Introduction

Periodontitis is a common disease affecting the tissues supporting and surrounding the teeth [1]. This disease is a major cause of tooth loss and is associated with a range of multiple long-term chronic conditions (e.g., diabetes [2, 3], cardiovascular diseases [4, 5], and cognitive impairments [6–9], which can negatively impact quality of life and increase the risk of mortality [10]. Because of the effect of periodontitis, periodontal treatment is critical to the patients and several recent periodontal treatment approaches has found more effectively reduce periodontitis [11] and also the inflammation mediator (C-reactive protein) that participated in both periodontitis and other systemic disease (i.e., cardiovascular disease) [12]. Despite of evolving periodontal treatments, how do we define the periodontitis is vitally important in either investigating the pathology but also in following treatment approaches. Although there have been consensus reports on universally accepted periodontitis definition such as for the 1999 classification of periodontal diseases [13–15]; and then later the 2018 AAP/EFP classification of periodontal and peri-implant diseases and conditions [16]), the actual definition of periodontitis employed in dental studies shows considerable heterogeneity.

Understanding of the pathology and aetiology of periodontitis is fundamental to improve the treatment approaches of periodontitis. Based on the current understanding, the pathology and aetiology of periodontitis is complex and multifactorial, including microorganism pathogens, environmental factors, lifestyle behaviours (such as nutrition, oral hygiene, and smoking) [2], epigenetic factors [17] and genetic factors [2, 18]. By focusing on the genetic impact on periodontitis, a recent systematic review on heritability of periodontitis suggested 7%-38% across different study designs (e.g., twin study, other family study, or genome-wide association study (GWAS)) [19]. GWAS comprehensively investigates the association between a trait or diseases and hundreds of thousands of genetic variants (most commonly single nucleotide polymorphisms, SNPs) across the genome [20]; the technique is considered agnostic in terms of not depending on any aspect of the disease biology. GWAS as a technique aims to identify SNPs statistically associated with the trait or disease of interest, so called genetic risk variants or loci. The statistical technique involves comparing the allele frequency differences between cases (with the trait in question) and controls (persons without the trait of interest). The GWA approach has been applied to most common diseases since technology allowed the conduct of such studies in 2005 [21]. Since then, this GWA technique has been applied to oral diseases including periodontitis, contributing to finding more SNPs/genes associated with periodontitis, complementing genetic studies based on biological mechanisms proposed to influence the likelihood of periodontitis.

Current GWAS of periodontitis face many challenges, most notably limited sample size, population stratification, variation in methodologies applied, and use of non-consensus definitions of periodontitis, that complicate interpretation of the consistency of the results [20]. Even though there are some reviews [17, 22, 23] on the genetic aspects of periodontitis, there have been limited attempts to systematically evaluate GWAS studies of periodontitis. To date, there is only one systematic review on the heritability of periodontitis [19], and one descriptive review of periodontitis [23].

Since more GWAS on periodontitis have been published recently, a systematic evaluation and, ideally meta-analysis, of the available evidence would improve the understanding of the genetic mechanisms of periodontitis, address the current research gap, and contribute to better design and analysis of future GWAS. The aim of this systematic review on periodontitis is to: 1) critically appraise the evidence of GWAS of periodontitis. 2) Synthesise the findings and results by summarizing SNPs identified by high quality GWAS, and 3) meta-analyse appropriate studies/SNPs if summary statistics of GWAS were available.

Materials and methods

The study was to systematically review the GWAS that explored the genetic risk factors associated with periodontitis in the general population. This study is registered at the PROSPERO platform (ID: CRD42023456388). The main amendment in this manuscript compares to the protocol is that the meta-analysis was not performed due to appropriate data’s unavailability.

Results

Study selection

After the systematic search in GWAS catalog and PubMed, there were 16 papers from GWAS catalog, 202 papers from PubMed, 491 papers from EMBASE, Global Health and Medline Via Ovid, and 38 papers from ScienceDirect retrieved. After removal of duplicate studies and screening the abstract, 88 papers’ full text were assessed for eligibility. Of these, 15 papers were deemed eligible [32–46]. All included studies were published between 2010 and 2023. Fig 1 displays the PRISMA flow chart, and PRISMA 2020 checklist is provided as supplement (S7 Table in S1 File and S8 Table in S1 Checklist).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The PRISMA flow chart of the study inclusion process.

https://doi.org/10.1371/journal.pone.0306983.g001

Quality assessment

The Q-Genie tool was applied to assess the quality of each eligible study. Overall, the studies have satisfactory quality (scores ranged from 38–50, Table 1). However, the quality varies regarding the classification of the outcome, description of comparison groups, whether the study is adequately powered, statistical methods, description of the test and inferences (scores varied from 2 to 5 in each area among included GWAS). It is worth noting that many studies have insufficient sample size, and only five studies [34, 37, 39, 44, 45] each had a total sample over 10,000 (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Quality assessment rating of all 15 included studies using Q-genie tool.

https://doi.org/10.1371/journal.pone.0306983.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Summary of all the 15 included studies.

https://doi.org/10.1371/journal.pone.0306983.t002

For quality control procedure performed, all included studies have had or reported quality control steps taken but varied from 3 steps reported to 11 steps reported (Table 3). There were 5 studies did not report the genomic inflation score or linkage disequilibrium score regression intercept for inflation check. The full details of the quality control step taken in each study can be viewed in S5 Table in S2 File. Most included SNPs have similar effect allele or minor allele frequency reported but few SNPs with mismatch allele as the 1000 Genomes project were also noted (e.g., rs4242220, rs12969041, rs2027756) and effect allele or minor allele from one study was not reported (Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Summary of quality control steps included in the 15 studies.

https://doi.org/10.1371/journal.pone.0306983.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. The top signal significant SNPs at 5e-6 significant level of periodontal disease extracted from the 13 included papers.

https://doi.org/10.1371/journal.pone.0306983.t004

Discussion

This systematic review has identified, critically evaluated, and synthesized genetics evidence from 15 publicly available GWAS studies on periodontitis, published between 2010 and 2023. The majority of studies had good quality, but 10 out of 15 were small studies with a sample size of less than 10,000. A total of 11 SNPs (rs242016, rs242014, rs10491972, rs242002, rs2978951, rs2738058, rs4284742, rs729876, rs149133391, rs1537415, rs12461706) at genome-wide conventional significant level (p<5x10^-8) and 41 SNPs at suggestive level (p<5x10^-6) were associated with the risk of having periodontitis. In which, three SNPs from three large studies (i.e., rs4284742 [G], rs11084095 [A], rs12461706 [T]) were reported in a same gene—SIGLEC5.

This systematic review has identified several risk variants associated with periodontitis from existing GWAS studies. However, there was huge heterogeneity among study designs and methodologies: sample sizes varied from hundreds to hundred thousand (with the proportion of cases varied from 7.2% to 73.4%, Table 2), ethnicity and population differences, especially, the periodontitis measurements and definitions which may have impact on the GWAS results. The periodontal measurements in the included studies varied from self-reported questionnaire, clinical examination to radiographs, full month examination to half mouth examination, and even studies using same criteria (e.g., CDC-AAP) employed measurements varied from on six sites per tooth to two sites per tooth. Although half mouth examination and fewer sites measurement could lead to more efficient measurement processes, risk of misclassification remains. By utilising questionnaire in measurements, self-reporting bias may also contribute to either underestimation or overestimation of the number of cases. For example, Shimizu et al. 2015 recruited and measured controls separately using self-reported health questionnaire, which may cause underestimation of cases from control participants and may contribute to reduced power to identify genome-wide significant SNPs. Although it is unclear to what extent these heterogeneities in periodontitis definition and measurements would lead to heterogeneity in the GWAS results across studies, future GWAS on periodontitis could benefit from more detailed measurements on periodontal conditions and use of standardised classification criteria. The 2018 periodontal status classification is advocated for use, and proposals have been made to further refine the use of current periodontitis classification to enhance epidemiological data collection and analysis [52]. In addition, more GWAS studies in different ethnic groups, with larger sample sizes and considering covariates are also important for observing the potential for ethnic differences on GWAS results or susceptibility of periodontitis and observe genome-wide significant SNPs.

Heterogeneity in the methods and results reporting was also noted. The current 15 studies incorporated multiple study designs such as inclusion of multiple GWA stages (e.g., 8 studies included discovery and replications). Meanwhile it has been noted that the replication stage had relatively small sample size than discovery stage with five studies included replication stage that have had more than 1,000 participants. However, insufficient sample size in the replication stage could lead to both type I and type II error in the replication results [53]. Meanwhile, it is also important to replicate the results from the combined analysis stage (meta-analysis of discovery and validation stage), which is also missing in the included studies. Future study could utilise better approaches to assess the reproducibility such as the meta-analysis model-based assessment [53]. In terms of results reporting, several included studies did not report the coefficient beta or odds ratio, the standard error or confidence interval, and effect allele of the reported SNPs. This missing information leads to difficulty during data synthesis. Genomic inflation scores were also missing in few studies and not all studies reported or not conducted all quality control steps selected from GWAS quality control guidance. A standardized guideline and consensus on GWAS reporting may be needed to uniform the GWAS report.

To date, there is no guideline in PRISMA on how to perform systematic review on GWAS studies, except Winkler 2014 suggested protocol for genome-wide association meta-analyses in terms of the quality control and conduction of such analyses, requiring the availability of full SNPs and associated statistics reported, allele frequencies, and population stratification [31]. Analytic tools like METAL might estimate the pooled effect for overlapping samples, but it needs genome-wide data to perform appropriate estimation. In addition, a standard and more detailed and widely acceptable quality assessment tool for GWAS is also needed. Although Q-genie tool has been used to assess validity and reliability here, it was designed for genetic association studies, not particularly for GWAS. An assessment tool specifically for GWAS with clear guidance on scoring would be beneficial not only for the assessors but also for the readers to better understand quality assessment criteria.

Form the included 15 studies, three studies based on populations of German, Dutch, European American, Turkish and Asian with sample size > 10,000 participants were commonly discovered three unique SNPs in the gene SIGLEC5, where the effect alleles of all three SNPs has been reported for their protective effect on periodontitis [34, 39, 45]. A recent study has investigated the three genome-wide significant SNPs in the region of SIGLEC5 and shown an impact on SIGLEC5 expression indicating that SIGLEC5 is indeed the target gene for the signal [54]. SIGLEC5 codes for sialic acid-binding Ig-like lectins as a transmembrane inhibitory receptor and is responsible for binding sialic acids and sialic acid-containing glycan ligands. It is expressed in cells in the innate immune system and plays a role in inflammation regulation, both in infection and wound healing [55]. They observed SNP rs11084095 at SIGLEC5 can influence ERG binding and enhancer activity [54], where ERG is important for endothelial homeostasis, including acute response to injury and repair of the endothelium [56]. Meanwhile, the SNP rs12461706 was found in complete linkage disequilibrium with rs11084095. In addition, SNP rs4284742 has been shown to affect MAFB binding affinity, with the common allele enhancing the binding affinity compared to the alternative allele [54]. MAFB is suggested to be associated with the activation of SIGLEC5 expression and contribute to early-onset periodontitis. Further investigation of SIGLEC5 in periodontitis pathologies and intervention targeting the biological pathway underpinned by SIGLEC5 may contributes to both aetiology understanding and disease treatment [57]. In addition to these three SNPs from the same gene region, the rest of the 8 SNPs may also contribute to the periodontitis, however, further GWAS replication with larger sample size may be needed.

The three SNPs in SIGLEC5, two of them were found from both aggressive and chronic periodontitis and one were found from mixed definition defined periodontitis (i.e., including both self-reported and also clinical definition defined periodontitis). This suggested that the SIGLEC5 and the three SNPs may play fundamental roles in all types of periodontitis instead of some specific periodontitis and investigation on SIGLEC 5 may contributes to all type of periodontitis treatment and common pathology understanding. According to the GWAS catalog, most of GWAS significant SNPs found on 15 studies were reported only for periodontitis, except rs2738058 was also found in kidney diseases in Chinese population (i.e., IGA glomerulonephritis) [58]. Meanwhile, several mapped genes of these GWAS significant SNPs were also found significant in IGA glomerulonephritis (e.g., DEFA9P and DEFA10P [58]), neuropsychological conditions (e.g., TMF1P1 [59]), despite of reporting uncommon SNP. These may suggest somewhat genetic similarity between periodontitis and these conditions but further investigation on their relationship with periodontitis still needed.

In comparison to the previous systematic review on the heritability of periodontitis [19], our focus lies more heavily on the methodology utilised in GWAS and synthesis of results. Moreover, comparing with the review article of genetics of periodontitis by Shaddox et al. 2021, we employed a systematic approach and included a greater number of studies than the prior review. It is noted that 6 out of 8 studies that used chronic periodontitis as disease phenotype did not meet the required sample size of >10K cases for such disease with low heritability, except the studies Munz et al. 2017 and Munz et al. 2019. For the 8 studies that used aggressive periodontitis with high heritability as disease definition, smaller sample size is usually acceptable but no studies showed any common risk variants, indicating a potential of false positive results.

The findings that NO common SNPs were consistently reported through all the included studies highlighted a significant level of heterogeneity in the results obtained from GWAS of periodontitis. Given this lack of repeatability in GWAS finding, any identified genetic variants must be interpretated with caution. Furthermore, we observed a reluctance within the dental research community to share GWAS results. Only one study (Shungin et al. 2019) provided a comprehensive list of SNPs statistics from their GWAS of periodontitis, while the remaining studies offered only a limited number of top-signaled SNPs, with some providing incomplete statistics such as lacking odds ratios or 95% confidence intervals. When we attempted to obtain full SNP statistics from these studies, they either declined or did not respond, underscoring a significant transparency issue in current dental research practices. In many common diseases within the medical field, guidelines exist mandating GWAS data sharing as a standard practice expected by funders and publishers. We call for similar guidelines to be established in dental research, requiring the sharing of GWAS statistics from published work to facilitate advancement within the field.

The current study has several strengths, such as summarizing existing risk variants in the literature and discussed the study design of current GWAS on periodontitis, and presenting each study clearly with its database used, population, and GWAS testing method used. However, there are also some limitations to consider. Publication bias should be kept in mind while interpreting the results, as only publicly available GWAS studies were included in the review [26]. Additionally, potential biases from the selected studies may exist, such as those from the periodontitis definition, and variation of methodology applied (e.g. number of covariates adjusted in the association analysis). The current study failed to obtain access to full GWAS summary statistics to perform meta-analysis which could contribute to resolve small sample size in many studies and detect risk variants in combined sample. For example, SNPs with consistently border-line association with periodontitis could have been missed, which may have in theory become statistically significant in meta-analysis if most studies had provided a full list of statistics for all SNPs. In addition, since the majority of included studies sampled were from white population (European, European American, Caucasian etc.) and only few of them included Latino, Black, Asian or mixed ethnicity, it is difficult to draw conclusion based on each ethnicity group. More studies investigating Black and Asian populations could help to understand the underlying ethnicity difference of periodontitis, and future GWAS should also provide summary statistics in repository such as GWAS catalog to facilitate further meta-analysis.

Conclusions

To conclude, our systematic review of 15 GWAS studies on periodontitis identified 11 SNPs were at genome-wide significance p<5x10^-8 level. Variants near or in the gene region SIGLEC5 were reported most frequently (i.e. in three large scale studies) for its potential role on periodontitis. These results imply potential therapeutic targets pathway underlined by the SIGLEC5. Further investigation on this gene could contributed to the periodontitis treatment approach design. However, the heterogeneity on study design, study sample size and target population between studies has been noted. To improve our understanding of periodontitis and support the development of effective treatment options, more high-quality and homogeneous methodology used in GWAS studies are needed. These studies should use standardized periodontitis definitions and assessment tools, have larger sample sizes, and include different ethnicities. Data repository of GWAS results should be made available so that further meta-analysis can be possible, especially in dental research, to ensure research transparency and reproducibility. These efforts will contribute to greater understanding of this oral disease and ultimately benefit public health.

Supporting information