The most common cause of type 1 diabetes (over 90% of cases) is T cell-mediated autoimmune destruction of the islet beta cells leading to a failure of insulin production. The exact aetiology is complex and still imperfectly understood. However, it is probable that environmental factors trigger the onset of diabetes in individuals with an inherited predisposition. Unless insulin replacement is given, absolute insulin deficiency will result in hyperglycemia and ketoacidosis, which is the biochemical hallmark of type 1 diabetes.
There is a striking variation in the incidence of type 1 diabetes between and within populations, with high frequencies in Finland (49 cases/100,000/year) and Sweden (32/100,000/yr), and low frequencies in areas of China and Venezuela (both 0.1/100,000/year) and the Ukraine (1/100,000/year). Marked differences also occur within the same country: the incidence in Sardinia (37/100,000/year) is 3-5 times that of mainland Italy. These differences in frequency suggest that environmental and/or ethnic-genetic factors may influence the onset of the disease….
The geographical variation within Europe has been highlighted by the EURODIAB epidemiology study. This survey found a 10- fold difference in the incidence of type 1 diabetes between Finland and Macedonia. The incidence generally falls along a north-south gradient, but Sardinia is a notable ‘hot spot’ with a much higher frequency than the surrounding Mediterranean areas. Interestingly, there are also different incidences in genetically similar countries such as Finland and Estonia, or Norway and Iceland. Moreover, the most recent report suggests an overall annual rate of increase in incidence in children aged below 15 years of 3.9% (range 0.6 – 9.3%). Prevalence is predicted to rise from 94,000 (2005) to 160,000 in 2020 in Europe. This suggests that environmental influences may predominate over genetic susceptibility in causing or triggering the disease.
Further evidence for environmental influences comes from studies that show a seasonal variation in the onset of type 1 diabetes in some populations, with the highest frequency in the colder autumn and winter months (Figure 6.4). This is often thought to reflect seasonal exposure to viruses, but food or chemicals might also be involved.
People who have migrated from an area of low to an area of high incidence for type 1 diabetes seem to adopt the same level of risk as the population to which they move. For example, children of Asian families (from the Indian subcontinent and Tanzania) who moved to the UK traditionally have a low frequency of type 1 diabetes but now have a rising incidence of the disease, which is approaching that of the indigenous population (Figure 6.5).
Familial clustering of type 1 diabetes provides evidence for complex genetic factors in its aetiology (Figure 6.6). In (European) siblings of children with type 1 diabetes, 5 – 6% have developed type 1 diabetes by the age of 15 years and 20% have diabetes if they are human leukocyte antigen (HLA) identical, compared with the population frequency of about 0.4%. However, only 10 – 15% of type 1 diabetes occurs in families with the disease (‘multiplex’) and most cases are said to be ‘sporadic’. The chance of a child developing type 1 diabetes is around 5% if one parent is affected or 15% if both have the condition. The risk is greater if the father is affected and there is also a small male preponderance in overall prevalence. The reasons for these sex differences remain unknown.
The incidence of type 1 diabetes is increasing in many countries. In Europe, the overall increase is 3.4% per year, but the increase is particularly notable in those diagnosed under the age of 5 years, where it is 6.3% per year and total numbers are likely to double by 2020 (Figure 6.7). Based on these figures, the prevalence of type 1 diabetes may be 70% higher in 2020 than in 1989. This sharp rise in frequency over a short period of time suggests changing environmental factors that operate in early life, as genetic factors would take much longer over several generations to make an impact.
Evidence for autoimmunity in the pathogenesis of type 1 diabetes comes from postmortem studies in patients who have died shortly after presentation and pancreatic biopsies from living patients. They have revealed a chronic inflammatory mononuclear cell infiltrate (‘insulitis’) (Figure 6.8 ) associated with the residual β cells in the islets of recently diagnosed type 1 diabetic patients. The infiltrate consists of T cell lymphocytes and macrophages. Later in the disease, there is complete loss of β cells, while the other islet cell types (α ,δ and PP cells) all survive. A major marker of insulitis is the presence of four circulating islet-related autoantibodies in patients with newly diagnosed type 1 diabetes; islet cells (ICAs), insulin molecule (IAAs), tyrosine phosphatase (IA – 2) and glutamic acid decarboxylase (GAD) antibodies.
However, not all those with islet autoantibodies go on to develop diabetes, which suggests that insulitis does not necessarily progress to critical β cell damage. Type 1 diabetes is manifest clinically after a prodromal period of months or years, during which immunological abnormalities, such as circulating islet autoantibodies, can be detected, even though normoglycemia is maintained.
In family studies, positivity to three or more autoantibodies confers a risk of developing type 1 diabetes of 60 – 100% over 5 – 10 years. Single positivity carries a much lower positive predictive value.
The autoimmune basis for type 1 diabetes is also suggested by its association with other diseases such as hypothyroidism, Graves’ disease, pernicious anemia and Addison’s disease which are all associated with organ-specific autoantibodies (Box 6.1 ). Up to 30% of people with type 1 diabetes have autoimmune thyroid disease.
The detection of ICAs and GAD antibodies in older persons with type 2 diabetes in Finland and the UKPDS, who were shown subsequently to be more likely to require insulin therapy, has led to the concept of latent autoimmune diabetes of adults (LADA). However, this concept has been challenged. Although GAD positivity had a specificity of 94.6% for early insulin use in the UKPDS, its sensitivity was only 37.9%. Moreover, the positive predictive value for GAD-positive antibodies was only 50.8% (i.e. only half of those positive went on to need insulin). Furthermore, people with LADA had a similar pattern of HLA haplotype (see below) as those developing type 1 diabetes in childhood. It is likely therefore that as LADA patients have some but not all of the immunological markers of type 1 diabetes of childhood, they represent part of a spectrum of autoimmune disease rather than a separate entity in their own right.
Genetic susceptibility to type 1 diabetes is most closely associated with HLA genes that lie within the major histocompatibility complex (MHC) region on the short arm of chromosome 6 (now called the IDDM1 locus). HLAs are cell surface glycoproteins that show extreme variability through polymorphisms in the genes that code for them. Both high- and low-risk HLA haplotypes have been identified. HLA DR/4, DQA1 * 0301 – DQB1 * 0302 and DQA1 * 0501 – DQB1 * 0201 account for over 50% of genetic susceptibility; whereas DQA1 * 0102 – DQB1 * 0602 and DRB1 * 1401 are protective.
Class II HLAs (HLA – D) play a key role in presenting foreign and self-antigens to T-helper lymphocytes and therefore in initiating the autoimmune process (Figure 6.11 ).
Polymorphisms in the DQB1 gene that result in amino acid substitutions in the class II antigens may affect the ability to accept and present autoantigens derived from the β cell. This is a critical step in ‘arming’ T lymphocytes, which initiate the immune attack against the β cells. Over 20 regions of the human genome are associated with type 1 diabetes, but most make only a minor contribution. IDDM2 corresponds to the insulin VNTR gene locus on chromosome 11, has a smaller effect than IDDM1 and acts independently. Together with IDDM 12 (CTLA – 4), it contributes around 15% of the genetic risk. The other 17 genes individually contribute little. The predisposing polymorphism in the IDDM 2 gene and how it influences the disease has yet to be determined.
Environmental and maternal factors
Although genetic factors are undoubtedly important, the relatively low concordance of < 50% in monozygotic twins together with the rapidly increasing incidence rates for type 1 diabetes at a younger age strongly suggest that external or environmental factors are playing a part. Much of the evidence that links environmental factors with the aetiology of type 1 diabetes is circumstantial, based upon epidemiology and animal research. The factors most often implicated are viruses, and diet and toxins, but a number of other influences, such as early feeding with cow’s milk and psychological stress, are being investigated.
Recent meta-analyses and pooled cohort studies have shown a link to birth weight (a 7% increase in risk for every 1 kg in weight); Caesarean section (a 20% increase); and maternal age (5% increase for each 5 years). These associations remain unexplained.
The viruses that have been implicated in the development of human diabetes have been deduced from temporal and geographical associations with a known infection. For example, mumps can cause pancreatitis and occasionally precedes the development of type 1 diabetes in children. Intrauterine rubella infection induces diabetes in up to 20% of affected children. Many people with recent-onset type 1 diabetes have serological or clinical evidence of coxsackie B virus infection, particularly the B4 serotype. Marked islet β cell damage has been detected in children who died from coxsackie B virus infection.
In a few cases, coxsackie viral antigens have been isolated in islets postmortem, and viruses isolated from the pancreas have been shown to induce diabetes in susceptible mouse strains. Electron microscopy of the pancreas in some subjects who died shortly after the onset of type 1 diabetes identified retrovirus-like particles within the β cells, associated with insulitis.
Viruses may target the β cells and destroy them directly through a cytolytic effect or by triggering an autoimmune attack (Figure 6.12 ). Autoimmune mechanisms may include ‘molecular mimicry’; that is, immune responses against a viral antigen that cross -react with a β cell antigen (e.g. a coxsackie B4 protein (P2-C) has sequence homology with GAD, an established autoantigen in the β cell). Also, anti-insulin antibodies from type 1 diabetic patients cross-react with the retroviral p73 antigen in about 75% of cases.
Alternatively, viral damage may release sequestered islet antigens and thus restimulate resting autoreactive T cells, previously sensitized against β cell antigens (‘bystander activation’). Persistent viral infection could also stimulate interferon-α synthesis and hyperexpression of HLA class I antigens, and the secretion of chemokines that recruit activated macrophages and cytotoxic T cells.
One model of β cell destruction is via the process of apoptosis or programmed cell death (Figure 6.13). This is effected by the activation of cellular caspases triggered by several means, including the interaction of cell surface Fas (the death-signalling molecule) with its ligand FasL on the surface of infiltrating cells. Other factors that induce apoptosis include macrophage derived nitric oxide (NO) and toxic free radicals, and disruption of the cell membrane by perforin and granzyme B produced by cytotoxic T cells. T cell cytokines (e.g. interleukin-1, tumour necrosis factor-α, interferon-γ) upregulate Fas and FasL and induce NO and toxic free radicals.
Wheat gluten is a potent diabetogen in animal models of type 1 diabetes (BB rats and NOD mice; see below), and 5 – 10% of patients with type 1 diabetes have gluten-sensitive enteropathy (coeliac disease). Recent studies have demonstrated that patients with type 1 diabetes and coeliac disease share disease-specific alleles. Wheat may induce subclinical gut inflammation and enhanced gut permeability to lumen antigens in some patients with type 1 diabetes, which may lead to a breakdown in tolerance for dietary proteins. Other possible diabetogenic factors in diet include N-nitroso compounds, speculatively implicated in Icelandic smoked meat, which was a common dietary constituent in winter months.
It has been suggested that early weaning and introduction of cow’s milk may trigger type 1 diabetes, but this remains controversial. Surveys have shown associations between both the consumption of milk protein and a low prevalence of breastfeeding with the incidence of type 1 diabetes in different countries (Figure 6.14). It is hypothesized that antibodies against bovine serum albumin may cross-react with an islet antigen (ICA69). The studies are inconsistent, perhaps because of variations in milk composition or the existence of a subset of milk-sensitive, diabetes-prone people. Immune tolerance to insulin might also be compromised by cow’s milk, which contains much less insulin than human milk.
The notion that there may be environmental β cell toxins is supported by the existence of chemicals that cause an insulin-dependent type of diabetes in animals. Examples are alloxan and streptozocin, both of which damage the β cell at several sites, including membrane disruption, enzyme interaction (e.g. with glucokinase) and DNA fragmentation. The rat poison vacor causes type 1 diabetes in humans, possibly because it has a similar action to streptozocin.
Spontaneous diabetes that resembles type 1 diabetes in humans occurs in some animals, notably the BioBreeding (BB) rat and the non-obese diabetic (NOD) mouse. These ‘animal models’ have many of the same characteristics as human autoimmune diabetes, including a genetic predisposition, MHC association, insulitis, circulating islet cell surface and GAD autoantibodies, a long prediabetic period that precedes overt hyperglycemia and environmental factors that trigger or accelerate the appearance of diabetes, such as wheat and cow’s milk proteins.
The increasing incidence of atopy as well as early-onset type 1 diabetes in Western societies may be a consequence of a lack of exposure to common pathogens such as mycobacteria, lactobacilli and helminth worms. Chronic exposure might include a more tolerant T cell response to antigens, while a cleaner, more sterile early environment would result in an exaggerated response. This hypothesis has increasing supportive associative data but remains unproven.
One model of the evolution of type 1 diabetes is that individuals destined to develop the disease are born with genes that confer predisposition and they outweigh any genes with protective effects (Figure 6.15). Environmental factors then act as triggers of the T cell-mediated autoimmune destructive process, which results in insulitis, β cell injury and loss of β cell mass. As β cell function declines, there is loss of the first-phase insulin response to intravenous glucose, subsequent glucose intolerance (pre-diabetes) and eventually the clinical onset of overt diabetes. An alternative view is that there is a chronic interaction between genetic susceptibility, cumulative exposure to environmental factors and immune regulatory processes over the entire period until a critical loss of β cell mass results in insulin deficiency and hyperglycemia.
These events are assumed to proceed more rapidly in children.
Next Week: Epidemiology and aetiology of type 2 diabetes