Viruses and type 1 diabetes
Viruses are frequently invoked as possible causal agents in type 1 diabetes, but the issue remains controversial after 50 years of debate. They are attractive candidates because they penetrate cells and alter the way in which these are recognised by the immune system; because some viruses show tropism for the pancreatic islets; because animal studies show that viruses can provoke diabetes, and because susceptibility to viral infection is often determined by HLA type, in common with autoimmune conditions. Notwithstanding, it has to date proved difficult to confirm a causal role for viruses in the genesis of human diabetes. One reason for this is that evidence of immune activation precedes clinical onset of a diabetes by many years. The best evidence of viral involvement in the pathogenesis of type 1 diabetes comes from animal studies and from the detection of viral sequences in the circulation of people with recently diagnosed diabetes, or in the islets of those who come to post mortem. Retrospective analysis cannot however prove a causal role for viral infection, and prospective studies from birth have not as yet been able to resolve the question. Viral infection does not explain the sustained rise in incidence of type 1 diabetes observed worldwide.
A viral cause for type 1 diabetes has long been considered: mumps was suspected almost a century ago, based upon the known ability of the virus to cause pancreatitis, and epidemiological associations with the rise of juvenile diabetes in Denmark. Diabetes later emerged as a feature of the congenital rubella syndrome, which appeared to offer proof of principle that a virus could cause diabetes, although this form of diabetes had distinctive features of its own. The autoimmune paradigm of the 1960s proposed that the link between organ-specific autoimmunity and human leucocyte antigen (HLA) susceptibility could be bridged by environmental triggers. Viral infection was an obvious candidate. The emerging paradigm for juvenile diabetes was formulated in 1974 as follows: 'one or more immune-response genes associated with HLA-A8 and/or W15 might be responsible for an altered T-lymphocyte response. The genetically determined host response could fail to eliminate an infecting virus (Coxsackie B4 and others) which in turn might destroy the pancreatic beta-cells or trigger an autoimmune reaction against the infected organ'. The viral hypothesis was supported by animal models of virally-induced diabetes and the observation that a virus might account for the seasonal onset of type 1 diabetes in children. It was later noted that a virus might also account for the increased incidence of diabetes in higher latitudes.
Early investigators assumed that juvenile diabetes resulted from acute viral infection, but prospective family studies made it clear that immune-mediated or type 1A diabetes has a long prodrome, indicating that any encounter with a causative virus must have occurred months or years before the onset of beta cell failure. Explanations based upon a causative virus must therefore account for this latent period. Another difficulty is that viral infection is unlikely to be responsible for more than a fraction of cases of type 1 diabetes, which makes it challenging to prove – or disprove – a possible link.
Acute viral infection is, in contrast, considered the likely cause of fulminant type 1 diabetes, a non-immune variant of the condition mainly reported in Japan and south-east Asia.
What types of evidence are available?
Epidemiological evidence has often been used to link an infection with a disease, but difficulties arise when, as with type 1 diabetes, there appears to be a long delay between the infection and the subsequent illness. The problem has been approached by means of cohort and case–control studies in which medical records or serological evidence of previous viral infection are compared in children prior to the onset of diabetes, compared with controls. An alternative has been to look for evidence of recent or ongoing viral infection in children with newly diagnosed diabetes, whether in terms of circulating viral antibodies or molecular markers of viral infection, or by examination of pancreas specimens obtained at autopsy from rare individuals who died following a diagnosis of diabetes. The other alternative line of approach has been by induction of viral disease in experimental animals, and an extensive literature has accumulated around this subject.
How reliable is the evidence that a virus can cause diabetes?
Robert KochRobert Koch suggested that evidence that an infective organism causes a disease should satisfy four postulates. These are that: (1) the microorganism should be present in all cases of the disease; (2) the microorganism must be isolated from a diseased person or experimental animal and grown in pure culture; (3) the cultured microorganism should cause disease when introduced into a healthy host; and (4) the microorganism must be isolated once again from the inoculated, diseased experimental host and confirmed to be identical with the original infective agent. These postulates were formulated with acute infection in mind, and clearly do not work so well when the postulated infection was encountered long before the disease. Koch's postulates have not been satisfied with respect to viral infection in type 1 diabetes or any other autoimmune condition, and other lines of evidence must therefore be considered.
How might a virus predispose to diabetes?
There are many possible ways in which a virus might be responsible for the development of type 1 diabetes, and the wealth of alternatives may have hindered the development of concrete testable hypotheses. The variables might be summarised as follows:
(1) Is a single virus responsible? On current evidence, a single viral strain is very unlikely to be responsible. Generic features of potentially 'diabetogenic' viruses therefore need to be sought.
(2) Is the infection inactive (i.e. historical), latent, or still active? Most virus infections are asymptomatic or trivial, yet these might nonetheless trigger an immunological chain reaction which results in diabetes long after all evidence of the original infection has vanished. Persistent virus infection might potentially hide in the nuclear material of the host cell without seeking to replicate, or it may be reactivated to pursue the pathway of replication and cell lysis.
(3) Does viral infection initiate, promote or precipitate progression to diabetes? As shown in the figure, a virus could potentially act in each of these roles, and the timing of infection may be of key importance. A causal role would thus imply exposure early in life (before or shortly after birth); alternatively, viral infection could promote an existing immune process or compromise beta cell function sufficiently to precipitate the onset of hyperglycaemia.
(4) What is the mechanism linking viral infection and immune-mediated destruction of beta cells? There are many potential mechanisms by which viral infection might promote autoimmunity. These include cell rupture with exposure of intracellular antigens to the immune system; alternatively persistent or latent infection might lead to altered recognition of the cell by the immune system, followed in turn by chronic low grade damage. Another postulated mechanism is molecular mimicry: according to this hypothesis, immune responses triggered by a molecular sequence presented by a virus could cross react with similar sequences on self tissues.
(5) Might susceptibility to viral infection simply be due to the immune defect associated with type 1 diabetes? Case–control studies have not to date provided clear evidence of viral infection as a trigger of autoimmunity. Earlier reports found an increase in viral antibodies in recent onset cases of diabetes, but this was not confirmed by meta-analysis. In contrast, meta-analysis of molecular studies, although open to criticism, has suggested that recent onset cases are more likely to show evidence of viral infection. If confirmed, this might however be explained by reverse causation (i.e. prediabetes may offer a more fertile soil for virus infection), or else the metabolic effects of viral infection might be sufficient to precipitate diabetes in someone already approaching beta cell failure. These observations, therefore, need not necessarily imply any role of the virus in the causation of diabetes.
Overview: Virus pro or con?
As noted above, there are three main lines of argument for viral involvement in the pathogenesis of type 1 diabetes: epidemiological studies and molecular studies in man, and animal studies.
Studies in animal models have clearly shown that viruses have the potential to induce diabetes in susceptible animal strains, but it is doubtful whether any of these models truly reproduce the postulated sequence of events in human immune-mediated diabetes. Animal studies furnish analogies and identify mechanisms, but cannot prove causation in other species.
Epidemiological studies, whether observational or case–control, provide inconsistent evidence of viral involvement, and there is considerable potential for publication bias in favour of positive studies. A recent review concluded that there is no consistent evidence of an association between recent viral infection and the induction of autoimmunity in prospective studies,8 and there is an equal lack of evidence from prospective studies when it comes to a role in precipitating the onset of diabetes. Despite many attempts, it has not proved possible to identify true 'epidemics' (i.e. a time-limited increase followed by a fall in incidence) of childhood type 1 diabetes, and the increasing incidence of diabetes in many countries is associated with improved hygiene and living conditions, which would be expected to reduce viral transmission. There is in fact an inverse association between viral infection and the incidence of diabetes. The 'polio hypothesis', discussed in the section on Enterovirus, attempts to resolve this paradox.
Molecular studies: The strongest arguments in favour of viral involvement derive from molecular studies in the newly diagnosed. Antibody studies have done little to advance the argument, for they do not discriminate between diabetogenic and neutral strains of virus, and there has been concern that differences between antibody responses or viral persistence in people with diabetes and controls may simply reflect patterns of immune responsiveness related to HLA type. In any event, meta-analysis of antibody based studies does not support an association between acute viral infection and new-onset diabetes.
Antibodies represent the footprints of departed viruses, and the actual presence of the virus in the circulation is best detected by PCR techniques. A number of studies have reported detection of viral protein in newly diagnosed or prediabetic children, most commonly coxsackie B4. PCR techniques have the limitation that the RNA sequences detected come from the untranslated part of the genome, and do not indicate the precise serotype, since this is based on variations in capsid proteins.
There are many technical limitations to the identification of trace amounts of viral material in human tissue, and a range of approaches has been attempted. This limits the value of meta-analysis of molecular methods of identification, but this has nonetheless shown strongly positive associations. Further developments in this area will depend upon methodological improvement and standardisation, and would be advanced by a workshop approach to this line of investigation.
Genetic analysis of type 1 diabetes lends some support to the viral hypothesis with the discovery that a locus in the viral RNA receptor gene interferon-induced helicase region (IFIH1) locus associates to type 1 diabetes.
In summary, there is at present conflicting evidence as to the role of early viral exposure in the pathogenesis of type 1 diabetes. Despite a mass of circumstantial evidence relating to serological changes suggestive of recent viral exposure in the newly diagnosed, no reliable conclusions can be drawn at this stage. PCR evidence of viraemia before or around the time of diagnosis continues to accumulate, but is hard to interpret. Large scale prospective studies from birth using improved methods of investigation should eventually resolve some of these issues. On present evidence, viral infection can only be implicated in a minority of cases of type 1 diabetes, and cannot explain the relentless linear rise in incidence noted from many parts of the world.
Enteroviruses, particularly coxsackievirus B4, are the most popular candidates for the causation of type 1 diabetes, and their potential role, together with that of rubella, cytomegalovirus and rotavirus, is discussed in more detail in the following sections.
^ Gundersen E. Is diabetes of infectious origin? J Infect Dis 1927;41:1975–9.
^ Gale EAM. Congenital citation virus or viral cause of type 1 diabetes? Diabetologia 2008;51:1559–66.
^ Nerup J, Platz P, Andersen O et al. HL-A antigens and diabetes mellitus. Lancet 1974;2:864–6
^ Gamble DR et al. Coxsackie viruses and diabetes mellitus. BMJ 1973;4(5887):260-2
^ Gale EAM. The rise of childhood type 1 diabetes in the 20th century. Diabetes 2002;51:3353–61
^ Imagawa A, Hanafusa T. Fulminant type 1 diabetes – an important subtype in East Asia. Diabetes Metab Res Rev 2011;27:959–64
^ Jankosky C et al. Viruses and vitamin D in the etiology of type 1 diabetes and multiple sclerosis. Virus Res 2012;163:424–20
^ Stene LC, Rewers M. The enterovirus link to type 1 critical review of human studies. Clin Exp Immunol 2012;163:12–23
^ Yeung WC et al. Enterovirus infection and type 1 diabetes systematic review and meta-analysis of observational molecular studies. BMJ 2011;342:d35
^ Viskari H et al. Relationship between the incidence of type 1 diabetes and enterovirus infections in different European populations - results from the EPIVIR project. J Med Virol 2004;72:610–7
^ Graves PM et al. The role of enteroviral infections in the development of IDDM. Limitations of current approaches. Diabetes 1997;46:161–8
^ Hyöty H, Taylor KW. The role of viruses in human diabetes. Diabetologia 2002;45:1353–61
^ Smyth DJ et al. A genome-wide association study of nonsynonymous SNPs identifies a type 1 diabetes locus in the interferon-induced helicase (IFIH1) region. Nat Genet 2006;38:617–9