Tongue coating microbiome characterization of Chinese preschool children with halitosis

Background: Halitosis may affect individuals of any age. However, epidemiological surveys and etiological research into halitosis among children are limited. To assess the bacterial characteristics and environmental factors associated with intra-oral halitosisamong Chinese preschool children. Methods: An epidemiological study was conducted among 273 preschool children (aged 3 to 4 years) in Shanghai, China,followed by collection of 16 samples from the tongue coating of caries-free healthy and halitosis participants according to their organoleptic score.The characterization of associated microbial communities was performed using 16S rRNA genesequencingon the MiSeq. Results:Halitosis was observed in 13.1% of the preschool children.Logistic regression analysis showed that less frequent intake of sweet snacks and mothers’ poor oral condition were associated with children’s halitosis positively. Further analysis detected the presence of 15 phyla, 24 classes, 35 orders, 53 families, 93 genera, and 226 operational taxonomic units (OTUs) in the microbial communities. In terms of microbial diversity, the tongue coating of control and halitosis groups showed no significant difference (P>0.05). Both groups shared a mass of common OTUs. The dominant bacterial genera observed in both groups included Prevotella_7(15.17% vs. 10.72%), Veillonella (13.52% vs. 10.86%), Streptococcus (13.95% vs. 18.68%),Neisseria (9.44% vs. 21.14%), Actinomyces(7.37% vs. 5.88%), Haemophilus (6.69% vs. 5.69%), andLeptotrichia (5.05% vs. 1.95%). At the genus level, the proportion of Gemellawas statistically larger in the control group (P=0.018). Conclusions: Overall, halitosis is associated with multi-microbial mutual

Due to the complex interaction of oral microbial flora, halitosis is mainly associated with mixed microbial infections rather than a single species infection [7,8]. Indeed, a wide range of oral microorganisms including Prevotella intermedia, Porphyromonas gingivalis, Treponema denticola, Fusobacterium nucleatum, and Tannerella forsythensis are associated with intra-oral halitosis [7,8]. Identification of the underlying oral bacterial species is integral within halitosis-related investigations. The development of next-generation sequencing (NGS) [9] technology enabled researchers to evaluate microbial communities including uncultured microorganisms associated with halitosis, leading to a deeper understanding of the micro-ecological changes related with halitosis in the pediatric population. The scientific evidence suggests that oral microbiomes differ in the oral environment of patients depending on whether halitosis is present. [10]. For instance, Ren et al. [11] reported that Prevotella shahii was frequently detected and was relatively abundant in saliva samples from children with halitosis.
Halitosis may affect individuals of any age. Indeed, parents frequently visit dentists complaining about the bad breath or odor coming from the oral cavities of their children. However, most halitosis studies [12][13][14] to date have investigated adults, and data from epidemiological surveys and etiological research into halitosis among children is limited [15][16][17][18][19]. Studies investigating the prevalence of halitosis in children have mainly focused on participants aged between 5 and 12 years or adolescents. Little is known of the prevalence, risk factors, and the microbiome associated with halitosis in preschool children. Since preschool children rarely have periodontal diseases, the microbial basis of halitosis in the oral environment remains largely unknown. Therefore, preschool children are considered good candidates for studying intra-oral halitosis and are likely to provide useful scientific data [11].
As no published articles had reported epidemiological data and microbial characteristics on halitosis among Chinese preschool children, the aim of this research was to assess the epidemiological features and environmental factors associated with intra-oral halitosis among Chinese preschool children. To investigate the bacterial variations associated with pediatric intra-oral halitosis in Chinese preschool children, tongue-coating samples were collected from selected children with and without halitosis. methods Subjects An epidemiological study was conducted on the oral health of pre-school children (aged 3 to 4 years) in Shanghai, China. Three kindergartens in Shanghai were randomly sampled: one from a central district and two from a suburban region. All children in the junior classes of these kindergartens were invited to participate in this study. Subsequently, 273 children underwent comprehensive oral examinations and were subjects of the study. The present study was reviewed and approval was obtained from the Ethical Committee of Ninth People's Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China (Ref No. 2015135). Prior to the examinations, all of the participants' guardians signed an informed consent form.
All procedures were conducted in accordance with the Declaration of Helsinki.

Questionnaire
To gather data related to halitosis, all of the participants' guardians were requested to complete a questionnaire one day before their children's clinical oral examinations. Information including sociodemographic characteristics (gender, age, height, weight, parents' education level, and parents' income); disorders or diseases during gestation (e.g., premature birth); physical conditions (self-reported oral malodor, systemic diseases, and discomfort); dietary habits (including frequency of fruit, sweets, and meat consumption); and oral hygiene habits (tooth brushing frequency, oral examination frequency, and parental assistance during tooth brushing) of the participants were collected using the questionnaires. Parents were also asked to disclose the medical history and any discomfort suffered by their children.
Methods, clinical oral examination, and halitosis assessment Clinical examinations were performed on each child during the morning in the kindergartens. Organoleptic assessment criteria were used for diagnosing halitosis as previously described [20]. Briefly, children were asked to keep their mouth shut for 3 minutes and then to expel air from their mouth slowly using a paper-tube placed 10 cm from the observer's nose. Organoleptic assessment scores (OS) were determined by a qualified and registered dentist using the ''0-5 Rosenberg  Subjects scoring 2 or greater (OS≥2) were considered to have halitosis [20].
Ten percent of the participants were reexamined to evaluate intra-examiner reliability. The Cohen's kappa values (κ) for all parameters (caries, indices, plaque accumulation, tongue-coating evaluation, and OS assessment) were >0.8. The participants were divided into two groups (halitosis and control) according to the preset selection criteria.

Sample collection
Tongue coating samples were collected from 16 participants by scrubbing from the dorsal to ventral surfaces of the tongue using a specialized tongue cleaning device (Shanghai, China). All data in this study has been submitted to SRA database (SRA accession: PRJNA548494).
Each 16S-rRNA gene sequence was classified and measured using the RDP Classifier algorithm (http://rdp.cme.msu.edu/) against the Silva 16S-rRNA database (threshold of 70% confidence level). Representative sequences were divided at different classification levels.
Tongue coating and bacterial richness diversity were evaluated in the sample using the alpha (α) index. The analysis of similarities (ANOSIM) was performed to compare the differences in tongue community composition among the children based on the weighted UniFrac distance. The community ecological package was used to obtain principal coordination analysis (PCoA) data. R-forge was applied to evaluate the beta diversity on the basis of weighted Unifrac and Braycurtis distance matrices (PCoA figure was generated by Vegan 2.0 package). The Circos software was used to construct the Circos graph. The linear discriminant analysis effect size (LEfSe) (http://huttenhower.sph.harvard.edu/galaxy/) at multiple levels was also applied to compare bacterial composition between the two groups. Wilcoxon rank-sum test was used to compare relative tongue coating microbiota abundance between two groups.

Statistical analysis
All data were processed using SPSS v20 software (IBM, NY, USA). The significance level was set at the 5% level and all the test were two-sided. Descriptive statistics were used for children's sociodemographic characteristics and clinical measurements. Student's t-test was applied to analyze the significance of difference in continuous variables between the halitosis and control group children.
Additionally, the Chi-square test was utilized for investigating differences both groups according to clinical and lifestyle factors. Logistic regression was conducted to estimate factors associated with children's halitosis. The differences in microbial diversity index were analyzed using the student's t-test. The Mann-Whitney U test was applied to assess variances of the bacterial compositions in tongue coatings.

General and clinical characteristics
The present study analyzed the questionnaire information and clinical data of 273 (143 male and 130 female) kindergarten children aged 3-4 years. There was no significant difference in clinical and physical characteristics such as caries and body mass indices between children with and without halitosis (P>0.05) ( Table 1). In the organoleptic assessment, the majority of children (n = 203) scored 0 while 34 had a barely noticeable odor (organoleptic assessment score, OS = 1). Thirty-six children produced an organoleptic score greater than 2 (OS≥2) and thus were considered to have halitosis. Moreover, all the children with halitosis showed no significant differences in dental caries compared to the control group children (Table 1).

Halitosis and associated lifestyle factors
The distributions of the main demographic and socioeconomic characteristics of participants showed no significant differences in the educational level or the monthly income of parents with regard to the prevalence of halitosis (Table 1). In terms of any association between halitosis and lifestyle factors, less sweet snack consumption was associated with a higher prevalence of halitosis (P = 0.047) ( Table   2). Furthermore, the study participants whose parents could always smell oral malodor in their children were observed to have a significantly higher percentage of halitosis (P = 0.040). Among child halitosis cases, the logistic regression analysis revealed a strong association between less frequent intake of sweet snacks and mothers' poor oral condition (Table 3).
Tongue coating samples and sequence characteristics Children (n = 8; 5 boys and 3 girls) who fulfilled the inclusion criteria for the halitosis group were selected. The control group also included 8 participants (3 boys and 5 girls) who fulfilled the required criteria. Both groups showed insignificant differences (P>0.05) on the basis of their sociodemographic background and oral conditions (Table S1).
The analysis of participants' tongue coatings showed a total of 594,710 reads after screening and a mean of 37,169 reads per sample following. Sequence OTU clustering and notation performed at a 3% divergence level detected 15 phyla, 24 classes, 35 orders, 53 families, 93 genera, and 226 OTUs. The OTU distributions were showed by the Venn diagram (Fig.S1A), and the majority of OTUs (187) were preserved and shared in both groups.

Microbial richness and diversity of tongue coating
The rarefaction curves of all samples indicated an adequate sequencing depth in the present study (Fig. S1B). The alpha (α) diversity (observed at OTU level) of five different indices were calculated (Table S2). There were no statistical differences (P>0.05) in the diversity of tongue microbial coatings between the halitosis and control groups, although the halitosis group showed lower indices by the Chao index and higher indices by the Shannon index ( Fig. 1A and Fig. 1B).
The comparison of overall microbial composition and structure between halitosis and control group was conducted using principal coordination analysis (PCoA). At the genus level, there was no significant separation in the bacterial community composition between halitosis and control group samples (the analysis of similarities, ANOSIM: R = 0.106, P = 0.115) (Fig. 1C).
Characterization of tongue coating microbiota and microorganisms associated with halitosis in preschool children At the genus level, genera with an average relative abundance greater than 1% were considered. The data showed that the dominant genera among samples of both groups were consistent (Fig. 2). A heat-map of the top 50 relatively abundant key bacteria genera of each sample was constructed on the basis of hierarchical clustering solution approach (Bray-Curtis distance metric & complete clustering).
The data showed the Z-transformed relative percentage for every genus (Fig. 3) The microbiota abundance of both groups was compared using the linear discriminant analysis effect size (LEfSe) analysis (Fig. 4A) and the linear discriminant analysis (LDA) scores (Fig. 4B). The distribution of Enterococcus was significantly higher in halitosis samples (P = 0.004), whereas the distribution of Gemella and Kingella showed a higher abundance in the control samples (P = 0.016, P = 0.001, respectively).
We applied the Wilcoxon rank-sum test for further comparison between both groups' tongue coating microbiota for relative abundance (%) of the top 25 genus. At the genus level, the proportion of Gemella was statistically larger in the control group (P = 0.018). Prevotella_7 and Veillonella showed higher relative percentages in the tongue coating samples of halitosis group compared to the control group, however, the differences were insignificant (P>0.05) (Fig. 4C). Comparisons of bacterial average relative abundances between the halitosis and control groups at the genus level are presented in Supplementary Table S3. Microbial correlation networks of tongue coating microbiota As Gemella was the only dominant genus that showed statistical difference in the comparison of relative abundance between two groups (Fig.4), to explore the potential role of Gemella in microbial communities associated with halitosis, the cooccurrence network of 25 most abundant bacteria at the genus level were established (Fig.5). The Gemella genus node was more central and more complex in halitosis group. Gemella had 4 positive (Oribaterium,Haemophilus,Porphyromonas and Fusobacterium) and 5 negative (Selenomonas,Campylobacter,Veillonella,Megasphaera and Atopotium correlations in the halitosis group (Fig. 5A). For the control group, Gemella showed a negative correlation with Lachnoanaerobaculum (Fig. 5B).

discussion
The present study investigated the epidemiological characteristics, environmental factors and bacterial variations associated with intra-oral halitosis among Chinese preschool children. A structured questionnaire was used to gather data on sociodemographic characteristics, lifestyle, medical history, and oral hygiene. The tongue coating samples were analyzed for microbial communities in both healthy (control group) and halitosis participants.
The prevalence of halitosis was observed clinically in 13.1% of 273 preschool children. The percentage of halitosis cases in the present study was significantly lower compared to other studies [14,25] reporting the prevalence of halitosis in adults. The previous studies reported that the prevalence of halitosis increases with age [16, 17,26]. The positive correlation between children's halitosis and age can be caused by the eruption of permanent teeth providing a greater surface area for bacteria accumulation.
Generally, there are two types of established methods used for the diagnosis of halitosis: organoleptic assessment (OA) and by sulfide monitoring equipment. OA is considered the gold standard treatment [27], and is time-efficient for in-school investigations. Therefore, we used OA methods in the present study to assess halitosis. For the halitosis that originates from the oral cavity among adults, the main contributory factors are poor oral hygiene, periodontal conditions (such as periodontitis, gingivitis), tongue coating, and dry mouth [2, 28]. Since there was a low incidence of periodontal diseases and dry mouth in preschool children, the present study investigated fresh insight into factors that play a pivotal role in causing halitosis in children.
Our results showed that the children consuming sweet foods more frequently had a lower prevalence of halitosis. These findings are in line with the result of our previous study [25]. In the oral environment, the degradation of peptides resulting in an acidic environment (low pH) due to a trypsin-like enzyme contributes to halitosis. The frequent consumption of sweet snacks (such as carbohydrates including glucose and sucrose) by children may inhibit the enzymatic activities of various enzymes hence reducing the production of halitosis-associated metabolites [8,29]. However, whether sweet foods cause halitosis in children needs to be investigated further. Our study also found that halitosis in children was related to the oral health status of their mothers. Indeed, parents' oral hygiene behaviors impact the oral health of their children [30,31]. Appropriate oral health education and oral hygiene instructions from parents can play a major part in maintaining a healthy oral cavity and preventing halitosis in children.
The qualitative nature of bacterial tongue coating is a well-known indicator of halitosis [15]. For instance, the deep fissures in the tongue coating provides a low oxygen environment for halitosis-associated anaerobic bacteria thus potentiating their pathogenic colonization. In this study we compared features of bacterial tongue coatings obtained from halitosis and healthy children (control) groups. In the present study, both α and β diversity indices based on the weighted UniFrac distance metric indicated that the overall microbial structure was similar between healthy and halitosis children. Seerangaiyan et al [32] showed similar findings in that the qualitative bacterial composition of the tongue microbiome in the halitosis patients was largely the same as that of the control group with only a few exceptions of certain bacterial species and genera. In contrast, Ren et al [11] reported that the tongue coating of children aged 4-5 years with halitosis contained a significantly higher quantity of bacteria than those without halitosis. A follow-up study may be required to explore halitosis in children of different ages to detect any differences.
In the present study, the dominant genera of both groups were Prevotella_7, Veillonella, Streptococcus, Neisseria, Actinomyces, Haemophilus, and Leptotrichia.
In addition, these bacteria were present in all of the tongue coatings and could adapt to the tongue coating environment of children. Neisseria accounted for more than a fifth (21.14%) of the bacterial community in the healthy group, and Prevotella_7 (15.17%) was the most abundant genus in the halitosis group.
Prevotella has been considered a pathogenic bacterium of halitosis and found to be positively correlated with VSC gases parameters [33][34][35]. As the major H 2 Sproducing bacteria, Prevotella showed a higher abundance in halitosis children in the present study; however, the difference was not significant and may have been due to the subjects' ages and severity of their halitosis. As a constituent of the oral microbiota, Leptotrichia is an opportunistic pathogen normally present in the mouth. suggested that Streptococcus salivarius contributed to mucin degradation and deglycosylate salivary glycoproteins, thus increasing the likelihood of halitosis.
However, the role of these bacterial species in halitosis and their pathological mechanisms remain uncertain.
In terms of the genus, the proportion of Gemella  a Obtained by Chi-squared test. * These variables have some missing data.
OS: Organoleptic assessment score.  Comparisons of the relative abundance of tongue coating bacterial communities between the