Causal Associations Between Oral Microbiota and Gestational Diabetes Mellitus: A Two-Sample Mendelian Randomization Study

Background

Gestational Diabetes Mellitus (GDM) is a disorder of glucose metabolism diagnosed for the first time during pregnancy and is among the most prevalent pregnancy complications, with a rising global prevalence.¹ Clinical characteristics of GDM include increased insulin resistance and impaired glucose control, which not only heighten the risk of complications such as preterm birth and preeclampsia in pregnant women but may also result in fetal macrosomia and neonatal hypoglycemia.² The development of GDM is associated with a multitude of factors, including genetics, obesity, age, ethnicity, and lifestyle.³ However, given the complexity and multifactorial nature of gestational diabetes mellitus (GDM), the current understanding remains incomplete. It necessitates continued exploration to identify new and previously unrecognized risk factors.

Oral microbiota is the community of bacteria, fungi, viruses, and protozoa that reside within the oral cavity. Oral microbiota may contribute to the development of GDM through multiple pathways. It can disrupt blood glucose metabolism, for instance, by triggering inflammation through the Toll-like receptor 2 (TLR2) pathway with certain bacterial overgrowth, thus impairing insulin signaling and causing glucose level fluctuations.⁴ Additionally, it can heighten systemic inflammation due to local oral inflammation, increasing GDM risk.⁵ Observational studies have revealed significant differences in the composition of oral microbiota between GDM patients and healthy pregnant women.⁶ Another study found that pregnant women with poor oral hygiene were at a higher risk of developing GDM.⁷ These findings suggest that oral microbiota may play a significant role in the pathogenesis of GDM. Inflammatory cytokines are key mediators in the interplay between oral microbiota dysbiosis and the development of GDM. Alterations in the oral microbiota of pregnant women with GDM often create a pro-inflammatory environment, which is characterized by increased levels of specific cytokines. These cytokines can influence both local oral health and systemic metabolic processes, contributing to the pathogenesis of GDM.^8,9

Mendelian Randomization (MR) is a method of causal inference based on genetic principles. The central assumption of MR is that genetic variants can be used as “instrumental variables (IVs)” which can randomly assign exposure factors, thereby avoiding the influence of confounding factors and reverse causality inherent in traditional observational studies, and providing a more reliable assessment of the causal association between exposure factors and disease.¹⁰ Previous studies have used a two-sample MR to reveal causal associations between the gut microbiota and type 1 diabetes.¹¹ However, the associations by which the oral microbiota may influence the pathogenesis of GDM remain unclear, and further studies are needed to explore these associations.

This study employs a two-sample Mendelian Randomization (MR) approach, integrating Genome-Wide Association Study (GWAS) data, to investigate the causal relationship between the oral microbiome and gestational diabetes mellitus (GDM). By identifying specific oral microorganisms that influence GDM pathogenesis, the research aims to optimize prevention strategies and advance personalized medicine. It also enhances diagnostic tools and innovates treatment options, potentially introducing microbial interventions as novel therapeutic targets. Furthermore, this work provides a scientific basis for public health policies and fosters interdisciplinary collaboration, ultimately improving health outcomes for pregnant women and their offspring.

Methods

This study was reported following the STROBE-MR checklist.¹² The checklist is available as Supplementary Materials.

Study Design

In our study, we carefully selected single-nucleotide polymorphisms (SNPs) from Genome-Wide Association Studies (GWAS) to serve as IVs. As shown in Figure 1, our two-sample Mendelian Randomization (MR) investigation is specifically focused on exploring the causal associations between GDM and oral microbiota-related characteristics, based on three core assumptions:¹³ The relevance assumption: There is a strong association between the IVs and the oral microbiota factor under investigation. The independence assumption: The IVs are not associated with any confounding variables that might affect the oral microbiota or the GDM. The exclusion restriction assumption: the IVs affect the GDM only through their influence on the oral microbiota factor, without any other causal pathways. The need for ethical approval was waived by the committee of the Second People’s Hospital of Nantong because the study used publicly available aggregated data that did not allow the re-identification of the original participants, as supported by Article 32 of the Measures for Ethical Review of Life Science and Medical Research Involving Human Subjects dated February 18, 2023, China, as well as various legislations around the globe.¹⁴

Figure 1 Flowchart of the two-sample Mendelian randomization analysis for the association between salivary microbial exposure and gestational diabetes. The flowchart illustrates the steps involved in conducting a two-sample Mendelian randomization (MR) analysis to investigate the causal relationship between specific salivary microbial taxa and the risk of gestational diabetes, while accounting for potential confounders. In this legend, bold text is used to emphasize and highlight the key components and crucial steps of the analytical process depicted in the flowchart, facilitating quick identification of the core elements of the MR analysis.

Data Sources

The outcome data for GDM were obtained from the FINN cohort data (ID: finngen_R10_GEST_DIABETES), which consisted of 14,718 cases and 215,592 controls. The exposure data were sourced from a published article¹⁵ that detailed the association between 43 oral microbial communities and host genotypes. These data were based on saliva samples from 610 adults in the Danish ADDITION-PRO cohort, who had the V4 region of the 16S rRNA gene sequenced on the Illumina HiSeq 2500 platform. Forty-three GWAS statistical abstracts were validated after stringent quality control measures. All study data were from European populations, with more specific details provided in Table 1.

Table 1 Detailed Information for the GWAS Data

Instrumental Variable Selection

In this study, the selection of IVs was based on the following criteria: (1) SNPs significantly associated with the whole genome of the oral microbiota were first identified with a threshold of P < 5 * 10⁻⁶.¹⁶ (2) Only SNPs with a minor allele frequency (MAF) greater than 0.01 were retained. (3) SNPs showing linkage disequilibrium (LD) were excluded using a threshold of R² < 0.001 and a window size of 10,000kb. (4) If the selected IV was not present in the summary statistics of the results, a SNP with a strong LD (R² > 0.8) to the IV was identified as a substitute and replaced with it.¹⁷ 5) An F value was calculated for each SNP in the IV to assess its strength and to mitigate possible bias caused by weak IVs associated with the exposure factor. The F value was calculated as F = R² * (N-2) / (1-R²), where R² is the exposure variance explained by the SNP in the IV. An F-value greater than 10 is required.¹² To ensure that the effective alleles are indeed from the same allele, we aligned the oral microbiota and GDM datasets, thereby removing any SNPs with discordant alleles and those with allele frequencies in the intermediate range.

MR Analyses

This study primarily employed the inverse variance weighted (IVW) method for analysis,¹⁸ calculating the odds ratio (OR) and its 95% confidence interval (CI) to evaluate the causal association between exposure and outcome risk. The IVW method, as the preferred approach for interpreting MR results, calculates the weighted average of effect sizes using the inverse variance of each SNP as weights. To assess the robustness of the results, we also used MR-Egger regression,¹⁹ weighted median,²⁰ and weighted mode methods.²¹ MR-Egger regression considers the intercept term, providing accurate causal effect estimates even in the presence of pleiotropic bias. The weighted median method assumes that half of the IVs are valid, analyzing the causal association between exposure and outcome. All analyses were conducted using the “TwoSampleMR” in the R software package version 4.0.5, and the results were visualized with scatter plots and sensitivity analysis graphs.

Sensitivity and Pleiotropy Analysis

Sensitivity analysis was performed to detect potential pleiotropy in MR studies. We assessed the heterogeneity among IVs using Cochran’s Q test,²² with a P-value greater than 0.05 indicating low heterogeneity, suggesting that the estimates of IVs are randomly varied and have a minimal impact on the IVW results. Considering that the pleiotropy of genetic variants might affect the estimation of effects, we used MR-Egger regression to explore the presence of horizontal pleiotropy. An intercept close to zero or statistically insignificant in the MR-Egger regression suggests the absence of pleiotropy. Additionally, we employed the MR Pleiotropy RESidual Sum and Outlier (MR-PRESSO) method to detect possible outliers and re-estimated the causal associations after their removal to correct for horizontal pleiotropy.²³ A leave-one-out analysis was conducted to evaluate the robustness and consistency of the results.

Result

Selection of IVs

In this study, we used IV analysis to gain insight into the oral microbiota community. After screening, we identified a total of 267 IVs associated with the abundance of the oral microbiota community and computed the corresponding F-statistics for each taxonomic unit. The mean of these F-statistics ranged from 21.51 to 27.10, with the minimum values ranging from 21.15 to 22.63, and the maximum values ranging from 21.51 to 38.10. To ensure the robustness of our analytical results, we specifically replaced SNPs inconsistent with summary data with suitable proxy SNPs in the analysis of oral microbiota community abundance (Table 2). For more detailed information, please refer to Table S1.

Table 2 Information for Instrument Variable (IV)

Mendelian Randomization Analysis

This study employed an MR approach to investigate the causal association between oral microbiota and GDM. The IVW analysis revealed a positive causal association between Genus Schaalia and GDM risk (Table 3 and Figure 2, OR = 1.03, 95% CI: 1.01–1.06; P=0.02), and a negative association between Genus Haemophilus and GDM risk (Table 3 and Figure 3, OR = 0.96, 95% CI: 0.93–1.00, P=0.034). However, no significant causal associations were identified between other oral microbiota and the risk of GDM.

Table 3 Genetics Predicts Associations Between Oral Microbiota and the Risk of Gestational Diabetes

Figure 2 The association between Genus Schaalia and the risk of gestational diabetes is presented in (A) a forest plot, (B) a leave-one-out sensitivity analysis, (C) a scatter plot, and (D) a funnel plot.

Figure 3 The association between Genus Haemophilus and the risk of gestational diabetes is presented in (A) a forest plot, (B) a leave-one-out sensitivity analysis, (C) a scatter plot, and (D) a funnel plot.

Sensitivity Analysis

A sensitivity analysis was performed to determine the reliability of the IVW results. The MR-Egger indicated no pleiotropic effects in the analysis (Table 4, P > 0.05). The Cochran Q test (Table 4) for heterogeneity showed no heterogeneity in the MR analysis between Genus Schaalia and Genus Haemophilus and GDM (P > 0.05) (Table 2). However, the MR analysis between Genus Neisseria, Genus Porphyromonas, and Species micronuciformis and GDM exhibited heterogeneity (P<0.05) (Table 4). Pleiotropy and heterogeneity disappeared after the exclusion of outlier SNPs for Species micronuciformis and Genus Neisseria identified by MR-PRESSO (Tables 4 and 5). Notably, after the removal of abnormal SNPs, the results did not change, remaining consistent with those obtained before the sensitivity analysis (Table 3).

Table 4 Heterogeneity Tests and Pleiotropy Tests for Instrumental Variables

Table 5 MR-PRESSO Results

Discussion

This study employed a two-sample MR approach to reveal the potential causal association between the oral microbiota and gestational GDM. The findings revealed a positive association between the genus Schaalia and the risk of GDM, while the genus Haemophilus was negatively associated with GDM risk. This finding suggests potential targets for future preventive and therapeutic strategies.

Schaalia bacteria, Gram-positive and facultative anaerobic microorganisms, predominantly colonize the oral ecosystems of humans and other animals. They play a pivotal role in biofilm formation, modulating host immune responses, and interacting with other microbes, which closely ties them to the pathogenesis of various infectious diseases.²⁴ Schaalia bacteria are commonly found as commensals in the oral cavity and other mucosal sites. While typically harmless, Schaalia species can become pathogenic when mucosal barriers are breached, leading to inflammatory responses, inducing the production of inflammatory cytokines.^25,26 Recent research indicates that in women with impaired glucose tolerance (IGT), a precursor to GDM, there is a significant reduction in gut microbiota diversity, along with an increase in the abundance of specific microbial species such as Schaalia turicensis. These alterations in microbial composition are closely associated with the development of IGT.²⁷ This finding is consistent with our findings suggesting that the dysbiosis of the oral Schaalia genus may be associated with an increased risk of GDM. Therefore, future research should delve into the specific role of Schaalia bacteria in the development of GDM and reveal more about the biological mechanisms of the disease, while also assessing the potential of the Schaalia genus as a target for the prevention and treatment of GDM.

Haemophilus, a group of gram-negative, facultative anaerobic bacteria, resides on the mucosal surfaces of humans.²⁸ They are well-known for their role in both localized and systemic infections, and their involvement in inflammation is central to their disease-causing potential and to trigger inflammation.^29,30 The study of the association between GDM and the oral microbiota has revealed a connection, with significant differences in β-diversity observed in the gut microbiota of pregnant women with GDM compared to controls, characterized by an increase in Haemophilus species and a decrease in the α-diversity of the oral microbiota.³¹ Other research has shown a notable rise in the relative abundance of Haemophilus among women with GDM.³² This increase may be related to the metabolic status of patients with GDM, as Haemophilus was associated with fasting blood glucose and lipid levels in participants.³³ Specifically, the rise in Haemophilus may be linked to the production of short-chain fatty acids, changes in incretin hormones, bile acid homeostasis, and the deficiency of peroxisome proliferator-activated receptor γ, which may affect insulin sensitivity and glucose metabolism.³⁴ These findings suggest that changes in the oral microbial Haemophilus profile could serve as a non-invasive biomarker for monitoring GDM during pregnancy. Future research needs to further explore the mechanism of Haemophilus in GDM and how oral microbiota regulation can prevent and treat GDM.

Oral dysbiosis promotes local inflammation in periodontal tissues, leading to the production and release of pro-inflammatory cytokines such as TNF-α, IL-6, and CRP. They play central roles in mediating the association between oral microbiota dysbiosis and GDM. These cytokines are produced in response to pathogenic oral bacteria, enter systemic circulation, and contribute to insulin resistance and glucose intolerance. A feedback loop exists where hyperglycemia and inflammation perpetuate each other, worsening both oral and metabolic health in pregnancy. TNF-α and IL-6 are known to antagonize insulin signaling, impairing glucose uptake and promoting insulin resistance (a central feature of GDM). Chronic elevation of these cytokines disrupts carbohydrate metabolism, increasing the risk of glucose intolerance and the development of GDM.⁸ This mechanistic understanding highlights the importance of oral health and inflammation control in the prevention and management of GDM.^8,9

This study used SNPs significantly associated with oral microbiota abundance as instrumental variables, which are controlled by genetic factors and unaffected by lifestyle and environmental influences.¹⁵ Thus, the MR method can effectively circumvent confounding biases inherent in traditional observational studies, providing a more accurate assessment of the causal association between the oral microbiota and GDM. However, MR also has its limitations, as its validity depends on the choice of instrumental variables. Weak associations or pleiotropy between the selected instruments and the exposure can lead to biased results.³⁵ In addition, this study, which is based primarily on data from European populations, may have limitations when applied to other populations. Therefore, subsequent studies should aim for validation and in-depth analysis in larger sample sizes.

Future research could further investigate the causal association between the oral microbiota and GDM and delve into its underlying mechanisms. For example, mechanistic studies through cellular and animal experiments could explore the specific pathways by which Schaalia and Haemophilus influence GDM. Moreover, the development of drugs targeting specific bacterial species for the prevention and treatment of GDM could be pursued. Clinical trials could assess the impact of probiotic or antibiotic interventions on the prevention and treatment of GDM and explore the optimal intervention strategies.

Conclusions

The results of this study indicate a causal association between the oral microbiota and GDM risk, particularly involving the genera Haemophilus and Schaalia. This provides new insights for the prevention and treatment of GDM and suggests that oral health may play a significant role in the occurrence and progression of GDM.