More than 90% of children presenting with type 1A diabetes express one of three anti-islet autoantibodies (autoantibodies reacting with insulin [IAA], an islet enzyme termed glutamic acid decarboxylase 65 [GAD65], or a molecule of unknown function termed ICA512 [or IA-2]) (4). These autoantibodies are typically present years prior to the development of diabetes. A relatively simple rule relates most of our current knowledge for prediction of diabetes, namely expression of two or more of the above autoan-tibodies is associated with a high risk of progression to diabetes (see Fig. 3) (4,44,45). Expression of a single anti-islet autoantibody is associated with a risk of approx 20%, whereas most individuals (but not all) expressing multiple anti-islet autoantibodies progress to diabetes over the next 5-10 yr. Insulin autoantibodies appear to be unique in that they are usually the first autoantibody to appear in the youngest children developing type 1A diabetes (46), and a high level of the autoantibody correlates with a more rapid progression to diabetes. For adults developing type 1A diabetes, GAD65 autoantibodies provide the highest sensitivity.
Although autoantibody testing has rapidly improved, the technology is relatively young, and when being applied to individuals, there are important caveats, as listed in Table 2. We currently do not utilize the cytoplasmic islet cell autoantibody test in that it has proven difficult to standardize, and when such antibodies are present in the absence of what are termed "biochemical" autoantibodies, diabetes risk is low. Insulin antibodies are induced by insulin therapy, even with human insulin; thus, several weeks postinsulin therapy, one may be measuring induced antibodies. Autoantibodies are not always static and they disappear in a few children who progress to diabetes prior to the onset of diabetes. One should not rely upon a single determination of autoantibodies. The presence of antibodies at more than one time-point has a much greater clinical relevance.
Although we have concentrated on autoantibodies, it is likely that (comparable to animal models of type 1A diabetes) the T-lymphocytes destroy islet P-cells (47,48), and
Caveats of Autoantibody Testing
Cytoplasmic islet cell antigen (ICA) assay is difficult to standardize.
Radio-binding assays rather than enzyme-linked immunosorbent assay (ELISA) should be utilized.
Laboratories differ in sensitivity and specificity: specificity should be set at >99%. Insulin antibodies develop after insulin therapy.
Transient antibodies may occur and appear to have little prognostic significance. Young prediabetic children have high frequency of IAA, whereas prediabetic adults have a high frequency of GAD65.
autoantibodies are of marginal pathogenic significance in humans. Their lack of pathogenesis is suggested by the absence of diabetes in infants born to mothers expressing anti-islet autoantibodies. At present, assays for autoreactive T-lymphocytes are in their infancy, and a recent T-cell workshop suggested that proliferation assays were not sufficient to distinguish autoreactive T-cells associated with diabetes (49). Two new techniques that are being applied offer some optimism, namely quantitation of individual T-cells producing cytokines (enzyme-linked immuno spot [ELISPOT]) (50) and analysis of T-cells with MHC complexed with peptides (termed "tetramer analysis") (51,52).
Was this article helpful?