Next-generation sequencing (NGS) approaches have been key in revolutionising microbiology, and dental caries microbial communities have been widely explored with these approaches. However, challenges remain when carrying out comparative data analyses due to the differences in methodologies and protocols employed. Recently, literature on carious biofilms investigated by 16S rRNA sequencing has grown considerably, however, much of the research up to now has been descriptive in nature. The existing accounts fail to resolve the contradiction between the abundance of several microorganisms in caries and in health. Studies on either 16S rRNA amplicon or shotgun sequencing allow the investigation of hundreds of taxa in large datasets, but a lack of biological connexions can be observed between those taxa and the disease development. Specific aspects of how some species collaborate to promote caries progression has been unmanageable by describing the microbiota in high abundance enriched in disease-associated biofilms. This is information that varies widely across studies. The enrichment of some species in a dysbiotic biofilm may reflect their unique capacity to exploit their niche, particularly in being favoured by the increase in specific nutrient availability. However, are there virulence factors that promote host damage in enriched taxa? Opportunistic pathogens may contribute to the compositional and/or functional shift towards dysbiosis. To answer that question, first, it would be imperative to identify who they are.

Meta-analyses of NGS data can enable the estimation of the global prevalence of specific members of the oral microbiome in caries for example. By gathering the existing evidence on enriched microorganisms present in caries, other than Streptococcus mutans, dysbiotic signatures may be detected and used as biomarkers or predictors in translational research. After understanding the pattern of microbial distribution, a deeper study of their physiology could infer on health-to-disease mechanisms and have significant clinical benefits. This could facilitate the development of novel treatments by focusing therapeutic strategies on a limited number of bacterial targets which may be able to stabilise the microbial community and revert or prevent its dysbiotic state [1].

Studies exploring the role of Veillonella spp. in oral biofilms are surprisingly infrequent despite their ubiquity and high abundance, and this has been credited to its challenging genetic manipulation [2]. Species from this genus are not among the most studied in dental caries, although their metabolic accountability in the transition of a homeostatic to dysbiotic oral microbial community is arguable as “middle” members of the carbon food chain, meaning that its substrates are derivates of metabolic processes of microbial partners. Veillonella spp. are involved in important inter-species interactions with other organisms such as streptococci through co-aggregation [3–6]. They are also capable of symbiotic relations by using the organic acids produced by several oral streptococci, including S. mutans, which benefits the growth of both species serving as an acid sink. Additional to the interactions in biofilms, a clue of the role of Veillonella in the second stage of root caries development was previously observed, with high expression of genes that code for bacterial collagenolytic proteases [7]. Exploring their abundance in biofilms in health and caries conditions could be relevant to recognise its ecological significance in the complex oral biofilms. Through this systematic review and meta-analyses, we aimed to determine the abundance of Veillonella species in dental caries in studies using NGS methods.

In light of recent technological advances in NGS and the accumulation of large and publicly available oral microbiome datasets, the need for meta-analysing data on caries microbiome is becoming essential. Developing a versatile analysis pipeline for the raw sequencing data of a wide range of studies would be particularly complex due to the heterogeneity of the datasets (differences in nucleic acid isolation, library preparation, sequencing platforms used, etc.) and their associated information on clinical data. Systematic reviews can, however, be a worthy strategy to achieve a cross-study comparison of the microbiome using NGS data, as we validated here. Although a meta-analysis approach is still subject to heterogeneity, it can nevertheless be feasible. As far as we know, this is the first meta-analysis of the NGS data on dental caries. After a comprehensive literature search, 39 studies presenting Veillonella data from caries samples were eligible, of which 38 and 6 were qualitatively and quantitatively analysed, respectively. We demonstrated a significant enrichment of Veillonella spp. in ECC samples when compared to samples from caries-free samples, yet with very low certainty. However, the most striking result here is lack of clarity in defining cases and controlling confounding factors, and the need for changes in the reports on the clinical data from the donors.

Here we showed differences in Veillonella proportion in different lesions and type of sampling, being the supragingival biofilms more prone to demonstrate differences in RA in health-to-disease. This higher abundance of Veillonella in supragingival biofilms and carious dentin compared to saliva may indicate its site-specialisation. They are evidently likely to benefit from the ecological conditions of low pH environment commonly observed in carious lesions. However, Veillonella spp. enrichment can also suggest their non-negligible role in caries dysbiosis. A relevant research question could perhaps consider Veillonella spp. to have virulence potential in caries, particularly in ECC. Veillonella spp. are considered commensal bacteria present in the oral cavity and the gastrointestinal tract of humans, due to being part of the core microbiome [49, 50]. They are known to be involved in cross-feeding and co-aggregation with acidogenic species in other environments [51], using their metabolic intermediates, which suggest that they act as an acid sink. However, in an in vitro study, it was shown that Veillonella, when associated with S. mutans, did not increase the biofilm pH as expected, as well as promote the growth and extracellular polysaccharide synthesis [49]. We believe that the identification of these virulence factors and a more detailed molecular and functional characterisation of Veillonella in caries-associated biofilms, in combination with a more thorough characterisation of other microorganisms, could be the key for the development of strategies to modulate the microbiota by using pre or probiotics. A more recent molecular characterisation of oral Veillonella revealed that the genomes of oral Veillonella species were remarkably diverse, and that these Veillonella have conserved pathways that utilise carbohydrates other than lactate as an energy source [50], which might be an interesting finding of their potential pathogenic traits as oral species. Furthermore, RNA sequencing seems to show higher abundances of Veillonella than DNA sequencing. Although the result should be interpreted with caution as no statistical analysis was achievable. In addition, it could also suggest some inactive or dead cells in the DNA studies. Benítez-Páez showed important differences in the RA of bacterial genera from metagenomic and metatranscriptomic data by analysing dental biofilms, and Veillonella was one of the most commonly found in the total DNA-based metagenome, however, with a slightly lower value in RNA-based metatranscriptomics [52].

Important limitations on the heterogeneity of the primary studies should be considered. There is a concerning pattern of lack of thorough definition and description of caries cases, even for the same condition, occurring in ECC which is defined differently throughout studies. Most studies used the standard criterion of the World Health Organisation (the classic DMFT), or solely the presence of cavities to define cases. However, this can add an important bias due to the overlooking of non-cavitated caries and caries activity. Furthermore, several studies did not describe whether the “caries-free” group had past caries experience, such as the presence of surfaces with inactive lesions, fillings, or missing teeth, making unclear if this could impact the microbial composition in the microbiome after caries management and arrestment. We hypothesise the presence of a “scarf of dysbiosis” in biofilms from individuals with past caries experience by observing findings from Corralo where microbial profiles of inactive caries-associated biofilms were unlike those of caries-free biofilms [33]. Different detection criteria should be used to assess dental caries in a more accurate way, such as the Nyvad Criterion [53] and the International Caries Detection and Assessment System (ICDAS) [54]. The criterion to determine cases plays an important role in investigating caries-associated microbiota. Furthermore, most studies did not evaluate disease in the same way and did not address the calibration of dental examiners, which would have improved the certainty of the evidence in the field. New studies on caries microbiome should have plans and strategies for a better definition of caries.

Despite advances in the 16S rRNA sequencing and the shotgun whole-genome metagenomic methods, there is an urgent need to define patterns in methodological data from studies on caries microbiology to make them comparable. Although our search found a considerable number of primary studies, it was extremely complex and difficult to make a synthesis within the current available data from the literature. The microbial composition was shown in different perspectives and taxonomic levels. Other sources of variation, such as the complexity of the related clinical elements in each study, may have a great impact on the overall microbiome investigation and is often omitted in many studies. Examples include the lack of the clarity at the selection criteria and the control of confounding factors. In the present study, 20 out of 39 studies were downrated at the JBI domain: “Were confounding factors identified?”. Those confounding factors could include a large range of age, gender, patient habits (including diet, smoking, oral hygiene practises, etc.), caries extent, caries activity, salivary or biofilm pH, salivary flow, other oral conditions (gingivitis, periodontitis, tooth loss, use of prosthesis), systemic diseases, etc. We believe that further study should provide those details as they will favour more rigorous and accurate analyses of the oral microbiome, particularly when saliva is sampled. Zaura et al. already described the need for improving the clinical characterisation in future NGS studies on the oral microbiome so that the quality of the scientific literature would be improved throughout more carefully designed papers [55].

We also identified a lack of standardisation in reporting this type of study, suggesting the need of a comprehensive guideline checklist development. Reports on NGS data following guideline checklists, such as the STROBE for observational studies [56] or the CONSORT for randomised clinical trials [57] promote research reproducibility, better-quality study designs, and help the development of meta-analyses. There is a need for a specific checklist for microbiological data as we faced difficulties in obtaining accurate Veillonella spp. RA data from studies. Authors were requested to share their data on Veillonella RA and the response rate was 34%, resulting in information losses that add a selective reporting bias. Although it would be more reasonable to meta-analyse several microorganisms as caries is a polymicrobial disease, it is not logistically easy due to the differences in these reports. In order to overcome this problem, additional tables containing the average RA of all species might be uploaded as a supplementary material in further studies. Besides, not all studies that presented data on Veillonella clearly defined the unit used for the RA. On a positive note, the laboratory experiments and bioinformatic pipelines were adequately described in all studies.

Zhou et al. lay emphasis on the need to optimise and standardise metagenomic studies [58] as technical or methodologic heterogeneity produces systematic biases that could obscure biologically meaningful information on the compositional differences. More homogeneous studies regarding the DNA/RNA extraction methods, amplicons region sequencing, pipeline (ASV or OTUs), etc. [59] ought to lead to additional meta-analyses. Another important bias control is to develop a sample size calculation, since none of the studies retrieved here presented, thus reducing their external validity (generalisation). Low sample sizes (median of 20 individuals sampled) can be explained due to the high costs of the NGS studies. However, power calculation methods for microbiome studies are now available [60], and specific sample sizes based on the prevalence of microorganisms can also be estimated.

In order to overcome the apparent geographical bias, with data originating from selected few countries (only 14), those with perhaps greater facilities and funds for conducting NGS research, new multicentric studies, in particular, should be developed with a wider interest in the microbiomes of all human populations and ethnicities. More meta-analyses of NGS data investigating other microorganisms (or a group of) in dental caries would help gather and confirm invaluable evidence of enriched microorganisms, such as Veillonella spp. Understanding the pattern of their distribution and physiology could help elucidate the mechanism of the health-to-disease transition, with crucial information on colonisation, survival, and interactions, which would help define research priorities in the field of caries microbiology.

Veillonella spp. are more abundant in individuals suffering with ECC when compared to caries-free controls (very low evidence certainty), and should be considered in further studies to better understand their metabolism and contribution to dental caries. There is an urgent need for a consensus in methodologies used to allow for more rigorous comparison, particularly including clinical data and details of caries diagnosis, which are currently scarce. Inconsistent reporting on the NGS data affected the cross-study comparison and the biological connexions of the relative abundances on caries microbiome.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

ND-T contributed to conception, design, data acquisition and interpretation, drafted, and critically revised the manuscript. AL contributed to design, data acquisition and interpretation, drafted, and critically revised the manuscript. TD contributed to conception, design, and critically revised the manuscript. CS contributed to conception, design, data acquisition and interpretation, performed statistical analysis, drafted, and critically revised the manuscript. All authors gave their final approval and agree to be accountable for all aspects of the work.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.