Paediatric Asthma and the Microbiome: A Systematic Review

Evidence from the literature suggests an association between the microbiome and asthma development. Here, we aimed to identify the current evidence for the association between asthma and the upper airway, lower airway and/or the gut microbiome. An electronic systemic search of PubMed, EBSCO, Science Direct and Web of Science was conducted until February 2022 to identify the eligible studies. The Newcastle–Ottawa Scale and the Systematic Review Centre for Laboratory Animal Experimentation risk of the bias tools were used to assess quality of included studies. Twenty-five studies met the inclusion criteria. Proteobacteria and Firmicutes were identified as being significantly higher in the asthmatic children compared with the healthy controls. The high relative abundance of Veillonella, Prevotella and Haemophilus in the microbiome of the upper airway in early infancy was associated with a higher risk of asthma development later in life. The gut microbiome analyses indicated that a high relative abundance of Clostridium in early childhood might be associated with asthma development later in life. The findings reported here serve as potential microbiome signatures associated with the increased risk of asthma development. There is a need for large longitudinal studies to further identify high-risk infants, which will help in design strategies and prevention mechanisms to avoid asthma early in life.


Introduction
Asthma is a chronic inflammatory disease that affects the respiratory system and leads to significant morbidity and mortality [1]. Individuals suffering from asthma exhibit an array of symptoms, from wheezing and coughing to chest tightness and shortness of breath [2]. These manifestations vary in time of onset and intensity between asthmatic patients [2]. The common triggers that may lead to asthma exacerbation include, but are not limited to, viral respiratory infections, air pollution, tobacco smoke and exercise [3]. Allergies, genetics, respiratory infections during infancy and environmental features are risk factors for asthma development [3]. However, the exact aetiology of asthma is not well understood.
Evidence from the literature suggests that there is an association between the human microbiome and the development of asthma [4]. Both human studies and studies performed on experimental animal models have linked the dysbiosis of the early-life gut microbiome to a greater risk for the development of asthma in individuals who are genetically susceptible to this disease [4][5][6][7]. The gut microbiome has been shown to regulate the immune responses associated

Materials and Methods
We initially performed a non-systematic search within relevant journals for asthma and microbiomes to identify the existing systematic reviews related to these topics. However, the available systematic reviews were generally limited to upper airway or gut microbiome investigations in humans and paid little attention to the lower airway microbiome and animal-based studies. The current review was developed based on the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [23]. The study team consisted of researchers with experience in microbiology, immunology and respiratory care. It also included a researcher with experience in systematic reviews who was familiar with searching variable databases. The Covidence software from Veritas Health Innovation, Melbourne, Australia (available at https://www.covidence.org/ and accessed on 31 January 2023), was used to manage the retrieved studies, track the status of each study and update the PRISMA flow diagram.

Eligibility Criteria
The eligibility criteria consisted of original articles published in English between inception and February 2022 that addressed asthma diagnosis as an outcome among children up to 18 years old and investigated microbial communities in the upper airway, lower airway or the gut in humans or animals. Studies that addressed asthma diagnosis as a subgroup analysis were also included. The exclusion criteria consisted of studies that examined environmental and/or pollutant microbiomes and asthma or that reported asthma symptoms and/or atopic/allergy diseases without an asthma diagnosis.

Information Sources and Search Strategy
We comprehensively searched the following major electronic databases from 3 to 5 March 2022: PubMed, EBSCO, Science Direct and Web of Science. The search strategy was applied as appropriate for each database. The general search keywords used were: (asthma) AND (microbiome OR dysbiosis OR microbiota). The following filters were applied: age (up to 18 years), language (English) and literature type (original/academic journals). More details on the search strategy are provided in Supplementary File S1.

Selection and Data-Collection Process
All studies were imported to EndNote version X9 and then uploaded to Covidence software. After duplicates were removed, two stages of screening were conducted. First, two independent reviewers screened the titles and abstracts of the imported studies. Second, two independent reviewers conducted full-text screenings for the studies included during the first stage of screening. Finally, independent reviewers performed data extraction based on a data collection form designed specifically to address the objectives of this review (Supplementary Materials Table S1). Conflicts in the screening stages and the data collection process were resolved through regular discussion meetings with all authors.

Data Items
The data collection form (Supplementary Materials Table S1) included the following variables that were extracted from each study: the citation and title of the article, the country where the study was conducted, the study type (human or animal based), the study design, the sample size for each group, the age for each group, the microbiome environment (the upper airway, lower airway and/or the gut), the type of specimen collected for the microbiome analysis, the time of specimen collection (one time point or different time points), the microbiome detection method, the genomic DNA extraction method, the sequencing platform used, the microbial community diversity assessment (α-diversity, β-diversity, or both), the bioinformatics pipeline used and the study findings.

Risk of Bias Assessment
The quality of the included human non-randomised studies was assessed using Newcastle-Ottawa Scale (NOS) tools adapted for each study's design. Three tools were used: (1) the NOS adapted for cross-sectional studies [24], (2) the NOS for case-control studies and (3) the NOS for cohort studies. The NOS tools were used to assess quality based on different items categorised into three domains (selection, comparability and exposure or outcome). Then, the quality of each study was rated as good, fair or poor by translating the results of the NOS to the Agency for Health Research and Quality standards, as described previously [22]. For animal intervention studies, the Systematic Review Centre for Laboratory Animal Experimentation (SYRCLE) risk of bias tool was used [25]. Details of the tools used are described in Supplementary Materials Table S2.

Synthesis Methods
Due to the nature of the present systematic review, the descriptive data were extracted using a data collection tool that was generated specifically to address the objective of this review (Supplementary Materials Table S1).

Results
The literature search resulted in a total of 1025 studies, which were uploaded to Covidence. After the duplicates were automatically removed (n = 339), 686 studies remained. The titles and abstracts were screened, as a result of which 477 studies were considered irrelevant to the aim of the current review and excluded. The full text of the remaining 209 studies was examined for eligibility. As a result, 184 were excluded for the reasons detailed in Figure 1. The screening phase resulted in 25 studies that met the inclusion criteria and were identified as eligible for inclusion in the present review. based on different items categorised into three domains (selection, comparability posure or outcome). Then, the quality of each study was rated as good, fair or translating the results of the NOS to the Agency for Health Research and Qualit ards, as described previously [22]. For animal intervention studies, the Systematic Centre for Laboratory Animal Experimentation (SYRCLE) risk of bias tool was u Details of the tools used are described in Supplementary Materials Table S2.

Synthesis Methods
Due to the nature of the present systematic review, the descriptive data w tracted using a data collection tool that was generated specifically to address the o of this review (Supplementary Materials Table S1).

Results
The literature search resulted in a total of 1025 studies, which were uploaded idence. After the duplicates were automatically removed (n = 339), 686 studies re The titles and abstracts were screened, as a result of which 477 studies were con irrelevant to the aim of the current review and excluded. The full text of the re 209 studies was examined for eligibility. As a result, 184 were excluded for the detailed in Figure 1. The screening phase resulted in 25 studies that met the inclu teria and were identified as eligible for inclusion in the present review.  Tables 1-3 show the quality assessment results of the included human stud 22) based on the NOT criteria for case-control, cohort and cross-sectional studies tively. Sixteen human studies out of twenty-two were classified as good quality four were classified as fair quality [42][43][44][45] and only two were classified as poor [13,46]. The limitations were generally related to the potential selection bias. The evaluation for the animal intervention studies (three out of twenty-five) is desc Table 4. The three animal intervention studies [47][48][49] generally indicated the p performance and detection bias in aspects specifically related to the blinding pro  Tables 1-3 show the quality assessment results of the included human studies (n = 22) based on the NOT criteria for case-control, cohort and cross-sectional studies, respectively. Sixteen human studies out of twenty-two were classified as good quality [26][27][28][29][30][31][32][33][34][35][36][37][38][39][40][41], four were classified as fair quality [42][43][44][45] and only two were classified as poor quality [13,46]. The limitations were generally related to the potential selection bias. The quality evaluation for the animal intervention studies (three out of twenty-five) is described in Table 4. The three animal intervention studies [47][48][49] generally indicated the potential performance and detection bias in aspects specifically related to the blinding procedures.    Twenty-two out of the twenty-five studies identified in this review were clinical and examined the upper airway, the lower airway and/or the gut microbiome in healthy controls and/or asthmatic children (Table 5). Ten studies (out of twenty-two) examined the upper airway microbiome [26,[29][30][31][33][34][35][36][37][38], while only three studies investigated the lower airway microbiome [13,39,46]. One study analysed both the upper and lower airway microbiomes in healthy controls and children with severe persistent asthma [27]. In a study conducted in 2021, both the upper airway and gut microbiome investigations were performed in healthy controls and asthmatic children [42]. Seven studies (out of twenty-two) analysed faecal specimens to characterise the gut microbiome [28,32,40,41,[43][44][45]. The specimen types used to examine the upper airway microbiome were nasal swab [27,29,35,42], nasal wash [33], hypopharyngeal aspirate [30], nasopharyngeal swab [34], nasopharyngeal wash [31], saliva [26] and throat swab [35][36][37][38]. In contrast, the specimen types used to study the lower airway were bronchoalveolar lavage (BAL) [13,27] and sputum [39,46], while faecal specimens were used to study the gut microbiome [28,32,[40][41][42][43][44][45].

Animal Intervention Studies
Three out of the twenty-five identified studies were conducted using animal models ( Table 6). All three studies used murine models consisting of BALB/c mice [47], Sprague-Dawley (SD) rats [48] and C57BL/6 mice [49]. Regarding asthma induction, for both the BALB/c mouse model [47] and the SD rat model [48], the animals were sensitised by intraperitoneal injections of ovalbumin (OVA) and then challenged by OVA aerosol inhalation. However, there were variations among the methods used in each study, including the frequency and dose schedule of OVA exposure. For the interleukin-13 (IL-13) transgenic (TG) C57BL/6 mouse model, asthma was induced by lung-specific IL-13 overexpression [49]. The first animal intervention study performed 16S rRNA sequencing on both the nasal lavage fluid and BAL to characterise the upper and lower airway microbiomes in mice with OVA-induced asthma [47]. The second study extracted the lung tissues from rats with allergic asthma to characterise the lower airway microbiome [48]. BAL, lung tissue and faecal specimens were collected from IL-13 transgenic mice simulating chronic asthma to examine both the lower airway and gut microbiomes [49].

Diversity Assessments
As shown in Tables 5 and 6, 18 out of the 25 identified studies (72.0%) assessed both the αand β-diversity of the upper airway, lower airway and/or gut microbiome. These studies have reported contradictory findings related to αand β-diversity. For instance, an insignificant difference was observed in both αand β-diversity between asthmatic children and non-asthmatics [29,40,42]. On the contrary, a significant difference in αand β-diversity of the upper airway, lower airway and/or gut microbiome was detected between asthmatic children and non-asthmatics [27,30,35,46,48]. Five studies evaluated only the α-diversity of the microbiome in the upper airway, lower airway and/or gut microbiome (20.0%) [13,26,37,43,44] and demonstrated conflicting data. For example, Espuela-Ortiz and colleagues (2019) reported a significant difference in the α-diversity of the upper airway microbiome between asthma cases and healthy controls [26]. Another study detected insignificant differences in the airway taxa composition between asthma patients and healthy controls [37]. However, Bisgaard and colleagues (2011) estimated band richness and conducted principal component analysis (PCA), which resulted in no association between the bacterial diversity of the infant's gut microbiome and asthma development in the first 6 years of life [32]. The total load of bacteria for asthmatic children and healthy controls was calculated, and the authors reported a higher bacterial load in asthmatic children than in the healthy control group [41].

Human Studies
The data presented in Table 5 indicates that the microbiome in the upper airways of asthmatic children has a significantly high relative abundance of Moraxella, Staphylococcus, Streptococcus, Haemophilus, Fusobacterium, Dolosigranulum, Corynebacterium, Veillonella and Neisseria elongate [26,31,33,34,36]. However, a significantly low relative abundance of Streptococcus, Moraxella, Dialister, Prevotella, Tannerella, Atopobium and Ralstonia was identified in the upper airways of asthmatic children [26,35,37]. An increased relative abundance of Haemophiles in children aged 2 to 13 months was significantly associated with a higher risk of asthma development [29]. An additional study reported that a high relative abundance of Veillonella and Prevotella at age 1 month was significantly associated with asthma development by age 6 [30]. However, a significantly high abundance of Lactobacillus at age 2 months was associated with a lower risk of asthma development, suggesting that this bacterium plays a protective role [29].
The lower airway microbiome indicated a significant increase in Protobacteria in asthmatic children, particularly in asthma exacerbation cases [13,39], while a significant decrease in Saccharibacteria and Actinobacteria was detected [39]. Moreover, asthma exacerbation was associated with a high relative abundance of Veillonella, Neisseria, Haemophilus, Fusobacterium, Oribacterium, Campylobacter and Capnocytophaga in sputum [39]. However, Saccharimonas, Rothia, Porphyromonas, Gemella and Actinomyces were detected with low significant relative abundance in asthma exacerbation cases [39]. A high relative abundance of Streptococcus, Moraxella and Staphylococcus was identified in asthmatic children, with the latter detected in difficult asthma cases [13,46]. A mycobiome analysis revealed a significantly low abundance of Penicillium aethiopicum and Alternaria spp. in sputum specimens collected from asthmatic children [46].
The gut microbiome studies that examined the faecal specimens of asthmatic children revealed a significant increase in the relative abundance of Clostridium, Escherichia and Enterococcus [32,40,44]. In addition, a higher load of E. coli, Helicobacter pylori, Streptococcus and Staphylococcus was detected in the faecal specimens of asthmatic children [41]. A lower load of Bifidobacterium and Lactobacillus was detected in the faecal specimens of the same group, indicating that these bacteria play a protective role [41]. The mycobiome analysis of faecal specimens obtained from infants revealed a high relative abundance of Candida and Rhodotorula, which were associated with a high risk of developing asthma [45]. In contrast, the relative abundance of Lachnospira, Faecalibacterium, Roseburia, SMB53, Dialister and Dorea was significantly decreased in asthmatic children [28,40,44]. Lachnospira, Veillonella, Faecalibacterium, Rothia, Bifidobacterium and Akkermansia were significantly decreased in high-risk children [43,45].

Animal Intervention Studies
A respiratory microbiome analysis identified an increase in the relative abundance of Pseudomonas spp. during the acute inflammatory stage, while Staphylococcus spp. and Cupriavidus spp. increased during the airway remodelling stage in mice with OVA-induced asthma [47]. The bacterial phylum Firmicutes were detected at higher levels in the lower airway (lung tissues) microbiomes of rats with allergic asthma [48]. Proteobacteria and Cyanobacteria phyla were identified at higher levels in the lungs of IL-13 TG mice [49]. The microbiome analysis of faecal specimens extracted from IL-13 TG mice reflected a lower level of Firmicutes and Protobacteria, whereas the lung microbiome indicated a low level of Bacteroidetes [49].

Discussion
The aims of the current study were to examine the association between asthma and the upper airway, lower airway and/or gut microbiome in humans and animals and identify the characteristics of the upper airway, lower airway and the gut microbiome commonly associated with asthma.
The data presented in this review demonstrated that the clinical specimens collected from both the control and asthmatic children were mostly from the upper airway (i.e., a nasal swab, nasal wash, hypopharyngeal aspirate, nasopharyngeal swab, nasopharyngeal wash, throat swab and saliva). Only three studies collected specimens from the lower airway (BAL and sputum) [13,39,46], and one contained specimens from both the upper and lower airways [27]. The limited number of lower airway microbiome studies might contribute to the difficulty in collecting lower airway human specimens (specifically from healthy children) as it is more convenient to collect specimens from the upper airway.
Evidence of the association between asthma and changes in the upper and lower airways and/or gut microbiome was synthesized. The phyla Proteobacteria (Haemophilus, Moraxella, Neisseria, Campylobacter, Escherichia and Helicobacter) and Firmicutes (Veillonella, Staphylococcus, Streptococcus, Dolosigranulum, Oribacterium, Alloiococcus, Clostridium and Enterococcus) were identified as being significantly higher in the asthmatic children [13,39] compared with the healthy controls. These findings confirm the previous observations that Proteobacteria (Haemophilus, Moraxella and Neisseria) and Firmicutes (Staphylococcus and Streptococcus) were the most abundant bacteria in asthmatic children [50].
A previous literature review performed in 2019 reported that the most dominant genera in the upper airways of infants are Corynebacterium, Dolosigranulum, Haemophilus, Moraxella, Staphylococcus and Streptococcus [51]. However, in this study, we found that the upper airway microbiome in 1-month-old infants indicated an increase in the relative abundance of Veillonella and Prevotella, which were associated with asthma development later in life [30]. Both genera were considered normal flora of the upper respiratory system and their increased abundance in infants suggests their potential involvement in asthma development later in life [30]. Furthermore, the upper airway microbiome in infants ranging in age between 2 and 13 months indicated a higher abundance of Haemophilus, which was associated with a higher risk of asthma development later in life [29]. This substantiates the results of a previous review, which highlighted that dysregulated Haemophilus was common in asthmatic children [52].
As shown in Table 5, the upper airways of asthmatic children have a significant high relative abundance of Moraxella, Staphylococcus, Streptococcus, Haemophilus, Fusobacterium, Dolosigranulum, Corynebacterium, Veillonella and Neisseria elongate and a high relative abundance of Streptococcus, Moraxella and Staphylococcus was determined in their lower airways. The above-mentioned bacteria are known as normal human microbiota in the respiratory tract [53]. Furthermore, Staphylococcus, Streptococcus and Haemophilus, followed by Moraxella and Veillonella were the most frequently reported bacterial genera in the respiratory system of asthmatic children (Table 5). Previous studies have highlighted that the clinical characteristics of asthma patients and the type of immune response stimulated by aeroallergens influence airway microbiome composition [54]. For instance, Moraxella catarrhalis, a species of Haemophilus, and Streptococcus were the predominant respiratory tract bacteria in patients with severe asthma and corticosteroid resistance [54]. The literature points to a lack of metabolomic investigations of the association between the metabolic characteristics of these dysbiotic bacteria and asthma phenotypes and treatment prognosis. For instance, asthma patients with steroid resistance might have a higher abundance of airway microbial communities that can degrade steroids [55].
It has been established in the literature that the dysbiosis of the normal gut microbiome plays an important role in the development of immune disorders, including asthma [56,57]. This is explained by the key role of the gut microbiome in shaping the human immune system [55]. Differences in the gut microbiome in terms of composition and diversity were previously reported between healthy and asthmatic children [52]. In this study, the high relative abundance of the genus Clostridium was detected in faecal specimens collected from asthmatic children in three studies [28,40,44]. Previous studies have shown that the Clostridium species have an impact on the host's immune system [7]. In addition, infant colonization with Clostridium species is associated with a higher risk of allergy development [7]. This substantiates the findings of the current review as a predominance of the Clostridium species was detected during early childhood and was associated with asthma development [28].
The studies analysed in this review lacked consistency in reporting their findings. Some of the studies on bacterial communities in airways and/or gut have identified most of the detected bacterial taxa at the phylum level [13,39,40,44,46], whereas the others have identified the detected taxa at the genus level [26][27][28][29][30][31][33][34][35][36][37][38]41,43,45] (Table 5). Due to this inconsistency, making an accurate comparison of these studies became challenging. Moreover, bacteria belonging to different genera under the same phylum might have different effects on a host. For instance, this review revealed that the genus Lactobacillus, which belongs to the phylum Firmicutes, is associated with a low risk of asthma development, suggesting that the bacteria under this phylum play a protective role in asthma. By contrast, other genera under the same phylum Firmicutes, such as Veillonella, are significantly associated with asthma development later in life, suggesting their contributory role in asthma development. Therefore, it has been recommended that the use of reporting guidelines (i.e., the Strengthening the Organization and Reporting of Microbiome Studies [STORMS] checklist) must be adopted in future human microbiome studies [58].
Contradictory findings on microbiome diversity were reported by the included clinical studies (Table 5). Of the 22 clinical studies, 15 determined both the αand β-diversity of the upper airway, lower airway and/or gut microbiomes, but they reported conflicting findings on αand β-diversity between the asthmatic and non-asthmatic children. As depicted in Table 5, the clinical studies were conducted in different geographic locations, including North America, Europe, Asia, and Middle East, and they analysed clinical specimens obtained from different ethnic groups. The literature highlighted that the gut microbiome composition is associated with ethnicity and geography [59]. Furthermore, the sample sizes in 16 clinical studies were heterogeneous, with minimum and maximum sample sizes of 20 [46] and 923 [29] children, respectively. This sample size variation might have contributed to the variations in the diversity metrics [60]. The clinical studies also varied with respect to technical protocols, next-generation sequencing platforms and bioinformatics pipelines, as described in Table 5, and these variations might have influenced the quality of the obtained microbiome data [61].
There is limited literature on the use of animal intervention studies to examine the association between asthma development and microbiomes. The criteria related to random housing, blinding and random outcome assessment may hinder the research on such studies as the validity might be compromised. The quality assessment of animal intervention studies included in this review [47][48][49] generally indicated the potential performance and detection bias in aspects related to the blinding procedures, which might influence the validity of the results of these studies [47][48][49]. Furthermore, the current review indicated a lack of microbiome data related to viruses, archaea and micro-eukaryotes (such as protozoa). The characterization of these rare microbiome components might have a valuable impact on our understanding of asthma development.

Conclusions
The phyla Proteobacteria and Firmicutes were identified as being significantly higher in the asthmatic children compared with the healthy controls. The high relative abundance of Veillonella, Prevotella and Haemophilus in the microbiome of the upper airway in early infancy was associated with a higher risk of asthma development later in life. Gut microbiome analyses indicated that a high relative abundance of the genus Clostridium in early childhood might be associated with asthma development later in life. The findings reported here serve as potential microbiome signatures associated with an increased risk of asthma development. There is a need for human studies targeting the lower airway as well as well-designed animal intervention studies to further identify high-risk infants, which will help in design strategies and prevention mechanisms to avoid asthma early in life.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.