The gut-brain axis indicates that the disorder of host intestinal microbiota may be one of the causes of ASD. Recently, Needham et al. (2020) summarized four approaches to how this “bottom-up” impact is carried out: vagus nerve, stimulation of endocrine cells (including enterochromaffin cells), immune-mediated signaling and transport of gut-derived metabolites from the circulation into the brain. And, they believe that all routes comprising the gut–brain axis are thought to be co-opted by the microbiota to impact brain activity and behavior, and signaling through any one of them may be intertwined with other routes (Needham et al., 2020). Based on these four pathways, this study listed the potential evidence of GM affecting ASD.

(a) The vagus nerve provides a direct neural communication pathway between the GM and the central nervous system (CNS), and promotes the regulation of the GM on the function of CNS. Previous studies have found that toxins produced by Staphylococcus and Bacillus (staphylococcus enterotoxin and glutenin) can send signals to the brain by stimulating the vagus nerve, so as to induce vomiting or other disease behaviors (Hu et al., 2007). Bravo et al. (2011) found that Lactobacillus rhamnosus could reduce anxiety and depression related behaviors only in mice without vagotomy, which further explained that neurotransmitters or other metabolites produced by GM could directly regulate vagal activity by stimulating vagal afferent sensory neurons.

(b) Studies have found that 90% of serotonin in the human body is produced by intestinal chromaffin cells, a secretory cell in the inner layer of the intestine (Gershon and Tack, 2007). Enterochromaffin cell production of serotonin impacts its circulating levels and has the potential to influence brain activity directly or indirectly (De Vedder et al., 2018). In addition, some studies demonstrated that some Bifidobacterium and Clostridium metabolites can also change the content of serotonin in the intestine (Yano et al., 2015; Tian et al., 2019). Improved performance in mouse models of depression have been shown by probiotic treatment with Bifidobacterium spp. in a study that concurrently observed an increase either in the levels of serotonin in the brain or in the secretion of serotonin precursor in enterochromaffin cells in vitro (Tian et al., 2019). Moreover, Colonic enterochromaffin cells do express receptors for, and respond to, various microbial metabolites, including microorganism-associated molecular patterns (MAMPs), short chain fatty acids (SCFAs), aromatic amino acid metabolites, and secondary bile acids (Kidd et al., 2008; Reigstad et al., 2015; Tsuruta et al., 2016; Lund et al., 2018).

(c) Studies have proved that there is a correlation between intestinal inflammation and immune dysfunction in ASD patients, such as abnormal balance of T cells in the intestine of ASD patients and increased GI problems in ASD patients (Navarro et al., 2016; Vuong and Hsiao, 2017; Rose et al., 2018). Recently, it has been clearly demonstrated that high concentrations of pro-inflammatory microbiota in the gut can lead to increased intestinal permeability and inflammation, resulting in mild systemic inflammation and immune dysregulation (Felix et al., 2018). In addition, lipopolysaccharide (LPS), as an effective endotoxin in the cell wall of Gram-negative bacteria, has also been proved to induce disease behavior, cognitive impairment and acute depression like behavior in mice by activating systemic inflammation, and affect fetal brain development (Needham et al., 2020). Emanuele et al. (2010) found that the serum LPS level of ASD patients was significantly higher than that of healthy peers and was related to social behavior disorders, which further indicated that immune inflammation may play an important role in the intestinal brain axis.

(d) Many microbial metabolites produced in the gut can pass into systemic circulation at varying levels and rates. One example is SCFAs, where previous studies have shown that any interference in this signal transduction may have a direct impact on the central nervous system, which may lead to neurodevelopmental disorders and neurodegenerative diseases (Borre et al., 2014; Hill et al., 2014). Moreover, it has been proved that SCFAs metabolized by intestinal microorganisms can enter the circulatory system to regulate immune and inflammatory reactions, and then affect the neural function and development of human brain (Foley et al., 2014; Frost et al., 2014; Chambers et al., 2015). In spite of many studies support the health benefits of SCFAs, such as energy supply for epithelial cells, restoring epithelial barrier function, anti-inflammatory, and immunomodulating activities (Richards et al., 2016). However, it is important to note that excessive quantities of propionic acid (the main SCFA produced by Clostridium, Bacteroides, and Desulfovibrio) have been reported in irritable bowel syndrome, and necrotizing enterocolitis (Wang et al., 2007; Tana et al., 2010). In addition, the study found elevated levels of SCFA in the feces of children with autism (Wang et al., 2014). Although it needs to be established whether these elevated intestinal levels of SCFA are high enough to reach substantial levels in the brain, studies in rats have shown that exposure to propionic acid leads to significant deterioration of social behavior, which has shown that propionic acid may has harmful effects on neurological function (Thomas et al., 2012; Foley et al., 2014).

Children with ASD face the problem of difficult early diagnosis. Most children show some abnormal behavior symptoms only at about 18–24 months, while other specific functional characteristics may only be found at an older age (Borghi and Vignoli, 2019). Current studies indicate that there are multiple subtypes of ASD, potentially caused by different routes of pathophysiology and each with diverse comorbid psychiatric and medical conditions (e.g., gastrointestinal symptoms, allergies, sleep disorders) (Huang et al., 2021). However, this heterogeneity is not addressed by the conventional DSM5-based behavioral diagnostic criteria. Accordingly, it is particularly essential to discover effective objective physiological indicators as the basis for clinical diagnosis and evaluation of ASD. In the past decade, as the importance of GM in ASD has been identified, researchers have begun to investigate the microbial diversity of ASD patients, seeking to identify certain gut microbial characteristics as biomarkers for ASD. Unfortunately, the results of two recent meta-analyses show that these cohort studies have produced inconsistent results in exploring the intestinal microbiota of ASD children (Xu et al., 2019; Iglesias-Vazquez et al., 2020). The interactions between ASD and GM may be influenced by complicated factors such as genetic background, daily diet and the physiological status of the host, which may explain the conflicting results of these studies. However, it is worth noting that most studies have found that the overall diversity of GM (composed of archaea, bacteria, fungi, and viruses) in ASD children increases, while its fungal diversity decreases (Kuehbacher et al., 2006; Finegold et al., 2010; Zou et al., 2021). This suggests that there may be too many harmful bacteria in ASD children, such as Clostridium and Desulfovibrio, which are more common in ASD patients, also considered to be potential pathogenic bacteria of ASD (Parracho et al., 2005; Finegold, 2011).

In order to verify the claims on previous research concerning changes in the gut microbiome associated with ASD, Wu et al. (2020) performed Machine-learning based on feature selection and classification evaluation which were performed in the training cohort, the validation cohort, and independent diagnosis cohorts to evaluate the potential of the gut microbiome as a non-invasive biomarker for ASD. The results showed that Prevotella, Roseburia, Ruminococcus, Megasphaera, and Streptococcus may be potential biomarkers of ASD, especially Prevotella has significant differences between ASD patients and typical neurodevelopers (Wu et al., 2020), but this result is not consistent with the prediction model established by Zhai et al. (2019). One possible reason for this inconsistency is that the composition of intestinal microbiota is affected by the in vivo and in vitro environmental factors of its host individual, and the other influence could be the calculation method used in establishing the prediction model. Besides, the quality control conditions and methods of sequencing data might affect the results of the prediction model as well. Therefore, further studies may be required to explore the GM characteristics of ASD. In addition, previous studies mainly focused on the differences of GM between ASD patients and normal people, but rarely analyzed the changes of these biomarkers from the perspective of intervention. Thus, this review discusses different treatment methods, compares the changes of intestinal flora before and after intervention, and further looks into whether intestinal flora has great potential in ASD screening.

Although much research has shown that antibiotics can improve the GI and behavioral symptoms of ASD children, there are still some disputes about antibiotic treatment. In principle, antibiotics not only kill potentially harmful bacteria, but also kill beneficial bacteria in ASD patients, thus increasing the probability of GI diseases in ASD children (Vargason et al., 2019). Therefore, antibiotic therapy may not be an optimal intervention for GM rebalancing.

Recently, dietary interventions in children with ASD are very popular. Previous studies have shown that a simple, light and nutritious Mediterranean diet impacts the GM and associated metabolome as well as cardiovascular diseases and neurobehavioral health outcomes (Atladottir et al., 2012; Liu et al., 2017). Therefore, we summarized the studies of dietary intervention in ASD (Table 1). Many studies have shown that the ability of a ketogenic diet (KD, i.e., a high fat diet that has demonstrated beneficial effects on mitochondrial dysfunction and epilepsy) to mitigate some of the neurobehavioral symptoms associated with ASD in an animal model (Verpeut et al., 2016; Castro et al., 2017). Improvements in seizure control and neurobehavioral symptoms have also been reported in ASD children with mild-moderate types of ASD as a result of following a KD (Evangeliou et al., 2003; Herbert and Buckley, 2013; El-Rashidy et al., 2017; Lee et al., 2018; Żarnowska et al., 2018). In addition, the gluten-free and casein-free (GFCF) diet is also one of the most popular dietary therapies for ASD. Some publications report favorable results in the core or peripheral symptoms of autism after a GFCF diet: communication and language, social interaction, stereotyped behavior, hyperactivity, and gastrointestinal symptoms (Knivsberg et al., 2002; Elder et al., 2006; Whiteley et al., 2010; Johnson et al., 2011; Pennesi and Klein, 2012; Herbert and Buckley, 2013; Navarro et al., 2015; Ghalichi et al., 2016; El-Rashidy et al., 2017). However, data on the efficacy of a GFCF diet as a treatment for ASD in children are limited (Pusponegoro et al., 2015; Hyman et al., 2016; Gonzalez Domenech et al., 2019; Josw Gonzalez-Domenech et al., 2020; Piwowarczyk et al., 2020). Particularly in recent years, there have been many reports of an absence of behavioral improvement after such diets. Even recently, researchers have shown that dietary interventions could potentially have a harmful effect (Fattorusso et al., 2019). For example, restrictive diets further limit the variety of food intake since individuals with ASD already exhibit picky eating behavior, so restrictive diets can result in macronutrient and micronutrient deficiencies. Moreover, the food taken by this kind of diet method is usually expensive, which imposes an additional burden on the families of ASD children, and the standard of dietary intervention is extremely strict and does not apply to all ASD patients.

Probiotics are defined as live microorganisms that, when administered in adequate amounts, benefit the host’s health. Prebiotics refer to non-digestible fibers, such as oligosaccharides, that promote growth and improve the functioning of the probiotics in the GI tract by acting as a specific substrate. Initial evidence suggests that supplementing probiotics and prebiotics may have a good preventive effect on neurological and mental diseases such as Alzheimer’s disease, Parkinson’s disease, depression, and autism spectrum disorder (Yang et al., 2021). Moreover, some research has discovered that since some common abnormal genes between ASD and GI diseases, the abnormal genes may cause abnormal GM in ASD patients (Niesler and Rappold, 2021). Considering the two-way communication of gut brain axis, we believe that probiotic intervention in ASD infants may affect the expression of related genes and decrease the prevalence of ASD. However, the specificity of GM in different patients suggests that precision medicine may be the hope of the future, where treatment protocols will be tailored for specific subpopulations of patients. Therefore, in order to explore the effectiveness of different probiotic and prebiotic therapies on behavioral symptoms and GI symptoms of ASD patients, we summarized the existing probiotic and prebiotic interventions, which can be divided into single strain intervention (Table 2), mixed strain intervention (Table 3), single probiotic and probiotic plus prebiotic intervention (Table 4).

Fecal microbiota transplant (FMT), consisted of transferring the fecal microbiota from healthy volunteers to patients with gut dysbiosis, may alleviate GI and neurobehavioral symptoms in children with ASD by rebalancing the physiological intestinal microbiota. We have summarized these studies in Table 5. Linda et al. (2016) used FMT to intervene ASD, and found that behavioral symptoms and GI symptoms improved in younger ASD patients, while there was no significant change in older patients (21 years old) before and after the intervention, proving the feasibility of FMT for the treatment of children with ASD. Based on this, Zhao et al. (2019) conducted a randomized controlled study of 48 patients with ASD. The FMT group received two FMT treatments (2 months apart) and the control group received only rehabilitation training. They found that after the first FMT treatment, the Childhood Autism Rating Scale (CARS) scores in the FMT group decreased by 10.8% (behavioral symptoms of ASD improved) compared with 0.8% (P < 0.001) in the control group. After the second FMT, the CARS scores in the FMT group continued to decrease slightly (P = 0.074), further demonstrating the efficacy of FMT. However, there are still some problems with FMT. For example, 7 cases (29.2%) in the FMT group had adverse reactions such as fever, allergy and nausea during the intervention. Therefore, some argue that the feasibility of this approach for all ASD children needs further validation.

In response to this, Kang et al. (2017) developed a modified FMT protocol (Microbiota Transfer Therapy, MTT) that consisted of 14 days of oral vancomycin treatment followed by 12–24 h of fasting bowel cleansing and then either oral or rectal administration of standardized human GM (SHGM) for 7–8 weeks. They found that 18 ASD children aged 7–16 years old not only improved their behavioral symptoms after treatment, but also increased the diversity of bacteria in their gut, with the increased abundance of Bifidobacterium, Prevotella, and Desulfovibrio. All of these changes persisted for at least 8 weeks after treatment ended. Furthermore, after 2 years of follow-up, the results showed that the ASD patients treated with MTT maintained a high diversity of gut bacteria and abundance of Bifidobacterium and Prevotella, and most of the improvement in gastrointestinal symptoms was also maintained. Importantly, behavioral symptoms were kept improved after 2 years of MTT treatment (Kang et al., 2019). The long-term effectiveness of this study shows that this treatment can maintain the remodeling of the gut of ASD patients, make their gut micro ecological system to achieve a healthy balanced state, and then improve the condition and the behavior level. The series of results confirmed an important role of overall gut microbes rebalancing during the process of intervention, which we believe to be an alternative and promising new approach for the treatment of GM dysbiosis in ASD.