This commentary is not about autoimmune diseases, and therefore I will not discuss why a significant proportion of individuals are prone to develop autoimmune diseases. Suffice to say that this failure to spare self might represent an advantage when fighting infections. In autoimmunity there is a clear genetic involvement with a strong association between these diseases and some HLA alleles,1but autoimmunity is not simply genetically controlled and environmental factors are essential. For example, in the syngenic NOD mice, the preferred animal model for type 1 diabetes, simple modifications in the “cleanness” of the housing conditions dramatically changes the incidence of the disease.2 More poignantly, in the context of a clinical setting, the concordance rate of autoimmune diseases in monozygous (genetically identical) twins is less than 50%.3 Hence environmental factor(s) are essential in induction of autoimmunity. Thus characterisation of “environmental” triggers and how they might promote autoimmunity is of paramount importance if we wish to understand, prevent, and eventually define strategies to treat these diseases. The gastrointestinal tract, with its vast surface of contact with the external world, represents the main door for the potential encounter of “environmental” triggers of autoimmunity.

To curb this risk, the gastrointestinal tract, as other mucosae, has devised means of inducing a protective response that however lacks the inflammatory flavour. In a simplistic way, antigens encountered on a mucosal surface trigger a Th2 (sometimes and depending on the authors Th3 or T regulatory) type of response instead of a proinflammatory Th1 response, dominant in autoimmune diseases.4 This mucosal characteristic has been exploited to protect or even treat autoimmune diseases via induction of mucosal tolerance.5

Some gastrointestinal diseases are associated with autoimmunity but coeliac disease (CD) has two important characteristics that make it a cut above the rest: the strongest HLA association and a single well defined trigger—gluten.6 CD itself is a “spurious” autoimmune disease as it induces a reaction against self (antibodies against tissue transglutaminase) but this self aggression resolves on gluten withdrawal and hence strictly speaking does not qualify as an autoimmune disease.

Because we can control the encounter with the environmental factor (gluten) in CD, this is an ideal condition to study the relevance of the environment in induction of autoimmunity. Several years ago Ventura and colleagues postulated that the longer a coeliac patient encounters gluten, the greater is the chance that they will develop an autoimmune disease.7 In that report duration of gluten exposure was measured in children from birth to the time of diagnosis when gluten is normally removed from the diet. This line of reasoning is not too difficult to comprehend as it straightforwardly states that the longer is the exposure to the “toxic” agent the greater the potential cumulative effect. The implications however are enormous as we would have defined a single environmental factor and more importantly we could visualise how this promotes autoimmunity. For instance, we could envisage that in CD a rescheduling of the mucosal microenvironment is induced with a switch from the customarily “anti-inflammatory” Th2 type milieu to a Th1 type, as observed in CD.8 In the long run, this could tip the balance towards an autoimmune response although other conceivable explanations, such as increased mucosal permeability, are possible. In this issue ofGut, this important topic, which trespasses the boundaries of CD, has been revised (see page 502).9Sategna Guidetti et al studied a group of adult coeliacs, rather than children, following a similar retrospective analysis implemented by Ventura and colleagues.7 The conclusions at first glance seem to be similar in the two reports: the later CD is diagnosed the greater the chance of a concomitant autoimmune disease. Sategna Guidetti et alhowever introduced a small “artifice” to reverse this apparently uncomplicated conclusion. Indeed, they argue that in a large group of coeliacs, diagnosis of the autoimmune disease preceded that of CD, and consequentially we should introduce this variable, among others, in analysis of the data. Surely if a patient becomes diabetic for example, 10 years before a diagnosis of CD, he/she did not require that extra time (10 years) of gluten challenge to develop diabetes. The authors thus reanalysed their data and surprisingly concluded that “actual gluten exposure”, a concept that at a superficial reading of the paper is not easy to grasp, was not different between coeliacs with or without other autoimmune diseases. Thus they challenge the main conclusion of Ventura's paper. In evaluating both of these papers we need to consider several points which may help reconcile the different conclusions. For example, the two studies did not compare the same populations of coeliac patients. One study was conducted in children and the other in adults, and to date there is no consensus that child and adult CD are the same condition. That age might have an unforeseen effect in the progression of an immune mediated disease is again proved by the study of monozygotic twins discordant for type 1 diabetes. In that study it was demonstrated that the concordance rate dramatically decreased if age at diagnosis of the diabetic proband twin was made after the age of 24.3 We also do not know whether, even in Ventura's study, the introduction of the concept “actual gluten exposure”, at least for the groups of children diagnosed later in life, would have changed interpretation of the data. Clearly, the only experiment that can answer this question is not a retrospective but rather a prospective study, maintaining one portion of the young early diagnosed children on a gluten containing diet, but this is obviously impossible. We also do not know whether patients that have been diagnosed with CD at 36 years of age have had CD all their life, as there is evidence suggesting that other concomitant factors may start the clinical manifestations of CD.

As is often the case, this paper raises more questions than answers. One that is immediately obvious is why, regardless of the duration of “actual gluten exposure”, coeliac patients with other autoimmune diseases are diagnosed later in life, implying that they may have a clinically less evident disease, making their diagnosis more difficult. Another question that has to be answered is whether or not the groups (with or without autoimmune diseases) are homogenous and what other parameters differentiate them apart from age at diagnosis. Indeed, we have to clarify why these late diagnosed patients are more prone to develop autoimmunity and how they differ compared with early diagnosed patients if we want to “make sense” of this finding. Therefore, functional, epidemiological, as well as genetic studies are required to unravel this puzzling question to shed light on the nature of the link between CD and autoimmune diseases and possibly to autoimmunity itself.