Candida albicans and Early Childhood Caries: A Systematic Review and Meta-Analysis

Early childhood caries (ECC) is the single most common childhood oral disease that disproportionately affects poor and minority children (<6 years of age), in the USA and worldwide [Dye et al., 2012; Kassebaum et al., 2015]. In addition, severe early childhood caries (S-ECC) occurs in children younger than 3 years of age and in children 4-6 years of age with elevated caries scores [Colak et al., 2013]. S-ECC often progresses rapidly leading to rampant and painful destruction of primary teeth. Treatment of S-ECC is most often provided under general anesthesia in the hospital operating room. As such, costs associated with treatment of ECC/S-ECC constitute a major public health expense [Hajishengallis et al., 2017].

ECC is a “family malady” in that the disease is infectious, transmissible, and is often associated with poor (sugar-laden) dietary habits [Douglass and Clark, 2015]. In addition to Streptococcus mutans and Lactobacillus species, other microorganisms also appear to be involved in the formation of cariogenic biofilms [Hajishengallis et al., 2017]. In this regard, the fungus Candida albicans is frequently detected together with S. mutans in the plaque/biofilms from children with dental caries [Marchant et al., 2001b; Hossain et al., 2003; de Carvalho et al., 2006; Rozkiewicz et al., 2006b; Raja et al., 2010]. This observation is intriguing, as C. albicans usually does not colonize teeth effectively on its own. Rather, C. albicans adheres mainly to oral mucosa or acrylic surfaces, while interacting with commensal streptococci to cause mucosal infections (oral candidiasis) [Xu et al., 2014; Pereira et al., 2017].

To date, the role of C. albicans in the pathogenesis of ECC remains unclear. A number of studies support a potentially positive association between oral Candida carriage and caries experience in children, with detection rates up to 89% in ECC children versus 2-22% in caries-free children [Marchant et al., 2001b; Hossain et al., 2003; de Carvalho et al., 2006; Rozkiewicz et al., 2006b; Raja et al., 2010]. Moreover, in vivo studies using rodent caries models have demonstrated the cariogenic potential of C. albicans[Klinke et al., 2011], particularly when coinfected with S. mutans. Coinfection has been shown to lead to rampant caries under experimental conditions conducive to ECC (e.g., with exposure to a sugar-rich diet) [Falsetta et al., 2014].

Conversely, some clinical studies have neither shown significant differences in oral Candida prevalence between clinically caries-free and caries-active populations [Neves et al., 2015; Thomas et al., 2016] nor a positive association between the presence of C. albicans and caries risk in children [Moreira et al., 2001; Peretz et al., 2011]. Additionally, a recent study indicated a 100% prevalence rate of salivary C. albicans in healthy children aged 12-71 months, regardless of caries status [Thomas et al., 2016].

Given the conflicting available evidence in the literature, this systematic review and meta-analysis aims to evaluate whether oral detection (saliva, plaque, and oral mucosal swab) of C. albicans is associated with ECC.

Database and gray literature searches were conducted in October 2016 and updated in March 2017 to identify published information on oral C. albicans as a risk factor for ECC. A medical librarian developed individual search strategies and retrieved citations from PubMed, Embase, Scopus, Web of Science, LILACS, Cochrane Library, and ClinicalTrials.gov. A combination of text words and controlled vocabulary terms were used (Candida, candidiasis, thrush, child, infant, breast feeding, newborn, dental caries). A detailed search strategy is found in the Appendix (see online suppl. Appendix; see www.karger.com/doi/10.1159/000481833 for all online suppl. material).

This systematic review included experimental and epidemiological study designs such as randomized controlled trials, nonrandomized controlled trials, quasiexperimental, before and after studies, prospective and retrospective cohort studies, case-control and analytical cross-sectional studies that examined the oral presence of C. albicans in children (age <72 months), with or without ECC. Statistical data from selected studies were reported as odds ratio (OR), relative risk (RR), prevalence ratio (PR), confidence intervals (95% CI), p values, or frequency of an absolute number of events/total number of individuals per group. In vitro and animal studies were excluded, as were papers with abstract only, literature reviews, letters to the editor, editorials, patient handouts, case reports, case series, or studies that included children with severe systematic diseases such as HIV and leukemia.

The independent reviewers were calibrated in accordance with inclusion/exclusion criteria using a sample of 20% of the retrieved studies. Agreement between reviewers was good (κ = 0.79). The inclusion and exclusion criteria were applied independently to the remainder of the studies, and any disagreement was resolved by consensus within the 4 reviewers.

Descriptive data, including clinical and methodological factors such as country, study design, subject recruitment site, dental examination and calibration, age of the subjects, sample sources, C. albicans isolation and identification methods, prevalence of C. albicans, as well as results from statistical analysis were obtained using an extraction form (see online suppl. Appendix).

Selected articles were assessed using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (National Heart, Lung and Blood Institute: http://www.nhlbi.nih.gov/health-pro/guidelines/in-develop/cardiovascular-risk-reduction/tools/cohort). Articles were scaled as “fair,” “good,” or “poor” following the protocol guidelines. The OpenMeta(Analyst) (http://www.cebm.brown.edu/openmeta/) software program was used for meta-analysis. Studies with similar designs (cross-sectional design) were included in the forest plot. Heterogeneity among the studies was evaluated using I² statistics. For categorical data, OR, 95% CI and p value were calculated in a forest plot using a fixed effects model (Mantel-Haenszel estimate). Subgroup analysis was performed based on the sample sources (saliva, plaque, and swab).

The literature analyses identified a total of 1,097 papers, including 1,095 articles from database searches and 2 articles from manual searching (Fig. 1). A total of 660 duplicate references were removed. The remaining 467 studies were imported into an Endnote Library for further review. From those, 425 studies were excluded after title/abstract screening. The remaining 42 articles were selected for a full-text review. Authors were contacted by e-mails when articles were not available. After the full-text analysis, 27 studies were eliminated based on the exclusion criteria and 15 articles were chosen for qualitative assessment. Nine articles were further assessed quantitatively using meta-analysis (1 article that used lesion site instead of tooth number to record caries was excluded, and 5 articles were excluded due to unspecified caries diagnostic criteria). The full list of excluded papers and meta-analysis results without exclusion of papers with unspecified caries diagnostic criteria are shown in the online supplementary Appendix.

From the 15 cross-sectional studies used for qualitative analysis (Table 1), 6 studies were assessed as “good” and 9 as “fair” using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies. These studies were all cross-sectional and were conducted in 11 different countries including Scotland, England, Germany, Brazil, Turkey, Iran, China, India, Poland, and the USA. Seven studies specified that the subjects were enrolled in a university dental clinic or nursing school setting. Four studies enrolled study subjects from local kindergartens, while another 4 studies did not detail the subject recruitment site. Children in all 15 studies were younger than 72 months of age, with the majority of the studies examining children from 1 to 5 years of age; 1 study from Scotland assessed children at an earlier age (<1 year old). Oral samples were collected from at least 1 of the following sources: saliva, plaque, oral mucosal swab, and carious lesions.

Salivary samples were collected in 6 studies. Among them, a tongue-loop method instead of harvesting whole saliva was used in 1 study [Radford et al., 2000]. Plaque samples were collected in 10 studies, while oral swab samples were collected in 2 studies. In addition to sound tooth surfaces, 4 studies examined samples collected from carious lesions (Table 1). Samples were plated onto Sabouraud dextrose agar and/or CHROMagar selective for C. albicans isolation. Microbiological and molecular methods were employed for C. albicans identification, including colony shape and color, germ tube test, Auxacolor Test (Sano® Diagnostics Pasteur, France), β-N- acetylgalactosaminidase assay, API 20C (BioMerieux), API ID 32C, and polymerase chain reaction.

All of the clinical studies used visual-tactile examination techniques; 3 studies performed intra- and interexaminer calibration. A κ value >0.8 was considered acceptable agreement. Different caries diagnostic criteria were used in the selected studies. Three studies utilized World Health Organization criteria, 1 study used the British Association for the Study of Community Dentistry criteria, 1 study used the International Caries Detection and Assessment System (ICDAS), and 1 study used the International Standardization for Caries. The remaining studies did not clearly specify the caries diagnostic method. The decayed (d), missing (m), filled (f), tooth (t) or surface(s) (dmft/s) index was used in most of the studies for caries severity. The American Academy of Pediatric Dentistry definition and classification of S-ECC was used in several studies (Table 1).

The C. albicans prevalence in ECC children ranged from 24 to 100% in saliva, 44 to 80% in plaque, 14.7 to 44% in swab samples, and from 60 to 100% in carious lesions. The C. albicans prevalence in caries-free children ranged from 10 to 100% in saliva, 7 to 19% in plaque, and 6 to 7% in swab samples. Statistical differences of C. albicans prevalence between ECC and caries-free children were examined in 11 studies using methods such as χ², Mann-Whitney U, and Pearson χ² with p < 0.05 (detailed in Table 1). Most studies (13 out of 15) found a higher prevalence of oral C. albicans in ECC children than in caries-free children.

In addition to the prevalence rate, C. albicans carriage was quantified in a few studies. Xiao et al. [2016] reported that colony-forming unit levels of C. albicans in saliva and plaque samples of S-ECC children were 3-log higher than in caries-free children. Thomas et al. [2016] found the median C. albicans count to be statistically greater in S-ECC groups than in the caries-free groups (except in the subgroup of children aged 1-3 years). Similarly, Rozkiewicz et al. [2006a] reported that carriage of C. albicans in caries-active children was significantly higher than in caries-free children.

Several studies compared C. albicans detection in samples collected from different sites. Plaque samples were found to yield a higher C. albicans detection than the salivary and swab samples [Xiao et al., 2016]. Plaque samples collected adjacent to carious lesion sites, especially cervical lesions, appeared to have a higher detection rate than those collected from sound tooth surfaces [Yang et al., 2012]. Furthermore, C. albicans was detected more frequently in infected dentin than in plaque close to carious lesions [Ghasempour et al., 2011]. Interestingly, plaque and infected dentin collected from proximal caries had a lower detection frequency of C. albicans than samples from cervical carious lesions [Ghasempour et al., 2011].

Meta-analysis results from 9 studies evaluated the odds of ECC experience (outcome) associated with the presence of oral C. albicans. The pooled estimate of OR (6.51) and 95% CI (4.94-8.57) indicated significantly higher ECC experience in children with C. albicans than those without C. albicans (p < 0.01) (Fig. 2). The odds of experiencing ECC in children with C. albicans versus those without C. albicans was 5.26 for salivary (Fig. 3a), 6.69 for plaque (Fig. 3b), and 6.30 for swab samples (Fig. 3c); all supported the association between C. albicans presence and greater ECC experience.

Several studies further indicated a positive correlation between C. albicans prevalence and/or carriage and ECC severity (p < 0.05) [Wu et al., 2015; Xiao et al., 2016; Lozano Moraga et al., 2017]. Wu et al. [2015] found that the C. albicans detection rate was positively correlated with ECC severity in terms of dmft. Xiao et al. [2016] reported a significant positive correlation between salivary/plaque C. albicans carriage and ECC severity (dmft/s) (p < 0.05). Results from Lozano Moraga et al. [2017] showed that C. albicans was more prevalent in the group with severe caries examined by means of ICDAS (p < 0.05). Ugun-Can et al. [2007] found the detection frequency of oral Candida to be statistically higher in children with moderate and high dft than that in caries-free children; however, there was no significant difference between low dft and caries-free children.

In this systematic review and meta-analysis, we noted a statistically significant difference between C. albicans prevalence in the oral cavity of children with ECC compared to those without ECC. Moreover, we found that individuals with oral Candida presence were associated with a >5 times odds of experiencing ECC. Despite the heterogeneity of the included studies with regard to the sample sources and C. albicans isolation/identification methods, as well as the relatively low evidence strength (e.g., cross-sectional study design, risk bias, small sample size of some studies), the nearly unequivocal conclusions of reported findings, and the magnitude of pooled OR estimates strongly support the association of C. albicans with caries experience.

C. albicans is by far the most commonly detected fungal organism on human mucosal surfaces [Cannon et al., 1995; Thein et al., 2009; Samaranayake and Matsubara, 2017]. It is considered an opportunistic pathogen that lives as a benign commensal organism in the mouths of healthy individuals, especially younger children [Thomas et al., 2016]. Its oral carriage can be affected by several factors, such as host age, diet, geographic location, socioeconomic status, gender, immunosuppression, and medication use [Cannon et al., 1995; Kadir et al., 2005; Samaranayake and Matsubara, 2017]. One attribute of C. albicans that makes it a successful opportunistic pathogen is its ability to adapt and proliferate in a broad range of host environments [Sherrington et al., 2017] such as acidic conditions [Cannon et al., 1995; Gunther et al., 2014; Sherrington et al., 2017]. In this context, the presence of C. albicans may just be serendipitous, coexisting with other oral microorganisms in biofilms or in carious lesions as a natural consequence of the acidified microenvironment. The majority of studies included in this review did not examine the effect of predisposing factors that might potentially be associated with C. albicans carriage in ECC children. The regression analysis from Xiao et al. [2016] showed that none of the factors such as antibiotic usage, birth weight, inhaler use, brushing frequency, and daycare attendance had a significant effect on the carriage of salivary and plaque C. albicans in S-ECC children.

Conversely, there is evidence from in vitro and in vivo mechanistic studies that strongly support the cariogenic properties of C. albicans such as: (1) an acidogenic and aciduric potential (even at pH 4.0) [Klinke et al., 2009] that is capable of dissolving hydroxyapatite [Nikawa et al., 2003] and causing caries in vivo (Klinke et al., 2011); (2) enhanced sucrose-dependent biofilm formation when cocultured with S. mutans in vitro [Pereira-Cenci et al., 2008; Gregoire et al., 2011; Metwalli et al., 2013; Sztajer et al., 2014; Kim et al., 2017] and in vivo [Falsetta et al., 2014; Hwang et al., 2017], and (3) capacity of causing advanced caries lesions in a rat model of ECC through synergistic interactions with S. mutans [Falsetta et al., 2014]. The cariogenic potential of C. albicans has also been supported by clinical studies showing S. mutans and C. albicans codetection in plaque, which was found to be strongly associated with ECC [Radford et al., 2000; de Carvalho et al., 2006; Neves et al., 2015; Xiao et al., 2016]. Further mechanistic and longitudinal studies are needed, however, to validate these observations, as some studies have shown no correlation with caries, while other studies consider C. albicans as a keystone commensal [Janus et al., 2016] with a possible protective role against dental caries development [Willems et al., 2016].

In addition to examining the association between oral Candida and ECC, a few other interesting findings emerged. For example, 1 study examined the maternal relatedness of C. albicans isolated from S-ECC children and found that the mothers of S-ECC children were also highly infected with oral C. albicans(>80% detection in both saliva and plaque samples) and more than 60% of the S-ECC children were carrying the same C. albicans strains as their mothers. This suggests that the mother might be a source for C. albicans acquisition in the oral cavity of children affected by the disease [Xiao et al., 2016] which, if validated, may have important implications for ECC prediction and prevention. Additionally, the genotypic distribution of C. albicans appeared to be associated with the caries experience of children, with the C. albicans genotypic subgroup A being the dominant strain in the plaque biofilm of children with S-ECC [Yang et al., 2012; Qiu et al., 2015; Wu et al., 2015]. Finally, in 1 study there was a strong correlation between oral and gastrointestinal C. albicans colonization [Hossain et al., 2003] suggesting that carious teeth may constitute an ecologic niche for C. albicans that can contribute to recurrent oral and nonoral candidiasis.

The findings presented here should be interpreted within the following limitations. (1) All the selected studies had a cross-sectional design instead of case-control or cohort design which was a weakness of the available evidence for the question our review attempted to answer. Without prospective cohort studies, it remains unclear whether C. albicans is a causative factor for ECC initiation or progression, or whether C. albicans presence is merely a consequence of an acidified oral environment following the development of ECC. (2) Small sample size was another limitation of most of the included studies. ECC is a multifactorial disease, with many predisposing factors other than fungal carriage. Limited sample size compromised the power of performing multiple regression analysis used in some studies. (3) The articles included in the meta-analysis were highly heterogeneous; the only 2 relatively comparable studies were the ones included in the swab sample subanalysis, with the I² = 0%, p = 0.49. (4) Less than half of the included studies were assessed as good quality, with the rest being of fair quality. (5) Methodologies for C. albicans identification were variable.

As the detection of C. albicans was the outcome measure in this systematic review, it is worth noting that clinical sample collection and processing methods can significantly affect C. albicans isolation outcome, especially in the case of dental plaque samples. None of the studies specified the amount of plaque collected or whether the viable counts were normalized. Additionally, cariogenic plaque is relatively sticky due to the rich content of extracellular polysaccharides, requiring adequate sonication to improve cell dispersion and cultivation. Among the selected literature, only 1 study described the sonication steps during plaque sample processing [Xiao et al., 2016]. Furthermore, there were multiple C. albicans isolation and identification methods that would provide different levels of sensitivity and specificity across the included studies. For instance, both Sabouraud dextrose agar and CHROMagar Candida were used in the studies, with the latter used more frequently. The yield (i.e., the number of colonies) and detection of yeast strains on CHROMagar Candida were shown to be greater than on Sabouraud dextrose agar, with a high C. albicans detection sensitivity (98.6%) and specificity (98.8%) [Coronado-Castellote and Jimenez-Soriano, 2013]. Molecular tools such as DNA-based identification were also used for enhanced precision of Candida detection at species level. These observations clearly emphasize the need for standardized methods for both identification and quantification to ensure comparable results while enhancing reproducibility and reliability of the data.

The evidence presented in this systematic review indicates that the prevalence of C. albicans in children with ECC is significantly higher than in caries-free children. In addition, children with oral C. albicans have higher odds of experiencing ECC compared to children without C. albicans. Further prospective observational cohort studies are needed to strengthen the evidence supporting the association between oral C. albicans and ECC, and to determine whether or not Candida detection can serve as a reliable risk factor or risk indicator for the development of ECC/S-ECC.

This work was supported in part by Jin Xiao's faculty start-up funds from the Eastman Institute for Oral Health, University of Rochester, and the National Institute for Dental and Craniofacial Research/National Center for Advancing Translational Sciences grant KL2 TR001999. The research in Dr. Koo's laboratory related to this study is supported by the National Institute for Dental and Craniofacial Research grant DE025220.

The authors declare no conflict of interests.

J.X. and H.K. contributed to the study design, J.X., X.H., N.A., H.A., S.A., D.A.C., F.C., J.D., T.T.W. and K.H. performed the data acquisition and analysis. J.X., X.H., F.C., T.T.W., D.T.K.-K., R.B., E.H., and H.K. contributed to the data interpretation, manuscript writing and critical revision of the manuscript.