In recent decades, CD has been proposed to play a possible role in the etiopathogenesis of ASD. This hypothesis emerged after some authors observed a high occurrence of gastrointestinal symptoms in presence of an increased prevalence of overt CD, and/or of autoantibodies against gluten, and/or an altered intestinal barrier/ gut-microbiota in the population with autism [22, 10, 15]. These findings fitted well with the XIX – XXI-century emerging theory of the “brain-gut-microbiota axis” [33] whereby dysbiosis may be responsible for an atypical pro-inflammatory mechanism causing an abnormal function of the blood-brain barrier and of the neural immune cells in the CNS, resulting in an anomalous CNS development and neurological and/or psychiatric disorders [18, 19]. Furthermore, some recent evidence shows that CD may lead to neuropsychiatric manifestation such as seizures, ataxia, irritability, hyperactivity, etc. [20, 21].
These findings were however not confirmed in several other studies [25 – 27]. The aforementioned debated theories, on the one hand, encouraged clinicians to largely screen ASD patients for CD autoimmunity and, on the other, promoted several research works with very controversial results [33, 34, 25, 26].
The aim of our study was to detect the CD seroprevalence and prevalence in young subjects with autism and verify the existence, or otherwise, of a relation between CD and autism spectrum disorder.
Regarding the past and current literature about CD in this set of patients, it is interesting to note that our cohort represents a large sample. We therefore carried out one of the very few mass ASD population study, in which data were collected prospectively and then compared with those of a mass screening in the Italian sample of healthy children (by Gatti et al.) [15].
Globally, we found a slightly lower percentage of ASD patients with gastrointestinal symptoms (19.90%) compared to Adams et al. (22.7%-70%) [22]. We retain that the lower rate is due to the very young age of our patients and therefore to their difficulty in expressing their possible gastrointestinal symptoms. Anyway, considering the subset of ASD subjects with overt-CD, symptomatic patients were more common than asymptomatic cases (60.0% vs 40.0%). Interestingly, atypical symptoms, and in particular constipation, prevailed both in the serologically positive/overt CD subsets and in the ASD controls, suggesting a possible etiology other than celiac disease and/or CD autoimmunity [22].
Like most published papers [36], our results validate the lack of a real association between ASD and CD, both with regard to CD autoimmunity, and in terms of fully-blown CD. Specifically, we found a CD seroprevalence of 4.08% against the 2.22% in the Italian cohort of healthy children by Gatti's et al., and a CD prevalence of 2.24% versus their 1.58% rate (P> 0.05 in both cases).
Looking at our data and those of other colleagues, we noticed some very important issues: first, ASD is more frequent in males than in females, but most research studies (like ours) compared the prevalence in ASD patients with the prevalence obtained from mass screenings for CD in healthy children, which were composed by similar rates of males and females; however, a female predominance in CD is widely recognized. This generates a selection bias. Secondly, the vast majority of studies in the literature are subject to other biases related to the low number and the mean young age of the patients enrolled. The large Confidence Intervals at 95% we obtained are primarily related to the relatively small number of subjects in our sample. Nevertheless, our ASD cohort is among the largest ever enrolled, and this provides the idea of the burden exerted by a singular case of overt CD on the calculation of the total CD prevalence, that we can summarize in a surveillance bias. The last relevant issue regards the mixed results achieved by different authors, that are partially related to changes in the diagnostic approach and guidelines for CD in the years, but especially to the different definitions of “serological CD”, “overt CD” and “doubtful cases” used by researchers. CD seroprevalence and/or CD prevalence rates are thus determined according to various methodologies depending on specific studies and can thus generate information biases.
In recent years, some studies supported the evidence of an increased immune reactivity and sensitivity against gluten in ASD subjects, even in the absence of full-blown CD. In their paper, de Magistris et al. found an increased prevalence of AGA-IgG and DPG-IgG in patients with autism spectrum disorder as compared to the healthy population [37]. A few years later, Józefczuk et al. demonstrated a lack of deterministic correlation between the presence of gluten-induced autoantibodies and an increased intestinal permeability in ASD patients, in opposition to the hypothesis of so-called “leaky gut” [35]. The authors maintained that, while intestinal Tissue TransGlutaminase 2 (TG2) is a specific autoantigen of CD, the Tissue TransGlutaminase 6 (TG6) is distributed in the central nervous system and seems to be related to neuropsychiatric disorders. Józefczuk et al. detected exclusively anti-TG6 IgA and IgG, and AGA-IgG respectively in 4 and 21 out of the 77 ASD patients evaluated. Neither anti-TG2, nor Intestinal Fatty Acid Binding Proteins (I-FABP) or zonulin, which are markers of an impaired gut barrier, were found in their sera [35]. These findings revealed an inconsistency in the hypothesis of protein translocation through the impaired intestinal epithelium to the systemic circulation and the blood-brain barrier, which would reach the brain and affect children’s behavior [35]. Not to mention that, both for TGA and anti-Deamidated Gliadin Peptide (DGP), a biochemical test seropositivity may be a transient phenomenon in young children that does not necessarily predict the development of CD, as it may be the case, for example, for subjects with HLA gene susceptibility who have ongoing infectious events [38].