Stacking the Deck: Studies of Patients with Multiple Autoimmune Diseases Propelled Our Understanding of Type 1 Diabetes as an Autoimmune Disease

Although symptoms of diabetes were first written about in 1500 bc, the etiology of the disease was a mystery until Minkowski and von Mering found that the pancreas governs glucose control (1). This led to the discovery and use of insulin in 1922 (2), enabling the survival of children with diabetes. Himsworth (3) subsequently described two types of people with diabetes: those who were insulin resistant and those who were insulin sensitive. The clinically defined terms insulin-dependent and non–insulin-dependent diabetes mellitus (4, 5) gave way by 1997, respectively, to type 1 diabetes (T1D) as a disease caused by immune-mediated β cell destruction and type 2 diabetes (T2D) as primarily a disease of insulin resistance (6).

The featured articles by Bottazzo et al. (7) and MacCuish et al. (8) in 1974 were instrumental in developing our understanding that the immune system is involved in T1D, because autoantibodies directed at the pancreatic islet cell were found for the first time. Interestingly, these seminal articles identified the autoantibodies in individuals with immune diseases affecting multiple organs. In both articles, the pivotal aspect of the research design was the selection of individuals with diabetes and one or more additional organ-specific autoimmune diseases. Many of these individuals likely had autoimmune polyglandular syndrome 1 (APS1 or autoimmune polyendocrinopathy candidiasis ectodermal dystrophy), a clinical constellation classically manifested by adrenal insufficiency, hypoparathyroidism, and mucocutaneous candidiasis but also associated with vitiligo, pernicious anemia, and T1D. These symptoms are now known to be caused by heterozygous or homozygous mutations in AIRE, which codes for autoimmune regulator (9–11). AIRE deficiency or related genetic defects lead to a failure of central T cell tolerance and the development of tissue-specific autoantibodies and, ultimately, multiorgan disease. Recent studies show that individuals who lack AIRE develop somatically mutated autoantibodies of very high affinity for conformational epitopes (12). Such high-affinity Abs are more easily detected than lower-affinity Abs, and this may explain why Bottazzo et al. and MacCuish et al. were able to identify islet Abs in this cohort. The study of these rare individuals shed light on the role of autoimmunity in T1D.

The articles by Bottazzo et al. and MacCuish et al., together with evidence from animal models and pancreatic specimens, along with the observed association of diabetes with other autoimmune diseases, selected HLA class II genotypes, and increased risk in family members, also contributed to the concept that insulin-dependent diabetes mellitus was immune mediated (13). Today, the strong HLA class II association in humans and animal models supports the concept that T1D is predominantly T cell mediated, although as our understanding of immunology grows, it is becoming clear that multiple cell types participate in disease pathogenesis (14). Although there is no evidence that autoantibodies themselves are pathogenic, the identification of islet-specific Abs in these seminal articles also raises the question of whether B cells are pathogenic or secondary to the immune pathology. The role of B cells in T1D has been controversial because of the ability of T cells to transfer disease in the absence of B cells (15) and the observed development of T1D in an individual with B cell deficiency due to a BTK mutation (16). However, the majority of studies supports a role for B cells in diabetes (17); insulin-specific B cells have been shown to play a role in disease development in the NOD mouse model (18), and B cell–depletion therapy was efficacious in preserving endogenous insulin secretion in individuals with T1D (19).

Recognition of the role of autoimmunity in T1D led to the first clinical trials attempting to interrupt the destructive immune process. The immune suppressant cyclosporine was shown to alter the disease course in the early 1980s (20–22). However, this therapy was abandoned because of toxicity with chronic use. More recent clinical trials have targeted several immune pathways and mechanisms, including T and B cell activation and inflammatory cytokines, among others (23). There are now four immune therapies with different mechanisms of action that have reasonable safety profiles and that temporarily alter the disease course of T1D (19, 24–26). Indeed, the effects of immune therapy in T1D are not dissimilar to those seen in other autoimmune diseases.

The Bottazzo et al. and MacCuish et al. findings in 1974 led to refinements of their islet cell autoantibody (ICA) assay, more widespread testing, and a search for the Ags involved. In 1983, Jerry Palmer and colleagues (27) described the presence of insulin-specific Abs prior to treatment with insulin (which were thus “auto” Abs). Baekkeskov et al. (28) identified Abs to glutamic acid decarboxylase by 1990, Abs to protein tyrosine phosphatases (islet Ag 2/ICA512) were described in the mid-1990s (29), and Hutton’s team (30) identified autoantibodies to zinc transporter 8 in 2007. The latter four are described as biochemical Abs and can be reliably measured in radioimmunoassays because the target Ags are defined. In contrast, ICAs are still measured by detection of reactivity of test serum to cadaver pancreatic islets. This labor-intensive process remains in use in research studies because ICAs have additional, as yet unknown, Ag specificities beyond insulin, glutamic acid decarboxylase, islet Ag 2, and zinc transporter 8. Indeed, recent articles describing hybrid peptides (31, 32) and posttranslationally modified Ags (33) suggest the existence of additional autoantibody specificities.

Of critical importance, soon after the 1974 articles, the leaders in the field, including George Eisenbarth, joined together to cross-compare and validate Ab measurements. Ab testing, validation, and standardization continue to be conducted through the Diabetes Antibody Standardization Program, now under National Institute of Diabetes and Digestive and Kidney Diseases leadership, as new assays and variations of assays are described that are aimed to reduce blood volume and cost or increase sensitivity and specificity. Using these validated assays, more than 90% of individuals identified clinically with T1D can be shown to have one or more autoantibodies at the time of diagnosis.

Interestingly, Bottazzo et al.’s article included several subjects without diabetes who were ICA⁺, including one who subsequently developed diabetes. This presages what is arguably the most important impact of autoantibody measurements. The presence of islet autoantibodies in individuals prior to clinical diabetes diagnosis has allowed the study of disease progression and pathogenesis and the conduct of clinical trials to alter disease course. Research studies measure Abs in genetically at-risk populations identified by virtue of screening family members or testing for high-risk HLA class II genotypes. Over the past 30 years, such studies have refined our understanding of disease progression such that we now consider T1D as starting when two or more autoantibodies are present, because almost all of these individuals progress to clinically overt disease (34). Individuals with two or more autoantibodies have stage 1 diabetes if their glucose tolerance is normal, stage 2 diabetes if they have two or more autoantibodies and abnormal glucose tolerance, and stage 3 diabetes if they have clinically overt disease (35). Immune therapies are being tested in early stages of disease to prevent or delay progression to clinical onset.

Although the association of autoantibodies with clinically typical T1D is clear, the lines separating T1D and T2D are increasingly blurred. Thus, T2D involves β cell dysfunction and can occur in children, and some individuals with T2D have autoantibodies and islet T cell autoreactivity. Conversely, insulin resistance can be a component of T1D, and many people with long-standing T1D have evidence of residual β cell function; thus, immune-mediated β cell destruction is not always complete. Moreover, recent years have seen a revival of studies on the role that the β cell plays in its own demise. But, as first articulated by Bottazzo et al., whether β cell injury and death are from suicide or homicide, there is no question that the immune system is involved in the pathogenesis of disease.

• Copyright © 2017 by The American Association of Immunologists, Inc.