Prediction not only reflects our understanding of the mechanisms underlying type 1 diabetes, but also makes it possible to plan and monitor interventions. In this respect our ability to predict type 1 diabetes probably exceeds that in any other complex disease. The disease process begins early in life, and prospective studies from birth have shown that islet autoantibodies appear in the circulation within the first years of life. The risk of progression to diabetes is largely confined to those with multiple autoantibody combinations, and these carry a >70% risk of clinical diabetes within 10 years. In the later stages, target organ function deteriorates, and the first sign of failing beta cell function is loss of the first-phase insulin response (FPIR) to intravenous glucose. Individuals with multiple autoantibodies and loss of FPIR have a >90% 5-year risk of clinical diabetes. Oral glucose tolerance begins to decline at around the same time, and impaired glucose tolerance is the next step on the road to overt diabetes.

What does 'prediction' mean?

Huge strides have been made in predicting the onset of type 1 diabetes. Risk models, constructed on Bayesian principles, allow the cumulative probability of diabetes development to be established. Thus, the baseline risk within a given population is increased ~15-fold by a first degree family history of early-onset type 1 diabetes, and similar levels of risk can be assigned to those with no family history who carry high-risk HLA haplotypes. These measures establish prior probability. Islet autoantibody measurement demonstrate the presence of autoimmunity directed against islet constituents, and prospective studies have shown that high risk of progression to diabetes is largely confined to those who have multiple autoantibodies, assumed to signify progressive expansion of the anti-islet immunity. Children with a family history of type 1 diabetes who carry three or more autoantibody types have a >90% risk of eventual progression to diabetes. Metabolic testing adds the dimension of target organ function, and allows the disease process to be staged. Thus, individuals who have lost their first-phase insulin response (FRIR) to intravenous glucose, or who have developed glucose intolerance, have an approximate 90% risk of progression within 5 years of testing. This degree of prediction is almost unequalled in other metabolic or immune conditions, and has provided the basis for trials of immune intervention before the onset of overt hyperglycaemia.

Islet autoantibodies

About 95% of newly diagnosed patients test positive for one or more islet autoantibodies. The presence of islet cell antibodies (ICA) was first detected by indirect immunofluorescence (Figure), and represented a composite of antibodies directed against a range of molecular entities. Direct measurement of these entities, which include insulin and proinsulin, glutamic acid decarboxylase (GAD), islet-autoantigen 2 (IA-2) and the zinc transporter ZnT8A, has now superseded the old test for ICA. These antigens have all been synthesised in recombinant form, which has allowed the development of sensitive high performance radiobinding assays. Conformational epitopes of these antigens have also been used in the attempt to improve the specificity of prediction.

Cell-mediated immune markers

Cellular immune mechanisms are considered to be responsible for beta cell killing, and it is therefore logical to expect that markers of cellular immunity directed against the beta cell would offer optimal prediction of diabetes. This potential has yet to be fully realised, largely because of the technical difficulty of the measurements involved. Both CD4+ and CD8+ T cells infiltrate the islets in the course of insulitis, and HLA class I–restricted CD8 T-cells would be the logical effector pathway, since these recognize peptide epitopes presented by HLA class I molecules on the surface of the cell. CD8+ cell counts have been noted to increase following beta cell transplantation, and to fall in the year following diagnosis of type 1 diabetes. To this expent they appear to mirror the extent of beta-cell destruction, but much more work will be needed before these assays could be considered of clinical value.[1]

Markers of beta cell function

Insulin secretory activity varies widely in the healthy population, since different individuals require different amounts of insulin to maintain similar levels of blood glucose. Measurement of circulating insulin or C-peptide is thus of limited value in the early prediction of diabetes. The insulin response to intravenous glucose shows a sharp first-phase increase, lasting for a few minutes only, followed by a slow sustained second phase response. Partial or total abolition of the first phase response is the earliest sign of impending secretory failure, and is widely used in diabetes prevention trials. This change, which is also seen in the early stages of type 2 diabetes, is also mirrored by progressive loss of tolerance to oral glucose. Many individuals pass through a stage of impaired glucose tolerance, typically lasting several months, before overt symptoms appear. The oral glucose tolerance test is also required to define the onset of diabetes (which is not necessarily symptomatic) in prevention trials.

Markers of beta cell mass

The total beta cell mass in a healthy individual is about the size of a pea, and is dispersed into millions of islets. Indirect measurement via tests of insulin secretory function can be deceptive, since a proportion of beta cells may be in the resting state at any one time, and it would therefore be highly desirable to be able to measure this mass directly, for example by scanning techniques. Unfortunately, the signal-to-noise ratio is too low for this to be a feasible procedure using current external scanning devices, and the beta cell mass at onset of type 1 diabetes is currently in some dispute.

See Also: Empirical risks


  1. ^ Roep BO. Islet autoreactive CD8+ cells in type 1 diabetes. Licensed to kill? Diabetes 2008;57:1156–7


Nobody has commented on this article

Commenting is only available for registered Diapedia users. Please log in or register first.