset atypical diabetes.