Type I (insulin-dependent) diabetes mellitus results from destruction of the beta cells of the islets of Langerhans, usually leading to absolute insulin deficiency. Although some patients who are insulin dependent have no immune changes, in the majority the destructive process is thought to be immune mediated [1]. The world-wide incidence of the disease varies, ranging from 1.7/100,000 person-years in Japan to 29.5/100,000 person-years in Finland [2]. In Western industrialised countries, Type I diabetes is the second most common chronic childhood illness after asthma. In the United States the life time prevalence of Type I diabetes approaches 0.4%[3]. Familial clustering of Type I diabetes emphasises the role of genetic factors in disease susceptibility. A major portion of this clustering is due to sharing of alleles at susceptibility loci in the major histocompatibility complex on chromosome 6, most notably HLADQB1 and HLA-DRB1, as well as at least 13 other loci on nine chromosomes [4]. None-genetic, probably environmental, factors are thought to operate in these genetically susceptible subjects over a limited period in early childhood to initiate the disease process [5]. The nature of these putative environmental factors is still not known; candidates include viruses, toxins and dietary factors [6]. At the clinical onset of Type I diabetes, 60±80% of islets are deficient in beta cells and the islets can be infiltrated with mononuclear cells [7]. Mononuclear cell infiltration occurs principally around islet cells containing insulin, but is not a consistent feature. These mononuclear cells include macrophages and T lymphocytes and CD8 positive T cells appear to be most prevalent [8, 9]. In the years before the clinical onset of Type I diabetes, immune and metabolic changes can be detected in peripheral blood [1]. The immune changes involve both cellular and humoral responses which persist over a prolonged period up to diagnosis of the disease [10]. The nature, intensity, extent and persistence of these immune changes distinguishes people who develop Type I diabetes from those who do not [1, 5, 10]. Immune and metabolic changes can, therefore, be predictive of the disease. The long prediabetic period and the potential for prediction has led to attempts to prevent the clinical disease [11]. It is widely believed that Type I diabetes is due to an autoimmune process [1]. Whereas the evidence for autoimmunity in humans is circumstantial, animal models of the disease, such as the non-obese diabetic (NOD) mouse and the Bio-Breeding rat, support an autoimmune process leading to diabetes [12, 13].

Identification of disease-associated autoantigens, that act as targets of the immune response, has dual importance: firstly, antibodies to the molecules could act as predictive markers, and secondly, modulation of the immune response to an antigen might alter the disease course. Islet cell autoantibodies (ICA), recognising islet cytoplasmic antigens, were detected many years ago as a feature of newly-diagnosed Type I diabetic patients and comprise autoantibodies to a number of antigens including glutamic acid decarboxylase (GAD) and protein tyrosine phos-