Genes and single nucleotide polymorphisms in the pathway of saliva and dental caries: a systematic review and meta-analysis

Abstract The aim of this systematic review and meta-analysis was to investigate the influence of single nucleotide polymorphisms (SNPs), related to genes in salivary composition and flow, on dental caries experience. Sixteen studies were included in the systematic review and ten in the meta-analysis. Forty-four SNPS, covering four genes (CA6, AQP2, AQP5, and MUC5B) were identified. Most of the SNPs were not associated with caries in meta-analysis. Homozygous TT genotype of the SNP CA6 rs17032907(C/T) was associated with caries [OR = 3.23(1.39–7.49)]. The pool effect of the SNPs assessed in AQP5 was associated with a reduction in the likelihood of caries [OR = 0.75(0.59–0.95)]. Considering all SNPs of salivary composition and flow, the effect allele was associated with a 75% increase in the likelihood of caries [OR = 1.75(1.06–2.89)] in the homozygous genotype. The present findings showed that the genes in salivary composition and flow can play an important role in dental caries experience.


Introduction
Dental caries is a biofilm-dependent disease with high global prevalence (Wen et al. 2022;Dutra et al. 2018), also reported as the main reason for dental restoration failures in permanent (Demarco et al. 2012), and deciduous  teeth. Although it affects a large proportion of the population, caries prevalence is disproportionately distributed among individuals, indicating a polarization in those who are, in some way, socially vulnerable (Marcenes et al. 2013). This is mainly due to the multifactorial etiology of caries, which exhibits a complex network of determinants and mediators (of variable intensity depending on the individual being evaluated) (Marcenes et al. 2013;Kassebaum et al. 2015). Thus, social, biological, and behavioral factors have been identified as the main causes of differences between populations in terms of the prevalence of dental caries (Marcenes et al. 2013;Torriani et al. 2014;Kassebaum et al. 2015;Piovesan et al. 2017;. However, recent studies have found significant associations between genetic polymorphisms and dental caries (Chisini et al. , 2022, which may explain why individuals with similar oral health-related behavior can exhibit different patterns of dental caries (van Loveren and Duggal 2001;Slade et al. 2013). Thus, genetic architecture could offer an additional, albeit smaller, influence on resistance/susceptibility to dental caries (Vieira et al. 2014b). Although genetic contributions to the occurrence of dental caries have been proposed since the late 1980s in twin studies, interest in this topic has grown in recent decades due to improvements in biological and molecular methodology, such as DNA sequencing, which has flourished with the advent of the Human Genome Project. Therefore, the association between genetic factors and dental caries has been investigated mainly through candidate gene studies (Vieira et al. 2014a;Anjomshoaa et al. 2015;Li et al. 2015) and by Genome-Wide Association Studies (GWAS) (Govil et al. 2018;Shungin et al. 2019;. With the aim of finding new genes involved in dental caries, only a small number of studies have used the GWAS strategy, while most of the studies investigating this association use the candidate gene approach (Vieira et al. 2014a). In this context, some Single Nucleotide Polymorphisms (SNPs) of salivary genes have been shown to have an influence on susceptibility to dental caries (Vieira et al. 2014b;Shimomura-Kuroki et al. 2018).
Indeed, saliva possesses components that can inhibit cariogenic bacteria, as well as containing calcium and phosphate, which are actively involved in the process of dental enamel demineralization and remineralization. In addition, saliva flow dilutes microorganisms and carbohydrates ingested by individuals, preventing their accumulation on dental tissue (Kidd and Fejerskov 2004). Accordingly, SNPs in the Carbonic Anhydrase 6 (CA6) gene have led to a decrease in salivary buffering capacity in a sample of Brazilian children (Peres et al. 2010). The CA6 SNPs' also influenced the teeth biofilm microbiota composition of Swedish adolescents (Esberg et al. 2019): the CA6 contributes to the neutralization of acids produced by bacteria, i.e., lactic acid, through the conversion of saliva HCO3 À to water and carbon dioxide, resulting in pH maintenance through the subsequent phase buffering (Bardow et al. 2000). This mechanism explains the increased Streptococcus mutans dental colonization observed in individuals with SNPs on CA6 (Esberg et al. 2019). In addition to the increase of Streptococcus mutans dental colonization, the general biofilm microbiota composition was affected by gene polymorphism in CA6. Aciduric species were associated with SNPs of CA6, leading to changes in the de/remineralization process and, consequently, influencing the caries risk of Swedish adolescents (Esberg et al. 2019). Mucin 5B gene (MUC5B) encodes a protein (MUC5B) responsible for mucus secretions present in saliva, decreases the bioadhesion of Streptococcus mutans, and inhibits the biofilm formation (Frenkel and Ribbeck 2015). A reduced amount of MUC5B in the dental pellicle can leave teeth vulnerable to Streptococcus mutans attachment and subsequently increase the risk of dental demineralization (Frenkel and Ribbeck 2015). In this context, SNPs of MUC5B were associated with high odds of biofilm presence and with an increase of dental caries in a Brazilian child sample (Cavallari et al. 2018). Similarly, the deletion of the Aquaporin 5 (AQP5) gene exhibited a reduction in saliva flow, increasing dental caries in mice (Culp et al. 2005). Several reviews have been published on this topic (Vieira et al. 2014b;Lips et al. 2017;Piekoszewska-Ziȩtek et al. 2017). However, they only reported the influence of SNPs on caries experience based on an overview of individual studies; the reviews did not pool the same genes and polymorphisms by way of an analytical approach, which might be an interesting strategy to better understand the real role of SNPs in caries susceptibility.
An understanding of which SNPs and genes in salivary composition and flow are involved in individuals' susceptibility to caries disease could provide a framework for the development of a viable approach to better comprehend these complex mechanisms. Therefore, the aim of this study was to perform a systematic review and meta-analysis to investigate the influence of single nucleotide polymorphisms related to genes in salivary composition and flow on the experience of dental caries.

Methods
The present systematic review was registered on PROSPERO (International Prospective Register of Systematic Reviews) under protocol number CRD42019121477. Furthermore, we reported the study according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline (Page et al. 2021).

Review question and searches
The research question-'Do the SNPs of the genes in salivary composition and flow influence the susceptibility of adults and children to dental caries'?-was structured following the PICO model: Participants/population: adults and children; Intervention/exposure: Minor allele/genotype frequencies from single nucleotide polymorphisms in the genes in salivary composition and flow. The effect allele in this study was standardized as the least frequent allele described in the studies. Whenever the minor allele frequency varied between studies, the effect allele was referred to as the minor allele in the majority of the studies. Also, to perform the estimates stratified by genotype, the minor homozygote and heterozygote models were chosen as the effect genotypes; Comparison/control: higher allele/genotype frequencies from single nucleotide polymorphisms in the genes in salivary composition and flow. Thus, the effect allele was compared to the reference allele, defined as that which is most frequent in the population. To perform the genotype analysis, the major homozygote was chosen as the reference; Outcome: Dental caries experience.
The search strategy was applied using appropriate keywords and entry terms related to MeSH Terms, depending on the database (Supplemental Material 1). Studies were searched using five different databases, namely Pubmed/Medline, Scopus, Web of Science, and BIREME-BVS Virtual health library and Scielo, up to the end of May 2022. All records retrieved were uploaded to EndNote Basic (www.myendnoteweb. com) with the aim of deleting duplicates. In this way, a virtual library was assembled. Two independent reviewers (LAC and RVC) read the titles and abstracts and evaluated all the papers found, in compliance with the inclusion/exclusion criteria. The eligibility criteria were defined as follows: a. Inclusion criteria: articles that aim to evaluate the association between genetic single nucleotide polymorphisms in genes in salivary composition and flow in children or adults. Only human studies with a cross-sectional, cohort, longitudinal or case-control design. No restrictions as to language or dates of publication were imposed. b. Exclusion criteria: studies designated as literature reviews, case reports or case series, conference abstracts, letters to the editor, and qualitative studies.
The articles selected at this stage were full-text assessed and evaluated again by the same reviewers. In the event of any disagreement in relation to the inclusion of a study, the reviewers discussed the matter to obtain consensus. If consensus was not reached, a third reviewer (MBC) made the final decision.
All the references in the included studies were investigated with the aim of identifying potentially eligible studies. The gray literature was also investigated using the same search strategy used in Pubmed/ Medline on Google (https://www.google.com/) and Google Scholar (https://scholar.google.com.br). The first 100 records were manually accessed for each of the databases.

Data collection
Data extraction was performed independently by the same reviewers on a predefined database (Kappa 0. 91). The following information was collected: Author, year, country, study design, sample, age, ethnicity of the sample (% for each ethnic group), percentage of the sexes in the sample, calculation of statistical power, evaluation of categorization of dental caries, analytical approach, data analysis (crude and adjusted analysis values and their respective confidence intervals), covariables, and main results.

Quality of studies
The quality of the studies was assessed using two instruments. Firstly, it was verified by means of the scale of the Appraisal Checklist for Observational Studies (Joanna Briggs Institute). This instrument has nine questions evaluating diverse arguments in the study, for which there are three options: 'No', 'not clear' or 'Yes'. Each 'Yes' answer is worth one point. Studies scoring between 0 and 3 were considered low quality; 4 to 6 were of medium quality; and 7 to 9 were considered high quality. Two reviewers (LAC and RVC) performed the evaluation independently. Secondly, we performed a complementary evaluation by way of an instrument specifically designed to evaluate genetic studies, adapted to a 10-point scoresheet (Clark and Baudouin 2006;Salles et al. 2017). This instrument comprises two different measures to estimate each of the 10 points (Yes ¼ 1) or (No/ undetermined ¼ 0). The same reviewer independently performed the evaluation.

Strategy for data synthesis
A qualitative analysis was performed describing the main results of the studies. In addition, aiming to synthesize the data in quantitative terms, a meta-analysis was designed. Initially, the plan was to perform a meta-analysis when the same polymorphisms were identified in at least three different studies. However, in view of the fact that the studies analyzed different SNPs in the same gene, we decided to perform a global meta-analysis, pooling the same SNPs across the studies, as well as pooling different SNPs in each gene across the studies.
Thus, for the meta-analysis, only SNPs discussed in at least two different studies were considered in the pooled SNPs and gene results. In addition, meta-analysis was performed pooled by gene, including the results of individual studies. In the various analyses, the effect allele and genotypes were compared to the reference allele and genotype, respectively. In the case of studies presenting more than one caries category, the DMFT/dmft ¼ 0 vs. DMFT/dmft !1 was chosen. Ideally, results adjusted for ethnicity were included.
When adjusted results were not reported, the unadjusted estimates were assessed or calculated and included in the analysis. In cases where results were only displayed by means of a stratified analysis, we included the group with the highest number of individuals. The odds ratio (OR) was used to measure effect size with a 95% confidence interval. The measurements of prevalence ratio were converted to OR using the formula proposed by Zhang and Yu: PR ¼ odds ratio/1 À risk0 þ risk0 Â odds ratio, where 'risk0' is the prevalence of the disease among nonexposed individuals (Zhang and Yu 1998;. To address the absence of ethnicity reporting, an investigation was conducted of allele frequencies stratified by populations, based on the human genome (GRCh37.p13).
To provide a robust analysis and avoid inconsistencies in the data analysis, data harmonization for palindromic SNPs was performed. If the palindromic SNP was present in two different studies, we only kept the SNP in the analysis if the study reported the DNA strand. If this information was missing in the study, the SNP was excluded from further analysis. In order to avoid biased estimates due to linkage disequilibrium (LD) in the gene pool analysis, a pruning exercise was carried out by LD for those studies that analyzed more than one polymorphism in the same gene. To this end, a pairwise comparison was carried out including only SNPs which were independent (r 2 < 0.3) of the others. For the SNPs in LD ! 0.3, the analysis included the one with the lowest p value for the association. Where the studies did not provide estimates of linkage disequilibrium, those retrieved from the 1,000 Genomes global population were considered as the benchmark. Thus, when the SNPs included in the meta-analysis (for the stratification of the genes) were extracted from the same study, they were only maintained in the analysis when the r 2 of equilibrium linkage was 0.30, depending on the investigated population. Considering the high degree of heterogeneity of observational studies (and the I 2 statistic observed across the studies), random models were tried out. The analyses were performed using Stata 16.0 software (StataCorp, College Station, TX, USA). To identify possible publication bias, the Egger's test and funnel-plot were assessed for allele, homozygote and heterozygote analysis. Accordingly, graph pooling by gene was also plotted.

Quality of evidence
The quality of evidence for the outcome (dental caries) was graded by two independent reviewers (LAC and RVC), according to allelic and genotype (homozygous and heterozygous) using the instrument developed by the Grading of Recommendations Assessment, Development and Evaluation (GRADE) working group of evidence, the GRADEpro GDT. The following aspects were considered: study design, risk of bias, inconsistency, indirectness, and imprecision.

Study selection
A total of 1,896 records were found in the initial search ( Figure 1). After exclusion of duplicates, 1,341 manuscripts remained for the screening of titles and abstracts in the digital library. Twenty-two full-text papers were assessed for eligibility, of which seven were excluded. One study was sourced from the gray literature. The studies and reasons for exclusion are shown in Table 1

Risk of bias within the studies
Regarding the quality assessment using the Critical Appraisal Checklist for observational studies (Joanna Briggs Institute), the majority of the studies (43.7%; n ¼ 7) were of low quality, followed by high quality studies (37.5%; n ¼ 6) (Table 2). Similarly, observing the methodological scoring protocol based on quality assessment for genetic studies, it was found that 50. 0% of the studies were classified as providing medium-quality evidence and 50.0% as low-quality evidence. The full results are displayed in Table 3.

Overview of single nucleotide polymorphisms
Forty-four single nucleotide polymorphisms, comprising four genes, were tested with regard to salivary composition and flow and their association with dental caries experience. Most of the SNPs were in intron regions (52.3%) and the most frequent functional impact was the protein coding (65.9%). Further information concerning the SNPs is available in Supplemental Material 3. Forty SNPs were included in the meta-analysis, without palindromic alleles. The SNPs rs142460367 and rs142460368 were not in agreement with the chromosome location reported in Table 1. Excluded studies and reasons for exclusion.

Qualitative analysis
Four different genes were found related to possible involvement with dental caries experience in adults and children: carbonic anhydrase 6 (CA6), aquaporin 5 (AQP5), aquaporin 2 (AQP2), and mucin 5B (MUC5B). A summary of the results of the studies, by gene and polymorphism, is displayed in Supplemental Material 4. CA6 SNPs were the most investigated among the included studies. CA6 rs2274327 (C/T) was investigated in nine studies (Peres et al. 2010 ]. In addition, an association between buffer capacity and the SNP CA6 rs2274327 (C/T) was observed. In a Brazilian population, the T Allele and TT genotype were less frequent in individuals with the highest buffer capacity (Peres et al. 2010). Similarly, in a Turkish population, the T allele and TT genotype of the same SNP were less frequent in individuals with the highest buffer capacity (Yarat et al. 2011).
The SNP CA6 rs17032907 (C/T) was investigated in four studies (Li et al. 2015;Esberg et al. 2019;Yildiz Telatar et al. 2020;Wu et al. 2022), being associated with an increase in caries risk in two studies (Li et al. 2015;Yildiz Telatar et al. 2020). When the TT genotype was compared to the CC, an increase in dental caries experience was observed in the Chinese population [OR ¼ 2.14 (1.10 À 4.20), although in the heterozygote analysis no association was observed (Li et al. 2015 CA6 rs10864376 (T/C) was assessed in three studies (Li et al. 2015;Esberg et al. 2019;Chisini 2020). While no association was found in the studies with Brazilian (Chisini 2020) and Chinese (Li et al. 2015) populations, an increase in caries experience was  Anjomshoaa et al. (2015) 2015 Li et al. (2015) 2015 Sengul et al. (2016) 2016 Zietek, et al. (2020) 2020 Wu et al. (2022) 2022 Yes; -No; /: Unclear; 1 ¼ Was the sample frame appropriate to address the target population? 2 ¼ Were study participants sampled in an appropriate way?; 3 ¼ Was the sample size adequate?; 4 ¼ Were the study subjects and the setting described in detail?; 5 ¼ Was the data analysis conducted with sufficient coverage of the identified sample?; 6 ¼ Were valid methods used for the identification of the condition?; 7 ¼ Was the condition measured in a standard, reliable way for all participants?; 8 ¼ Was there appropriate statistical analysis?; 9 ¼ Was the response rate adequate, and if not, was the low response rate managed appropriately? observed in individuals with the C allele in the Swedish population (Esberg et al. 2019). Moreover, the G allele of CA6 rs12021597 (A/G) [OR ¼ 1.90 (1. 10 À 3.40)], C allele of CA6 rs3737665 (T/C) [OR ¼ 2.30 (1.20 À 4.40)] and the C allele of CA6 rs12138897 (G/C) [OR ¼ 1.80 (1.10 À 2.90)] were risk factors for dental caries in the Swedish population (Esberg et al. 2019).
Regarding the AQP2 gene, both SNPs investigated were associated with dental caries experience. Anjomshoaa et al. (2015), in a multicentric study, found that AQP2 rs467323 (A/C) and AQP2 rs10875989 (T/C) were associated with dental caries (USA, Turkey, Argentina, Brazil). Considering the AQP2 rs10875989 (T/C), associations with caries were only observed in the recessive model in the same populations. However, Chisini (2020)  Of the four SNPs assessed in the AQP5 gene, three were associated, in at least one study, with dental caries. In a study carried out with children in the USA (4 to 7 years old), the AQP5 rs1996315 (G/A) exhibited an association with a decrease in dental caries . In this study the sample comprised Caucasian (95%), Afro-descendants (2%) and other racial/ethnic groups (3%). Two other studies that investigated the same association in Chinese populations did not find any association Wu et al. 2022).
The CC homozygous genotype of AQP5 rs923911 (C/A) was associated with an increase in dental caries in a Chinese population [OR ¼ 2.95 (1.56 À 5.59)], although the heterozygote did not exhibit any association [OR ¼ 0.77 (0.58 À 1.03)] . The same SNP was investigated in a USA population but no associations were observed . Similarly, AQP5 rs3759129 (A/C) was positively associated with dental caries experience in a multicentric study carried out in the USA, Turkey, Argentina and Brazil. This study included a large sample size (n ¼ 1,383) with populations of different ages (Anjomshoaa et al. 2015). No associations were observed for AQP5 rs3759129 (A/C) in the Brazilian (Chisini 2020) and Chinese (Wu et al. 2022) populations in other studies.  performed haplotype analysis by conducting associations between dental caries experience and haplotypes in two SNPs within the same gene. Thus, when the C allele of rs923911 Table 3. Methodological scoring protocol based on quality assessment for genetic studies. (AQP5) was combined with the A or G allele of rs1996315 (AQP5), both haplotypes (CA and CG) exhibited a protective effect against caries for all forms of caries (USA population) ). Moreover, Generalized Multifactor Dimensionality Reduction approaches (GMDR) found epistatic interactions between AQP2 rs2274333 (A/G) and AQP5 rs3759129 (A/C) (Chisini 2020). Regarding MUC5B, five SNPs were assessed in two studies (Cavallari et al. 2018;Chisini 2020) in the Brazilian population. However, the studies presented contrasting results. Although Chisini (2020) did not find any associations, Cavallari et al. (2018) found that the CT genotype [OR ¼ 2.14 (1.11 À 4.12)] and TT genotype [OR ¼ 6.69 (2.79 À 16.02)] were associated with an increase in dental caries experience in the MUC5B rs2735733 (C/T

Synthesis of results (meta-analysis)
Ten studies were included in the meta-analysis. A summary of individual and meta-analysis results (at least two studies evaluating the same SNP), according to allele and genotype models, is displayed in Table 4. To perform the analysis pooled by gene, the results of single articles were considered and polymorphisms in linkage disequilibrium were excluded. Supplemental Material 5 presents the data of SNPs in linkage disequilibrium and the SNPs excluded from the metaanalysis. Thus, for gene evaluation, only 15 SNPs remained after exclusion of SNPs in linkage disequilibrium.
Turning to the CA6 gene, five SNPs were evaluated in the meta-analysis. The TT homozygous genotype of SNP CA6 rs17032907 (C/T) was associated with an increased likelihood of 3.23 in dental caries experience [OR ¼ 3.23 (1.39 À 7.49)]. Also, although the summarization of the allele analysis of CA6 did obtain an association [OR ¼ 1.18 (1.06 À 1.31)], this association disappeared when SNPs with linkage disequilibrium were excluded [OR ¼ 0.86 (0.72 À 1.07)]. No other association was observed. The summarization of the APQ5 gene analysis found an association after evaluation of linkage disequilibrium. Thus, the pool of effect SNPs assessed in AQP5 was associated with a 25% reduction in the likelihood of caries experience [OR ¼ 0.75 (0.59 À 0.95)]. No MUC5B SNPs remained associated in meta-analysis and the results of the gene pool show any association when SNPs in linkage disequilibrium were excluded from the analysis. However, overall, considering the grouping of all the SNPs related to saliva composition and flow, it was observed that, in the homozygous genotype, the effect allele was associated with a 75% increase in the likelihood of dental caries [OR ¼ 1.75 (1.06 À 2.89)].

Risk of bias across studies and quality of evidence
Funnel plot results showed significant publication bias across the studies in the allelic and genotype heterozygote model. The Egger's test confirmed these observations (allele [p ¼ 0.015] and genotype-homozygote [p ¼ 0.246] and heterozygote [p ¼ 0.049]-analysis) ( Figure 2). Regarding the quality of evidence assessed by GRADE (Table 5), very low evidence was considered for the outcome of dental caries when allelic or genotype (homozygous or heterozygous) expositions were investigated.

Discussion
Our systematic review identified 44 different polymorphisms present in four different genes (CA6, AQP2, AQP5, and MUC5B) which are related to caries experience in children and adults. These genes may influence the salivary flow, buffering capacity, biofilm microbiota composition, and bacterial attachment, which are directly related to the individual's caries-risk assessment. Although some SNPs did present an association with caries experience in the original studies, most of the associations disappeared when these SNPs were included in the meta-analysis. Only the SNP CA6 rs17032907 (C/T) in heterozygosity remained associated with caries experience in the meta-analysis. Moreover, when all effect SNPs were pooled in the global meta-analysis, excluding SNPs in linkage disequilibrium, an association with caries experience was observed in the homozygote model, while none was observed in the heterozygote model.
Most of the SNPs investigated in this review were related to the Carbonic Anhydrase 6 (CA6) gene. CA6 polymorphisms were associated with Streptococcus mutans colonization and aciduric microbiota profile, leading to changes in dental caries risk in a Swedish sample (Esberg et al. 2019). The protein encoded by this gene is an isozyme of carbonic anhydrase, which is only found in salivary glands and saliva. In addition, it plays an important role in the reversible hydration of carbon dioxide, although its function in saliva is still not well established (Li et al. 2015). A decrease in the buffering capacity of saliva in healthy children has been shown to be related to SNP CA6 rs2274327 (C/T), which is a missense variant and can change the protein coding. The T allele and TT genotype of this SNP were less frequent in individuals with the highest buffer capacity in a Brazilian sample (Peres et al. 2010). However, significant variations in caries experience were not observed in this population (Peres et al. 2010). In a sample of Turkish students, the pH and buffering capacity of the saliva were evaluated, and an association between buffer capacity and the CA6 rs2274327 (C/T) was also observed. Similarly, the T allele and TT genotype were less Table 4. Summarization of results (meta-analysis and individuals) according by allelic and genotype (homozygous and heterozygous) analysis, pooled by gene.  Aquaporin 5 (AQP5) is an integral membrane protein responsible for transporting water, codified by the respective gene (AQP5) and is localized in the 12q13 region. In particular, the AQP5 water channel protein plays a role in the generation of tears, pulmonary secretions and saliva, expressed in the apical  membranes of serous acinar cells, in both the salivary and lacrimal glands (Funaki et al. 1998). In this context, the hypothesis is that some influence caused by the SNPs of this gene may change the natural salivary composition or flow and upset the homeostasis. In fact, initial observations in models using mice, which targeted deletion of the gene encoding AQP5, exhibited a reduction in saliva flow and, hence, an increase in dental caries (Culp et al. 2005). Similar studies observed that water permeability to determine the flow and ionic composition of mice saliva are controlled mainly through the AQP5 gene (Krane et al. 2001). However, few studies have investigated the effect of related SNPs in dental caries in populational studies. The meta-analysis found an association between AQP5-SNPs and dental caries in a heterozygote model. Similarly, aquaporin 2 (AQP2) is also a gene that encodes water channel protein being localized in the 12q13 region, very close to AQP5. Both SNPs [AQP2 rs467323 (A/C) and AQP2 rs10875989 (C/T)] from AQP2 related-genes were associated with dental caries experience in a multicentric study (Anjomshoaa et al. 2015) and AQP2 rs10875989 (T/ C) in a Brazilian population (Chisini 2020). AQP2 rs467323 (A/C) and AQP2 rs10875989 (C/T) are an alteration in a three-prime untranslated region (3 0 -UTR), which contains regulatory regions that posttranscriptionally influence gene expression, thereby explaining the present results.
MUC5B encodes a member of the mucin family of proteins, which are highly glycosylated macromolecular components of mucus secretions. This family member is the major gel-forming mucin in mucus. Therefore, it is a primary contributor to the lubricating and viscoelastic properties of whole saliva (Dijkema et al. 2012). MUC5B does not seem to change the Streptococcus mutans growth. However, MUC5B can reduce their attachment and biofilm formation on the tooth surface (Frenkel and Ribbeck 2015). Mucins encoded by MUC5B in a sucrose-supplemented medium inhibited the biofilm formation on the dental surface due to the maintenance of bacteria in the planktonic state (Frenkel and Ribbeck 2015). In the present review, only two studies (Cavallari et al. 2018;Chisini 2020) investigated MUC5B-related SNPs, both of which were in Brazilian populations, but which produced contrasting results. While Cavallari et al. (2018) found a high number of SNPs associated with dental caries experience, Chisini (2020) did not find any SNP association. Thus, in the meta-analysis, no associations were observed.
Although specific SNPs related to salivary composition and flow have shown potential associations with dental caries experience in children and adults, a wide variation was observed in the methodologies employed in the SNPs investigated. Moreover, most SNPs were only evaluated in a few studies. Limited reports of power calculations in the included studies should be interpreted with caution; the sample size of the studies ranged from 43 (Koç € Ozt€ urk et al. 2012) to 1,383 (Anjomshoaa et al. 2015) individuals, and the lack of reporting on sample size calculation may lead to a decrease in power, since non-significant results could be just a problem with the sample and not a lack of association. This could lead us to false-negative type inferential errors. Similarly, few studies Cavallari et al. 2018;Chisini 2020) reported the ethnicity/ancestry information of populations. Considering all SNPs investigated in the present systematic review, significant variances between allele frequencies and sample ethnicity have been detected when reported SNPs were explored in a complementary database. The need to carry out an adjustment for ancestry information should be emphasized because genetic effect sizes can also vary among populations, at least for some traits. In addition, the funnel plot and Egger's test identified publication bias among studies included in allelic and heterozygotic models. These limitations reflected the low-quality studies evaluated by two instruments, also contributing to the very low level of evidence obtained by GRADE. Thus, these pieces of information must be considered in interpretating the results.
Moreover, it is important to highlight that only gene candidate studies were included in this systematic review and meta-analysis, and this must also be considered in the interpretation of the results. Other as yet unknown and unstudied pathways related to salivary flow and composition may influence dental caries experience. Consequently, we would reinforce the importance of conducting more gene candidate studies, with meticulous, methodological rigor and quality, as well as studies of genomic scales. The number of genomic-wide association studies evaluating caries experience has grown in recent years. However, no SNP identified in the present review with a potential association with caries experience was associated in GWAS studies available in the literature Shaffer et al. 2013;Zeng et al. 2013;Morrison et al. 2016;Govil et al. 2018;Haworth et al. 2018;Orlova et al. 2019;Shungin et al. 2019;Alotaibi et al. 2021;Olatosi et al. 2021).
Despite the limitations of our study, some strengths should be highlighted. A broad review was performed, with several control filters (for palindromic SNPs and linkage disequilibrium), aimed at reducing possible bias in our estimates. Moreover, the reference genotype was standardized and all SNPs in linkage disequilibrium were excluded from the analysis, thereby according robustness to our results. We would stress that this topic is extremely new and further studies with different populations are needed to provide more robust estimates. Further studies should include representative samples of the target populations with sample calculations and also address different ethnic samples. Collaborations between research groups could be an interesting strategy aimed at combining databases to increase sample sizes and/or replicate the results. Genome-wide association studies are important strategies and an alternative approach to help with the identification of new SNPs and pathways associated with caries experience, also essential as a basis for understanding the polygenic trait and genetic architecture of this phenotype.
Even considering the increase in the number of studies in the last decade, the clinical application of genetic approaches to caries disease is still not usual, in part due to the limited evidence available, the polygenic nature of caries genetics, and the high costs of the approach. Understanding which SNPs or mutations may be involved in individuals' susceptibility to caries could support the development of a viable approach to identify risk groups. A better understanding of these complex mechanisms may increase the potential for clinical translation. For diseases where the impact of genetic mutations are well known, such as mutations on the BRCA1/2 gene for breast cancer, genetic investigations have been recommended for individuals with familial risk. However, the actual impact of genetic screening in reducing the incidence of cancer has not yet been investigated (Nelson et al. 2019). In this context, when we better understand the genetic influence of caries disease, it may be possible to include such genes within the wide genetic assessments for a high number of diseases. Thus, reducing the costs of such approaches and enabling the development of measures aimed at preventing different diseases.
In conclusion, the results showed that most of the SNPs related to genes in salivary composition and flow were not associated with dental caries. However, when the effect of all SNPs was pooled, a 75% increase in the likelihood of caries experience was observed in the homozygous genotype. The meta-analysis also showed that the SNP CA6 rs17032907 (C/T) in heterozygosity was associated with dental caries experience. Studies with a higher quality of reporting and a better methodological approach should be performed to support and confirm the present results. Hence, interpretations must be treated with caution. The main limitations of the present study are related to the designs and methodologies of the included studies, which varied widely. Most studies did not consider ethnicity of the evaluated population, and only gene candidate studies were included in this systematic review.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This study was conducted in a Graduate Program supported by CAPES, Brazil.