Causal Association Analysis of Periodontitis and Inflammatory Bowel Disease: A Bidirectional Mendelian Randomization Study

Abstract

Background

Periodontitis has been reported to be associated with inflammatory bowel disease (IBD), including ulcerative colitis (UC), and Crohn’s disease (CD). However, the causality of these 2 diseases remains unclear. We conducted bidirectional Mendelian randomization (MR) to investigate the causal relationship between periodontitis and IBD.

Methods

We obtained the genome-wide association study (GWAS) summary data of European populations from FinnGen database (for IBD) and a published article (for periodontitis), from which independent single nucleotide polymorphisms were selected as instrumental variables. Inverse variance–weighted (IVW), MR-Egger, and weighted median (WM) methods were utilized for MR analysis. Heterogeneity or pleiotropy was detected through Cochran’s Q test and MR-Egger intercept, respectively. Outlier was identified with MR-PRESSO (Mendelian Randomization Pleiotropy RESidual Sum and Outlier) and leave-one-out analysis. All statistical analyses were performed with R 4.2.1 and the packages of TwoSampleMR version 0.5.6.

Results

Genetic prediction showed that periodontitis was the risk factor of UC (odds ratio [OR], 1.13; 95% confidence interval [CI], 1.01-1.26; P = .027), rather than of CD (OR, 0.92; 95% CI, 0.74-1.15; P = .456) and IBD (OR, 0.96; 95% CI, 0.81-1.13; P = .619). To the contrary, CD, not UC or IBD, resulted in exacerbating periodontitis in terms of the results of the IVW (OR, 1.09; 95% CI, 1.01-1.17; P = .021) and WM (OR, 1.10; 95% CI, 1.01-1.20; P = .030) methods. Heterogeneity or pleiotropy was acceptable.

Conclusions

Our results indicated that CD was the risk factor for periodontitis; conversely, periodontitis was responsible for the exacerbation of UC, enhancing the existence of mouth-gut axis. Patients with UC should pay more attention to periodontal health, while patients with periodontitis should actively pay close heed to intestinal health.

Lay Summary

A bidirectional Mendelian randomization study indicated that Crohn’s disease was the risk factor for periodontitis; conversely, periodontitis was responsible for the exacerbation of ulcerative colitis, enhancing the existence of the mouth-gut axis and suggesting paying attention to oral health for patients of inflammatory bowel disease.

periodontitis, IBD, causal relationship, Mendelian randomization

What is already known?

The association between inflammatory bowel disease and periodontitis has been reported, but the causal relationship remained unclear.

What is new here?

Our results confirmed a bidirectional link between the oral cavity and the gut.

How can this study help patient care?

Our results suggested that physicians should pay attention to oral health.

Introduction

Inflammatory bowel disease (IBD) is a recurrent progressive disease that composed of ulcerative colitis (UC), Crohn’s disease (CD), and IBD–unclassified. The former 2 are the main types of IBD, which mainly characterized by abdominal pain, diarrhea, and hematochezia.^1,2 This annoying disease affects approximately 0.2% of the European population, imposing a heavy economic burden and poorer quality of life on patients.³ The pathogenesis of IBD is related to heredity, environmental factors, gut microbes, and immunity disorders.^1,2 A variety of environmental risk factors, for instance, smoking, industrial food additives,^4,5 mental disorders,⁶ and vitamin D,⁷related to IBD were reported.⁸

The oral cavity was considered as one of the extraintestinal sites of IBD, and the concept of mouth-gut axis was proposed.⁹ Periodontitis is a periodontal disease closely associated with microorganism and the disorder of the immune function.¹⁰ Preclinical study showed that periodontitis aggravated the intestinal inflammation in colitis mice.⁹ Several case-control studies have shown that the prevalence of periodontitis was higher in IBD patients than in non-IBD patients.^11-14 Nevertheless, the research method cannot determine the causal relationship between IBD and periodontitis. The results of 2 cohort studies indicated that periodontitis was a risk factor for UC, rather than for CD.^15,16 As far as we concerned, the effect of IBD (including CD and UC) on periodontitis has not been studied. In a word, the causality between IBD and periodontitis has not been well understood.

Mendelian randomization (MR) is a common method to elucidate the causality between exposure and outcome. Preliminary conclusions were deduced through available genome-wide association study (GWAS) with lower manpower, money, and time. In addition, this method could avoid reverse causality and confounding to a certain extent. MR was undertaken to illustrate the causal relationship between depression and IBD,¹⁷ periodontitis and hypertension,¹⁸ and periodontitis and psoriasis¹⁹ in previous studies. The purpose of this study was to explore the relationship between IBD (including UC and CD) and periodontitis by utilizing 2-sample MR.

Methods

The causal association between periodontitis and IBD were estimated by a bidirectional 2-sample MR. Instrumental variables (IVs) took the place of exposure in this research. The hypothesis of this method is shown in Figure 1.

Figure 1.

Hypothesis of Mendelian randomization. IV, instrumental variable.

Study Overview

Periodontitis was taken as the exposure factor and IBD as the outcome in this study; then, it was the other way around. The specific flow chart is presented in Figure 2.

Figure 2.

Flow chart of the study. IVW, inverse variance weighted; MR-PRESSO, Mendelian Randomization Pleiotropy RESidual Sum and Outlier; WM, weighted median.

Data Sources

The periodontitis data were derived from a GWAS meta-analysis,²⁰ which included 17 353 European periodontitis patients and 28 210 control individuals. IBD data were obtained from a publicly available GWAS database. The sample sizes of periodontitis and IBD are presented in Table 1.

Table 1.

The sample of IBD and periodontitis.

Disease type	Patients	Control individuals
Periodontitis	17 353	28 210
IBD	5671	303 191
UC	3407	303 191
CD	1021	301 234

Abbreviations: CD, Crohn’s disease; IBD, inflammatory bowel disease; UC, ulcerative colitis.

Selection of IVs

As no periodontitis-associated single nucleotide polymorphism (SNP) met the recommended value (P < 5 × 10⁻⁸), SNPs with P < 5 × 10⁻⁶ were selected as IVs referred to the previous study,¹⁹ while SNPS with P < 5 × 10⁻⁸ were selected as IVs for IBD. Removing linkage disequilibrium for the independent SNPs, the parameters were set to r² = 0.001 and kb = 10000. Eventually, we excluded potential confounding factors concerned with the outcome in Phenoscanner, a website regarding mental factor, blood pressure, blood glucose, blood lipid, and so on.²¹

Instrumental Strength and Power Calculation

Instrumental strength was estimated as the F value, which was computed as (beta/SE)². F >10 was conventionally considered sufficient, and the SNP was removed if the value of F was ≤10.

Merge

We utilized merge to find the SNPs in outcome that were identical to the exposure. If an SNP was not found in the outcome, it was discarded in MR analysis.

Harmonization

To ensure the allele alignment, we removed the palindromic alleles owing to the absence of effect allele frequency by harmonize.

MR Analysis

We utilized inverse variance–weighted (IVW), MR-Egger, and weighted median (WM) methods for MR analysis. IVW was the primary method, which assumed that all genetic variants were valid IVs. It was considered as the most efficient method of analysis. MR-Egger allows for pleiotropy of IVs, but it should meet the assumption of InSIDE (Instrument Strength Independent of Direct Effect); in other words, the effect of IVs of exposure is independent of the direct effect of IVs on the outcome, but this method is less efficient.²² The WM assumes that over 50% SNPs are effective; in this case, the results are reliable and robust.²³ If these 3 methods did not result in the same direction, the IVs were reselected with a more stringent P value.²⁴

Sensitivity Analysis

Cochran’s Q test was used for the heterogeneity; the MR-Egger intercept test determined whether there was horizontal pleiotropy; the MR-PRESSO (Mendelian Randomization Pleiotropy RESidual Sum and Outlier) test detected the outliers. Leave-one-out analysis was conducted to evaluate the robustness in this research and detect outliers. We removed the outlier, then reanalyzed.

Statistical Analysis

We utilized IVW, MR-Egger, and WM methods for MR analysis. All statistical analyses were performed with R version 4.2.1 (R Foundation for Statistical Computing) and the packages of TwoSampleMR version 0.5.6.

Results

Periodontitis to IBD

A total of 8 eligible SNPs were selected. All F values of SNPs were >10, indicating the strong IVs. After harmonization with IBD, UC, and CD, respectively, the palindromic SNP (rs4811024) was removed because of the absence of the value of effect allele frequency. The remaining 7 SNPs, as detailed in the Supplementary Table 1, were included in the MR analysis.

Periodontitis to CD

For CD as outcome, the result showed periodontitis was not the risk factor for CD supported by the results of IVW (odds ratio [OR], 0.92; 95% confidence interval [CI], 0.74-1.15; P = .456), MR-Egger (OR, 0.91; 95% CI, 0.69-1.21; P = .540), and WM (OR, 0.95; 95% CI, 0.74-1.21; P = .668) methods (Table 1 and Supplementary Figure 1). Heterogeneity or pleiotropy was acceptable (Table 1). The results of MR-PRESSSO or leave-one-out analysis indicated that there were no outliers (Table 1 and Supplementary Figure 2).

Periodontitis to UC

The causal relationship between periodontitis and UC was identified through the IVW method (OR, 1.13; 95% CI, 1.01-1.26; P = .027); however, the result was not supported by the MR-Egger or WM methods (Table 1 and Supplementary Figure 1). Heterogeneity or a pleiotropic effect was not found (Table 1). No outlier was picked up by MR-PRESSO (Table 1). The result of the leave-one-out analysis is displayed in Supplementary Figure 2.

Periodontitis to IBD

No causal relationship was detected between periodontitis and IBD strongly supported through the IVW (OR, 1.05; 95% CI, 0.97-1.15; P = .221), MR-Egger (OR, 1.09; 95% CI, 0.99-1.20; P = .154), and WM (OR, 1.09; 95% CI, 0.98-1.21; P = .101) method; the visualizations are shown in Supplementary Figure 1. No heterogeneity or pleiotropy was detected (Table 1). No outlier was explored with MR-PRESSO (P = .529), but 1 outlier (rs138868497) was explored with leave-one-out analysis (Supplementary Figure 2). The causality between periodontitis and IBD was not changed after removing the outlier (Tables 2 and 3 and Supplementary Figure 3).

Table 2.

The result of analysis of periodontitis to IBD.

Disease type	Number of SNPs	IVW	MR-Egger	WM	Cochran’s Q test		MR-Egger intercept P value	MR-PRESSO global test P value
					IVW P value	MR-Egger P value
CD	7	0.92 (0.74-1.15), .456	0.91 (0.69-1.21), .540	0.95 (0.74-1.21), .668	.285	.194	.892	.450
UC	7	1.13 (1.01-1.26), .027a	1.18 (1.04-1.34), .055	1.15 (0.99-1.33), .063	.641	.712	.299	.541
IBD	7	1.05 (0.97-1.15; .221	1.09 (0.99-1.20), .154	1.09 (0.98-1.21), .101	.666	.763	.277	.529

Values are odds ratio (95% confidence interval), P value, unless otherwise indicated.

Abbreviations: CD, Crohn’s disease; IBD, inflammatory bowel disease; IVW, inverse-variance weighted; MR, Mendelian randomization; MR-PRESSO, Mendelian Randomization Pleiotropy RESidual Sum and Outlier; SNP, single nucleotide polymorphism; UC, ulcerative colitis; WM, weighted median.

^aindicates P value < 0.05..

Table 3.

The result of analysis of periodontitis to IBD after removing the outlier (rs138868497).

Disease type	Number of SNPs	IVW	MR-Egger	WM	Cochran’s Q test		MR-Egger intercept P value	MR-PRESSO global test P value
					IVW P value	MR-Egger P value
IBD	6	0.96 (0.81-1.13), .619	1.00 (0.77-1.30), .993	0.98 (0.80-1.20), .839	.808	.714	.704	.847

Values are odds ratio (95% confidence interval), P value, unless otherwise indicated.

Abbreviations: IBD, inflammatory bowel disease; IVW, inverse-variance weighted; MR, Mendelian randomization; MR-PRESSO, Mendelian Randomization Pleiotropy RESidual Sum and Outlier; SNP, single nucleotide polymorphism; WM, weighted median.

IBD to Periodontitis

CD to Periodontitis

For CD as exposure, a total of 7 SNPs were selected, and the F value varied from 30 to 49. One SNP (rs41562816) was not found in outcome; no palindromic SNP was deleted after harmonization. Therefore, the remaining 6 SNPs were included in MR analysis (Supplementary Table 2).

The IVW (OR, 1.09; 95% CI, 1.01-1.17; P = .021) and WM (OR, 1.10; 95% CI, 1.01-1.20; P = .030) methods indicated that CD resulted in aggravating the periodontitis, while the MR-Egger method (OR, 1.09; 95% CI, 0.76-1.56; P = .656) did not support such a result (Table 4 and Supplementary Figure 5). The MR-Egger intercept test or Cochran’s Q test did not find pleiotropy or heterogeneity, respectively (Table 4). No outlier was detected by MR-PRESSO and leave-one-out analysis (Table 4 and Supplementary Figure 5).

Table 4.

The result of analysis of IBD to periodontitis.

Disease type	Number of SNPs	IVW	MR-Egger	WM	Cochran’s Q test		MR-Egger intercept P value	MR-PRESSO global test P value
					IVW P value	MR-Egger P value
CD	6	1.09 (1.01-1.17), .021a	1.09 (0.76-1.56), .656	1.10 (1.01-1.20), .030a	.242	.151	1.000	.349
UC	15	0.98 (0.93-1.05), .627	0.99 (0.71-1.38), .956	0.98 (0.90-1.06), .552	.523	.444	.974	.589
IBD	27	0.99 (0.92-1.06), .728	0.94 (0.62-1.41), .751	0.97 (0.88-1.06), .450	.093	.075	.794	.117

Values are odds ratio (95% confidence interval), P value, unless otherwise indicated.

Abbreviations: CD, Crohn’s disease; IBD, inflammatory bowel disease; IVW, inverse variance weighted; MR, Mendelian randomization; MR-PRESSO, Mendelian Randomization Pleiotropy RESidual Sum and Outlier; SNP, single nucleotide polymorphism; UC, ulcerative colitis; WM, weighted median.

UC to Periodontitis

For UC to periodontitis, 23 SNPs totally met the conditions for IVs. The SNP (rs9617090) associated with type 2 diabetes mellitus was removed, as previous studies have shown that diabetes mellitus was a possible risk factor for periodontitis.²⁵

Two SNPs (rs11209026 and rs28695223) were not found in the outcome. Removing 5 palindromic SNPs (rs117115824, rs2571400, rs6658353, rs72755097, and rs9268492) after harmonization, 15 SNPs were included in MR analysis eventually (Supplementary Table 3).

No causal relationship between UC and periodontitis was explored in the favor of IVW (OR, 0.98; 95% CI, 0.93-1.05; P = .627), MR-Egger (OR, 0.99; 95% CI, 0.71-1.38; P = .956), and WM (OR, 0.98; 95% CI, 0.90-1.06; P = .552) methods (Supplementary Figure 4). The results of heterogeneity or pleiotropy are shown in Table 4. No outlier was detected using MR-PRESSO and leave-one-out analysis (Table 4 and Supplementary Figure 5).

IBD to Periodontitis

For IBD to periodontitis, we selected 36 independent SNPs, and 3 SNPs (rs9617090, rs55884413, and rs191615076) were deleted due to the possible causal relationship to the outcome or the absence of the outcome. A total of 27 SNPs were involved in MR analysis after removing 6 palindromic SNPs (rs11580078, rs2304442, rs2427516, rs2524299, rs4817983, and rs9517691) (Supplementary Table 4).

Genetic prediction proved that no causal relationship between IBD and periodontitis in terms of the result of the IVW, MR-Egger, and WM methods (Table 4). There was no pleiotropy or heterogeneity according to Cochran’s Q test and the MR-Egger intercept (Table 4). MR-PRESSO and leave-one-out analysis did not find outliers (Table 4 and Supplementary Figure 5).

Discussion

Our findings indicated that periodontitis was responsible for the exacerbation of UC, rather than CD or IBD; CD increased the deterioration of periodontitis, enhancing the existence of bidirectional mouth-gut axis and suggesting paying more attention to the periodontal health of patients with IBD and intestinal health of patients with periodontitis.

Our results suggested that periodontitis was the risk factor for UC, which was in line with 2 large sample cohort studies conducted in Korea and Taiwan province in China,^15,16 although the GWAS summary statistics we utilized derived from European participants and cannot be applied to other human species. In addition, a preclinical study has shown that periodontitis increased the secretion of inflammatory factors such as tumor necrosis factor-α (TNF-α), interleukin-6 (IL-6), and IL-1β in the colon of UC model mice, and aggravated histopathological damage of colon.⁹ The results of these researches were in accord with our study.

The results of our study were in contrast to those of another similar MR Analysis²⁶, which showed that UC could increase the risk of periodontitis and periodontitis increase the risk of CD. We have several advantages over another study. First, our results were tested by the leave-one-out method and showed robustness; our results were consistent with clinical and basic research. Second, no pleiotropy was found in our results.

The pathophysiological mechanisms of the mouth-gut axis have not been elucidated clearly. Microbes and immune disorders were possible to play a significant role. A CD model mouse, called SAMP1/YitFc, developed periodontal disease and its severity was positively correlated with ileitis.²⁷ The salivary microbiome of IBD patients enriched the germs associated with oral diseases and was inconsistent with that of healthy control individuals.^28–30 This could be one of the reasons why IBD patients were more susceptible to periodontitis. These studies revealed that intestinal inflammation imposed a negative impact of on periodontal health. In turn, oral diseases had an adverse impact on intestinal health. A variety of oral microbiota have been shown to promote intestinal inflammation. Klebsiella isolated from the saliva of CD patients colonized the colon and induced T helper 1 cell proliferation in the lamina propria, leading to colonic damage.³¹ Wild-type mice were treated with saliva by gavage from patients with periodontitis or microbes (ie, Porphyromonas gingivalis) associated with periodontitis, then the gut microorganism was similar with those substances taken by gavage.^32,33 In addition, intestinal barrier dysfunction was observed with the increase of messenger RNA expression of proinflammatory factors (IL-6, IL-1β, interferon γ, etc.) and the decrease of anti-inflammatory factor IL-10 in the colon,^32,33 indicating that saliva flora could colonize the gut and result in low-grade intestinal inflammation. Enterobacteriaceae, including Klebsiella aerogenes, K. pneumoniae, and K. variicola, from the oral cavity of mice with periodontitis could translocate to the gut and produce IL-1β through activating inflammasome in macrophages, exacerbating colitis.⁹ In parallel, oral-derived pathological T helper 17 cells migrated to the gut, then participated in the inflammatory response after being activated by oral-derived pathogenic microorganisms.⁹ In a word, preliminary evidence existed to support for a bidirectional connection between the oral cavity and gut through microbes and immunity. Further studies are needed to elucidate the mechanism of the mouth-gut axis.

To obtain strong IVs, we used SNPs with F >10 and removed linkage disequilibrium. The Cochrane Q test and MR-Egger intercept test were performed to assess heterogeneity or pleiotropy for the reduction of confounding. The leave-one-out method was used to verify the robustness of the MR analysis, and we reanalyzed the data after removing the outlier to obtain reliable results.

Several limitations should be recognized. First, we performed MR analysis through taking advantage of GWAS from European sample and cannot be generalized to other ethnic groups. Second, data on exposure and outcome should be independent of each other, preferably without sample overlap. The rate of sample overlap could not be calculated in this study. Data on IBD were obtained from FinnGen, while periodontitis data were obtained from a meta-analysis study, which included a cohort born in 1966 in Northern Finland. However, no patients with periodontitis were diagnosed in this cohort, which minimized sample overlap. Third, the diagnostic criteria of the participants with periodontitis were not completely consistent. Although the majority of subjects were diagnosed based on probing depth and/or clinical attachment level, the cutoff values of probing depth and/or clinical attachment level were inconsistent. Fourth, the GWAS summary data of the severity of periodontitis was not distinguished, hence subgroup analysis was not likely to be performed to recognize the relationship between severity of periodontitis and IBD.

Our results provided supporting evidence for a bidirectional connection between the gut and mouth and suggested paying more attention to oral health of the patients with UC and intestinal health of the patients with periodontitis. Further studies should be conducted to explore the bidirectional connection pathway of the mouth-gut axis.

Conclusions

This is the first MR study to investigate the causal relationship between periodontitis and IBD. We found that periodontitis has a negative effect on UC, whereas CD promotes the development of periodontitis. Gastroenterologists should make multidisciplinary diagnosis and treatment with stomatologists to strive for more benefits for patients.

Acknowledgments

The authors thank FinnGen (https://www.finngen.fi/en) for providing the genome-wide association study data.

Author Contribution

X.Q. and C.Z. designed the study and finished the manuscript. Z.Z. helped for data acquisition and analysis. T.Z., L.W., S.F., T.J., X.L., and Y.Y. committed to data interpretation. W.W. devoted himself to the conceptualization.

Funding

The study was not funded.

Conflict of Interest

All authors declare no conflicts of interest.

Data Availability

All data are publicly available.