The rapid increase in the incidence of type 1 diabetes in genetically stable populations implies an important role for environmental factors. Since autoantibodies directed against beta cell constituents and predictive of future diabetes typically appear within the first few months or years of life, interaction with the environment seems most likely to have begun before or shortly after birth. Prenatal influences are suggested by the observation that babies with higher birthweight and the children of older mothers have a slightly higher risk of diabetes. Leading post-natal environmental candidates include exposure to enteroviruses, early diet, levels of vitamin D, and routine vaccinations. Other more general factors have been proposed. These include lack of immune stimulation due to an antigen-free environment (the hygiene hypothesis) or childhood overnutrition resulting in increased insulin resistance (the accelerator hypothesis).
There are a number of indirect indications that events before birth may influence a child's subsequent risk of diabetes. One is that a child whose mother has type 1 diabetes is less likely to develop the condition than when the father is affected. The risk is slightly increased with higher birthweight, in the offspring of older mothers, and in children with a higher birth order. Pre-eclampsia does not appear to increase the risk.
A seasonal onset of type 1 diabetes has been observed, with increased incidence in the spring and autumn, and the reverse phenomenon is seen in the southern hemisphere. This might be consistent with a role for viral infection in precipitating clinical onset of diabetes. There appears to be no clear association with season of birth, population density or spatial clustering, and no consistent findings have emerged with respect to socioeconomic status.
Viruses are attractive aetiological candidates. They penetrate cells and alter the way in which these are recognised by the immune system; they show tropism for tissues including the pancreatic islets, and animal models provide good evidence that viruses can provoke autoimmune responses. In addition, susceptibility to viral infection is often determined by HLA type in common with a whole range of autoimmune conditions. The difficulties of pin-pointing a viral aetiology for an autoimmune disease have been discussed by Rose. Most virus infections are asymptomatic or trivial and causal exposures might precede clinical disease by years or even decades. Common viruses are ubiquitous and frequently exchange genetic material by recombination; the timing or frequency of exposure or the particular strain of virus involved may be critical. There are in addition many potential mechanisms by which viral infection might promote autoimmunity or interfere with beta cell function. At one end of the spectrum these may cause acute cell rupture with exposure of intracellular antigens to the immune system; at the other, persistent or latent infection might modify the affected cell in ways that lead to altered recognition by the immune system, followed in turn by chronic low grade damage.
Timing of exposure is another key issue. Prenatal viral exposure has been considered potentially diabetogenic in congenital rubella, cytomegalovirus (CMV) and enteroviral infections, but the case is far from proved. Alternatively, viral infection might trigger autoimmune responses against the pancreatic islets, and evidence for this has been sought in prospective studies from birth, but remains circumstantial. Viruses might also act by precipitating clinical onset when superimposed upon an established chronic autoimmune process involving the islets. Finally, the possibility of reverse causation (i.e. that children at increased risk of diabetes are also more susceptible to viral infection or to viral invasion of damaged beta cells) cannot be excluded.
EnterovirusEnteroviruses, including coxsackie B4, are the most widely considered viral candidates, and have been implicated by analysis of circulating viral RNA and by post mortem examination of pancreases from children who died soon after the onset of diabetes.
Cow's milk and other early nutritional factors
Coeliac disease and type 1 diabetes have common features, including a degree of overlap between HLA haplotypes conferring susceptibility. Dietary gluten is the environmental agent responsible for coeliac disease, but the condition typically develops years or even decades following first exposure. Dietary constituents encountered early in life might potentially play a similar role in the pathogenesis of type 1 diabetes. There has been controversy for 15 years as to whether early exposure to cows' milk predisposes to childhood diabetes; similar arguments have taken place in other diseases, including asthma and multiple sclerosis, but these remain equally inconclusive. Animal studies have shown that elimination of milk proteins (present in standard laboratory chow) from the early diet greatly reduces the risk of diabetes in the BioBreeding (BB) rat, with similar results in the non-obese diabetic (NOD) mouse. Other dietary constituents, including wheat and soya beans have also been implicated, however, and cow's milk is only one among a number of possible dietary candidates.
A systematic review of the epidemiological evidence in humans found a weak (odds ratio 1.5) but detectable positive effect of early exposure to cow's milk proteins, but subsequent reports have yielded conflicting results and the area remains controversial.  As with virus exposure, prospective analysis of high-risk populations from birth has great advantages over other means of analysis, and will allow exposure to milk and other dietary constituents to be correlated with formation of islet autoantibodies. Improved and standardised methods of analysis of cellular immune responses to dietary constituents are badly needed, and should be corrected for HLA type, but the issue will only be resolved beyond doubt by intervention studies.
Vitamin D is present in food or generated in the skin in response to sunlight; its metabolically active form is 1,25(OH)2D3. Many studies have shown that young people with type 1 diabetes have lower circulating levels of this metabolite than controls, and lack of sunlight correlates well with the increased incidence of type 1 diabetes at higher latitudes. Three key genes involved in 1,25(OH)2D3 metabolism are associated with increased risk of type 1 diabetes, and functional studies confirm that this metabolite is under genetic control but set at lower levels than in control populations. 1,25(OH)2D3 receptors are expressed in pancreatic beta cells and in immunocytes, and high doses of 1,25(OH)2D3 can reduce the incidence of diabetes in the NOD mouse. It has, however, yet to be demonstrated that administration of vitamin D or its analogues can delay the onset of of type 1 diabetes or influence its clinical course in humans. Once again, intervention studies are needed to resolve these issues.
Despite much speculation, there is little or no evidence to suggest that the risk of childhood diabetes is influenced by routine vaccination against childhood infections.
Non-specific environmental factors
It has been suggested that non-specific environmental factors might also influence the incidence of type 1 diabetes. For example, the hygiene hypothesis suggests that a cleaner environment, with less antigenic stimulation in early childhood, might predispose to the development of dysfunctional patterns of immune response, thus linking the rise of type 1 diabetes to the rise of asthma and other childhood atopic disorders. An alternative is the accelerator hypothesis, which proposes that increasing childhood obesity promotes both insulin resistance and progression towards type 1 diabetes.
^ Cardwell CR et al. Birthweight and the risk of childhood-onset type 1: A meta-analysis of observational studies using individual patient data. Diabetologia 2010;53:641–51
^ Cardwell CR et al. Maternal age at birth and childhood type 1: A pooled analysis of 30 observational studies. Diabetes 2010;59:486–94
^ Cardwell CR et al. Birth order and childhood type 1 diabetes a pooled analysis of 31 observational studies. Int J Epidemiol 2011;40:363–74
^ Henry EB et al. A meta-analysis of the association between pre-eclampsia and childhood-onset type 1 diabetes mellitus. Diabet Med 2011;28:900-5
^ Rose NR. The role of infection in the pathogenesis of autoimmune disease. Immunology 1998;10:5–13
^ Wing-Chi G Yeung et al. Enterovirus infection and type 1 diabetes systematic review and meta-analysis of observational molecular studies. BMJ 2011;342:d35
^ Scott FW. Milk and type 1 diabetes. Diabetes Care 1996;10:379–83
^ Gerstein H. Cow’s milk exposure and type 1 diabetes mellitus? A critical overview of the clinical literature. Diabetes Care 1994;17:13–9
^ Harrison LC, Honeyman MC. Cow’s milk and type 1 diabetes. The real debate is about mucosal immune function. Diabetes 1999;48:1501–7
^ Ellis TM, Atkinson MA. Early infant diets and insulin-dependent diabetes. Lancet 1996;347:1464–5
^ Cooper JD et al. Inherited variation in vitamin D genes is associated with predisposition to autoimmune type 1 diabetes. Diabetes 2011;60:1624-31
^ Salemi S, D'Amelio R. Could autoimmunity be induced by vaccination? Int Rev Immunol 2010;29:247–69
^ Wilkin TJ. The accelerator a review of the evidence for insulin resistance as the basis for type 1 as well as type 2 diabetes. Int J Obes 2009;33:716–26