The aim of this study was to carry out a comprehensive analysis of markers of celiac disease and gluten sensitivity in a group of children with autism who had been diagnosed according to strict criteria and defined instruments. Our data indicate that children with autism have higher levels of IgG antibody to gliadin compared to healthy controls. In addition, among patients with autism, the antibody response to gliadin was greater in those with GI symptoms. However, in contrast to patients with celiac disease, no association was observed between the elevated anti-gliadin antibody level and the presence of highly specific serologic markers of celiac disease or HLA-DQ2/DQ8. The findings indicate that the observed anti-gliadin immune response in patients with autism is likely to involve a mechanism that is distinct from celiac disease, without the requirement for TG2 activity or antigen presentation through DQ2/DQ8 MHC molecules 
The data from this study should be interpreted with caution. Most importantly, the observed increased IgG antibody response to gliadin does not necessarily indicate sensitivity to gluten or any pathogenic role for antibodies to gliadin in the context of autism. In addition, the results do not rule out the possibility of moderately increased prevalence of celiac disease among children with autism, especially as duodenal biopsy, the gold standard for definitive diagnosis of celiac disease, was not performed. However, considering the excellent sensitivity and specificity of anti-TG2 and (and to a lesser extent anti-deamidated gliadin) antibodies, as well as the high negative predictive value of HLA-DQ2/DQ8 markers for celiac disease, it can be concluded with high certainty that the overwhelming majority of autism patients with elevated antibody to gliadin do not have celiac disease. If future studies prove the existence of sensitivity to gluten in a subset of patients with autism, the gluten-associated symptoms in such individuals may fall within the spectrum of “non-celiac gluten sensitivity” 
Compared to previous reports examining the link between celiac disease/gluten sensitivity and autism, this study is unique in several ways. First, a shortcoming in earlier studies has been the lack or incompleteness of suitable age-matched healthy control groups necessary for this type of analysis. In this work, the antibody levels in children with autism were compared to two separate pediatric control groups: unaffected siblings of the same patients, as well as a larger cohort of unrelated healthy children. Second, previous reports have used specimens from more heterogeneous groups of patients generally recruited at local hospitals or clinics, and while most report the use of DSM diagnostic criteria, it is unclear which test(s) informed the final diagnosis of autism. In contrast, the samples in this study were acquired from a well-recognized repository of biomaterials (AGRE), which is managed by the world’s largest autism advocacy organization and has been utilized in various past research projects. The associated AGRE database includes information about family pedigree, scores from various tests and questionnaires, and medical histories for many of the patients for which biospecimens are available. Patients in this study were selected only if they were identified as having autism according to two separate instruments, ADOS and ADI-R, thus greatly increasing the likelihood of accurate diagnosis.
A limitation of this study is that we could not control for geographical distribution, socioeconomic status, or diet of the research participants. These factors may contribute to levels of antibodies against dietary and other antigens in patients and controls. In addition, information on GI symptoms was available only for some patients and none of the controls. Access to such data would have strengthened the study’s finding regarding the association between GI symptoms and anti-gliadin antibody levels. As such, the conclusions of this study should be considered preliminary, requiring further confirmation in larger and better-characterized cohorts of patients and controls.
We can consider some possibilities to explain the higher anti-gliadin antibody levels found in the cohort of children with autism. Previously, associations between autism and increased GI symptoms, as well as impaired intestinal permeability, have been reported 
. Increased intestinal permeability resulting from damage to the intestinal epithelial barrier in those with autism may be responsible for increased exposure of the immune system to partially digested gluten fragments, resulting in the detected increase in antibody response. The observation here that anti-gliadin antibody reactivity is elevated in patients with GI symptoms lends some support for this idea. At the same time, the fact that the higher anti-gliadin antibodies in autistic children were limited to the IgG isotype, without a concomitant rise in IgA, may imply a non-mucosal and/or gluten-independent origin for the observed antibody reactivity. One possibility is that the IgG-specific antibody response in children with autism would have been triggered by ingested gluten at some point in the past, but no longer dependent on continuous mucosal exposure to the proteins. Alternatively, the detected anti-gliadin antibodies may be unrelated to gluten as the immunogen. Various immune abnormalities have been demonstrated in autistic children, including increased antibody reactivity to autoantigens 
. It is conceivable that certain autism-associated autoantibodies, the exact targets of which are yet to be identified, would cross-react with one or more gluten proteins and contribute to the detected difference in anti-gliadin antibody level between patients and controls. Circulation levels of such antigen-independent or gluten cross-reactive antibodies would not be expected to respond to dietary gluten restriction.
Results of this study are intriguing in the context of disease pathophysiology and biomarker identification. The observed increase in antibody reactivity to gliadin in over one fifth of the autism cohort points to potential shared genetic and/or environmental associations in a sizable subset of patients. As such, the generated data provide an impetus to further examine the affected patient subset for additional immunologic and genomic clues. It is possible that, in a subset of children with autism, the condition is associated with antibody reactivity to a unique set of gluten proteins that would be significantly different from the pattern of anti-gliadin antibody response in celiac disease and other conditions. This specific pattern of antibody reactivity may be useful as a source of biomarkers. A unique antibody response to particular gluten molecules could also be associated with specific HLA genes in that disease subset.
In conclusion, the increased anti-gliadin antibody response in autism and its association with GI symptoms points to a potential mechanism involving immunologic and/or intestinal permeability abnormalities in a subset of patients. The observed antibody reactivity to gliadin in most children with autism appears to be unrelated to celiac disease. Therefore, the heightened immune response to gluten in autism deserves further attention and research in determining its utility as a source of biomarkers and clues regarding disease pathophysiology. Better understanding of this immune response may offer novel markers for the identification of subsets of patients who would be responsive to specific treatment strategies.