Salivary Proteins as Dental Caries Biomarkers: A Systematic Review

Dental caries (further referred to as caries) is a highly prevalent chronic infectious oral disease in humans, afflicting a large proportion of the world’s population [Selwitz et al., 2007]. It is a multifactorial, sugar-driven, biofilm-mediated dynamic disease that is characterized by the phasic demineralization and remineralization of tooth hard tissues [Arshad et al., 2020]. Caries is a result of imbalance cariogenic microorganisms present in the oral biofilm which ferments dietary carbohydrates to generate acid, causing loss of minerals from dental hard tissues and consequently the destruction of tooth structure [Gao et al., 2016]. The treatment costs and complications of caries inflict a heavy toll on people and the community as a whole [Petersen et al., 2005; Moussa et al., 2022]. The interaction between host susceptibility, diet, and microorganisms determines whether caries will develop [Featherstone, 2003; Selwitz et al., 2007]. Since teeth are continuously bathed in saliva, the characteristics and constituents of the saliva play a vital part in the development and progression of caries [Siqueira et al., 2012]. Saliva is considered to be one of the most essential host factors and an important regulator mediating the direction and speed of the cariogenic pathway [Fejerskov et al., 2015].

Recently, there has been an increasing interest in the utility of saliva as a diagnostic fluid due to the simplicity and noninvasiveness of sample collection [Piekoszewska-Ziętek et al., 2019]. Salivary proteins are successfully utilized to monitor diseases affecting the oral cavity, adjacent tissues as well as the entire body such as endocrinological or autoimmunological, oncological, or infectious diseases [Castagnola et al., 2011]. Salivary proteins contribute to the maintenance of tooth integrity and caries prevention via many mechanisms: (a) generation of acquired enamel pellicle (AEP) for continuous protection against tooth wear; (b) inhibition of exposed tooth surfaces’ demineralization; and (c) enhancement of enamel remineralization via the attraction of calcium ions; (d) antimicrobial properties such as the secretion of antimicrobial peptides (AMPs), accumulation and clearance of microorganisms from the oral cavity, and inhibition of cariogenic species attachment to the enamel surface [Wang et al., 2019].

Numerous salivary proteins play a defensive role in the oral cavity including carbonic anhydrase, lactoferrin, immunoglobulins (Igs), proline-rich proteins (PRPs), and mucins [Kivelä et al., 1999; Van Nieuw Amerongen et al., 2004; Tao et al., 2005]. Mucins are the most frequently occurring glycoproteins of unstimulated saliva containing around 20–30% of whole salivary proteins. Two classes of mucins exist: (1) MG-1 (MUC5) and (2) MG-2 (MUC7). A hike in salivary MG-1 is noticed in individuals with high intensity of caries, whereas MG-2 is mostly found in patients with low intensity of caries [Bennadi et al., 2014]. MG-1 is responsible for increasing salivary viscosity, whereas MG-2 contributes to the agglutination of microorganisms, such as cariogenic bacteria [Bennadi et al., 2014]. A parallel function is designated to PRPs [Renuka et al., 2013]. Lactoferrin is associated with the inhibition of biofilm production and bacterial growth by attaching and chelating iron ions [Van Nieuw Amerongen et al., 2004]. Carbonic anhydrase VI (CAVI) has characteristics essential for oral health owing to its impact on dental erosion and caries formation. CAVI possesses the capability of infiltrating into dental plaque and neutralizing the bacterial acids bicarbonate buffer system [Kimoto et al., 2004]. Cathelicidins (also known as histatins [HTNs]) and defensin are the category of salivary AMPs that have a synergistic impact and carry out a high activity against gram-negative or gram-positive bacteria, together with fungal organisms (primarily Candida albicans) [Tao et al., 2005]. High salivary α-defensin HNP-1-3 levels are found in children without caries [Renuka et al., 2013]. However, it is believed that the gene for β-defensins is associated both with low and high caries risk [Renuka et al., 2013]. HTNs possess antibacterial features, and they destabilize bacterial cell membrane which causes the cell to disintegrate and facilitate wound healing [Jurczak et al., 2015].

Considering the role of salivary proteins in the pathophysiology of caries, a boom of relevant studies over the past decade has given rise to controversy and irreproducible or conflicting findings, primarily because of variations in methodological designs, statistical analysis, and interpretation of study results. Some studies have reported an association between salivary protein levels and caries [Piekoszewska-Ziętek et al., 2019], whereas some reports have not found any such relationship [Martins et al., 2013]. Performing a systematic review for evaluating the available evidence in salivary proteomic studies would allow future improvement in this rapidly advancing discipline and might aid in the development of saliva as a diagnostic tool. Hence, the purpose of this systematic review was to determine if there is an association between caries and salivary proteins by comparing individuals with and without caries experience. Thereby, assessing the possibility of whether salivary proteins can be regarded as biomarkers for caries. The present systematic review was conducted to update our previous systematic review published in 2013 [Martins et al., 2013] that used the same research question.

The present update of a systematic review [Martins et al., 2013] was performed following the guidance of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklists (https://www.prismastatement.org) [Hutton et al., 2015] and was registered at the National Institute for Health Research PROSPERO, International Prospective Register of Systematic Reviews (https://www.crd.york.ac.uk/PROSPERO, registration number: CRD42021289462).

Is there an association between caries and salivary concentration of proteins for determining whether salivary proteins can be used as biomarkers for caries?

The inclusion criteria comprised observational studies (i.e., cohort, cross-sectional, and case-control studies) published between July 2012 and March 2022, without any language restriction, comparing the concentration of salivary protein(s) in systematically healthy individuals not consuming any medication that may influence the composition of saliva, with (DMF > 0) or without (DMF = 0) caries experience with mixed or permanent dentition. Exclusion criteria comprised (1) studies that did not use any index for assessing caries and (2) studies that assessed root caries or early childhood caries (children aged <72 months).

An electronic search of the literature was performed to update our previous systematic review published in 2013 that used the same research question [Martins et al., 2013], using different electronic databases including PubMed (National Library of Medicine), EMBASE, Clarivate Analytics’ Web of Science, and Elsevier’s Scopus. The search was carried out between July 2012 and March 2022 since the present review is an update of our previous systematic review [Martins et al., 2013]. We updated the literature search protocol from the original systematic review including additional databases, without language restriction [Martins et al., 2013]. The following Medical Subject Headings terms (https://www.nlm.nih.gov/mesh/meshhome.html) were employed in the search. We used a combination of Medical Subject Headings and free terms (“dental caries,” “saliva,” “peptides,” and “proteins”), including the primary salivary proteins mentioned in the literature [Levine, 1993]. The complete search strategy is described in online supplementary Table 1 (for all online suppl. material, see www.karger.com/doi/10.1159/000526942). A supplementary search in the gray literature was undertaken including OpenGrey (https://www.opengrey.eu) and Google Scholar. In addition, we examined the reference list from retrieved full-text articles and used existing review articles to identify additional studies. EndNote X7 (Thomson Reuters) was employed for collecting references and removing duplicates.

Study selection was performed by two independent investigators (P.A. and A.H.) in the following phases:

• Initial screening of citations using titles and abstracts against the inclusion criteria for identifying potentially suitable publications selected by at least one investigator.

• Screening of potentially relevant citations using full-text articles.

Any disagreement regarding the eligibility of citations was discussed and resolved by reaching a consensus. The agreement level between the investigators was measured using kappa statistics for the 2nd phase of screening.

Two investigators extracted data from the eligible studies (P.A. and A.H.), and a third researcher cross-checked and confirmed data accuracy (W.S.). Any disagreement was resolved through discussion until reaching a consensus among the three investigators. When data were incomplete or poorly reported, we contacted the corresponding author of the eligible article to retrieve the missing information. The following data were extracted from the included studies: (a) study population (country); (b) age of subjects; (c) salivary sample size; (d) caries index; (e) type of saliva; (f) saliva collection time; (g) saliva amount; (h) method for assessing concentration of salivary protein(s); (i) salivary protein assessed; (j) levels of salivary protein(s); (k) statistical significance; (l) main findings; and (m) conclusion.

Two investigators (P.A. and A.C.-L.) evaluated the quality of studies (cohort or case-control studies) utilizing the Newcastle-Ottawa Quality Assessment Scale (NOS) [Stang, 2010]. The NOS is a validated tool to conveniently assess the risk of bias (RoB) in nonrandomized and observational studies. The NOS employs a star system to evaluate the studies from 3 broad perspectives: (a) the selection of the study group (four stars); (b) the comparability of the groups (two stars); and (c) the ascertainment of either exposure or outcome of interest (three stars) [Stang, 2010].

In total, 428 studies were initially identified using different databases. After title and abstract screening and removing duplicates, 52 articles were selected for full-text examination and 22 studies proved eligible (Kappa = 0.95%). Figure 1 depicts the PRISMA flowchart of the selection procedure and excluded publications with reasons.

Table 1 depicts the primary characteristics of the included 22 studies [Golpasand Hagh et al., 2013; Pal et al., 2013; Priya et al., 2013; Gabryel-Porowska et al., 2014; Gornowicz et al., 2014; Yang et al., 2015; Castro et al., 2016; Yassin, 2016; Jagadesh Babu et al., 2017; Picco et al., 2017; Sitaru et al., 2017; Haeri-Araghi et al., 2018; Monea et al., 2018; Pyati et al., 2018; Szkaradkiewicz-Karpinska et al., 2018; Nawaz et al., 2019; Soesilawati et al., 2019; Al-ani et al., 2020; Razi et al., 2020; Stojković et al., 2020; Angarita-Díaz et al., 2021; Khan et al., 2021]. Twelve studies were conducted in Asia [Golpasand Hagh et al., 2013; Pal et al., 2013; Priya et al., 2013; Yang et al., 2015; Yassin, 2016; Jagadesh Babu et al., 2017; Haeri-Araghi et al., 2018; Pyati et al., 2018; Nawaz et al., 2019; Soesilawati et al., 2019; Razi et al., 2020; Khan et al., 2021], followed by Europe (n = 6) [Gabryel-Porowska et al., 2014; Gornowicz et al., 2014; Sitaru et al., 2017; Monea et al., 2018; Szkaradkiewicz-Karpinska et al., 2018; Stojković et al., 2020], South America (n = 3) [Castro et al., 2016; Picco et al., 2017; Angarita-Díaz et al., 2021], and North America (n = 1) [Al-ani et al., 2020]. A total of 1,551 subjects were recruited in the included 22 studies, of which 574 were males and 547 were females, while 8 studies did not report male-to-female distribution [Pal et al., 2013; Priya et al., 2013; Gabryel-Porowska et al., 2014; Gornowicz et al., 2014; Jagadesh Babu et al., 2017; Picco et al., 2017; Sitaru et al., 2017; Khan et al., 2021]. Approximately half of the participants (n = 848, 54.7%) were diagnosed with carious lesions, whereas 703 (45.3%) individuals were diagnosed as CF, with ages ranging between 6 and 70 years. All included studies evaluated unstimulated saliva, with two exceptions, which assessed stimulated saliva [Picco et al., 2017; Soesilawati et al., 2019].

The caries index used in all included studies was DMF (DMFT/DMFS), except for 5 studies that employed the ICDAS [Haeri-Araghi et al., 2018; Angarita-Díaz et al., 2021; Khan et al., 2021], CAMBRA [Al-ani et al., 2020], and visual detection method [Monea et al., 2018]. Six studies considered DMFT ≥ 5 as a high risk of caries [Golpasand Hagh et al., 2013; Yang et al., 2015; Yassin, 2016; Picco et al., 2017; Nawaz et al., 2019; Stojković et al., 2020], one study considered DMFT = 3–5 as high risk of caries [Sitaru et al., 2017], whereas four articles considered DMFT >10 as high risk of caries [Gabryel-Porowska et al., 2014; Gornowicz et al., 2014; Szkaradkiewicz-Karpinska et al., 2018; Razi et al., 2020]. On the other hand, twelve studies considered DMFT = 0 as a low risk of caries [Golpasand Hagh et al., 2013; Pal et al., 2013; Priya et al., 2013; Castro et al., 2016; Yassin, 2016; Jagadesh Babu et al., 2017; Picco et al., 2017; Pyati et al., 2018; Nawaz et al., 2019; Soesilawati et al., 2019; Razi et al., 2020; Stojković et al., 2020], while five studies considered DMFT between 0 and 5 as a low risk of caries [Gabryel-Porowska et al., 2014; Gornowicz et al., 2014; Yang et al., 2015; Sitaru et al., 2017; Szkaradkiewicz-Karpinska et al., 2018. This finding suggests a crucial methodological variability source in outcomes across varying articles.

A variety of salivary proteins were assessed in the included studies including (1) α-amylase; (2) APRP-1; (3) CAVI; (4) fibronectin; (5) hBD-2; (6) HTN-5; (7) HNP-1; (8) IgA; (9) IgG; (10) lactoperoxidase; (11) cathelicidin LL-37; (12) MUC1; (13) MUC5B; (14) MUC7; (15) proteinase-3; (16) statherin; and (17) total proteins (Table 2). A statistically significant increase in the salivary levels of α-amylase [Sitaru et al., 2017; Monea et al., 2018], APRP-1 [Szkaradkiewicz-Karpinska et al., 2018], HTN-5 [Gornowicz et al., 2014], lactoperoxidase [Gornowicz et al., 2014], and MUC1 [Gabryel-Porowska et al., 2014] was found in CA patients, while the salivary levels of CAVI [Picco et al., 2017], proteinase-3 [Yang et al., 2015], and statherin [Angarita-Díaz et al., 2021] were observed to be significantly increased in CF subjects. Whereas a statistically nonsignificant association was observed between the salivary levels of 7 proteins, including cathelicidin LL-37 [Stojković et al., 2020; Angarita-Díaz et al., 2021], fibronectin [Stojković et al., 2020], APRP-1 [Szkaradkiewicz-Karpinska et al., 2018], HNP-1 [Stojković et al., 2020], hBD-2 [Stojković et al., 2020], MUC5B [Gabryel-Porowska et al., 2014], and MUC7 [Gabryel-Porowska et al., 2014], among individuals with and without caries. Conflicting results were found regarding the salivary levels of Igs (IgA and IgG) and total proteins among CA and CF subjects. Five of the included studies reported a statistically significant increase in the salivary levels of IgA in CA patients [Priya et al., 2013; Gornowicz et al., 2014; Haeri-Araghi et al., 2018; Nawaz et al., 2019; Khan et al., 2021], while seven studies reported a statistically significant increase of IgA in CF subjects [Golpasand Hagh et al., 2013; Pal et al., 2013; Castro et al., 2016; Yassin, 2016; Jagadesh Babu et al., 2017; Soesilawati et al., 2019; Razi et al., 2020]. However, two studies reported a nonsignificant association of IgA between CF and CA subjects [Al-ani et al., 2020; Angarita-Díaz et al., 2021]. Regarding total proteins level in saliva, two studies reported a statistically significant increase in CA patients [Pyati et al., 2018; Razi et al., 2020], while one study reported a statistically significant increase in CF subjects [Castro et al., 2016]. Similarly, comparable salivary concentrations of HNP-1 [Stojković et al., 2020], hBD-2 [Stojković et al., 2020], IgA [Al-ani et al., 2020], IgG [Razi et al., 2020], cathelicidin LL-37 [Stojković et al., 2020], MUC5B [Gabryel-Porowska et al., 2014], MUC7 [Gabryel-Porowska et al., 2014], and APRP-1 [Al-ani et al., 2020] were found between individuals with and without caries. Interestingly, a discrepancy was found in one of the included studies [Stojković et al., 2020], in which the authors reported the incorrect odds ratio of hBD-2 (i.e., odds ration: 9,169,167 [95% confidence interval: 0.000–47.231]). The first author of the present review (P.A.) tried to contact the corresponding author of the included study, but the attempt was unsuccessful. Owing to significant heterogeneity in relation to study protocols and reported outcomes among the included studies, a meta-analysis could not be performed.

The included studies were classified into high, moderate, and low RoB. Of the 22 included publications, 15 were categorized as low RoB [Golpasand Hagh et al., 2013; Priya et al., 2013; Castro et al., 2016; Yassin, 2016; Jagadesh Babu et al., 2017; Picco et al., 2017; Haeri-Araghi et al., 2018; Monea et al., 2018; Pyati et al., 2018; Nawaz et al., 2019; Soesilawati et al., 2019; Al-ani et al., 2020; Razi et al., 2020; Stojković et al., 2020; Khan et al., 2021], 4 as medium RoB [Pal et al., 2013; Sitaru et al., 2017; Szkaradkiewicz-Karpinska et al., 2018; Angarita-Díaz et al., 2021], and 3 as high RoB [Gabryel-Porowska et al., 2014; Gornowicz et al., 2014; Yang et al., 2015] (Table 3). The RoB as evaluated by the NOS varied significantly across the included articles, ranging between 4/9 and 8/9.

This systematic review aimed to determine the association between the presence of caries and salivary proteins/total protein concentration by comparing individuals with and without caries experience. Hence, evaluating the possibility of whether salivary proteins can be considered biomarkers for caries. Our findings show a statistically significant association between the salivary concentrations of nearly half (57%) of the proteins assessed in the included studies. The upregulation of 5 salivary proteins was found in CA, whereas 3 salivary proteins were upregulated in CF subjects. Conflicting results were found regarding the salivary levels of Igs (IgA and IgG) and total proteins among CA and CF subjects. Nonetheless, of the 22 included articles, the majority revealed evidence of an association between caries and concentrations of salivary proteins, which is in opposition to a similar systematic review that failed to find evidence of an association [Martins et al., 2013]. This might be due to a greater number of studies (n = 22) and assessed individual salivary proteins (n = 13) included in the present systematic review, which found an association between caries susceptibility and salivary proteins levels than the number of included studies (n = 7) and assessed individual salivary protein (n = 7) in our previous systematic review [Martins et al., 2013] because of growing research interest in this domain over the past 10 years. In addition, a statistically significant association between the salivary levels of total proteins among subjects with and without caries and found high evidence that the total protein concentration increases with caries experience. Due to significant heterogeneity in the type of protein, cutoff values, and reported outcomes among the included studies, a meta-analysis could not be performed.

Immunoglobulins play a critical part in the immune system mechanism since they are a constituent of humoral immunity and permit bacterial neutralization, which is vital in caries development process [Colombo et al., 2016]. The findings of the present review revealed that the salivary IgA was upregulated among CF individuals in 8 of the included studies. This outcome is in agreement with the previous studies by Doifode and Damle [2011] and Kuriaskose et al. [2013], which reported higher salivary levels of IgA in CF children. Contrarily, 5 included studies reported an upregulation of the same salivary protein in CA patients. This outcome is consistent with other studies conducted by Ranadheer et al. [2011], Al Amoudi et al. [2007], and Bagherian et al. [2008], which reported considerably greater levels of salivary IgA in CA patients as compared to CF subjects [Colombo et al., 2016]. The published literature suggests that CF individuals have naturally higher levels of salivary IgA, whereas a reduced concentration of salivary IgA leads to caries susceptibility [Rose et al., 1994; Bhatia et al., 1996]. Contrarily, Watanabe and colleagues [1997] failed to affirm any association between caries and salivary IgA and concluded that salivary flow might be more important than its constituents. Although Ig can contribute to controlling the development of caries, the number of reports regarding this topic is still unequivocal.

Three included studies were regarding the salivary levels of AMPs [Gornowicz et al., 2014; Stojković et al., 2020; Angarita-Díaz et al., 2021]. This class comprises peptides that possess the ability to inhibit or kill bacterial growth (i.e., HTNs, cathelicidins, defensins) [Colombo et al., 2016b]. One included study reported upregulation of cathelicidin LL-37 in CF individuals [Angarita-Díaz et al., 2021], while another study revealed comparable levels of cathelicidin LL-37 in individuals with and without caries [Stojković et al., 2020]. Similarly, comparable levels of salivary HNP-1 and hBD-2 in CF and CA subjects were revealed [Stojković et al., 2020]; however, increased levels of HTN-5 were found in CA patients in the included studies [Gornowicz et al., 2014]. The literature suggests increased salivary levels of HNP-1 [Malcolm et al., 2014], cathelicidin LL-37 [Davidopoulou et al., 2012], HTN-5 [Jurczak et al., 2015], and hBD-2 [Jurczak et al., 2015] in CA patients. Such outcomes were not found by Toomarian et al. [2011] and Colombo et al. [2016b]. Contrary to the findings of the present review, Tao et al. [2005] found significantly increased levels of salivary α-defensin in CF subjects. Hence, in the light of the presented evidence, the monitoring of salivary levels of AMPs might be useful in the risk assessment of caries. Nonetheless, future longitudinal investigations in this domain are necessary for evaluating this dependence in varying dentition types.

Four included studies in this review were concerning salivary enzymes (i.e., amylase, proteinase-3, and CAVI) [Yang et al., 2015; Picco et al., 2017; Sitaru et al., 2017; Monea et al., 2018]. Picco et al. [2017] found increased levels of salivary CAVI in CF subjects, which is in agreement with the outcome of a study performed by Makawi et al. [2017]. This might provide evidence of the protective role of CAVI via stimulating oral buffers, accelerating acid neutralization, and regulating salivary pH [Leinonen et al., 1999]. In this review, levels of salivary α-amylase were found to be increased in CA subjects than in CF individuals [Sitaru et al., 2017; Monea et al., 2018]. This finding is in opposition to the results of a study conducted by Borghi et al. [2017], which found raised levels of salivary α-amylase in CF subjects (children aged 2–4 years) than in subjects with early childhood caries, suggesting a negative association between the salivary enzyme and caries. Contrarily, Vitorino et al. [2006] found a positive association between the levels of salivary amylase and the number of DMF among male adult subjects. Amylase binds to oral bacteria present in dental plaque, whereas free salivary amylase might help hydrolysis of dietary starch for providing supplementary low-molecular weight carbohydrates for metabolism via dental plaque bacteria. Later, the consequent acid formation might be incorporated into the pool of acid in plaque for accelerating tooth demineralization and further progression of caries [Arya and Taneja, 2015]. To date, however, no conclusive evidence is available regarding the association between caries susceptibility and salivary amylase levels.

One of the included studies in the present review was pertaining to major salivary glycoproteins (i.e., mucins) [Gabryel-Porowska et al., 2014], which reported increased salivary levels of MUC1 in high-intensity caries subjects (DMF > 11) as compared to low DMF individuals. Comparable levels of MUC5B and MUC1 in saliva were observed in CA and CF subjects. MUC1 has been demonstrated to be expressed by minor and major salivary glands, together with oral epithelial cells [Offner and Troxler, 2000; Sengupta et al., 2001]. Previous reports have shown that MUC1 is a vital component of the mucosal barrier to infection [Kardon et al., 1999; DeSouza et al., 2000; McAuley et al., 2007], and its upregulation proceeds infection with a bacterial pathogen [McAuley et al., 2007]. It has also been indicated that the protective effects of MUC1 might be particularly vital in oral cavity, where epithelial surfaces are continuously exposed to an array of both commensal and pathogenic microbes [Li et al., 2003; Pramanik et al., 2010]. In fact, MUC1 upregulation in oral epithelial cells has been found to cause from Porphyromonas gingivalis infection or increase in proinflammatory cytokines including tumor necrosis factor-alpha, interleukin-1B, and interleukin-6 [Li et al., 2003]. Hence, it is postulated that the expression of MUC1 in the oral mucosal epithelial cells is influenced by alterations in the salivary levels of proinflammatory cytokines.

In the present review, levels of salivary statherin were found to be increased in CF subjects than in CA patients [Angarita-Díaz et al., 2021]. These findings are in agreement with the outcomes of a study conducted by Vitorino et al. [2005], which reported a strong association between increased concentration of salivary statherin in CF subjects. Moreover, Wang et al. [2018], via iTRAQ analysis, revealed that salivary levels of statherin were significantly raised in CF individuals. These outcomes suggest a protective role of statherin against caries. This might be justified by the fact that statherin is known as a potential precursor of the AEP owing to its strong affinity to hydroxyapatite. It binds calcium and maintains the supersaturated saliva via inhibition of spontaneous precipitation of phosphate and calcium salts, which protects the tooth integrity by promoting enamel remineralization [Humphrey and Williamson, 2001]. Moreover, negative charges in the N-terminal domain of statherin might inhibit S. mutans adsorption onto the hydroxyapatite surface [Shimotoyodome et al., 2007].

In the present review, levels of salivary APRP-1/2 were found to be increased in CA subjects than in CF subjects [Szkaradkiewicz-Karpinska et al., 2018]. These outcomes are in opposition to the findings of studies conducted by Banderas-Tarabay et al. [2002], Vitorino et al. [2006], and Wang et al. [2018], which found significantly increased salivary levels of PRPs in CF individuals than CA subjects. The APRPs have a 30-amino acid N-terminal domain rich in glutamate and aspartate with some serine phosphate residues, which contribute to the AEP formation and act as salivary receptors for many dental plaque-forming bacteria [Amano et al., 1994]. Additionally, APRPs have a vital role in the protection of the tooth enamel via inhibiting the precipitation of calcium phosphate and hence enhancing calcium homeostasis in the oral cavity [Vitorino et al., 2007].

Regarding the salivary total proteins assessed in the included studies, this review found a statistically significant association between salivary concentration of total proteins and caries; however, this outcome was not substantiated by previous reports [Shimotoyodome et al., 2007; Roa et al., 2008]. Therefore, we agree with Banderas-Tarabay and colleagues [2002] that further evidence related to the role of salivary proteins in modifying the risk for caries is required.

In agreement with previous reports [Al-Tarawneh et al., 2011; Allen et al., 2019], a high variation in the saliva sample size was observed in the included studies ranging between 33 and 213 subjects. Unlike the previous report [Al-Tarawneh et al., 2011], this review noticed a greater homogeneity regarding the saliva sample collection procedure. All studies, except for two [Picco et al., 2017; Soesilawati et al., 2019], collected whole non-stimulated saliva, immediately centrifuged (with varying speed), and fast-frozen the samples at −70°C to −82°C. Debris removal via centrifugation and rapid freezing is acceptable and known protocols for preserving protein biomarkers [Al-Tarawneh et al., 2011]. Apart from centrifugation, some other methods have been employed in the literature including filtration and concentration of the sample [Ellias et al., 2012], utilization of protease inhibitors [Zhang et al., 2012], and application of albumin/immunoglobulin/amylase depletion protocol [Kaczor-Urbanowicz et al., 2017].

The fluctuation of saliva collection in a specific collection time may affect the findings significantly. We observed that each included article had its procedure of when to collect salivary samples [Ferguson and Fort, 1974]. Nearly all included studies performed morning saliva sample collection between 8 a.m. and 11 a.m., prior to eating, drinking, and hygiene. This is, probably, the best method to avoid contamination of saliva with external factors together with minimizing any possible impact from eating. These findings are in opposition to a recently conducted similar systematic review which reported that most of the articles adopted the convenience of clinic and collection time [Allen et al., 2019].

A considerable challenge is the difference in total protein levels which, in healthy individuals, can range from a minimum of 1,000–4,000 μg/mL [Oppenheim et al., 2007] and is significantly increased by exercise [Dawes, 1981]. Numerous salivary proteins, including statherin (range: 2–12 μg/mL), might exhibit a difference of >500% [Dawes, 1981]. Moreover, the salivary proteome is made up of proteins having a dynamic range from pg/mL to mg/mL. This huge range requires to be taken into consideration when a particular protein or group of proteins is investigated for a putative biological role. For instance, it is unlikely that a salivary protein present at pg/mL levels yields any crucial function against an oral disorder including caries. However, these minute levels do not disregard these salivary proteins as potential biomarkers for any disease. Many proteomic methods, including immune-affinity depletion, might be utilized for highlighting low-abundance salivary proteins by first eradicating high-abundance salivary proteins including amylase and mucin [Siqueira and Dawes, 2011].

A crucial issue in saliva collection that might affect proteomic biomarkers is the utilization of resting versus stimulated saliva as the biological sample. Whole saliva can be resting or stimulated, which varies in areas including the proportion of major salivary glands’ contribution and level of specific ions, proteins, and water [Ohshiro et al., 2007]. Resting saliva represents an equilibrated condition, with less impact from salivary glands. One of the two included studies in this review used whole stimulated saliva [Picco et al., 2017] since the authors determined the salivary pH together with the levels of salivary CAVI. Although it is considered that the salivary flow rate affects the salivary pH, the authors selected to assess the salivary pH in whole stimulated saliva primarily owing to the level of bicarbonate in saliva is greatly increased at increased flow rates [Dawes, 1996]. Furthermore, no variation between caries-free and caries-active subjects has been demonstrated when resting saliva pH has been investigated [Preethi et al., 2010; Abbate et al., 2014].

The limitations of the present systematic review are important to recognize. First, most of the included studies used cohort or case-control design. Hence, it is essential to perform further research, particularly cohorts with sufficient follow-ups, which will formulate and systematize further knowledge on this research area before we can regard salivary proteins as adequate biomarkers for caries. In the future, it is probable that the assessment of salivary protein levels might be established as an element of caries risk assessment, which in modern dentistry is an inseparable part of caries management. Second, varying types of salivary proteins assessed in the included studies resulted in the heterogeneity of the data that was not suitable to perform a meta-analysis. Moreover, the heterogeneity of the study participants might be an issue since, when it is over-controlled, it could yield outcomes that are only applicable to a particular population. The majority of the included articles attempted to reduce internal heterogeneity as protein investigations/salivary proteomics could suffer significantly from minute fluctuations of participants [Al-Tarawneh et al., 2011]. It might also yield outcomes specific to one small population that might not be an adequate representation of the general population. Finally, this study is also limited by the restrictions of the included articles, which might result in potential bias and are associated with sampling protocols and sample sizes, measures applied to collect information, lack of longitudinal observations, and missing data. Current literature suggests that increasing sample size and applying a standardized/controlled procedure can improve future salivary proteomic research.

In conclusion, salivary levels of protein(s) might be a useful biomarker for caries diagnosis, especially alpha-amylase, acidic PRP-1, HTN-5, lactoperoxidase, mucin-1, carbonic anhydrase 6, proteinase-3, and statherin. However, their diagnostic value must be verified by large-scale prospective studies.

An ethics statement is not applicable because this study is based exclusively on the published literature.

The authors have no conflicts of interest to declare.

This study was funded by the Canadian Institutes of Health Research # 106657 and # 400347.

Paras Ahmad participated in the conceptual design of the work, data acquisition, and drafting of the manuscript. Ahmed Hussain participated in the conceptual design of the work and drafting of the manuscript. Alonso Carrasco-Labra and Walter L. Siqueira participated in data acquisition and revision of the manuscript.

All data generated or analyzed during this study are included in this article, its online supplementary material files, and the article included in the review. Further inquiries can be directed to the corresponding author.