Autism spectrum disorder (ASD) is a complex developmental disorder encompassing a heterogeneous patient population. It is generally agreed that there are at least two types of ASD with regard to disease development; abnormal cognitive development evident from birth (classical autism) and developmental regression, usually between 18–24 months of age, following apparent normal development (regressive autism) [1].

In addition to behavioral symptoms, co-morbid clinical conditions such as gastrointestinal (GI) symptoms are frequently noted in ASD children. A high prevalence of GI symptoms, which often improve after dietary intervention, has been reported by parents. This has led to speculation that there may be a high prevalence of food allergy (FA) in ASD children. Although IgE-mediated FA does not appear to be prevalent in ASD children, our previous study indicated a higher prevalence of non-IgE mediated FA (NFA) in ASD children. Specifically, our results revealed increased tumour necrosis factor-α (TNF-α) production against cow's milk protein along with correlating clinical features consistent with NFA [2]. In these ASD children with NFA, we also observed excessive production of TNF-α in response to LPS, an agonist of Toll-like receptor 4 (TLR4) [3].

Apart from FA, we have also encountered a number of ASD children who suffer from recurrent infection (typically viral syndromes) accompanied by exacerbations of behavioral symptoms (hyperactivity, temper tantrums, irritability and self-stimulatory behaviors) despite good responses to dietary intervention. Such behavioral changes were pointed out by teachers/therapists/care takers independent of parents. Immune insult via microbial infection caused by various pathogens appears to counter-act beneficial effects of behavioral, dietary, and other intervention measures in these ASD children. However, atopy or primary immunodeficiency (PID) does not appear to be a major factor in these children. The above-described clinical observations led us to hypothesize that in these ASD children, antigen non-specific (innate) immune responses are altered, leading to dysregulated neuro-immune interactions apart from PID or atopy.

Among 133 ASD children evaluated in our clinic, we identified 26 ASD children with the above-described clinical features (ASD test group). Interestingly, they are all documented to have shown regression at the onset of symptoms. In this study, we assessed prevalence of atopy, FA, CRS/ROM, and PID in the ASD test and control groups as well as in the non-ASD control groups (FA, CRS/ROM, and normal controls). In addition, we assessed responses to a panel of TLR agonists in 19 ASD test group children and compared these to responses for ASD and normal case controls.

Subtle immune abnormalities in ASD children have been reported by many researchers [13–15]. However, results are rather variable partly due to the small number of study subjects, heterogeneous patient populations, and lack of careful immunological characterization of the study subjects in these studies. Moreover, reported results are conflicting; some studies indicate type 1 T-helper (Th1)-skewed responses while other studies indicate Th2-skewed responses [13–16]. A high prevalence of GI symptoms in autistic children also raised possibilities of GI autoimmune conditions and FA. However, to date, the role of the immune system in the onset and progression of ASD remains unclear.

The effects of environmental factors in genetically vulnerable individuals have also been implicated in the development of autism. There is ample evidence indicating that there is increased oxidative stress in autistic children [17]. The 'redux/methylation hypothesis' postulates that environmental factors and genetic factors play pathogenic roles in some autistic children [17]. It has been postulated that exposure to xenobiotics in subjects with a genetic predisposition for impaired methylation and/or increased susceptibility to oxidative stress may result in the neurological deficits observed in certain ASD children [18, 19]. Oxidative stress also activates innate immunity. Interestingly, the presence of low-grade chronic inflammation has been reported in brain tissue of individuals with autism [20, 21]. This 'redux/methylation hypothesis' may also be associated with skewed Th1 or Th2 responses in certain conditions [22]. However, how such genetic/environmental factors modulate the immune system in autistic children is not well understood.

For a number of ASD children, parents report frequent infections [recurrent pharyngitis and viral syndromes, ROM, and CRS] and an unusually prolonged course of such illnesses. Following microbial infection, parents repeatedly describe exacerbations of behavioral symptoms and even loss of previously acquired cognitive skills, an occurrence also documented by care-takers/teachers/therapists. This study focused on this subset of ASD children (ASD test group) as defined in the Methods section. Our first hypothesis is that clinical features of the ASD test group are not associated with atopy, asthma, FA, or PID.

To test this hypothesis, we retrospectively reviewed 133 ASD children referred to the Pediatric Allergy/Immunology Clinic for evaluation of atopy/immune abnormalities. Given the nature of our clinic, the ASD children reviewed may have had more medical issues than ASD children in general, as evidenced by their need to visit the clinic. Nevertheless, it is advantageous to analyze these children, since extensive data had already been generated by thorough allergy and immune workups. We also have the advantage of having control non-ASD children with chronic airway or GI inflammation who underwent a similar workup (control CRS/ROM and FA groups) that enabled us to address the presence of any distinguishing clinical features in the ASD test group.

Prevalence of atopic disorders and asthma was equivalent to that reported in general population in both test and control ASD children. A lower prevalence of atopy in the control FA patients may reflect their young age, since atopic disorders develop with age. Likewise, a younger median age of the control ASD children indicates that atopic diseases may develop with age. Nevertheless, given the frequency of atopic disorders in the ASD test group, who are slightly older in median age than the ASD controls, it is very unlikely that atopy is associated with the clinical features of the ASD test group.

NFA was highly prevalent in both ASD groups, consistent with the high prevalence of GI symptoms in ASD children (Table 2). As summarized in Table 2, a major difference in the clinical features of the test vs. control ASD children was the higher frequency of CRS/ROM and SPAD in the ASD test group children. However, diagnoses of CRS/ROM and SPAD were only seen in 6/26 and 5/26 of the ASD test group children, respectively. All of the ASD test group children with SPAD suffered from CRS/ROM (Table 2). SPAD was most prevalent in CRS/ROM children (Table 2). These findings taken together suggest that the distinguishing clinical features of the ASD test group are unlikely to be associated with FA, asthma, or SPAD in most of the ASD test group children.

In CRS patients, it was postulated that transmission of inflammatory mediators via post-nasal dripping may be associated with chronic GI irritation and predispose them to FA [23]. It was also speculated that frequent antibiosis and resultant dysbiosis can disrupt gut mucosal immune homeostasis, resulting in sensitization to food proteins. However, our results revealed a higher prevalence of NFA in ASD children than in CRS/ROM children, who were likely to have had more frequent antibiosis than ASD children (Tables 2). Asthma was prevalent in CRS/ROM children, which is not surprising since CRS is one of the major triggers of asthma [12]. Our findings indicate CRS/ROM children do not display clinical features similar to those observed in ASD test or ASD control children, making it unlikely that CRS/ROM account for the high prevalence of NFA in ASD children in our study.

It is of note that we did not find a high prevalence of known PID in control ASD children, although the test group ASD children revealed a few cases of SPAD (Table 2). Therefore, extensive immune workup focusing on PID is likely to yield negative results in most ASD children, especially those without concerning clinical features, as reported previously [24].

In summary, our results support our initial hypothesis that the distinct clinical features of our ASD test group are not associated with atopy, asthma, FA, or presence of known PID. However, our results have limitations. The number of the subjects and a potentially skewed population due to the fact that the ASD study subjects were those evaluated in the Pediatric Allergy/Immunology Clinic have some bearing on our results. A larger population study will be informative to further elaborate our initial findings.

In this study, we identified 19 ASD test group children without SPAD among 133 ASD children (14.3%). Pathogens affecting these ASD test group children appeared various, indicating the possibility of impaired antigen-independent immune responses – innate immunity. Innate immunity mounts the first line, antigen-independent immune defense by recognizing microbial by-products or those from damaged tissue cells via pattern recognition receptors including TLRs [25–27]. TLR-mediated responses lead to the production of various soluble mediators that can signal the brain [28–31]. Such signalling events help the central nervous system restore autonomic homeostasis and provide inhibitory regulatory signals to prevent excessive immune responses [28]. Subtle changes in innate immune responses can affect the intricate neuro-immune network that is mediated by innate immunity. Since NFA prevalence was similar in both the ASD groups, it was the frequent occurrence of viral syndromes in the ASD test group children that led us to our 2^nd hypothesis. Namely, we theorized that TLR responses, especially those sensing viral by-products, are altered in the ASD test group.

Among TLRs, TLR2 is important for sensing encapsulated bacteria and intracellular pathogens such as mycobacterium [32, 33]. TLR4 is important for sensing gram negative bacteria, common found in the GI tract [32]. TLR7/8 are important for sensing single stranded RNA derived from RNA viruses, which are common causes of childhood viral infections including measles, rhinovirus (the most common cause of 'cold'), and influenza [34]. This study assessed responses to TLR2/6, 4, and 7/8 agonists by measuring production of a panel of cytokines in the test group ASD children without SPAD and comparing those to children with ASD and to normal case controls. It is important to mention that since many FA patients outgrow this condition with age, the FA control children are younger than the ASD children evaluated in the study. In addition, the number of non-ASD controls with FA or CRS/ROM is low due to the fact that the study was limited to patients evaluated/treated in our clinic. Thus the study is lacking non-ASD case controls with FA or CRS/ROM for evaluation of innate immune responses.

Several abnormalities were found in the ASD test group: lower production of IL-6 with the TLR2/6 agonist, lower production of IL-1β with the TLR7/8 agonist, and higher production of IL-23 with the TLR4 agonists than case controls in the absence of LPS pre-treatment. These cytokines which were found to be altered in production in the ASD test group are the key regulators in the neuro-immune network [28, 35, 36]. Thus our findings indicate that children in the ASD test group may be less capable of controlling microbial infection in the initial stages, leading to ineffective signalling to the brain. It is also intriguing to find increased production of IL-23 with the TLR4 agonist in the ASD test, since IL-23 is associated with development and maintenance of Th17 cells, a recently defined T-helper cell subset [37, 38]. It is of note that Th17 cells are implicated in various autoimmune and chronic inflammatory diseases including multiple sclerosis and inflammatory bowel diseases [37–39]. Thus the ASD test group children may be more prone to Th17-mediated inflammatory responses.

When immune cells were pre-treated with LPS, a TLR4 agonist, subsequent TLR responses were suppressed; this phenomenon is called LPS tolerance [40]. This is thought to be important for immune homeostasis in the gut, and especially in the colon where bacteria exist at high density and hence innate immune cells in the mucosa are likely to have frequent exposures to endotoxins (LPS) [40]. We found lower IL-1β production by PBMCs with TLR4 or TLR7/8 agonists following LPS pre-treatment in the ASD test group. IL-1β is one of the major mediators of the neuro-immune network for maintenance of homeostasis [35, 36, 41]. Thus suboptimal production of IL-1β may impair neuro-immune signaling, while excessive IL-1β can be toxic to the brain [35, 41, 42]. Our finding may indicate a risk of suboptimal neuro-immune signalling in the ASD test group.

sTNFRII is an important counter-regulatory cytokine for TNF-α as made evident by the marked therapeutic effects of exogenous sTNFRII in the treatment of certain autoimmune diseases [43]. We also found lower sTNFRII production by PBMCs with TLR 7/8 agonist following LPS pre-treatment in the ASD test group, indicating a possibility of blunted counter-regulatory measures in the ASD test group at the time of viral infection. However, sTNFRII production in this condition also appeared to be lower in the ASD control group except for 2 high outliers. Thus this finding may not be specific for the ASD test group children.

Taken together, our findings of altered TLR7/8 responses with or without endotoxin (LPS) pre-treatment may indicate a predisposition to recurrent viral infections in the ASD test group as compared to ASD and non-ASD case controls. The above-described, altered TLR responses may also affect neuro-immune interactions in the test group. Our findings may be useful further defining this subset of ASD children in larger scale studies.

Our results indicate the presence of a distinct subset of ASD children. This subset, the ASD-test group, is characterized by frequent infections and by recurrent loss of previously acquired cognitive skills with worsening behavioral symptoms following infection. Clinical features of this subset were not associated with atopy, asthma, FA, or PID in this study but may be associated with altered TLR responses important for neuro-immune interactions.