The Role of Gut Microbiota in Gastrointestinal Symptoms of Children with ASD

Background and objectives: Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by impaired communication, social interaction disorder, and repetitive behavior. Dysbiotic gut microbiota (GM) could be a contributing factor to the appearance of ASD, as gastrointestinal (GI) symptoms are comorbidities frequently reported in ASD. As there is a lack of reviews about the role played by GM in the GI symptoms of ASD, this work aimed to carry out a systematic review of current studies comparing the GM of children with ASD and GI symptoms with those of healthy controls in the last six years. Materials and Methods: The systematic review was performed following the PRISMA guidelines. The databases chosen were Web of Science, Scopus, PubMed, and PsycINFO, and the keywords were (gut* OR intestine* OR bowel* OR gastrointestinal*) AND (microbiota* OR microflora* OR bacteria* OR microbiome* OR flora* OR bacterial* OR bacteria* OR microorganism* OR feces* OR stool*) AND (autistic* OR autism* OR ASD*). Results: A total of 16 articles were included. Ten articles performed correlations analysis between GI symptoms and ASD. Among those 10 articles, 7 found differences between the GI symptoms present in children with ASD and healthy controls. The most common GI symptom was constipation. Among the seven articles that found differences, three performed correlations analysis between GI symptoms and gut microbe abundance. Candida, Prevotella, Streptococcus, and Veillonella showed higher and lower abundance, respectively, in children with ASD and GI symptoms in more than one article. Bacteroidetes, Firmicutes, Actinomyces, Dorea, Lactobacillus, Faecalibacterium prausnitzii, and Bacteroidetes/Firmicutes ratios showed abundance discrepancies. Conclusions: It is still too early to draw a conclusion about the gut microbes involved in GI symptoms of ASD. Future research should consider the relationship between ASD behavior, GM, and GI symptoms in a multidisciplinary way and homogenize sample characteristics.


Introduction
Autism or autism spectrum disorder (ASD) is defined in the Diagnostic and Statistical Manual for Mental Disorders (Fifth Edition (DSM-5)) as a heterogeneous group of neurodevelopmental disorders featuring the three main characteristics of impaired communication, social interaction disorder, and repetitive behavior [1]. Autism includes Kanner's autism [2], Asperger's syndrome [3], and pervasive developmental disorder not otherwise specified (PDD-NOS), and is characterized by persistent deficits in social communication interaction and restricted-repetitive patterns of behavior, interests, or activities [1]. It is worth mentioning that people without neurological, medical, or psychiatric diagnoses are defined as neurotypical in the autism community [4].
GI symptoms and diet problems in ASD children, especially idiosyncratic feeding behavior, has been also reported, and ASD children suffering from multiple GI symptoms tend to be those who also have dietary problems [23].
Traditionally, characterization of GM has been carried out using culture-dependent techniques. One of the great limitations of these techniques is that it is impossible to ensure that the phenotypic behavior of a microorganism in the laboratory is identical to its behavior in vivo, due in part to the processes of metabolic symbiosis between microorganisms. In addition, the number of microorganisms that can be characterized with this technique is limited [24]. However, the emergence and rapid development of culture-independent or molecular techniques has made culture-dependent techniques nearly obsolete, and the characterization of GM is currently largely carried out using culture-independent techniques, as they allow scientists to easily identify a large proportion of bacterial diversity and provide rapid results [25,26]. This has led to many studies that compare the diversity of the GM of ASD children with healthy controls [27][28][29][30][31], of which a bibliographic review can be carried out. In this sense, several bibliographic reviews about GM and ASD can be found in the scientific literature. For example, Hughes et al. [32] and Liu et al. [33] studied the dysbiosis of GM in ASD, and Andreo-Martínez et al. [34] studied the gut microbes belonging to the Fungi kingdom in people with ASD. In addition, there have been few studies that have analyzed the correlation between GM composition and GI symptoms in people with ASD. For example, in a meta-analysis about GI symptoms in ASD [9], performed on studies published until 2012, there was only one study [35] which reported that GI symptoms in people with ASD were correlated with high levels of Clostridium (p = 0.001). However, to the author's best knowledge, there have been no bibliographic reviews that have attempted to relate the gut microbes involved in the GI symptoms of people with ASD. Therefore, the aim of the present work was to carry out a systematic review on the studies that have compared the GM of people with ASD and GI symptoms with those of healthy controls, continuing the meta-analysis line of McElhanon, McCracken, Karpen, and Sharp [9], during the last six years, to try to find some bacterial abundance patterns and to identify future research strategies.

Materials and Methods
The design of the present systematic review was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [36]. This systematic approach under the PRISMA perspective covers the developments in the field of the GM of people with ASD and GI symptoms by providing critical analysis of key parameters, such as the role that gut microbes play in the GI symptoms of people with ASD, and gut microbe abundance of people with ASD with GI symptoms.
The bibliographic search was performed for works published between 2012 and February 2019, and the four comprehensive databases chosen to carry out the bibliographic search were: Web of Science, Scopus, PubMed, and PsycINFO. The Boolean strings entered to carry out the search were: (gut* OR intestine* OR bowel* OR gastrointestinal*) AND (microbiota* OR microflora* OR bacteria* OR microbiome* OR flora* OR bacterial* OR bacteria* OR microorganism* OR feces* OR stool*) AND (autistic* OR autism* OR ASD*), and the searches included works published in all language. Scopus search options were: title, abstract, and keywords; Web of Science search option was "theme," while for PsycINFO and PubMed, the search option selected was all fields.
The inclusion criteria were: (1) articles with human populations; (2) articles published from 2012 to February 2019; and (3) articles comparing the GM of children with ASD and GI symptoms with those of control groups. The exclusion criteria were: (1) descriptive and systematic reviews; (2) books and book chapters; (3) editorial material, letters to the editor, shorts reports, and proceedings of conferences; (4) animal model and in vitro articles; (5) other diseases; (6) articles studying only metabolites in blood, plasma, or urine; and (7) articles published in a language other than English.
The 1540 works obtained by the four databases were passed through EndNote X7 software (Thomson Reuters, New York, USA) to delete possible duplicated works. According to the PRISMA methodology, A.E.M.-G. and P.A.-M. formed the review team and independently screened titles, abstracts, and full texts of the works for potential inclusion. These two authors evaluated them according to the inclusion or exclusion criteria, and disagreements on whether a given reference should be included or not were resolved through discussion. In addition, the reference lists of the selected works were also examined in search of potentially relevant papers, not finding any additional article in this step. It is noteworthy that some of the works could be included in more than one elimination group, but the final criteria were agreed by the review team. Finally, a total of 16 articles were found to be eligible for the present systematic review and they are shown in Table 1. The risk of bias for each included study was evaluated using the Methods Guide for Comparative Effectiveness Reviews [37]. The forms of bias considered are presented in Table 2. Figure 1 shows the process of identifying articles for inclusion in the present systematic review. formed the review team and independently screened titles, abstracts, and full texts of the works for potential inclusion. These two authors evaluated them according to the inclusion or exclusion criteria, and disagreements on whether a given reference should be included or not were resolved through discussion. In addition, the reference lists of the selected works were also examined in search of potentially relevant papers, not finding any additional article in this step. It is noteworthy that some of the works could be included in more than one elimination group, but the final criteria were agreed by the review team. Finally, a total of 16 articles were found to be eligible for the present systematic review and they are shown in Table1.The risk of bias for each included study was evaluated using the Methods Guide for Comparative Effectiveness Reviews [37]. The forms of bias considered are presented in Table 2.  Table 1 shows the sixteen studies included in the present systematic review and the results of the analysis of the samples in children with ASD and GI symptoms.  ASD (n = 23) divided into ASD w GI symptoms (12/23) and ASD w/o GI symptoms (11/23) Range Age 3-10 y (15/23 aged 3-5 y; 6/23 aged 6-7 y; 2/23 aged 8-10 y)
Desulfovibrio with ASD severity in the ADI restricted/repetitive behavior subscale score. There is a correlation of the autism severity with the severity of GI dysfunction. (p = 0.01) No correlation of GI symptoms and Desulfovibrio abundance.

Assessment of Risk of Bias for Each Included Article
The 16 articles were identified and assessed as medium to high quality (Table 2). When assessing the quality of selection, the studies presented some limitations in terms of sample size, information on comorbidity of neurodevelopmental disorders, diet of study participants, blinding of participants and personnel, and mean scores and standard deviations in the analysis of study data. Many of these studies were also open-label trials and relied on qualitative, self-reported questionnaires and surveys to gauge treatment response, inviting potential bias into the studies. There might also be various difficulties experienced by the parents in evaluating these aspects, especially owing to the communication deficits typical of children with ASD. More randomized, controlled studies with a larger study population and the use of clinician scoring may lead to more robust studies and results [17].

Relationship between GI Symptoms and ASD
Among the 16 articles found in the present systematic review, ten of them performed analysis to try to find correlations between GI symptoms and ASD. Seven articles found statistically significant differences between the occurrence of GI symptoms in children with ASD and healthy controls [28,30,31,[42][43][44]48], and three articles found no such differences [27,39,41]. The remaining six articles did not perform statistical analysis to analyze correlations between GI symptoms and ASD [29,38,40,[45][46][47].
The most common GI symptom in children with ASD was found to be constipation, results in line with those reported elsewhere [10]. In this sense, five articles reported statistically significant differences in children with ASD suffering constipation compared to healthy controls [30,[42][43][44]48].

Relationship between Gut Microbe Abundance and GI Symptoms in Children with ASD
Three articles tried to find correlations between the GI symptoms of children with ASD and gut microbe abundance [31,42,43]. Two of those articles found correlations between certain gut microbes and GI symptoms, specifically constipation [42,43], and irritable bowel syndrome, functional constipation, aerophagia, and abdominal migraine [43]. In light of those results, some genera of bacteria are involved in constipation, but the results between the studies are not coincident ( Table 1). The remain article reported no correlation between Desulfovibrio and GI symptoms [31].

Gut Microbiota Dysbiosis found in children with ASD and GI Symptoms
Seven articles reported statistically significant differences in the presence of GI symptoms in children with ASD compared to healthy controls [28,30,31,[42][43][44]48]. These articles also showed significant differences in the abundance of some gut microbes in children with ASD suffering GI symptoms (Table 1) compared to their respective control groups. The gut microbes that were found by more than one article to show significant differences in abundance are listed below.
Candida was found to be the only gut microbe that showed significantly higher abundance in children with ASD and GI symptoms compared to their control groups by more than one article [30,42].
Lower abundance of Prevotella [27,44], Veillonella [42,48], and Streptococcus [44,48] in children with ASD and GI symptoms compared to their control groups was also found to be statistically significant.

Limitations
An important limitation of the studies is the absence of a psychometric analysis of the relationship between the severity of behavioral ASD symptoms with GM abundance and GI symptoms. Similarly, none of the studies indicated whether children with ASD presented a diagnosis of intellectual disability, although cognitive difficulties are a determining factor in the severity of ASD symptoms [49].
On the other hand, the present systematic review has the same limitations as other systematic reviews, such as the databases used, the established inclusion/exclusion criteria, database selection, keywords, or time-frame chosen. In addition, the lack of sufficient articles made it impossible to combine the results into a meta-analysis [9].

Discussion
The present systematic review found 16 articles that studied the GM in children with ASD and GI symptoms, and, as discussed before, among the 10 articles that analyzed the correlation between GI symptoms and ASD, 7 of them found significant differences. These results are similar to those reported by Adams et al. [50], who found a strong correlation between the 6-GSI score and the Autism Treatment Evaluation Checklist score (Pearson's correlation coefficient r = 0.60; p < 0.001). Those results are also in line with a meta-analysis which included 15 articles comparing heterogeneous groups (siblings, ASD and healthy control) [9]. Among the 15 selected articles, 10, 12, 9, and 8 reported general GI concerns (p < 0.0001), diarrhea (p < 0.0001), constipation (p < 0.0001), and abdominal pain (p = 0.016), respectively. In conclusion, there seems to be evidence that there is a greater prevalence of GI symptoms among children with ASD compared with control children.
On the other hand, three studies [31,42,43] analyzed the correlation between gut microbe abundance and GI symptoms in children with ASD and GI symptoms. Although the studies are very promising, there is not yet enough scientific evidence to determine the relationships between gut microbe abundance and GI symptoms in children with ASD, as the gut microbes found to be correlated with GI symptoms in children with ASD were different between the different studies (Table 1). In this sense, the results are very premature, and the number of articles found was very low. Therefore, as discussed before, it was not possible to perform a meta-analysis of the relationships between gut microbes and GI symptoms in children with ASD [9].
Regarding the abundance of some gut microbes in children with ASD and GI symptoms, Candida was found to be more abundant by two studies [30,42], Candida albicans being the most frequently identified species [30]. In fact, it has been reported that a high abundance of Candida albicans can cause lower absorption of carbohydrates and minerals, and can generate higher toxins levels in the GI tract that can contribute to some ASD behaviors [51]. It has been also reported that the immunological responses to Candida in the GI tract can be modulated by some Lactobacillus species [42,52]. In this sense, it is possible that a dysbiotic GM in ASD may lead to a Candida population expansion, as can be observed by the lower abundance of Lactobacillus found in Italian children with ASD and GI symptoms compared to their control group [30]. However, a higher abundance of both Lactobacillus and Candida was also found in Italian children with ASD and GI symptoms compared to their control group [42]. Therefore, these results are not conclusive, and, as can be observed, there is a clear lack of investigation of the whole Fungi kingdom in children with ASD, and little is still known about the role of Candida and other types of fungi in both GI and ASD symptoms [34].
Lower abundances of three gut microbe genera (Prevotella [27,44], Veillonella [42,48], and Streptococcus [44,48]) in children with ASD and GI symptoms were also reported by more than one article included in the present systematic review. Lower abundance of Prevotella may act as an indicator of Westernization, and this fact can also result in an altered immune system [53]. In addition, the presence of Prevotella together with other Bacteroidetes is also associated with colon health, and it has been hypothesized that ASD can be triggered by a dysbiosis where Prevotella decreases and Sutterella increases [54]. In this sense, the lower abundance of Prevotella found in American children with ASD was associated with the presence of ASD symptoms instead of GI symptoms [27]. These authors did not find this association in a later work [28], although they found that GI symptoms were significantly more severe in children with ASD compared to their control group.
Streptococcus, together with Lactobacillus, Bifidobacterium, and Lactococcus, are lactate producing bacteria [48]. It is known that children with ASD show elevated levels of both lactate and GI symptoms [55], suggesting an elevation of glycolysis through the phenomenon of aerobic glycolysis in ASD, since the dysregulation of this balance has been also been proposed as a trigger for ASD [56]. In this sense, as Veillonella is also able to ferment lactate, a lower abundance of Veillonella may also disturb the fermentation of lactate in children with ASD [48]. However, it cannot be stated that the elevated levels of lactate in people with ASD and GI symptoms can be attributed only to the GM, since, as discussed before, there are discrepancies in the abundance of other lactate-producing bacteria, such as Lactobacillus. Regarding the implications of Veillonella in GI symptoms presented by children with ASD, Veillonellaceae, together with Prevotellaceae, Prevotella, and Coprococcus, has been included in a probiotic mixture for treating ASD [57], and it is known that the differences in GM composition (especially in Veillonella) of children with ASD may be a consequence of diet [27,46].
The discrepancies found, at a statistically significant level, in the abundance of Bacteroidetes [42,48] and Firmicutes [31,48] phyla; Actinomyces [44,48], Dorea [42,43], and Lactobacillus [30,31,42] genera; Faecalibacterium prausnitzii species 28,43]; and Bacteroidetes/Firmicutes ratio [31,42,48] can be explained by several factors such as ASD heterology, age of participants, nature of control groups, sample location, small sample sizes, different nationalities, inter-individual differences, or different bacterial identification methods [14,31,53,58]. For example, one study used a culture-dependent method to detect Lactobacillus [30], and it was found to be less abundant in children with ASD and GI symptoms compared to the control group. However, two studies [31,42] that used culture-independent methods found opposite results, and although some authors have postulated that both methods yield similar results [59], culture-independent methods are relatively new and they are still developing [26]. Another study found a higher abundance of Dorea in stool samples of children with ASD and GI symptoms [42], while other studies found opposite results taken from rectum mucosal biopsy samples, and it is known that the place where the sample is taken can be relevant for mammalian physiology [14], as there are differences in microbial composition between stool and the GM [60,61]. This was confirmed in a larger-scale study showing that the populations recovered in the stool seemed to combine bacteria derived from the mucous membrane and the luminal part of the intestine, either adherent or not adherent to transient organic matter [62]. It is assumed that microorganisms seen in the stool reflect the microbiology of the colon, particularly the descending colon and rectum. Finally, it is important to highlight that there are still existing gaps in knowledge regarding the interactions between the microbiome and the host in vivo-and the pathways of metabolites-and how their metabolites influence the microenvironment. Therefore, further mechanistic studies involving "omics" technologies, as adapted from previous studies [63], might help shed light on these questions, and future studies should try to homogenize, as much as possible, the characteristics of the samples to be compared, including the place where the sample is taken from and bacterial detection methods, among others.

Conclusions
Although recent scientific literature has provided evidence of the relationship between GI symptoms and ASD, it is still too early to draw a conclusion about the gut microbes involved in GI symptoms of children with ASD, due to the limited number of studies reporting correlations between them. Furthermore, the correlation between certain gut microbes and GI problems do not determine the causality of the symptoms. However, it can be said that children with ASD and GI symptoms show higher abundance of Candida, lower abundance of Prevotella, Veillonella, and Streptococcus, and also show discrepancies in the abundance of other gut microbes due to ASD heterology, age of participants, nature of control groups, sample location, small sample sizes, different nationalities, inter-individual differences, or different bacterial identification methods. In addition, we consider that ASD has a wide phenotypic variability, so future studies should consider the relationship between GM, GI symptoms, and behavior of ASD in a more integral and multidisciplinary way. Finally, we argue that future research should homogenize sample characterization in order to develop a meta-analysis on the GM involved in ASD children with GI symptoms. Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.