A systematic review and network meta-analysis of the association between periodontitis and inflammatory bowel diseases

Abstract

Objectives

Several earlier studies have shown that IBD (including its two subtypes, ulcerative colitis (UC) and Crohn’s disease (CD)) increases the risk of periodontal disease. This study aimed to evaluate the relevance among periodontitis and IBD subcategories.

Methods

This study was conducted based on PRISMA guidelines. The Web of Science, PubMed, Google Scholar, and Scopus databases were searched up to February 2024 using pertinent keywords. Case series, review articles, and animal studies were excluded. The risk of bias in this research was evaluated through the Joanna Briggs Institute (JBI) criteria. The meta-analysis was conducted using R statistical software.

Results

A total of 9134 patients within 13 studies after the screening process were evaluated. Our study has shown that periodontitis is significantly more prevalent among IBD patients (UC and CD). According to prior meta-analyses, PD morbidity was found to be significantly high among CD patients (OR: 4.30; 95% CI: 3.72–4.98; I2 = 0%). Similarly, UC elevated PD risk (OR: 4.55; 95% CI: 3.76–5.50; I2 = 0%). The risk of periodontitis was not significantly different between CD and UC patients (OR: 0.96; 95% CI: 0.65–1.43; I2 = 34%).

Conclusions

UC and CD patients were more likely to develop periodontitis, with low heterogeneity between studies, while the prevalence of periodontitis among UC and CD patients was not meaningfully different.

Clinical relevance

The higher risk of periodontitis in patients with IBD indicates the necessity of screening for periodontitis. Considering the various oral manifestations and poor quality of life associated with IBD, it is important to be aware of the symptoms of periodontitis.

Introduction

Inflammatory bowel disease (IBD) represents a persistent inflammatory disorder that predominantly affects the gastrointestinal tract. IBD is a worldwide public health issue that is receiving more scholarly attention due to its rising incidence, mortality, and life years with disabilities [1,2,3]. It comprises two principal categories: Crohn’s disease (CD) and ulcerative colitis (UC) [4]; they can be distinguished by their histopathology and clinical appearance (symptoms and site) [5]. Concisely, any part of the gastrointestinal system can be affected by CD, while UC mostly induces pathologies in the rectum and colon [6]. In addition to intestinal inflammation and associated difficulties, numerous patients may experience extraintestinal diseases affecting other areas, such as chronic kidney disease [7], celiac disease [8], acute coronary syndrome [9], and Rosacea [10, 11]. Despite the unknown etiology of IBD, it is widely acknowledged that integrating multiple genetic and environmental variables modifies the immune system’s reactions to microorganisms [12].

Periodontitis is one of the most common illnesses among humans, with a prevalence of 50% to 90% in developing countries and varying from 4 to 76% in developed countries [13]. In this inflammatory disease, the alveolar bone and periodontal ligament undergo progressive destruction, which is indicated by recession and increased probing depth, or both, and at an advanced level, it contributes to gingival infections, tooth loss, and resorption of the alveolar bone [14, 15]. Periodontitis is a multifactorial disorder. The tooth surface biofilm is the primary cause of periodontitis. Numerous variables can either exacerbate or improve disease manifestations and progression, including the host response, calculus, plaque, environmental variables, systemic health, lifestyle behaviors, and several social factors [16, 17].

Several prior original and meta-analytic studies have established a positive correlation between periodontitis and IBD, indicating that IBD is associated with a greater risk of developing periodontitis [18,19,20,21,22,23]. In contrast, a cohort study directed by Buchbender et al. in 2021 demonstrated no evidence of clinical impairment to the periodontal tissue in IBD cases. A further analysis reported that the periodontal data exhibited no obvious gap between the IBD patients and the control group [24].

Our research sought to perform a systematic review and meta-analysis to evaluate the relationship between periodontitis and inflammatory bowel disease (IBD) statistically (consisting of UC and CD). We established this goal to conduct precision research and explore contradictions. Moreover, we included recent studies that had been absent in former reviews.

Materials and methods

This systematic review is reported using PRISMA standards [25]. Following the standards, the protocol for the greater review was registered and documented in PROSPERO: CRD42024571071. A librarian developed a search strategy for inflammatory bowel disease and periodontitis.

Literature search

All searches included PubMed, Google Scholar, Scopus, and Web of Science, conducted in February 2024. The methodology employed an extensive array of terms related to periodontitis and inflammatory bowel disease for a complete search strategy. In the course of a systematic search, the subsequent search term was employed: (“IBD” OR “inflammatory bowel disease” OR “Crohn’s disease*” OR “Crohn’s Disease*” OR “Ulcerative Colitis*”) AND (Periodontitis OR Pericementitis) and their MeSH terms. A more database-specific search strategy is available within the supplementary material (See Table S1).

Study selection and data collection

After the methodical inquiry was executed, the citations were incorporated into a bibliographic management application (EndNote X9, Clarivate Analytics, Philadelphia, PA, USA) and evaluated for redundancies, which were eliminated both automatically and through manual verification. Following the exclusion of duplicate studies, the selection was executed autonomously by two evaluators, initially by article title and abstract. Subsequently, the complete text of manuscripts satisfying our exclusion parameters was examined. All phases of the title interface by data generalization were finalized by two autonomous team members, and discordances were resolved through discussion.

We included studies in which periodontal conditions of cases with known IBD diagnoses and their control group in the absence of IBD were studied. Those with a background of a systematic disease or pharmacologic intervention use that may have an impact on periodontal tissues were excluded.

The following are the inclusion criteria:

• 1- Studies on human subjects;

• 2- Cross-sectional studies, case–control studies, and cohort studies examining patients with identified IBD and no IBD;

• 3- Studies that defined periodontitis as a predominant observation;

• 4- English-full text studies;

• 5- Studies that reported original data.

We excluded case series, reports, animal studies, review articles, in vitro studies, posters, abstracts, letters, and editorials from the selection. A more detailed PECOS strategy for inclusion/exclusion is explained in Table 1.

Table 1 PECOS Strategy

Based on the criteria for inclusion and exclusion, two independent researchers completed each phase of the title screen through data abstraction. The following data were extracted: the first author’s surname, the date of publication, the study’s design, duration of follow-up, the number of samples collected, subtype of IBD, demographic information such as the average age of study participants, exposure evaluation, study quality score, modifications, and measured risk estimates (HR, OR, or RR).

Risk of bias

We evaluated methods of randomization, treatment allocation, and blinding. Joanna Briggs Institute (JBI) criteria [26] were approached to evaluate internal validity. Two reviewers were involved in the quality assessment, and the disagreement between them was reconciled by a third reviewer.

Statistical analysis

The odds ratio for the difference in PD treatment in IBD patients was calculated using a mixed treatment comparison random-effects frequentist network meta-analysis. We used the Netmeta package of the software program R statistical version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria) for our analysis. Pairwise treatment effects were estimated using a random-effects network meta-analysis and reported as odds ratios (ORs) featuring a 95% confidence interval (CI). In the case in which the 95% confidence interval excludes 1, it signifies a statistically significant variation in the odds. Funnel plots were generated to estimate the risk of publication bias.

Additionally, in this study, with the help of the NMAoutlier library in R, a forward search (FS) algorithm was used to find studies exhibiting extreme outcomes in the Network meta-analysis (NMA) model.

Results

A detailed process of study selection is demonstrated in Fig. 1. A total of 1162 studies were found through searching in databases. After duplicate article removal, six hundred and thirty-three articles entered the title and abstract screening phase; the process led to having seventy-two studies left. From these, fifty-nine studies were excluded due to various reasons. For example, ten studies were in non-English languages, seventeen studies were abstracts or poster presentations, and despite having relevant topics, ten studies did not assess periodontal parameters numerically suitable and were therefore excluded. Finally, our study encompassed thirteen studies: nine case–control [20, 21, 23, 27,28,29,30,31,32], two cross-sectional [2, 33], and two cohort studies [14, 34].

Fig. 1

PRISMA flowchart for identification of eligible studies

Included Study Characteristics

Among the included studies, 13 investigated the odds ratio (OR) of PD prevalence in IBD patients (CD, UC, or both) compared to healthy people controls. Among these, three studies compared three groups against the normal group, five studies compared two groups against the normal group, and the rest only compared two groups. The total number of pairwise comparisons in the network was equal to 29. Also, the total number of people present in this network was equal to 51,044 people.

Before starting the meta-analysis network analysis, we used the FS test to investigate the studies with outliers from the network. Table 2 depicts the summary of indicators related to the FS test. According to this test, pairwise comparisons 1, 23, 11, 10, 13, 14, and 28 were excluded from the network due to an unreasonable increase in heterogeneity. Figure 2 shows the funnel plot before and after removing studies from the network, respectively. After removing the studies identified as outliers, the heterogeneity of the network meta-analysis random effects model decreased from I² = 91% (89.7–94%) to I2 = 0% (0–48%). After this, the characteristics of 13 of the encompassed studies were gathered (see Table 3).

Table 2 The progression of the study entered the FS algorithm iterations

Fig. 2

Funnel plot for base NMA with all studies (A) and after removing outliers detected by FS (B)

Table 3 Summary of the included studies

Quality assessment of the included studies

According to JBI critical appraisal checklists for experimental studies, the result was an average quality score of 81% for case–control studies (see Table S2), 100% for cross-sectional studies (see Table S3), and 68% for cohort studies (see Table S4). These scores highlighted the proper quality of the included studies [26].

Pairwise meta-analysis

Figure 3 shows the results of OR effect size for studies to compare the prevalence of PD in one of the subtypes of IBD. According to these results and the estimated value of direct pooled OR for comparing CD versus control, it shows that the chance of PD in CD patients is 4.3 times higher compared to the control (95% CI: 3.72–4.96). This value is equal to 4.55 (95% CI: 3.76–5.5) and 1.86 (95% CI: 1.63–2.13) for comparing UC against control and CD + UC against control, respectively. However, comparing the odds ratios of PD prevalence in the two groups of CD and UC did not differ statistically between the two groups (pooled OR = 0.96).

Fig. 3

Forest plot for pairwise comparisons (IBD patient subgroups versus control). CI, confidence interval; P, P-Value; df, degrees of freedom; CD, Crohn’s disease; UC, ulcerative colitis; χ², chi-square test of heterogeneity; I², a statistical measure of study heterogeneity

The degree of heterogeneity in the comparison subgroup of CD against UC was estimated to be 34%. In other subgroups, the degree of heterogeneity was 0.

Geometry of network meta-analysis

The geometry of the network meta-analysis illustrates the relationships and comparisons across the included studies, providing a comprehensive view of the evidence. In this study, we constructed a network diagram (Figure S1) to compare the prevalence of periodontitis (PD) across different groups: Crohn's disease (CD), ulcerative colitis (UC), CD + UC, and healthy controls. Each node in the network signifies a distinct group, and the magnitude of the nodes is commensurate with the quantity of constituents within each group. The connections linking the vertices denote the availability of direct comparisons among defined groups, with the thickness of the connections signifying the comparative volume of evidence for each comparison.

The network comprises four major nodes: CD, UC, and controls. Direct comparisons were available for CD versus controls, UC versus controls, CD + UC vs controls, and CD versus UC. The OR for each pairwise comparison was derived based on both direct and indirect evidence, ensuring a robust synthesis of the data.

Network meta-analysis results

The direct effects (green color) and network effect sizes (yellow color) for pairwise comparisons are presented in Supplementary Table 5. Except for the comparison of CD versus UC, the rest of the comparisons show a significant difference in the odds of PD prevalence. It also shows the direct, indirect, and net effects of these comparisons. According to this diagram, for the comparisons of CD against control, UC against control, and CD against UC, the OR is equal to 4.39, 4.40, and 1, respectively (See also Figure S1).

In addition, Fig. 4 shows the network of included studies to compare the PD morbidity rate among IBD patients and healthy individuals.

Fig. 4

Forest plot of the direct, indirect, and pooled network effect sizes. CI, confidence interval; OR, odds ratio; CD, Crohn’s disease; UC, ulcerative colitis; I², a statistical measure of study heterogeneity

Discussion

Periodontitis is known to be a chronic multifactorial inflammatory condition [35]. It is a highly observed oral disease worldwide, and it is deeply intertwined with numerous systemic diseases due to analogous etiological factors [36,37,38]. Meanwhile, inflammatory bowel disease, encompassing UC and CD, is also a chronic disease with a wide spectrum of genetic and immunologic factors as etiology [39]. The interaction existing between the two aforementioned diseases is yet to be understood thoroughly, but several links have been identified. This systematic review and meta-analysis investigated and summarized the prevalence of periodontitis among IBD (both CD and UC) patients. Records from the total number of 9134 IBD patients and 31,099 controls extracted from the included studies were analyzed. It was demonstrated that periodontitis is significantly more prevalent in the presence of UC and CD. Also, it seemed that there was no significant difference in PD morbidity between UC and CD patients.

As mentioned, it was demonstrated that UC elevated PD, which was in accordance with some former meta-analyses [6, 18, 19, 22, 40,41,42]. As with UC patients, PD morbidity was also reported to be significantly high among CD patients, like in previous meta-analyses [6, 18, 19, 22, 40]. Both results were associated with low between-study heterogeneity, and their odds ratios were improved compared to the latest meta-analysis on this topic, showing the impact of the latest included studies [40]. In a recent meta-analysis of 2 studies by Wang et al., it was demonstrated that IBD was associated with a higher prevalence of PD (hazard ratio = 1.37), however, the association between its subtypes and PD was not clear [43]. Moreover, in other meta-analyses of 9 and 6 studies by Lorenzo-Pouso et al. and Nijakowski et al. respectively, despite assessing the association of PD with IBD (and its subtypes) and finding it significant, no comparison of PD morbidity within IBD subtypes was performed [6, 42]. Repeatedly, in a meta-analysis of 6 studies by She et al., both UC and CD subgroups were thought to be associated with PD (ORs = 5.37, 3.64), but no comparison was performed between them, as the meta-analysis of 8 studies by Domokos et al. [19, 22]. Finally, Zhang et al. analyzed 6 studies and found that IBD (and both of its subtypes) are associated with PD (ORs for IBD, CD, and UC = 2.1, 1.72, and 2.39) [18]. In our analysis, not only we have included 13 studies and assessed OR within IBD and its subgroups, but also we compared OR for periodontitis among two CD and UC patients via a network analysis to comprehend the interplay more thoroughly.

To explain this connection, Yu et al. performed a bi-directional, two-sample Mendelian randomized study, which examined the connection between IBD (UC and CD) genetic markers and periodontitis [44]. It was shown that this connection was more a kind of correlation rather than a causality; however, another Mendelian randomization study conducted by Wang, Z. et al. showed the opposite: a more causal relationship [45]. This difference was probably due to different datasets and populations included.

Several microbiologic and immunologic similarities can justify the interplay between two diseases. Microbiota alternation can disrupt the normal balance on every mucosal surface. Said, H.S. et al. discovered that there are some alternations associated with IBD in salivary microbiota, including an increase in the genus Prevotella, Leptotrichia, and Saccharibacteria (TM7) and a decrease in the genus Streptococcus. This unsteady state leads to opportunistic bacteria growth, especially in subgingival plaques, and a loss of diversity among the oral microbiome [46]. This loss of diversity is found to be more significant in CD patients by a predominance of genus Wolinella and Streptococcus mutans [47, 48], and therefore, a sensible variance may be observed among periodontitis patterns. It is also understood that the oral microbiome can affect gut microbiota and therefore disrupt the gut floor, which may exacerbate intestinal manifestations [49, 50]. A mucosal inflammatory reaction to dysbiosis among oral and gut microbiota is a shared characteristic that reveals another connection between IBD and periodontitis [51]. Arias-Bujanda et al. compared cytokine expression patterns in IBD and periodontitis patients, and higher expressions of MMP-8, IL-17A, and INF-γ were detected [52]. Also, in another similar study conducted by Finamore et al. to elucidate differences between saliva immunologic profiles in CD and UC patients, several clues may be available: for example, UC patients have higher TGF-β[1] and nitric oxide levels [53], and CD patients have elevated levels of salivary TNF-α, IL-1β, and IL-6, resulting in variant oral lesions [54]. After these findings, it was better understood how an anti-TNF pathway can mediate periodontal manifestations [55] as well as IBD [56]. Besides these microbiologic and immunologic interplays between the two diseases, transcriptomic-linked hub genes IGKC and COL4A1 were detected in a bioinformatic analysis by Xiong et al. [57].

Behavioral and demographic risk factors can also influence the comorbidity rate independently or between these two diseases; for example, smoking, poor oral hygiene, socioeconomic status, lack of exercise, and dietary habits are among the most repeated behavioral and lifestyle-related risk factors [14, 20, 28, 33, 58]. To assess the demographic effect, numerous studies agree that an age of more than 50–60 years in IBD patients can be a noteworthy risk factor for developing periodontitis [20, 58, 59]. Such an agreement is not observed regarding sex as a risk factor, while some studies did not recognize sex as a risk factor [28, 33, 58]. Yu et al. demonstrated that female IBD cases are at higher risk of developing periodontitis [14]. Paradoxically, in a study by Kang et al., a significantly higher rate of periodontitis development was observed among male IBD patients [59].

Finally, great population size and proper heterogeneity are major advantages of the present meta-analysis. Due to the larger size of the population compared to previously conducted meta-analyses, our study provides a more reliable result, and as a network meta-analysis, it demonstrates a deeper understanding of the network between IBD subtypes and PD. Despite several previously mentioned advantages, our study has faced numerous limitations. A substantial limitation of our study is the observational design of included studies, as observational studies may be prone to confounding factors and sensible risk of bias, and therefore can induce heterogeneity. Additionally, the most frequent age group evaluated within the included studies was the adult population, however, the study by Koutsochristou et al. evaluated a pediatric population, which could induce heterogeneity among the results [29]. However, we conducted a pairwise meta-analysis after the exclusion of Koutsochristou’s study, and the results were the same (See Figure S2). Moreover, the lack of proper geographic distribution may also be a limitation, as most studies were performed in Europe and East Asia, which may affect our findings' generalizability. At last, various guidelines for detecting periodontitis may also be a source of inconsistency within the results.

Conclusion

In summary, it was concluded and confirmed that the abundance of periodontitis is significantly higher among UC and CD patients, and low heterogeneity is observed between studies, while the prevalence of periodontitis between UC and CD patients did not differ meaningfully, and moderate heterogeneity was calculated between studies.