Diabetes Mellitus: A Fundamental and Clinical Text
3rd Edition

In 1849, Addison (6) first described the clinical and pathologic features of adrenocortical failure in patients, some of whom also appeared to have pernicious anemia. Ogle (7) reported the first instance of coexisting diabetes and adrenal insufficiency in 1866. In 1908, Claude and Gourgerot (8) suggested a common pathogenesis for the simultaneous expression of polyglandular insufficiencies involving the pancreatic islet, thyroid, gonads, adrenal gland, and anterior pituitary hypophysis, a fascinating and correct assertion. Parkinson (9) in 1910 was the first to note an association between pernicious anemia and IMD, although Addison himself had asserted an association between adrenal insufficiency and pernicious anemia. Mononuclear leukocyte infiltrates of goitrous thyroid glands were first noted by Hashimoto (10) in 1912; a similar lesion of pancreatic islets, termed insulitis, was first described by von Meyenburg (11) in 1940. The association between adrenocortical failure and thyroiditis was documented by Schmidt (12) in 1926, and the syndrome
complex was extended by Carpenter et al. (13) in 1964 to include IMD. It was not until 1956 that the autoimmune pathogenesis of these disorders was first documented, beginning with the discovery of circulating precipitating autoantibodies to thyroglobulin in patients with Hashimoto thyroiditis by Roitt et al. (14).

The ability to detect organ-specific humoral antibodies with methods developed by Anderson et al. in 1957 (15) and Blizzard and Kyle in 1963 (16) confirmed the clinical association between diabetes and idiopathic (autoimmune) adrenalitis. Solomon et al. in 1965 (17) demonstrated the coexistence of adrenal atrophy in diabetic patients with thyroid and adrenal dysfunction. Irvine et al. in 1970 (18) reported that both pernicious anemia and thyroid disorders occur with significant frequency in first-degree relatives of diabetic patients. Autoantibodies to specific thyroid and gastric antigens (19,20,21,22,23) as well as to adrenal and islet cell antigens (24,25,26,27,28,29) in diabetic patients have been studied extensively. Cellular autoimmunity to thyroid antigens in diabetes has also been reported (30). Neufeld et al. and our group (31,32) distinguished the two major autoimmune polyglandular syndromes associated with Addison disease (APS-1 and -2), summarized in Table 36.2. APS-1 and APS-2 are well circumscribed entities, whereas APS-3 (defined as no Addison disease) represents a clinical grouping that may be a subset of APS-2 or vice versa. Various studies have now identified the gene called autoimmune regulator (AIRE) responsible
for APS-1 and the many mutations that lead to it. The AIRE gene, which is located on the long arm of chromosome 21, was first identified by two individual groups (33,34). This gene consists of 14 exons and encodes a protein of approximately 545 amino acids.

APS-1 is a rare childhood disease that affects both sexes equally but is more prominent in certain races, such as in Finns (35) and Iranian Jews (36). APS-1 is diagnosed when a patient presents with at least two of its three cardinal clinical features: hypoparathyroidism, chronic mucocutaneous candidiasis, and hypoadrenocorticism (Addison disease). Patients with chronic mucocutaneous candidiasis (usual onset in early infancy) and hypoparathyroidism are at high risk for the subsequent development of Addison disease, which may only unusually appear before the development of hypoparathyroidism (31). The symptoms of hypoparathyroidism, however, can be masked by the development of Addison disease and manifested only after the start of steroid replacement therapy (37). The other diseases that might be associated with APS-1 include alopecia universalis, hypogonadism, chronic active hepatitis, pernicious anemia, vitiligo, and hypophysitis, but less commonly IMD. In Finland, where the prevalence rates of IMD are unusually high, IMD more commonly is seen in the context of APS-1. Many more patients with APS-1 develop islet cell autoantibodies (including those to glutamic acid decarboxylase) than will ever develop IMD. Patients with APS-1 therefore should be followed for the development of all of the associated diseases.

APS-2, which most commonly occurs in women, is characterized by the presence of autoimmune Addison disease together with autoimmune thyroid disease or IMD. APS-2 usually occurs in early to middle adulthood, with a peak age at onset of 30 years for Addison disease (31). Addison disease can occur concurrently with IMD and autoimmune thyroid disease, or may follow many years later. Therefore, patients presenting with IMD or autoimmune thyroid disease should also be screened for marker antibodies for Addison disease such as to the 21-hydroxylase antigen. Graves disease, Hashimoto thyroiditis, and pernicious anemia also frequently occur in APS-2. APS-3, on the other hand, is characterized by the presence of autoimmune thyroid disease in association with one of the other organ-specific autoimmune diseases such as pernicious anemia, vitiligo, or IMD, but no Addison disease (31). Associations among IMD, myasthenia gravis (38), and vitiligo (39) have also been documented.

The clinical associations between IMD itself and autoimmune diseases are impressive, as are findings of a strikingly high prevalence of certain organ-specific autoantibodies in type 1 diabetic sera (Table 36.3). The presence of these additional autoantibodies suggests that patients with IMD have a generalized tendency toward autoimmunity involving multiple endocrine glands and specific organs. It remains unclear, however, whether these apparently unrelated humoral specificities play any direct role in the pathogenesis of the disease. Most evidence in fact suggests that T-lymphocyte–mediated processes play a more important role. In 1974, two independent studies confirmed the existence of an antibody directed against the islet cells [islet cell autoantibodies (ICAs)] of the endocrine pancreas in the sera of diabetic patients who had other coexistent autoimmune endocrine disorders (40,41). Insulin autoantibodies were also described, first in untreated newly diagnosed diabetic patients (42) and subsequently in first-degree relatives of patients, with the use of either radioimmunoassay (43) or enzyme-linked immunosorbent assay (44).

In 1982, Baekkeskov et al. (45) first described a pancreatic islet cell protein found to be an islet cell autoantigen recognized by disease-specific antibodies and autoreactive T cells. This 64-kd protein could be precipitated from human and rodent islets using sera containing autoantibodies from patients with IMD and prediabetic subjects (46). In 1990, this protein was identified to be a lower-molecular-weight isoform of the glutamic

acid decarboxylase (GAD) enzyme, the enzyme responsible for the conversion of glutamine amino γ-aminobutyric acid (47).

Kaufman et al. (48) demonstrated that two specific isoforms of GAD, GAD₆₅ and GAD₆₇ (65 and 67 kd, respectively), could be immunoprecipitated using the sera of patients with IMD. Human islets have been found to express high levels of GAD₆₅, but lower levels of GAD₆₇ (49).

In summary, the study of the etiopathogenesis of type 1 diabetes and human autoimmunity in general has attracted a great deal of interest among both immunologists and diabetologists. The common associations of multiple endocrine disorders affecting single patients, their families, or both, as well as the association of multiple endocrine disorders with genes in the immune response region, specifically the HLA-DR and HLA-DQ genes, have been used as evidence of a common autoimmune origin for these multiple disorders.

Immune-mediated diabetes is determined by a combination of environmental and genetic factors. The MHC region contains multiple susceptibility loci, including the MHC class 2 genes (together called IMD locus 1 or IMD1). In addition, multiple non–HLA-linked diabetes susceptibility loci have been found to be associated with IMD. Some HLA class 2 (DR/DQ) haplotypes predispose to the disease, whereas others protect. IMD can be separated into two phases: (a) an inactive insulitis, characterized by islet-reactive T cells and autoantibodies but with normal islet function and without clinical symptoms, followed by (b) an invasive insulitis leading to an initially reversible, but later irreversible, destruction of β-cells (Fig. 36.1). Progression from phase 1 to 2 is related to class 2 loci. However, in terms of prognosis, serum autoantibodies to islet cells are still the best predictors of IMD, whereas people with DQB1*0602 rarely acquire the disease but may develop islet autoantibodies. The mechanisms by which HLA-DQ allelic proteins elicit susceptibility and protection in IMD are not known. Defining the biochemical properties of proteins encoded by HLA-DR/DQ alleles will provide clues for elucidating these mechanisms.

Much research has focused on identifying IMD susceptibility genes in an attempt to enhance prediction of who is at high risk for development of the disease, as well as to understand the disease’s pathogenesis. Rapid progress has been made in identifying genes that predispose to IMD. The major genetic component, which resides in the HLA class 2 region on chromosome 6p21.3 (IMD1), was discovered in 1974 (50). This HLA gene region plays a role in antigen presentation and initiation of immune responses. Multiple loci in this region contribute to susceptibility: HLA-DRB1, DQB1, DQA1, DPB1, and perhaps others (51). In human IMD, there is a hierarchy of genetic associations, of which HLA-DQA1*0301/DQB1*0302 are the predominant HLA class 2 alleles associated with susceptibility in IMD, and HLA-DQA1*0102/DQB1*0602 are the predominant HLA class 2 alleles associated with protection, even in people carrying the susceptible HLA-DQA1*0301/DQB1*0302 (52,53). Other HLA-DQ genotypes, such as HLA-DQA1* 0301/DQB1*0301 and HLA-DQA1*0102/DQB1* 0604, are not associated with IMD, even though these later genotypes are
structurally similar to the susceptible allele, DQA1*0301/ DQB1*0302, and the protective allele, HLA-DQA1*0102/ DQB1*0602, respectively (54). Analysis of the DRB1*04 subtype is particularly informative because it encodes a risk of IMD that is greatly variable depending on the population studied. DRB1*0401 and 0404 have been shown to be associated with IMD in Norwegians (55), DRB1*0402 and 0405 in French and in Mexican Americans (56), and DRB1*0401 and 0402 in Australians (57). DRB1*0405 confers a strong risk in Sardinians, Spaniards, North Africans, blacks, and Japanese (58). Conversely, protection from the disease is provided by DRB1*0403 in Sardinians, Belgians, and Chinese (58,59,60), by DRB1*0404 in French (60), by DRB1*0403,0408, and 0411 in Mexican Americans (56), and by DRB1*0406 in Asians (60). One study showed that DRB1*0401 subtypes represent almost 74% of all DR4 alleles and confer a significant risk for IMD when combined with DQB1*0302 (61). Among the various DRB-DQB allelic combinations, a single dose of protective DR or DQ allele, such as DRB1*0403, is sufficient to provide dominant protection.

Another component (IMD2), located near the insulin gene on chromosome 11p15, was identified in 1984 by association analysis (62). IMD2 has been identified as the variable number of tandem repeats immediately 5′ to the insulin gene, which appears to play a role in the level of gene expression (63). IMD1 and IMD2 contribute at levels of 42% and 10%, respectively, to the assignable risk from familial inheritance of the disease (λs), and although many other IMD susceptibility genes have been described, none have the large contribution ascribed to IMD1 (64). Reports of at least five additional loci influencing IMD predisposition (IMD3, IMD4, IMD5, IMD7, and IMD8) represent significant advances in our understanding of the genetic complexity of this disorder. IMD3 was localized to chromosome 15q26 (65), IMD4 to the 11q13 region (66), and IMD5 near ESR in chromosome 6q25 (67), whereas IMD7 has been localized to chromosome 2q31-q33 (67). More recently, another locus on chromosome 6 near the marker D6S264, distal to IMD5, has been reported that has been designated IMD8 (68). The exact genes responsible for the increased susceptibility to IMD, however, have not been identified.

A search for type 1 diabetes susceptibility genes in families from the United Kingdom revealed that two of the regions of the genome outside IMD1/MHC showed evidence of linkage: chromosome 10p13-p11 and chromosome 16q22–16q24. These and other novel regions, including chromosome 14q12-q21 and chromosome 19p13–19q13, could potentially harbor disease loci (69).

Various animal studies have shown that there are additional genetic associations besides HLA. These genes may alter the immune responses nonspecifically, and include polymorphism of the immunoglobulin constant region (Fc) receptors and the associated cytotoxic T-lymphocyte adhesion ligand (CTLA-4) (70). It has been postulated that increased susceptibility of islet cells to the induction of apoptosis by cytotoxic T cells also may be responsible for the facilitated death of islet β-cells (71). IMD12 (CTLA-4 gene) is one of the confirmed susceptibility loci located on chromosome 2q33. Initial studies provided evidence of linkage and association of a polymorphic marker in the CTLA-4 gene with IMD in Italian and Spanish families (72). Additional studies in French, Mexican-American, and U.S. white families confirmed these results (72). There is also an association of CTLA-4 polymorphisms with Graves disease and Hashimoto thyroiditis (73). The male:female (M:F) ratio in patients with type 1 diabetes is near unity. A survey showed that high-incidence countries (mainly European) have a high M:F ratio, and low-incidence ones (Asian and African) have a low M:F ratio. There are two main IMD1 susceptibility haplotypes, HLA-DR3 and -DR4, which are present in 95% of white cases. A study by Cucca et al. (74) revealed that in the medium/ high-incidence white population from the United States, United Kingdom, and Sardinia, the bias in male incidence is largely restricted to the DR3/X category of patients (where X is not DR4), with a M:F ratio of 1.7, compared with a ratio of 1.0 in the DR4/Y category (where Y is not DR3).

Whereas the HLA class 2 associations with IMD are well described, less is known about the genetics leading to the development of the IMD-associated autoimmune diseases. APS-2, characterized by the presence of autoimmune Addison disease together with an autoimmune thyroid disease or IMD, follows an autosomal dominant pattern with incomplete penetrance. Addison disease, a major component disease of APS-2, is strongly associated with HLA-DR3, and whether non–HLA-related genes are involved in the disease remains to be seen. Patients with APS-2 often share the same susceptible HLA alleles with those patients with only an individual component disease (75). For example, DR3-DQB1*0201/DQB1*0302 are associated with APS-2 when IMD is present (76). In addition, multigenetic involvement in the development of individual component diseases of APS-2 has been proven, including in IMD, which is reportedly linked to more than 10 loci in non-HLA genomic regions (77), or autoimmune thyroid disease, which appears to be polygenic as well (78). Patients with APS-3 lack definitive genetic features except for their association with its component diseases and their associated HLA alleles. For example, DQB1*0301 is increased in Hashimoto thyroiditis, DRB1*03 and DRB3 genes are increased in Graves disease, and DRB1*13 is increased in vitiligo. Addison disease occurring outside of APS-1 has been reported to be associated with DR3 and DR4 (78), and Graves disease has been associated with DR3 and DR5 (79). Other investigators have also reported that Hashimoto thyroiditis is associated with DR4 and DR5 (80), and that vitiligo is associated with DR4 (81).

The DR3 molecule confers susceptibility to multiple autoimmune endocrine diseases, including IMD, Addison disease (in APS-2), and Graves disease, whereas the susceptibility conferred by DR4 and DQB1*0302 is specific for β-cell autoimmunity. However, the strength of the association between DRB1*04-DQB1*0301 and Hashimoto thyroiditis needs to be clarified (82) (Table 36.4). Interestingly, CTLA-4 association has also been reported for Graves disease (83). Because IMD
sometimes occurs with autoimmune thyroid diseases, it was investigated whether CTLA-4 gene polymorphism favored the development of polyendocrine disorder in patients with IMD. However, no major role of the CTLA-4 gene was found in the association of autoimmune thyroid diseases with IMD. A study by Tomer et al. (84) has suggested the presence of a major Graves disease susceptibility gene on chromosome 20q11.2.

Knowing the autoantigen targets in an organ-specific autoimmune disease is essential to understanding its pathogenesis. In IMD, several β-cell autoantigens have been identified, but their importance in the diabetogenic process is unknown. Immunoprecipitating antibodies are present in most people with β-cell destruction and development of IMD. GAD₆₅, the tyrosine phosphatases insulinoma antigens 2 and 2β (IA-2 and IA-2β), and glima 38 (85) represent three distinct targets of immunoprecipitating IgG autoantibodies associated with β-cell destruction and IMD. Autoantibodies to IA-2 are detected in most newly diagnosed patients with IMD, compared with less than 2% in normal control subjects (86). Thirty-five percent to 50% of patients with newly diagnosed IMD have autoantibodies to another protein tyrosine phosphatase, IA-2β, which has been shown to be highly similar to IA-2, compared with 1% of normal control subjects (87). Over 95% of the IMD sera that react with IA-2β also react with IA-2, but only 50% to 80% of the IMD sera that react with IA-2 also react with IA-2β, suggesting that IA-2 and IA-2β share common epitopes, but also possess unique epitopes. An understanding of the nature of the T-cell repertoire and studies on the T-cell antigen-specific responses in IMD are still at a very early stage. An increased T-cell response in IMD has been reported with the intracellular fragments of IA-2 and GAD₆₅ (88), but antigen-specific differences in T-cell responses between patients with IMD and control subjects have not been consistently demonstrated. A number of laboratories have reported variable T-cell reactivities in diabetic patients against GAD₆₅ and IA-2 (89,90,91,92). Cellular proliferative responses to determinants common to GAD₆₅ were noted and a shared proliferative response marked 25% of 16 newly diagnosed patients with IMD, but none of the 13 healthy control subjects showed a positive response. A set of overlapping synthetic GAD₆₅ peptides was used to study the most reactive T-cell determinants in subjects at increased risk for IMD (i.e., autoantibody-positive, first-degree relatives of patients with IMD). Elevated in vitro T-cell responses were observed to GAD₆₅ peptides (amino acids 247–266 and 260–279) in newly diagnosed diabetic patients and autoantibody-positive, at-risk subjects. In one of the more recent studies, the epitope spanning amino acids 805 to 820 elicited a maximum T-cell response in all at-risk relatives from a total of 68 overlapping synthetic peptides encompassing the intracytoplasmic domain of IA-2 (93).

Cytokines have been shown to play a crucial role in autoimmune diseases and their outcomes. The discovery of polarized subsets of CD4⁺ T cells that differ in their cytokine secretion pattern and effector functions has provided a basis for understanding the diversity of T cell–dependent immune responses (94). CD4⁺ T cells can be distinguished, based on their pattern of cytokine production into two major types. T-helper type 1 (Th1) cells are characterized by secretion of interferon (IFN)-γ, interleukin-2 (IL-2), and tumor necrosis factor (TNF)-β, and they promote cell-mediated immunity to eliminate intracellular pathogens. Th2 cells selectively produce IL-4 and IL-5 and are involved in the development of humoral immunity protecting against extracellular pathogens (95). Uncontrolled Th1 and Th2 responses can cause chronic inflammatory autoimmune (delayed hypersensitivity) diseases and antibody-mediated autoimmunities or allergies, respectively (96).

Type 2 helper T-cell cytokines appear to be protective in type 1 diabetes, whereas IL-12–driven Th1 cells are pathogenic. In humans, it was found that T cells from diabetic patients preferentially produced a Th1 cytokine pattern, whereas their healthy siblings did not. Various studies in NOD mice have shown that the β-cell destruction was observed in conjunction with infiltration of local IFN-γ–positive T cells, whereas lesions dominated by IL-4–producing T cells did not result in islet cell destruction. Furthermore, diabetes has been transferred by CD4⁺ T cells expressing a Th1 cytokine profile in NOD mice, whereas treatments with IL-4 or IL-10 protected these mice from the disease. A Th1-dominated infiltration of T cells has
also been observed in type 1 diabetic patients who died at their disease onset.

Natural killer (NK) T cells (NKT cells) are a subpopulation of T cells expressing NK 1.1 (97), many of which also express CD4. They are capable of making substantial amounts of IL-4 promptly on in vivo stimulation with anti-CD3 (98). These cells play a role in determining the cytokine environment when conventional T cells make their initial encounter with antigen. NKT cells were first identified in mice, where they can be distinguished from conventional T cells by their expression of the NK locus-encoded C-type lectin molecule NK1. These cells have a restricted T-cell receptor (TCR) repertoire, with most expressing an invariant TCRα chain structure (Vα14-Jα281) paired preferentially with Vβ8,7, or 2 (99,100). It has been demonstrated that the human immune system contains a population of T cells that show striking conservation of most of the key features of murine NKT cells (101,102). The human NKT cells express an invariant Vα24-JαQ TCRα chain, which is highly homologous to the invariant TCRα chain of murine NKT cells (114). Human NKT cells also express a restricted Vβ repertoire (103) and several NK cell locus-encoded C-type lectins, including NKR-P1A (CD161), which is the homologue of murine NK1 (104). The function of NKT cells in the immune response remains incompletely resolved; however, recent experimental findings have generated much interest in subpopulations of T cells. Murine and human NKT cells are able to produce large amounts of both IL-4 and IFN-γ rapidly on activation, and may represent the major source of early IL-4 in certain immune responses driving Th2 differentiation (105). Several mouse models of autoimmunity are characterized by decreased Vα14 T cells. Mieza et al. (106) showed that a specific decrease in invariant Vα14⁺ NKT cells correlated strongly with the development of diabetes, suggesting that NKT cells regulate autoimmune responses. Studies also suggest major roles for murine NKT cells in the rejection of malignant tumors (107) and in regulating autoimmunity (106).

Similarly, human studies have suggested that NKT cell deficiencies are associated with various T cell–mediated autoimmune diseases like systemic sclerosis (108), multiple sclerosis (109), and rheumatoid arthritis (110), including IMD (111,112). A defect of NKT cells was also observed in autoantibody-positive first-degree nondiabetic relatives of IMD patients, suggesting that a primary defect in NKT cells leads to disease onset (112) rather than arises as a consequence of diabetes. Wilson et al. showed that NKT cells cloned from nondiabetic twins of IMD patients had altered gene expression polarized to Th2 type (IL-4 secreting), while in patients there was a downregulation in the expression of several genes encoding proteins normally involved in the recruitment, activation, and differentiation of dendritic cells (DCs) and macrophages (113). However, we (AK) reported that Vα24⁺ enriched NKT cells in IMD patients are defective in their secretion of both IFN-γ and IL-4 (112) as compared with normal controls. Our findings are similar to data from NOD mice that have a lack of immunoregulatory functions of NKT cells correlated with a dramatic defect in proliferation and differentiation toward IFN-γ secreting phenotype upon their stimulation (114).

NOD mice (115) and patients with IMD have low numbers of circulating resting CD4⁺/CD25⁺ Treg cells (116). This was also not secondary to diabetes in the patients, because low numbers were also found in islet cell autoantibody–positive but nondiabetic relatives of patients with IMD, but not in patients with overt type-2 diabetes. The defect appears to be required but is not itself sufficient to develop IMD.

Given the interrelationships between IMD and other endocrinopathies in the polyglandular syndromes, how should the clinician monitor the patient with IMD? Clearly, accurate and early diagnosis of other endocrinopathies, particularly Addison disease, can significantly reduce morbidity and mortality. All patients with IMD should optimally be screened for the presence of adrenal, celiac, steroidal cell, gastric parietal, and thyroid relevant autoantibodies at the time of their diagnoses, which itself should be confirmed with ICA, GAD₆₅A, IA-2A, and IAA testing when clinical presentations are not classical for type 1 diabetes (Fig. 36.2). Family members of a proband with IMD should also undergo these autoantibody studies, especially when the proband is found positive. Positive autoantibodies should be followed for the relevant hormonal evaluations and treatments. The necessity for further workup of patients with IMD or their family members depends on their panel of autoantibody results. ICA screening should be repeated annually in children under 10 years of age and every 2 years in relatives under 20 years of age because these age groups are at higher risk for development of pancreatic β-cell autoimmunity; the risk for conversion from negative to positive decreases markedly with increasing age.

Many patients with IMD acquire autoimmune thyroiditis (e.g., Hashimoto thyroiditis, Graves disease), most commonly recognized after the diagnosis of IMD (117), although such autoimmunities arise simultaneously (118,119). The consequences of untreated hypothyroidism include congestive heart failure, dyslipidemia, infertility, growth retardation, and rarely, slipped femoral epiphysis. Graves disease can cause significant weight loss, weakness, congestive heart failure, and exophthalmos/proptosis. The prevalence of thyroperoxidase antibodies is nearly 25% in girls and 10% thyroglobulin antibodies and higher still in older patients (120). Hashimoto thyroiditis leads to gradual loss of thyroid function at a rate of 5% per year (121).

We strongly advise screening for thyroid microsomal antibodies in all patients with IMD at their diagnosis, and if negative, every 2 or 3 years thereafter until adulthood. Furthermore, all patients with positive thyroid autoantibody tests should have thyroid panel determinations performed annually. Thyroid-stimulating hormone (TSH) levels are most helpful diagnostically. Those with hypothyroidism should be treated by thyroid hormone replacement therapy. If a patient is found to have thyroiditis,

family members also should be screened for thyroid autoantibodies and goiter.

The prevalence of Addison disease in patients with IMD is around 0.5% to 1% (122,123). By definition, IMD with Addison disease presents more frequently, in up to 20% in cases of APS-2 (124). Autoimmune Addison disease can develop slowly over many years before symptoms appear, making screening of patients with IMD and autoimmune thyroiditis for adrenal autoantibodies valuable for early identification and treatment of the disease. Some 1 in 200 to 300 patients with IMD will develop antibody-positive Addison disease. Addison disease is readily treated when diagnosed (125). However, the disease can be fatal and has been occasionally diagnosed only at autopsy in IMD patients. Autoantibodies to 21-hydroxylase (21-OH) occur in 2% of patients with IMD and in 3% of patients with Graves disease (126). After the appearance of antibodies to the adrenal cortex or to 21-hydroxylase (21-OHA), the first evidence of adrenal insufficiency is usually an increase in plasma renin activity after patients have been recumbent for more than half an hour. The elevated renin level is due to a failing zona glomerulosa with salt loss, with low-normal or low plasma aldosterone concentrations (123). A history of weight loss, hyperpigmentation (often described as “dirty” pigmentation of the neck or elbows) with increased pigmentations in skin creases, easy fatigability, muscular weakness, dehydration, emesis, or malaise should be sought at each clinical visit (Table 36.5). Patients should be educated to recognize and report these symptoms immediately. Zona fasciculata dysfunction can become evident months to years later, first by increased afternoon serum adrenocorticotrophic hormone (ACTH) levels, then by decreasing serum cortisol responses to ACTH stimulation, and finally by decreasing morning serum cortisol concentrations and the appearance of symptoms (Fig. 36.3). Thus these autoantibodies are useful markers for the prediction of the development of Addison disease, particularly for children (127). There are two types of antibodies detected by microscopic indirect immunofluorescence. Adrenal autoantibodies (AAs) react with only the adrenal cortex, whereas steroidal cell antibodies
(SCAs) react with all steroid hormone-producing cells. One autoantigen is involved in reactions of AAs is the protein of the P-450 21-OH enzyme, with epitopes in the central segment of the enzyme and the C-terminal portion (26,128). Component antigens for SCAs have been recognized to be other P-450 enzymes, 17-α-hydroxylase (17-OH) (25), and side-chain cleavage enzyme (SCC) (28). Almost all patients with both Addison disease and gonadal failure have positive 17-OH and SCC antibodies in the context of APS-1. Autoantibodies against 3-β-hydroxysteroid-dehydrogenase (3βHSD) have been reported in 23% of IMD, frequencies significantly higher than that observed among healthy individuals (5%). These 3βHSD autoantibodies are reported to be associated with HLA-DQB1*0301 and with DQB1*0603 (129).

Any patient with IMD with a family pedigree including Addison disease should be screened for adrenocortical (21-OH) and steroidal (17-OH) cell autoantibodies. Similar screening should be sought in patients with unexplained weight loss, refractory hypoglycemia, hyperpigmentation, or unexplained dehydration. Patients with adrenocortical antibodies should have annual determinations of afternoon plasma ACTH, morning serum cortisol, and supine renin levels. Electrolyte abnormalities are a late finding, and normal electrolytes do not exclude disease. An increase in potassium level is the hallmark biochemical finding, and patients may die from salt loss without hyponatremia necessarily becoming evident. Patients with steroidal cell antibodies may contract both Addison disease and gonadal autoimmunity. Thus, antibody-positive patients should be urged to recognize the symptoms of adrenal insufficiency described previously. They should also be monitored for secondary amenorrhea, pregnancy loss, signs of infertility, and rare testicular failure in men. Annual determinations of follicle-stimulating hormone and luteinizing hormone may permit prediction of gonadal failure. Patients also should be examined for evidence of mucocutaneous candidiasis or for hypoparathyroidism if these autoantibodies are detected.

One of the most devastating problems accompanying IMD can be autoimmune neurologic involvement, which appears to be antibody mediated and thus β-cell/plasma cell mediated, because it is associated with high titers of autoantibodies to the NH2 terminus of glutamic acid decarboxylase. This neurologic involvement can have heterogeneous presentations such as stiff man syndrome (SMS) (130,131), therapy-resistant epilepsy (132), partial epilepsy (133), cerebellar ataxia (134), and peripheral and autonomus nervous system involvement such as double vision and bladder dysfunction (135,136).

Glutamic acid decarboxylase is one of two isoenzymes responsible for converting glutamate to gamma-aminobutyric acid (GABA). GAD₆₅ is a major target of humoral autoimmunity in SMS and subacute cerebellar ataxia (137,138). Recently, it was described that anti-GAD autoantibodies in the cerebrospinal fluid of an ataxic patient selectively suppressed GABA-mediated transmission in cerebellar Purkinje cells without affecting glutamate-mediated transmission (139). Cerebellar Purkinje cells, which are affected in most types of cerebellar degeneration, contain high levels of GAD. It seems unlikely that GAD autoantibodies are merely an epiphenomenon, because GAD autoantibodies are not present in most other types of cerebellar degeneration. GAD₆₅ autoantibodies in SMS and IMD differ in several respects. In SMS, their levels are higher and they usually react by immunoblotting since the reactive epitopes are linear. In IMD the autoantibodies are found at lower concentrations but react to conformational epitopes and thus only to undenatured or native GAD antigens. In addition, GAD₆₅ autoantibodies in SMS recognize epitopes of the GAD₆₅ autoantigen other than those in IMD (140,141). This could explain why patients with GAD autoantibodies develop varying diseases. However, to our knowledge there are a limited number of studies to differentiate between the reactive epitopes of GAD autoantibodies found in cerebellar ataxia, SMS, and IMD.

IA-2 protein represents one of the members of the protein tyrosine phosphatase (PTP) family, expressed in pancreatic islet and brain tissues. Tyrosine phosphatase–like receptor proteins span the membrane of dense core vesicles of neuroendocrine tissues. They are of interest as molecular components of the secretory machinery and as major targets of autoimmunity (142). Furthermore, SMS has both anti-GAD₆₅ and anti–IA-2 humoral immunoreactivities. GAD₆₅ and IA-2A autoantibodies are present in other neurologic diseases, supporting their autoimmune origins. There was a higher prevalence of such antibodies in Lambert-Eaton myasthenic syndrome (LEMS) (GAD₆₅A 35%; IA-2A 21%; double positivity 18%), amyotrophic lateral sclerosis (18%, 12%, and 12%, respectively), and multiple sclerosis (10%, 3%, and 3%, respectively). In LEMS, the humoral reactions were more frequent or appeared earlier in the paraneoplastic forms. The detection of such autoantibodies in patients with small cell lung carcinoma (SCLC) without LEMS suggests that these autoantigens, GAD₆₅ and IA-2, could be produced by SCLC tissue as well (143).

The incidence of IMD in patients with multiple sclerosis (MS) has been estimated to be approximately 9 in 1,000 in Israel (144). The association of MS with IMD is rare as predictable on the basis of HLA genetics in that the MS-susceptible allele, DQB1*0602, is a known protective allele in IMD. Studies in NOD mice indicated that in this animal, autoimmunity directed against myelin and islet antigens coexists (145). Myelin proteolipid protein, myelin oligodendrocyte glycoprotein, myelin basic protein, and myelin-associated glycoprotein are self-antigens in the myelin sheath that are involved in MS.

Patients with MS, partial seizures, and SMS with strong family histories of IMD should be screened for GAD₆₅ and IA2 antibodies. Patients with IMD and cognitive deterioration, seizures, and stiffness should be screened for GAD₆₅ titers, which are usually much higher in case of autoimmune neurologic involvement.

Pernicious anemia is caused by autoimmune destruction of gastric parietal cells and presence of antibodies to intrinsic factor, which lead to malabsorption of vitamin B₁₂ with resultant megaloblastic anemia. Antibodies to gastric parietal cells are targeted to the H⁺/K⁺-ATPase, the parietal cell proton pump (23). Autoantibodies to intrinsic factor lead to vitamin B₁₂ malabsorption in the terminal ileum and inhibit the binding of vitamin B₁₂ to intrinsic factor, contributing further to vitamin B₁₂ malabsorption. Vitamin B₁₂ deficiency leads to a megaloblastic anemia, atrophic glossitis, and paresthesias from peripheral neuropathies. Any patient with these autoantibodies should have yearly determinations of ferritin and vitamin B₁₂ levels. Parietal cell autoantibodies are positive in up to 18% of IMD and 11% of their relatives (146). Up to 7% of IMD patients, especially those with DQA1*0501 and DQB1*0301 HLA genotypes with positive parietal cell autoantibodies, will develop pernicious anemia (147). Because atrophic gastritis is usually subclinical, patients present clinically with iron deficiency or vitamin B₁₂ deficiency. Patients with IMD with (a) strong family histories of thyroiditis or pernicious anemia, (b) megaloblastic anemia, or (c) unexplained fatigue should be screened for gastric parietal cell antibodies and antibodies to intrinsic factor. Autoimmune gastritis is β-cell/plasma cell mediated with a future increased risk for gastric carcinoma and carcinoid tumors. If vitamin B₁₂ deficiency is found, monthly vitamin B₁₂ injections should be administered. Most patients reach midlife before this becomes necessary. Any patient with IMD with new onset of paresthesia should notify his or her endocrinologist promptly.

Celiac disease is an immune-mediated intolerance to ingested wheat gluten, particularly to the gliadin component of it. Prevalence of celiac disease is 1 in 100 to 250 in the United States (148). The prevalence of celiac disease in patients with IMD has been reported to be 3.9% to 13.5% (149,150). Siblings of patients with celiac disease have been found to have celiac disease themselves by biopsy in 5.5% to 18% cases (151). The haplotype (A1, B8, DR3, DQA1*0501, DQB1*0201) characteristic for in celiac disease is similar to that in immune mediated diabetes (152).

Gliadin autoantibodies (IgA, IgG) are seen in patients with celiac disease but have low sensitivity and specificity for the disease since they can be found in other gastrointestinal diseases and in healthy individuals. Endomysial autoantibodies and transglutaminase autoantibodies have the best sensitivities and specificities for celiac disease (153). Unlike antibodies to gliadin, the titers of these autoantibodies reflect immunizations following the ingestion of gliadin. The target autoantigen of the endomysial antibody is tissue transglutaminase (tTG) (154). It forms complexes with gliadin, which initiates an immune response directed against both transglutaminase and gliadin. Children with a genetic risk for celiac disease (HLA-DR3) express antibodies to tTG with a positive predictive value of 70% to 83%. Furthermore, those with high serum levels of these antibodies had a 100% positive predictive value for celiac disease (155). Patients suspected of having celiac disease can have confirmation of their disease by duodenal biopsy or by clinical trial of withdrawal of gliadin from the diet. Duodenal biopsies will identify an individual as having the characteristic findings of villous atrophy and intraepithelial lymphocytes.

Recurrent or severe hypoglycemia, unexplained weight loss, growth retardation, symptoms of malabsorption or with minimal or no gastrointestinal symptoms, poor growth, osteoporosis, hypocalcemia, and increased frequency of fractures can all be presenting features of celiac disease in patients with diabetes (156).

The symptoms and histologic findings resolve with removal of gluten from the diet, which is currently the only treatment. Numerous complications associated with even subclinical celiac disease clarify the importance of treatment of silent celiac disease. In addition to IMD, dermatitis herpetiformis, psoriasis, alopecia areata, autoimmune hepatitis, primary biliary cirrhosis, gluten ataxia, and epilepsy with cerebral calcifications are seen in association.

Vitiligo affects 1% of the population in the United States and up to 3% to 4% of Asian Indians. Vitiligo has been reported in 7% of patients with autoimmune thyroid disease, whereas thyroid autoimmunity occurs in as many as 50% of patients with vitiligo (39). Vitiligo and psoriasis are the most commonly observed skin lesions in IMD, occurring in approximately 9% of such patients. Vitiligo is caused by an immune response to proteins situated in the membrane and the cytoplasm of melanocytes and characterized by positive antimelanocyte antibodies. However, the diagnostic sensitivity of autoantibodies for vitiligo is low so far (157). Tyrosinase, the rate-limiting enzyme for melanin formation, was the first target antigen identified in vitiligo (39). Later antibodies to tyrosine-related protein 1, tyrosine-related protein 2, melanocyte-specific protein Pmel117 (158), the transcription factor SOX9 and SOX10 proteins (159), antibodies to the melanin-concentrating hormone receptor 1 (160), and autoreactive T cells recognizing melanocyte antigens, including MelanA/MART-1 and Pmel117 (161), were recognized. Vitiligo skin demonstrates chronic epidermal and dermal infiltration of lymphocytes and loss of melanocytes (162).

Alopecia is associated with other autoimmune diseases, and alopecia universalis is especially common in APS-1 (43). AA is associated with diabetes as well as APS—present in up to 20% of cases of APS-1 and 0.5% of cases of APS-2 (163). In all stages of AA there is a lymphocytic peritubular infiltrate composed mostly by T-helper cells (164). The autoantibody responses are directed to different hair follicle structures. The family of tetrahydrobiopterin-dependent hydroxylases consists of the three highly homologous enzymes: tryptophan hydroxylase
(TPH), tyrosine hydroxylase (TH), and phenylalanine hydroxylase. Antibodies to TPH, which is highly expressed in the gut, are coupled with gastrointestinal dysfunction; and antibodies to TH, expressed in hair follicles, are correlated with alopecia areata. TH was cloned from a human scalp complementary DNA library, and a statistical correlation is found between TH antibodies and alopecia in APS-1 patients (165). TPH-containing enterochromaffin cells in the small intestine are absent in biopsy samples from APS-1 patients with TPH autoantibodies and intestinal symptoms, suggesting an autoimmune attack specifically directed toward these cells (166).

Immune-mediated diabetes and thyroid disease are found in approximately 10% to 35% of patients with autoimmune hepatitis (AIH) (167). Antinuclear antibodies (ANAs), smooth muscle antibodies (SMAs), antibodies against liver/kidney microsomes (LKMs), and antisoluble liver protein antibodies are diagnostic (168). Autoantibodies directed against against single-strand DNA, double-strand DNA, snRNPs, transfer RNA, cyclin A, and lamins A and C are ANAs. SMA antibodies are directed against the structures in cytoskeleton, with autoantibodies to F actin representing the major subset of SMAs. Cytochrome P-450 2D6, an enzyme active in many detoxification reactions, is a major antigen for LKM1 antibodies (169). Soluble liver protein autoantibodies (SLAs) directed against cytokeratins 8 and 18 are also reported (170). AIH type 1 is characterized by ANA and SMA antibodies (168,171). AIH type 2 is characterized by microsomal antibodies LKM and AIH type 3 by autoantibodies directed against SLAs (anti-SLAs) (172). Patients with AIH-1 are younger, have more severe disease, and frequently progress to cirrhosis (173).

Mucocutaneous candidiasis presents in 73% to 100% of patients with APS-1 (174) and is a frequent finding in IMD, especially Candida vaginitis. In the former, candidiasis results from an intrinsic defect in cell-mediated immunity to the yeast, whereas in diabetes, cellular immunity to the yeast reflects the metabolic derangement from lack of diabetes control. Candidal esophagitis presents with underlying dysphagia, along with substernal pain with subsequent risk of stricture. Abdominal pain and meteorism can be signs of intestinal candidiasis. Elevated glucose levels (175), lower amounts of production of proinflammatory cytokines upon stimulation of monocytes with lipopolysaccharide (176), and defective Th1 and Th2 cell responses (177) are the major factors leading to mucocutaneous candidiasis.

Appreciating that IMD often occurs with other autoimmune endocrinopathies and skin autoimmunities, we recommend screening for those diseases both in patients with IMD and in their relatives (Fig. 36.2B). Such screening is not an esoteric exercise. The consequences of undiagnosed endocrinopathies can involve significant morbidity or mortality. It could be argued that the clinician certainly would recognize the onset of a new endocrinopathy in a patient with IMD or the symptoms of endocrine insufficiency in a relative; however, disease onset may be insidious and easily missed unless specifically and routinely targeted and tested for.

Immune-mediated diabetes is one of a group of autoimmune diseases that have in common an association with specific HLA alleles. Its most common presentation with other autoimmune disorders is as a component of APS-3, which is the constellation of IMD and autoimmune thyroiditis, sometimes with pernicious anemia, vitiligo, or hypogonadism. The association of IMD with Addison disease in APS-2 carries the greatest risk for mortality and requires rapid recognition and specific replacement therapies to prevent fatalities. There appear to be no common antigens shared uniquely among organs targeted by the autoimmune processes underlying APS. Indeed, the putative antigens of pathogenic importance appear to be distinctly different (Table 36.3). Furthermore, no common genes are known to explain the concomitant development of the autoimmune endocrinopathies seen in APS, nor do the HLA associations offer any explanation. Furthermore, the unknown basis of specific target organ responses (thyroid or β-cells) in these autoimmune diseases is crucial to the understanding and ultimate prevention and cure of the disease.

In IMD, identification of the involvement of the insulin gene may be the first instance in which an organ-specific region has been identified. It is likely that other β-cell–specific genes will be defined that affect IMD susceptibility, such as genes coding for transcription factors or those affecting β-cell maintenance or replication. The identification of the AIRE gene in APS-1 is another milestone in the understanding and classification of autoimmune endocrinopathies. Mutations of a single recessive gene (AIRE) have been shown to be responsible for APS-1; the test for mutations of this gene is a crucial advance in predicting the development of autoimmune endocrinopathies. Patients with AIRE gene mutations should also be screened for component diseases, such as 21-hydroxylase for Addison disease.

Substantial time and additional resources will be required to identify and characterize the next generation of IMD susceptibility genes and antigens. We see IMD as an immune deficiency disorder, whereby tolerance to self-antigens encompasses tissues far beyond those of the pancreatic β-cells. It thus seems clear that sorting out the genetic basis underlying each pathogenetic form is essential if prevention and cure of IMD are to realized.