ABSTRACT

Type 1A diabetes (T1D) represents an autoimmune disorder that can affect individuals from within a year of birth until age 60. A number of genes strongly influence the development of disease, including genes found within the human lymphocyte antigen (HLA) complex. The role of non-HLA genes is being defined in recent studies, and we are beginning to identify pathways that lead to autoimmunity and eventually pancreatic islet cell destruction. Although genes can predispose one to type 1A diabetes, environmental factors may also play a significant role in the pathogenesis. These as-yet-undefined factors appear to have accelerated the onset and markedly increased the frequency of disease in many populations around the world over the last 30 years. The development of ever more sophisticated immunoassays to detect antibodies directed against pancreatic antigens have helped define the autoimmune nature of the disorder, but as importantly have also provided an opportunity to identify those individuals with prediabetes and to stratify their risk of developing overt hyperglycemia. Immunologic assays as well as intervention trials are allowing us to learn more about the immune pathways that are disordered and offer hope for future therapeutic approaches to prevent and reverse type 1A diabetes. For extensive review of all related areas of Endocrinology, visit WWW.ENDOTEXT.ORG.

INTRODUCTION

Type 1A diabetes mellitus is defined as immune mediated diabetes mellitus.(1-6). It can become manifest with hyperglycemia presenting in the first days of life or in adults over the age of 60. Current estimates indicate that immune mediated diabetes represents approximately 5 to 10% of the diabetes developing in adults and that approximately as many individuals develop this form of diabetes as adults as do children(7-9). In the United States the great majority (>90%) of Caucasian children developing diabetes have type 1A diabetes; whereas, approximately 50% of African American and Hispanic American children developing diabetes lack the autoantibody and immunogenetic markers of typical type 1A diabetes(10-12). Most of these latter children appear to have variants of type 2 diabetes with a small number having specific characteristic genetic syndromes (e.g. MODY: Maturity Onset Diabetes of Youth) with identified mutations of genes such as glucokinase and HNF (Hepatic Nuclear Factors)(13). In addition, studies of the pathology of the pancreas of Hispanic and African American children who lack islet autoantibodies show that all islets have some beta cells, but in decreased numbers(11). In contrast, in the pancreas of patients with type 1A diabetes, there is lobular loss of beta cells (termed pseudoatrophic islets)(11).

When an individual presents with type 1A diabetes it indicates that they and their relatives have an increased risk of having or developing a series of autoimmune disorders(12). Celiac disease, hypothyroidism, hyperthyroidism, Addison's disease and pernicious anemia are some of the most prominent associated diseases. For example approximately 1/20 patients with type 1 diabetes have celiac disease(14,15). Most of these patients are asymptomatic and the disorder is only discovered if anti-transglutaminase autoantibodies are measured and individuals with positive antibodies biopsied. In that the therapy for celiac disease, namely gluten avoidance, is highly effective, at the Barbara Davis Center we routinely screen all type 1 diabetic patients. We also screen for thyroid disease, which has an incidence of approximately 20% in type 1 diabetes, with yearly TSH measurements and for Addison's disease (21-hydroxylase autoantibodies)(16).

Studies of the pathogenesis of type 1 diabetes have blossomed during the past two decades and there are now complete books devoted to this subject (e.g., Immunology of Type 1 Diabetes available at www.barbaradaviscenter.org with appended "Teaching" slides). In terms of summarizing the pathogenesis of this disorder it is convenient to divide the disease into a series of stages (Figure 1) beginning with genetic susceptibility and ending (from an immunologic standpoint) with complete islet beta cell destruction(16). This is only a general description of the disease process. It is likely for instance that genetic determinants influence many of the stages of disease progression and are important determinants of individuals who express anti-islet autoantibodies but do not progress to diabetes (e.g., a major subset of those with anti-islet autoantibodies and the protective HLA allele DQB1*0602 allele)(17).

Figure 1

Hypothetical stages and loss of beta cells in an individual progressing to type 1A diabetes. From Eisenbarth, NEJM, 1986 (18)

STAGE I. GENETIC SUSCEPTIBILITY

An expert committee of the American Diabetes Association has divided type 1 diabetes into type 1A (immune mediated) and 1B (not immune mediated but with profound loss of insulin secretion)(19). The great majority of patients with insulin dependent diabetes have type 1A, and firm examples of type 1B are either lacking or controversial. One recent example is a description from Japan of patients who extremely rapidly developed type 1 diabetes (fulminant diabetes) such that they had marked hyperglycemia, but at diabetes presentation their HbA1c was near normal, suggesting very recent onset of hyperglycemia(20). These patients lacked anti-islet autoantibodies, a subset had elevated serum pancreatic enzymes, but many had HLA alleles associated with type 1A diabetes. On biopsy beta cells were destroyed, there were lymphocytes in the acinar pancreas, but no typical insulitis (invasion of islets by lymphocytes). Whether this is truly type 1B or an extremely rapid variant of type 1A is currently debated.

Figure 2

Genetic heterogeneity of immune mediated type 1 diabetes with monogenic, polygenic, and oligogenic forms. Modified from teaching slides www.barbaradaviscenter.org

Type 1A diabetes is itself heterogeneous, with several forms of immune mediated diabetes with known genetic causes as parts of autoimmune syndromes (thus likely to be classified as other Specific Forms of Diabetes). In particular, patients develop immune mediated diabetes when they have mutations of the AIRE (Autoimmune Regulator) gene(21) and the human gene homologous to the mutated gene causing Scurfy in mice (22)(figure 2: monogenic disorders). Mutations of the AIRE gene result in Autoimmune Polyendocrine Syndrome Type I (23,24). Mutations of the Scurfy homologue lead to overwhelming neonatal autoimmunity (IPEX syndrome: Immune dysregulation, Polyendocrinoptahy, Enteropathy, X-linked)(21). A mutation of the FoxP3 genes, an essential transcription factor for CD4+CD25+ regulatory T cells, is the cause of the IPEX syndrome(25,26). These children can develop type 1 diabetes in the first days of life and illustrate the importance of T cell regulation. Most other forms of type 1A diabetes are polygenic in etiology, and polymorphisms of genes within the major histocompatibility complex (HLA genes) play a major role in determining disease susceptibility(27,28). Such heterogeneity is also apparent in three spontaneous animal models of type 1A diabetes, the BB rat (Biobreeding), the NOD (Non-obese diabetic), and the Tokushima rat (Figure 3). For all three strains polymorphisms of genes homologous to HLA DR and DQ of man are essential for disease(29). In addition, more than 15 other loci play an important role in diabetes susceptibility of the NOD mouse, but each locus contributes relatively little (polygenic inheritance) by itself.

Figure 3

Three rodent strains that develop immune mediated diabetes. Modified from teaching slides www.barbaradaviscenter.org

For the BB rat and Tokushima rat there are major loci outside of the MHC contributing to disease (Oligogenic inheritance)(30,31). It is not likely that human type 1 diabetes is less heterogeneous than these few animal models.

Identical twins of patients with type 1A diabetes have an overall risk of developing type 1 diabetes approaching seventy percent.(32) Consistent with heterogeneity, that risk varies dramatically with the diabetic twin's age of diabetes onset. If one identical twin develops diabetes prior to age 5, the risk for the other twin exceeds 50%. In contrast if the twin develops diabetes after age 25, the other twin’s risk is less than 10%(29). The risk of developing diabetes is approximately 1/20 for a sibling or offspring of a patient with type 1A diabetes. The U.S. population risk for type 1A diabetes is approximately 1/300 and the country with the highest incidence in the world (Finland) has a risk of approximately 1/100. As will be discussed subsequently, genetic polymorphisms greatly influence disease risk. A sibling of a patient with type 1 diabetes who has the highest risk genotype DR3/4 DQB1*0302 has an almost 50% risk of developing anti-islet autoantibodies and progressing to diabetes(33).

Polymorphisms of genes within the major histocompatibility complex contribute to disease of both humans and rodent models.

Figure 4

Schematic of the human major histocompatibility complex approximately 4 million base pairs in length. Redondo MJ, et al. Recent Prog Horm Res, 56: 69-89, 2001.(34) Copyright The Endocrine Society.

Figure 5

Hierarchy of HLA terminology with specific alleles, groups of alleles of different genes on the same chromosome, and finally genotype. Modified from teaching slides www.barbaradaviscenter.org

The alleles of different HLA genes (e.g., DRB1 and DQB1) are non-randomly associated with each other, such that with DRB1*0401 one usually finds one of three DQ alleles (e.g., DQB1*0301, DQB1*0302, DQB1*0303) rather than any one of more than forty different DQB molecules. Such non-random association of alleles of different genes on the same chromosome is termed linkage dysequilibrium.The histocompatibility complex is divided into three regions, class II, class III and class I (Figure 4). The most important determinants of type 1 diabetes are the HLA DQ and DR alleles. These molecules on the surface of antigen presenting cells (e.g., macrophages) bind and present short peptides that are recognized by T cell receptors of T lymphocytes(27,35,36). They are termed immune response genes in that the specific amino acid sequence of these molecules determines which peptides will be bound and to a large extent determine which peptides an individual will respond to. Each different amino acid sequence is given a number. For the DQ molecules both its alpha and beta chain gene are polymorphic, and thus to specify a DQ molecule one must specify both chains. For DR molecules only the DRB chain is polymorphic and thus only this chain is specified. Each number after the star indicates a specific amino acid sequence of the HLA allele (Figure 5) and the letters and first number the gene (e.g., DRB1*0401, DR B chain gene number 1, allele 0401).

Figure 6

Hierarchy of diabetes risk with examples of haplotypes that lead to diabetes susceptibility, are neutral, or protective. Modified from teaching slides www.barbaradaviscenter.org

There is a tremendous spectrum of diabetes risk associated with different DR and DQ genotypes(37-39)(Figure 6). For Caucasians with type 1A diabetes the most common diabetes-associated haplotypes are DR3 and DR4. More than 90% of patients with type 1A diabetes have one or both of these alleles versus approximately 40% of the general U.S. population. With the finer sequence information that is now available, DR4 haplotypes are subdivided based on specific variants of DRB1 and DQB1. The highest risk DR4 haplotypes have DRB1*0401, DRB1*0402, DRB1*0405, while DRB1*0403 is moderately protective. The highest risk DR4 haplotypes have DQB1*0302, with DQB1*0301 and DQB1*0303 of lower risk. Thus both DR and DQ alleles contribute to diabetes risk. DR3 haplotypes are almost always conserved with DRB1*03 combined with DQA1*0501, DQB1*0201(40). The highest risk genotype have both DR4/DR3 DQB1*0302/DQB1*0201. This genotype occurs in 2.4% of newborns in Denver, Colorado, and between 30 and 50% of children developing type 1A diabetes. Approximately 50% of children developing type 1A diabetes early (i.e., less than age 5) are DR3/4 heterozygotes versus 30% of young adults presenting with type 1A diabetes.

There are three HLA molecules that provide dominant protection. The most common is DQB1*0602 that occurs in approximately 20% of U.S. individuals(41-43). Protection is not absolute, but less than 1% of children with type 1A diabetes have this molecule. DQA1*0201 with DQB1*0303 and DRB1*1401 also provide dramatic protection, rarely being found in patients with type 1 diabetes and rarely transmitted from a parent with the alleles to their diabetic offspring(38,39). It is noteworthy that both DR and DQ alleles can protect. The specific mechanism underlying both susceptibility and protection are not fully understood. One attractive hypothesis is that protective alleles when expressed within the thymus lead to deletion of T cells with receptors that recognize a critical islet peptide(44). With deletion of such T cells, the risk of diabetes would be reduced. In addition it is likely that high-risk HLA alleles present specific peptides of target islet molecules to T lymphocytes(28).

Figure 7

Summary of subset of confirmed loci from whole genome screens associated with type 1A diabetes (from Teaching Slides www.barbaradaviscenter.org) www.barbaradaviscenter.org. Modified from Todd et al. Robust Association of four new chromosomes regions from genome-wide analyses of type 1 diabetes, Nature Genetics, June 6, 2007. (45)

Multiple additional loci (Figure 7) have been implicated with estimates that approximately 50% of the familial aggregation of type 1 diabetes is attributable to the HLA region, perhaps 10% to the insulin locus, with all other loci contributing much less, though in aggregate their contribution is important. In the Cox analysis (Figure 7) of approximately 700 sibling pairs the only significant LOD score was for a locus on chromosome 16q that was not given an iddm designation with earlier genome screens. Several areas implicated in the past had suggestive scores, but there is overlap with the families from which the original evidence was generated. It is likely that contributing loci may differ between populations contributing to the initial difficulty of replicating putative loci in different studies(56,57).More than 40 genetic loci contributing to diabetes risk have been implicated (Figure 7). Polymorphisms of the insulin gene are well established as contributing to risk. A repeat sequence upstream (5') of the insulin gene termed a Variable nucleotide tandem repeat or VNTR, is divided into three general repeat sizes with the longest set of repeats associated with protection from diabetes(46-48). This set of alleles is also associated with greater thymic production of insulin messenger RNA(49), leading to the hypothesis that greater thymic message and presumably greater proinsulin production dampens anti-insulin autoimmunity(49-51). A functional polymorphism of the LYP gene (Lymphocyte Specific Phosphatase; PTPN22- Protein Tyrosine Phosphatase) has been associated with type 1 diabetes, rheumatoid arthritis, and lupus erythematosus(52-54). The R620W missense mutation (tryptophan replacing arginine) disrupts the binding of the phosphatase to the molecule Csk and this blocks its ability to down-regulate T cell receptor signaling. With an odds ratio of between 1.7 and 2.0 of the "autoimmunity” allele which is relatively common (5-10% allele frequency) there is a large genetic effect that is much greater than CTLA-4 polymorphisms associated with diabetes risk(55). Combining known diabetogenic polymorphisms of LYP, the insulin gene, alleles of DP, DQ, and DR class II immune response genes, as well all of the new loci account for approximately 48% of the familial aggregation of type 1A diabetes, with DR and DQ loci accounting for 41% of this 48%(45). A recent study suggests that for a major subset of individuals with the highest risk HLA genotype (DR3/4-DQ2/DQ8 heterozygotes) who share both HLA haplotypes with a diabetic sibling, risk of activating anti-islet autoimmunity is as high as 80%(33).

STAGE II: TRIGGERING

Anti-islet autoimmunity (i.e., insulin autoantibodies), insulitis, and immune mediated diabetes can be triggered in a number of animal models by a large number of immunologic and genetic manipulations(58-60). Perhaps the most relevant of these manipulations is the administration of poly-IC (poly inosinic cytodylic acid) to normal rat strains that have the diabetes susceptible major histocompatibility RT1-U haplotype(61). Administration of poly-IC in some of these normal strains leads to insulitis while in others it leads to overt diabetes with islet beta cell destruction. Poly-IC interacts with Toll 3 receptors of the innate immune system leading to a cascade of intracellular and cytokine mediated events. Infection of Diabetes Resistant BB rats, a strain of rats that does not develop diabetes and lacks the lymphopenia gene of the spontaneously diabetic BB rat strain, leads to diabetes probably by a pathogenic mechanism similar to the effects of poly-IC(62). Poly-IC is a mimic of viral double stranded RNA and thus it is easy to envision that many common RNA viral infections may induce diabetes in genetically susceptible patients. Normal mouse strains such as Balb/c mice rapidly develop insulin autoantibodies if challenged with an insulin peptide (B chain peptide, amino acids 9 to 23)(58). If the peptide is administered with poly-IC, insulitis is induced and in genetically susceptible mice diabetes can be induced. These and many other studies in animal models indicate that normal animals harbor autoreactive B and T lymphocytes that can be expanded and activated, with resultant diabetes. Though these strains are "normal” they have variants of MHC molecules that determine disease susceptibility, by influencing T cell responses to relevant peptides. The class II MHC molecules (equivalent of DR and DQ of man) appear to be most important, and these molecules probably influence disease either by the peptides they bind and present to T cells within islets and draining lymph nodes and by their influence on the thymic T cell repertoire(63). It has recently been appreciated that the position or “register” of insulin B chain amino acids 9-23 presented by the class II MHC molecule activates insulin specific T cells towards islet autoimmunity in a spontaneous mouse model of autoimmune diabetes. With the remarkable homology between mouse and humans (e.g., similar MHC class II molecules contributing genetic risk, identical amino acid sequences for insulin B chain amino acids 9-23, and similar T cell receptors responding to insulin), the underlying biology of insulin presentation may be identical in mouse and humans(64,65).

In humans, environmental factors that trigger anti-islet autoimmunity are largely unknown. Congenital rubella infection is associated with a risk of type 1 diabetes that exceeds 1/5(66,67). It is however only congenital infection that increases the risk of diabetes, and leads to a series of autoimmune disorders (e.g., thyroid autoimmunity). It is thus likely that the congenital infection damages the developing immune system leading to relatively broad disease susceptibility(68).

Figure 8 lists a number of additional environmental factors that may impact on the development of type 1A diabetes. The enteroviruses are probably the most extensively studied(69,70). Investigators have associated for instance antibodies to Coxsackie viruses and Coxsackie viral RNA with type 1A diabetes(71). An important study from Finland has provided

Figure 8

Environmental factors that may increase or decrease diabetes risk. Congenital rubella infection is bolded as it is the only clearly defined factor.

evidence that enterovirus infection may be associated with the activation of anti-islet autoimmunity as measured by the appearance of anti-islet autoantibodies. Such studies have been difficult to replicate, and the DAISY study from Denver, Colorado, that also follows newborns from birth for evidence of anti-islet autoimmunity has not found an association with the enteroviral infection(70).

Figure 9

Secular trend in the incidence of type 1A diabetes in Finland subdivided by the age of appearance of diabetes. Modified from teaching slides www.barbaradaviscenter.org

Dietary factors that may contribute to diabetes are being extensively studied. One hypothesis is that bovine milk ingestion, particularly in the first months of life, is associated with development of diabetes(72). Again there is conflicting data with reports that bovine milk ingestion increases the development of anti-islet autoantibodies, and reports from Denver, Munich, and Melbourne, that it does not(73). Recently two studies (Germany and Denver) have implicated early (<3 months) introduction of cereals as a risk factor for type 1 diabetes(74,75). In the Denver study, late introduction (>7 months) also increased risk. At present it is my view that the data do not allow any firm recommendation in terms of changing infant diet, except to emphasize the current standard recommendation for introduction of cereals between 3 and 7 months.

Environmental factors might not only increase the development of diabetes but may also provide protection. There is a very wide range in the risk of type 1 diabetes ranging from an annual incidence of less than 1/100,000 in China to approximately 50/100,000 in Finland(76). Much of this difference may relate to genetic factors, but there is strong evidence that environmental factors are influencing diabetes risk. The strongest evidence comes from a marked secular trend in terms of increasing diabetes incidence in multiple populations(77). As shown in figure 9, the incidence has increased dramatically particularly for children developing diabetes prior to age 5, increasing more than 3 fold over the past three decades. Such a rapid change in disease incidence cannot be due to changes in gene pool. Something that increases risk has been added, or more likely an environmental factor that decreases risk, has been removed from the population. With increasing public health, a "hygiene” hypothesis has been advanced, particularly directed at asthma and type 1 diabetes(78). It is hypothesized that as the environment becomes "cleaner” the normal development of the immune system is disrupted (e.g., regulatory T cell development is subnormal) resulting in increases of both presumed Th2 (asthma) and Th1 (Type 1 diabetes) mediated diseases. For instance, one review discusses decreasing pinworm infection as a potential factor(78).

STAGE III AUTOIMMUNITY

The assays for anti-islet autoantibodies have improved remarkably over the past three decades(79). A series of anti-islet autoantibody workshops where sera is sent blinded to multiple laboratories throughout the world has stimulated assay improvements and standardization(69). Such workshops have evaluated not only anti-islet autoantibodies of humans but also of the NOD mouse model of type 1 diabetes. In the mouse model the only specific autoantibody detected reacted with insulin(80) and similar to studies in humans, fluid phase radioassays provided better sensitivity and specificity compared to ELISA. Newer assays using electrochemiluminescence to detect islet autoantibodies has allowed for higher sensitivity and specificity and the potential to multiplex each individual assay.(81,82)

The autoantibodies that are primarily measured in humans react with insulin, glutamic acid decarboxylase 65, and ICA512 (IA-2) and ZnT8 (zinc transporter 8) (Figure 10). GAD67 autoantibodies are primarily a subset of antibodies that cross-react with GAD65, and similarly IA-2beta antibodies are predominantly a subset of ICA512 autoantibodies.

Figure 10

Five characterized islet autoantigens with specific/sensitive autoantibody assays. Modified from teaching slides www.barbaradaviscenter.org

Insulin autoantibodies are usually the first autoantibody to appear in children followed from birth for the development of type 1A diabetes(84,85). These autoantibodies can appear in the first six months of life. Once insulin autoantibodies appear in such young children there is a high risk of development of additional anti-islet autoantibodies and progression to diabetes. More than 90% of children developing type 1A diabetes prior to age 5 have insulin autoantibodies while less than 50% of children developing diabetes after age 12 have such autoantibodies(86). Therapy with human insulin induces insulin antibodies that cannot at present be distinguished from insulin autoantibodies. Thus if an individual has been treated with insulin for more than several weeks, positive insulin autoantibodies are not interpretable. For all autoantibodies measured in the first 9 months of life, the antibodies may be transplacental in origin, a particular problem if a mother has type 1 diabetes and is treated with insulin.There are a number of important caveats in the utilization of anti-islet autoantibody assays. The field developed from the initial observation that patient's sera "stained” islets of cut sections of human pancreas, the cytoplasmic islet cell antibody (ICA) assay(83). This assay, given its utilization of human pancreas from cadaveric donation and subjective reading of slides, has proven the most difficult to standardize(69). The assay predominantly detects antibodies reacting with GAD65, IA-2 and ZnT8, but does not detect anti-insulin autoantibodies. Given the difficulty in standardization, reliability over time, and major overlap with defined autoantibody assays, a number of investigators no longer utilize this assay. For research purposes and potentially in older adults with what has been termed LADA (latent autoimmune diabetes of adults) the ICA assay may have utility in that there is evidence of one or more additional autoantibodies detected with this assay and not with GAD65, IA-2, ZnT8 and insulin autoantibody determination.

Figure 11

Progression to diabetes of first-degree relatives of patients with type 1 diabetes subdivided by the number of autoantibodies expressed (of GAD65, IA-2, and insulin). Adapted from Verge et al, Diabetes 1996 (87)

A single autoantibody, even when present on multiple occasions, is associated with only a modest risk of progression to diabetes: approximately 10%(87,88). Once two or more anti-islet autoantibodies are present in children, progression to diabetes is very high, approaching almost 100% after 15 years of follow up(89) (Figure 11). In addition, once multiple autoantibodies are present it is very unusual for an individual to lose all expression of autoantibodies prior to the development of overt diabetes. Following the development of diabetes, IA-2 and more slowly GAD65 (over decades) autoantibodies wane. Following islet or pancreatic transplantation expression of GAD65 and IA-2 autoantibodies can be induced in patients with long-standing diabetes(90).

The most specific of the autoantibodies react with the molecule IA-2, but IA-2 autoantibodies are usually detected following the appearance of insulin and/or GAD65 autoantibodies(84). Even with IA-2 autoantibodies, however, there are apparent "false” positives in terms of diabetes risk. We evaluated approximately 10 individuals with either transient IA-2 autoantibodies or normal controls with IA-2 autoantibodies. None of these individuals expressed an additional anti-islet autoantibody. In contrast to patients diagnosed with or developing type 1 diabetes, the ICA512/IA-2 autoantibodies of nine out of ten of these normal individuals did not recognize multiple ICA512 epitopes and did not react with the dominant ICA512 autoantigenic domain(91). This indicates that even with a highly specific radioassay, if one screens tens of thousands of sera, one can find sera that presumably by chance cross-react with some epitope of the IA-2 molecule. It is much less likely to find an individual with antibodies that by chance react with two different islet autoantigens using fluid phase radioassays set with specificity at the 99th percentile of controls.

STAGE IV. LOSS OF INSULIN SECRETION

Not all individuals with two or more autoantibodies are destined to progress to type 1 diabetes. For instance the diabetes risk is unknown for individuals with expression of two or more anti-islet autoantibodies with the protective HLA molecule DQA1*0102, DQB1*0602. Figure 12 is a life table of progression to diabetes of first-degree relatives with high titer cytoplasmic autoantibodies with a subset of relatives having the protective HLA allele DQB1*0602. Two of the DQB1*0602 relatives expressed multiple "biochemical” anti-islet autoantibodies (one with two autoantibodies and the other three) and neither of these individuals has yet progressed to diabetes. In that approximately 1% of patients with type 1 diabetes have DQB1*0602 it is possible that these individuals will eventually progress to diabetes.

Figure 12

Lack of progression to diabetes of cytoplasmic ICA positive first-degree relatives that have the protective HLA allele DQB1*0602. Modified from teaching slides www.barbaradaviscenter.org

There are three general hypotheses in terms of progression to type 1A diabetes (Figure 13). At present, beta cell mass is not readily measured over time in humans, so it is not possible to absolutely define progression of beta cell loss. There is however no doubt that measurable anti-islet autoimmunity precedes the development of diabetes in terms of anti-islet autoantibodies in humans, and autoantibodies and T cell invasion in animal models. In the NOD mouse there is evidence of some beta cell destruction and beta cell regeneration prior to the onset of diabetes (92). There is also evidence for a change in the immune system close to the time of onset of diabetes (i.e., Th2 to Th1)(93-96). This change is associated with more rapid disease progression, ability to transfer diabetes by T cells, and a time window during which a specific immunotherapy (monoclonal anti-CD3 antibodies) is effective(97). In humans the best evidence for progressive loss of beta cell function comes from studies of insulin and C-peptide secretion(98). C-peptide, the connecting peptide of proinsulin, is secreted in equimolar amount to insulin, but C-peptide is not present in insulin preparations utilized to treat diabetes. Thus C-peptide has become an important indicator of remaining beta cell function. Following the onset of diabetes it has long been appreciated that C-peptide secretion progressively declines, until for most patients with type 1 diabetes C-peptide becomes non-detectable, associated with true insulin dependence. In a similar manner, first phase insulin secretion following a bolus of glucose on intravenous glucose tolerance testing is progressively lost for relatives followed to the development of type 1 diabetes(99). Such metabolic abnormalities may result in part from functional inhibition of beta cell secretion, but pathologic studies indicate that beta cell mass is normal for identical twins of patients that have not activated anti-islet autoimmunity, and for new onset patients that bulk of beta cells are destroyed(100). Within the pancreas of a patient with type 1 diabetes there is heterogeneity of islet lesions, with most islets lacking all beta cells and with no lymphocytic infiltrates (pseudoatrophic islets), few normal islets with no infiltrates, and few islets with remaining beta cells and infiltrates. This is perhaps analogous to the progressive development of vitiligo in patients, with patches of skin with all melanocytes destroyed, whereas other skin is normal.STAGE IV. LOSS OF INSULIN SECRETION

Figure 13

Three hypotheses of progression to type 1A diabetes. Modified from teaching slides www.barbaradaviscenter.org

STAGE V. OVERT DIABETES

The development of type 1 diabetes is usually perceived as an abrupt event, and some individuals may rapidly manifest severe hyperglycemia. Now that we can follow individuals to the development of type 1A diabetes, we can see that anti-islet autoantibodies can precede hyperglycemia by years, and there is usually some deterioration in glucose tolerance more than one year prior to diabetes onset (particularly with intravenous glucose tolerance testing)(101). The majority of individuals identified to be diabetic following autoantibody testing are found to have a diabetic 2-hour glucose on oral glucose tolerance testing (>200mg/dl) rather than fasting hyperglycemia. The acute presentation with severe hyperglycemia and ketoacidosis is life threatening, and it is estimated that approximately 1/200 children die at the onset of type 1 diabetes(102,103). Such children typically have a medical history where the first health care providers have failed to make the diagnosis of diabetes; the child then presents again later and dies with cerebral edema. The classic symptoms of polyuria, polydipsia, and weight loss are usually present but the initial diagnosis is still missed. The alternative diagnosis of nausea and vomiting due to viral illness is the most common mistaken diagnosis, and with the ready availability of glucose determination from a finger or heel stick, there should be a low threshold in emergency rooms and physicians’ offices for ruling out diabetes. Though transient hyperglycemia can occur, such children obviously need close follow up. We usually arrange glucose monitoring for children thought to have transient hyperglycemia, and measure anti-islet autoantibodies(104). Of those with anti-islet autoantibodies and transient hyperglycemia, almost all progress to type 1 diabetes within several months.STAGE V. OVERT DIABETES

A report by Barker and coworkers illustrates the potential importance of screening for anti-islet autoantibodies and monitoring high-risk children. Less than 5% of children in such intensive follow-up who developed diabetes were hospitalized with ketoacidosis compared to 40% of children not being screened. Figure 14 illustrates the blood glucose for the two groups at the time of diagnosis with markedly elevated levels in the non-screened children, including values greater than 1,000 mg/dl associated with severe metabolic decompensation(105).

Figure 14

Glucose at diagnosis in children screened for anti-islet autoantibodies and metabolically monitored, versus children from the general population presenting with diabetes(105). Modified from teaching slides www.barbaradaviscenter.org of Type 1 Diabetes: Molecular, Cellular and Clinical Immunology.

At the onset of type 1 diabetes, almost all individuals have residual insulin secretion, and there is convincing evidence that residual insulin secretion as measured by C-peptide secretion is of clinical benefit (less hypoglycemia, less microvascular complications, and much easier diabetes management). With many immunologic therapies it is possible to prevent diabetes in animal models, and as reviewed, the disorder is predictable in humans. At present there is no proven therapy to either prevent progression to type 1 diabetes or to halt beta cell loss after presentation with diabetes. One arm of the DPT-1 prevention trial, where low doses of subcutaneous insulin were administered, did not delay progression to diabetes(106). The DPT-1 trial of oral insulin indicated that overall, oral insulin did not delay progression to diabetes. A subset analysis of antibody positive relatives entering the trial with elevated levels of insulin autoantibodies, however, suggests a delay in progression to diabetes of approximately 4 years. A repeat trial of oral insulin for diabetes prevention is underway by TrialNet(107). Several studies of anti-CD3 monoclonal antibody modified to reduce cytokine release have shown promise in patients with new onset and recent onset diabetes(108-111), but these monoclonal antibodies only transiently delay the loss of c-peptide secretion(112-114). Other therapies include anti-CD20 monoclonal antibody (rituximab targeted to B cells). Treatment with CTLA-4 Ig and alefacept (anti-CD2) similarly transiently delays the loss of c-peptide(115-117). In contrast, trials of a GAD/alum vaccine did not delay the loss of endogenous c-peptide production nor did therapies targeting the IL-1 pathway(118-120). It is our belief that more specific therapies targeting autoreactive immune cells such as recently reported by Roep and colleagues(121) used in combination will allow for disease prevention and potentially a cure of type 1 diabetes when done in conjunction with islet replacement

.SUMMARY

The National Institutes of Health has created a cooperative trial network to develop therapies to prevent beta cell loss in prediabetics and new onset patients, termed TrialNet. Relatives of type 1 patients throughout North America can be screened for diabetes risk, and trials are underway to preserve beta cell function in new onset patients. A contact number for TrialNet is 1-800-HALT-DM1. In addition, the Immune Tolerance Network is evaluating therapies in a series of disorders designed to restore "tolerance,” including type 1A diabetes. They are seeking applications to test innovative therapies at immunetolerance.org.

Acknowledgments: Our research is supported by grants DK32083, DK095995, DK085509, DK097681, AI097346, Diabetes Endocrine Research Center (P30 DK57516), Autoimmunity Center of Excellence (UM1AI110503) and Human Islet Research Center (UC4DK104223) from the National Institute of Health, the Juvenile Diabetes Research Foundation, Brehm Coalition, Helmsley Foundation and a grant from the Children's Diabetes Foundation at Denver. We wish to acknowledge the late Dr. George Eisenbarth whose numerous contributions to this chapter and the field, have been invaluable.