Periodontal disease-associated oral and gut microbiome changes in female rheumatoid arthritis patients

Abstract

Background

Although oral and gut microbiota dysbiosis is implicated in rheumatoid arthritis (RA), their coordinated alterations across RA–periodontitis strata remain unclear. This case-control study investigated oral and gut microbiome perturbations in female RA patients stratified by periodontitis severity and their clinical correlations.

Methods

Thirty-two female RA patients and thirty-three matched healthy controls (stratified into mild [MP]/severe periodontal disease [SP] groups) were included. 16 S rRNA sequencing of saliva, subgingival plaque, and stool samples was performed. Microbial composition and functional pathways were analyzed via QIIME2 and PICRUSt2. Spearman correlation analysis was performed to assess correlations between microbe abundances and clinical indices. Statistical analyses were conducted using SPSS (v22.0) and GraphPad Prism (v9), with statistical significance set at P < 0.05.

Results

The final cohort comprised 65 female participants, with RA patients having a mean age of 48.09 ± 10.96 years and healthy controls 47.97 ± 11.79 years. Based on RA status and periodontal condition, participants were stratified into four groups: 17 RA patients with mild periodontitis (RA-MP), 15 with severe periodontitis (RA-SP), 16 healthy controls with mild periodontitis (N-MP), and 17 with severe periodontitis (N-SP). Analysis of microbial α-diversity revealed no significant differences between groups across all sample types. In contrast, β-diversity analysis showed significant separation in the salivary microbiota between RA patients and controls (PERMANOVA: R²=0.039, P = 0.008), with this effect being most pronounced in the SP subgroup (R²=0.077, P = 0.003). At the genus level, LEfSe analysis identified significant enrichment of Prevotella and Streptococcus in RA saliva, while controls were enriched in Neisseria. Notably, correlation analysis demonstrated that saliva-specific Porphyromonas abundances showed positive correlations with both periodontal and rheumatologic parameters, highlighting their potential clinical relevance as non-invasive biomarkers. In contrast, the abundance of the gut genus Fusobacterium was negatively correlated with RA-related parameters, while gut Prevotella showed no significant association.

Conclusion

Saliva-specific Porphyromonas are positively correlated with RA markers, highlighting the clinical relevance of the oral microbiota as potential noninvasive biomarkers in RA–periodontitis comorbidity.

Trial registration

ChiCTR2500108690. This clinical trial was registered with the Chinese Clinical Trial Registry (ChiCTR) on 3 September 2025 under the registration number ChiCTR2500108690.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12903-026-07831-8.

Introduction

Periodontitis is a chronic inflammatory condition triggered by plaque-induced microbial dysbiosis. This condition is modulated through microbe-host-environment interactions and progressively destroys periodontal supporting tissues via inflammation-mediated bone resorption and connective tissue degradation [1, 2]. The clinical manifestations of periodontitis include gingival erythema, probing depth ≥ 4 mm, and pathological tooth mobility. It holds the rank of the world’s sixth most prevalent disease; in 2019, severe periodontitis affected approximately 1.1 billion people worldwide [3, 4]. Beyond its localized oral manifestations, periodontitis exacerbates systemic inflammation through microbial metabolite dissemination (e.g., lipopolysaccharides) and neutrophil extracellular trap (NET) overactivation, contributing to distant tissue damage [5, 6].

The bidirectional association between periodontitis and rheumatoid arthritis (RA), a chronic systemic autoimmune disease with an estimated worldwide prevalence of 1% among adults [7], has garnered particular attention. Both diseases share the following pathogenic hallmarks: (1) tumor necrosis factor (TNF)-α/interleukin (IL)17-driven osteoclast activation, (2) matrix metalloproteinase (MMP) 8-mediated collagenolysis, and (3) overlapping modifiable risk factors (tobacco use, obesity, and HLA-DRB1 polymorphisms) [8–10]. Notably, shared epitope (SE) susceptibility alleles amplify autoimmune responses under both conditions by enhancing citrullinated peptide recognition [11].

Meta-analyses have demonstrated that periodontitis increases the risk of RA by 69%, whereas compared with healthy controls, patients with RA exhibit doubled edentulism rates [12]. Mechanistic studies have revealed that periodontal scaling reduces RA disease activity scores through Porphyromonas gingivalis clearance and gut microbiota remodeling [13, 14], whereas methotrexate therapy improves periodontal parameters via the suppression of IL-6/IL-1β activity [15, 16].

Although gut dysbiosis has been implicated in RA pathogenesis through impaired intestinal barrier function (increased zonulin and decreased occludin expression) and Th17/Treg imbalance [17], recent single-cell RNA sequencing studies revealed oral mucosal dendritic cells as critical intermediaries in the transport of periodontal pathobionts to synovial tissues [18, 19]. Zhang et al. reported concurrent ecological collapse in the oral (subgingival plaque) and gut microbiomes of patients with RA, marked by enrichment of Prevotella intermedia and depletion of Faecalibacterium prausnitzii [20]. Despite advancements, critical limitations in current research hinder a comprehensive understanding: notably, the inability to differentiate RA-specific microbial signatures from periodontal-associated dysbiosis conflates disease-specific mechanisms [21], while isolated analyses of either subgingival or gut microbiomes fail to elucidate their synergistic roles in RA pathogenesis [10]. These gaps preclude holistic insights into microbiome-mediated RA-periodontitis interactions, limiting mechanistic and therapeutic exploration.

To bridge critical knowledge gaps in the interplay between RA and periodontitis, we designed a case‒control clinical trial to systematically characterize periodontitis-driven oral/intestinal microbiome alterations and their integration with multidimensional clinical datasets. The aim of this study was to identify potential etiological pathways at the oral–RA interface and ultimately inform personalized strategies for periodontal management as a disease-modifying adjunct in RA therapeutics.

Materials and methods

Clinical subject recruitment

This research was case‒control format, matched Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [22] and performed in accordance with the relevant ethical principles of the Declaration of Helsinki and the National Institutes of Health guidelines for the use of clinical specimens. The study was conducted within the Rheumatology and Immunology Department and Stomatology Center of the Second Affiliated Hospital of Nanchang University from March 2024 to December 2024.

Sample size estimation was conducted a priori using G*Power software (version 3. 1) based on the primary outcome of detecting significant differences in oral microbial β -diversity (assessed by PERMANOVA) between RA patients and healthy controls, stratified by periodontitis severity. Assuming a moderate effect size (Cohen’s *f* = 0.35) for microbiome-wide comparisons, as reported in comparable RA-periodontitis microbiome studies [23], a minimum of 30 participants per group (RA vs. control) was required to achieve 80% power at a significance level of α = 0.05 for multivariate analysis of variance (MANOVA)-based tests.

The participant recruitment, stratification, and subsequent analysis workflow are detailed in Supplementary Fig. 1. This study included 32 female patients with RA and 33 age- and sex-matched healthy controls (N group), collectively defined as the AM group (all matched participants), for comparative analysis. Patients were recruited on the basis of the 2010 RA classification criteria [23], and ethical approval was obtained from the Ethics Committee of the Second Affiliated Hospital of Nanchang University (Approval No. O-MedResEthRev No. (17)). Additionally, the trial was registered with the Chinese Clinical Trial Registry (registration number: ChiCTR2500108690). All recruited participants were 18 years or older, and written informed consent was obtained from each one.

The exclusion criteria for RA patients were smoking, malignancies, stage II periodontitis, severe oral diseases, pregnancy/breastfeeding, recent periodontal treatment or antibiotic use, other systemic diseases affecting periodontitis, a body mass index (BMI) ≥ 28, and inability to complete the trial. Healthy controls were matched by age and sex. Demographic, medication history and behavioral data, including the rheumatoid factor (RF) and anti-citrullinated protein antibody (ACPA) titers, C-reactive protein (CRP) levels, and erythrocyte sedimentation rate (ESR), were retrieved from medical records.

Periodontal status and temporomandibular joint assessment

Standardized comprehensive periodontal examinations were conducted by two calibrated senior periodontists at the Stomatology Diagnosis and Treatment Center of the Second Affiliated Hospital of Nanchang University. Examinations were conducted under consistent artificial lighting with the participant seated in a dental chair. Examiner calibration was performed prior to the study, achieving an inter-examiner reliability coefficient of κ > 0.85 for all temporomandibular function assessment and periodontal measurements through duplicate examinations on 10 patients not included in the study cohort. The assessment of temporomandibular function comprised measurements of the maximum mouth opening, observation of the mouth opening pattern, and documentation of joint sounds (clicking or crepitus). Clinical parameters, including theplaque index (PLI; Greene & Vermillion, 1964), probing depth (PD, recorded at six sites per tooth), bleeding on probing (BOP), bleeding index (BI; Saxer & Mühlemann, 1975) and clinical attachment level (CAL), were assessed using a standardized manual periodontal probe.

Periodontitis staging was performed according to the 2018 classification system jointly established by the American Academy of Periodontology (AAP) and the European Federation of Periodontology (EFP) [24]. Gingivitis and stage I periodontitis were categorized as mild periodontal disease (MP), while stages III and IV were categorized as severe periodontal disease (SP). Patients with stage II periodontitis were excluded to maintain clearer distinction between disease severity groups. A blinded dual stratification protocol was implemented on the basis of RA status and periodontal health conditions, and this stratification yielded four study cohorts: (1) RA with MP (RA-MP), (2) RA with SP (RA-SP), (3) controls with MP (N-MP), and (4) controls with SP (N-SP).

Biological sample collection

Saliva, subgingival plaque, and fecal samples were obtained under strict aseptic protocols. Unstimulated whole saliva was collected into sterile tubes. After at least 2 h of fasting, the participants maintained a seated position (head tilted 15 forward) for passive drooling into prechilled 50 ml tubes at 2-minute intervals. Samples for which the volume of saliva reached 1 ml within 15 min were temporarily placed in an ice box and stored at -80 °C within 2 h [25]. Subgingival plaque collection was conducted after food debris removal, which was followed by the scraping of supragingival plaque and soft deposits with a probe. The area was dried with a sterile cotton ball, and subgingival plaque was scraped with sterile Gracey curettes to harvest plaque from four sites (the mesiobuccal, midbuccal, distobuccal, and midlingual surfaces) of the index teeth (1 anterior and 1 molar). Fecal samples were collected as the midstream portion of the first morning stool, with a volume equivalent to the size of a soybean. All the samples were preserved at -80 °C pending subsequent 16 S rRNA sequencing.

Microbial DNA extraction and 16 S rRNA gene sequencing

Total DNA extraction was carried out from biological samples using the E.Z.N.A.^® Soil DNA Kit (Omega Biotek, USA) in accordance with the supplier’s guidelines. DNA concentration and purity were assessed using a NanoDrop2000 spectrophotometer, with extraction quality confirmed by 1% agarose gel electrophoresis. The V3-V4 variable region of the bacterial 16S rRNA genes was amplified via the primers 338F (5’-ACTCCTACGGGAGGCAGCAG-3’) and 806R (5’-GGACTACHVGGGTWTCTAAT-3’). PCR products were purified, quantified, pooled at equimolar concentrations, and sequenced using the Illumina MiSeq PE300 platform (Illumina, San Diego, USA).

Bioinformatic analysis of microbiome data

The raw FASTQ files were processed for quality control using fastp (version 0.23.2). High-quality sequences were clustered into operational taxonomic units (OTUs) at a 97% similarity threshold using Vsearch (version 2.22. 1), generating OTU feature tables. Representative OTU sequences were classified taxonomically using the RDP classifier (v2.13) at a 70% confidence threshold against the Greengenes2 [26].

All microbiome bioinformatics analyses were performed in R 4.1.3. Alpha diversity metrics, including observed OTUs, the Chao1 index, and the Shannon index, were computed and compared across groups using the Wilcoxon rank-sum test (two-group comparisons) or the Kruskal‒Wallis test (multigroup comparisons). Beta diversity patterns were visualized via principal coordinate analysis (PCoA) based on unweighted UniFrac distance matrices, with statistical significance assessed through PERMANOVA (999 permutations). Taxonomic differences between groups were identified using the linear discriminant analysis effect size (LEfSe) method, with thresholds of P < 0.05 (Kruskal‒Wallis test) and a linear discriminant analysis (LDA) score > 2.0. Spearman rank correlation analysis, adjusted for multiple comparisons via the Benjamini‒Hochberg false discovery rate (FDR < 0.05), was used to evaluate associations between microbial taxa and clinical parameters. Functional profiling to predict metabolic pathways from the OTU tables was conducted using PICRUSt2, with differential pathway analysis performed via STAMP (Welch’s t test; FDR-adjusted P < 0.05).

Statistical analysis

Statistical analysis of clinical demographic and periodontal parameters was conducted using SPSS software (v22.0; IBM) and GraphPad Prism (v9; GraphPad Software). Clinical demographic and periodontal parameters with missing values were excluded from the analysis on a per-analysis basis. Continuous variables were tested for normality via the Shapiro-Wilk test: normally distributed data are presented as the means ± standard deviations (SDs), and homoscedasticity was confirmed by Levene’s test; intergroup comparisons were then performed via independent t tests; nonnormally distributed data are expressed as medians with interquartile ranges (IQRs) and compared using the Mann-Whitney U test. Categorical variables with sample sizes < 40 were assessed with Fisher’s exact test. Statistical significance was set at P < 0.05 using two-tailed tests, and bar graphs displaying the mean ± SD error bars were generated in Prism. To isolate the associations between microbial taxa and RA-related parameters from those linked to periodontitis alone, we conducted Spearman correlation analyses solely within the RA patient cohort.

Results

Study population and clinical characteristics

A cohort of 32 female RA patients and 33 demographically matched healthy controls (N group) were included. The RA group was divided into the RA-MP (n = 17) and RA-SP (n = 15) groups, whereas the N group was divided into the N-MP (n = 16) and N-SP (n = 17) groups. We first compared demographic and periodontal/RA-related clinical parameters between different disease groups (Table 1). A significant difference in BMI was observed between the N-SP and RA-SP groups (P = 0.006). Further analysis revealed significant differences in BMI within the N group, with N-SP patients having a higher BMI than N-MP patients did (P = 0.014). RA patients in both the MP subgroup and the SP subgroup had significantly greater PLIs compared to their respective matched controls (P = 0.014 and 0.004, respectively). Compared with patients in the N-SP group, patients in the RA-SP group reported greater joint tenderness (P = 0.001). Periodontal clinical examinations revealed statistically significant differences (P < 0.05) between the N group and the RA group regarding the PLI and the percentages of sites with PD ≥ 4 mm and ≥ 6 mm (Table 2). These parameters (PLI, and the percentages of sites with PD ≥ 4 mm and PD ≥ 6 mm) demonstrated a graded increase in mean values/percentages across the study groups, in accordance with increasing periodontal disease severity. The parameters associated with different periodontal disease severity levels and the RA-associated index were also compared. No statistically significant differences were observed in temporal mandibular joint assessments (including maximum mouth opening and joint sounds) or RA-related biomarkers (RF and ACPA titers, CRP level, and ESR) across the study groups (Table 2).

Table 1.

Demographic, periodontal, and temporomandibular clinical characteristics of participant

	AM			MP			SP
	RA (n = 32)	N (n = 33)	P value	RA-MP (n = 17)	N-MP (n = 16)	P value	RA-SP (n = 15)	N-SP (n = 17)	P value
Age	48.09 ± 10.96	47.97 ± 11.79	0.965	45.71 ± 11.41	43.94 ± 13.87	0.691	50.80 ± 10.11	51.76 ± 8.14	0.767
BMI (kg/m2)	20.82 ± 2.79	22.74 ± 2.10	0.007	20.42 ± 3.32	21.69 ± 1.74	0.007	21.34 ± 1.90	23.79 ± 1.94	0.006
Periodontal Clinical Indicators
PLI	1.73 ± 0.45	1.32 ± 0.42	< 0.001	1.57 ± 0.42	1.15 ± 0.42	0.014	1.91 ± 0.42	1.48 ± 0.37	0.004
BI	2.29 ± 0.81	2.19 ± 0.88	0.627	1.85 ± 0.70	1.90 ± 0.83	0.862	2.79 ± 0.62	2.47 ± 0.86	0.229
BOP＋sites (%)	77.56(67.45,84.44)	87.04 (53.68,97.83)	0.546	63.25 ± 20.86	65.16 ± 24.09	0.808	86.66 ± 8.45	89.06 (56.52, 98.21)	0.850
PD ≥ 4 mm sites (%)	3.87(0.52, 13.54)	8.54(1.34,23.61)	0.593	1.33(0.00, 3.27)	0.94(0.00, 4.96)	0.897	22.45 ± 16.17	19.93 ± 13.60	0.636
PD ≥ 6 mm sites (%)	0.00(0.00, 1.36)	0.00(0.00, 1. 18)	0.409	0.00(0.00,0.00)	0.00(0.00,0.00)	0.332	2.38(0.56, 5.33)	0.54(0.00, 3.65)	0.101
Temporomandibular Joint Examination Indicators
Limited Mouth Opening	2(6.25%)	0(0.00)	0.238	1(5.88%)	0(0.00%)	1.000	1(6.67%)	0(0.00%)	0.469
Deviation on Mouth Opening	17(53.13%)	10(30.30%)	0.208	8(47.06%)	4(25.00%)	0.282	8(53.33%)	7(41.18%)	0.723
Joint Clicking	12(37.50%)	10(30.30%)	0.363	6(35.29%)	6(37.5%)	1.000	6(40.00%)	4(23.53%)	0.267
Joint Tenderness	11(34.38%)	1(3.03%)	0.001	3(17.65%)	1(6.25%)	0.601	8(53.33%)	0(0.00%)	0.001

Data for categorical variables are presented as numbers (percentages), whereas data for continuous variables are presented as the means ± SDs or medians (25th, 75th percentiles), as appropriate

Table 2.

Correlation analysis between subject characteristics in the RA and N groups and the severity of periodontitis

	RA			N
	RA-MP (n = 17)	RA-SP (n = 15)	P value	N-MP (n = 16)	N-SP (n = 17)	P value
Age	45.71 ± 11.41	50.80 ± 10.11	0.194	43.94 ± 13.87	51.76 ± 8.14	0.055
BMI (kg/m2)	20.42 ± 3.32	21.34 ± 1.90	0.382	21.69 ± 1.74	23.79 ± 1.94	0.014
Periodontal Clinical Indicators
PLI	1.57 ± 0.42	1.91 ± 0.42	0.041	1.15 ± 0.42	1.48 ± 0.37	0.050
BI	1.85 ± 0.70	2.79 ± 0.62	0.001	1.90 ± 0.83	2.47 ± 0.86	0.063
BOP＋sites (%)	63.25 ± 20.86	86.66 ± 8.45	< 0.001	65.16 ± 24.09	89.06(56.52, 98.21)	0.058
PD ≥ 4 mm sites (%)	1.33(0.00, 3.27)	22.45 ± 16.17	< 0.001	0.94(0.00, 4.96)	19.93 ± 13.60	< 0.001
PD ≥ 6 mm sites (%)	0.00(0.00,0.00)	2.38(0.56, 5.33)	< 0.001	0.00(0.00,0.00)	0.54(0.00, 3.65)	0.001
Temporomandibular Joint Examination Indicators
Limited Mouth Opening	1(5.88%)	1(6.67%)	1.000	0(0.00%)	0(0.00%)	0.465
Deviation on Mouth Opening	8(47.06%)	8(53.33%)	1.000	4(25.00%)	7(41.18%)	0.270
Joint Clicking	6(35.29%)	6(40.00%)	1.000	6(37.5%)	4(23.53%)	0.311
Joint Tenderness	3(17.65%)	8(53.33%)	0.062	1(6.25%)	0(0.00%)	0.485
Rheumatoid-Related Parameters
RF (IU/mL)	135.00(53.60, 218.00)	46.50(20.25, 513.00)	0.155	/	/	/
ACPA (U/mL)	509.10(46.10, 800.00)	281.90(71.48, 800.00)	0.982	/	/	/
CRP (mg/L)	10.88(2.22, 46.00)	6.17(3.11, 9.98)	0.286	/	/	/
ESR (mm/h)	50.00(16.50, 95.50)	46.73 ± 30.25	0.777	/	/	/

Data for categorical variables are presented as numbers (percentages), whereas data for continuous variables are presented as the means ± standard deviations (SDs) or medians (25th, 75th percentiles), as appropriate

Distinct microbial diversity shifts characterize RA-Associated periodontitis

Microbial α diversity was compared among RA patients with different periodontal health statuses (AM/MP/SP subgroups) via observed OTUs, the Chao1 index, and the Shannon index. The analysis revealed no significant differences (P > 0.05; Fig. 1) in these metrics for saliva, subgingival plaque, or the fecal microbiota between RA patients and healthy controls with different periodontal health statuses. These findings suggested similar microbial richness and evenness between individuals with RA and healthy control subjects, indicating the limited impact of periodontal health status on α-diversity. However, as demonstrated below, significant differences in microbial community composition and function were still observed despite these similar α-diversity metrics.

Fig. 1.

Comparative analysis of microbial α diversity in saliva, subgingival plaque and fecal samples from RA patients with different periodontal health statuses. A, B, C Boxplots showed α diversity indices (observed amplicon sequence variants [OTUs], the Chao1 richness estimator, and the Shannon diversity index) of the salivary microbiota in AM, MP and SP subgroups. D, E, F α Diversity of subgingival plaque microbiota. G, H, I α Diversity of the fecal microbiota. A two-sided Wilcoxon rank-sum test was performed for comparisons between the MP and SP subgroups within each RA or non-RA (healthy control [N]) group. *P < 0.05

PCoA based on unweighted UniFrac distances revealed distinct microbial community composition patterns. In the AM group, significant differences in β diversity were detected in the salivary microbiota (PERMANOVA: R²=0.039, P = 0.008), whereas the subgingival plaque and fecal microbiota did not differ significantly between patients with RA and healthy controls (P > 0.05) (Fig. 2A, D, G). These findings suggest that patients with RA may develop salivary niche-specific dysbiosis. Within the SP subgroup, more pronounced microbial disparities emerged: salivary β diversity differences increased (R²=0.077; P = 0.003), accompanied by significant subgingival plaque variations (R²=0.058; P = 0.032), whereas the fecal microbiota remained homogeneous compared with that of healthy controls (P = 0.214) (Fig. 2C, F, I). These results implied that severe periodontitis may act as a “second hit” factor that synergistically exacerbates oral microbiome dysregulation in patients with RA. Notably, the β diversity of the MP subgroup did not significantly differ across all sample types (saliva/subgingival/feces) (P > 0.05) (Fig. 2B, E, H), indicating that limited periodontal inflammation may be insufficient to induce characteristic microbial structural shifts in RA patients. Consequently, given the absence of significant saliva/subgingival microbiome stratification between RA patients and controls in the MP subgroup, we focused mainly on the AM and SP subgroups for further analysis.

Fig. 2.

Comparative analysis of microbial β diversity in saliva, subgingival plaque and fecal samples from RA patients stratified by periodontal status. A, B, C PCoA plots of salivary microbiota β diversity based on Bray‒Curtis dissimilarity for AM, MP and SP subgroups. D, E, F β Diversity of subgingival plaque microbiota. G, H, I β Diversity of the fecal microbiota. The PCoA plots showed the first two principal coordinates (PC1 and PC2) with explained variance as a percentage. Permutational multivariate analysis of variance (PERMANOVA) was conducted to assess the differences in β diversity between groups

Differentially abundant microbes in the AM and SP groups

To provide a structured overview of the dysbiosis patterns, we present the results of the LEfSe analysis, which identifies significant biomarkers across the entire taxonomic hierarchy. For enhanced clarity, the following description is organized by body site and primarily highlights the most resolved taxonomic level (genus), while noting higher-level consistencies where appropriate.

In the AM group comparison, the salivary microbiota of RA patients was significantly enriched with the genera Streptococcus, Veillonella, and Prevotella. Conversely, the control group was dominated by Neisseria, Porphyromonas, and Fusobacterium (Fig. 3A). This dysbiotic pattern in RA was further accentuated in the SP subgroup, which showed additional enrichments in Prevotella, Veillonella, Nanosynobacter, and Leptotrichia. The control SP group, in contrast, remained enriched in Neisseria and Porphyromonas, alongside Granulicatella and Lautropia (Fig. 3B). The consistent enrichment of Neisseria in healthy controls across both comparisons underscored its potential role as a health-associated commensal.

Fig. 3.

LEfSe-based discriminative features of the oral and gut microbiota across clinical groups. LEfSe revealed differences in characteristic biomarkers in (A, B) salivary (LDA score > 3.5, P < 0.05), C, D subgingival plaque (LDA score > 3.5, P < 0.05), and (E, F) fecal microbiota (LDA score > 2, P < 0.05) between the AM group and SP group

In subgingival plaque, the AM group of RA patients exhibited a significant enrichment of the genus Actinomyces. This signature shifted in the SP subgroup, where the RA-associated dysbiosis was characterized by an increase in the family Actinomycetaceae (which includes the genus Actinomyces) and the genus Campylobacter (Fig. 3D). Healthy controls, in both AM and SP comparisons, were consistently enriched in the genera Neisseria and Capnocytophaga (Fig. 3C, D), which reinforcing the site-specific depletion of these taxa in RA.

Analysis of the gut microbiome revealed distinct, group-specific signatures. In the AM group, RA patients showed increased abundances of the genera Gemmiger, Barnesiella, Ligilactobacillus, Coprobacter, and Enterococcus (Fig. 3E). In contrast, the fecal microbiomes of the healthy control group were significantly enriched in the genera Lachnospira, Fusobacterium, Phocaeicola, Fusicatenibacter (Fig. 3E). Within the SP subgroup, the gut microbiota of RA patients was specifically enriched in Barnesiella and Eubacterium (Fig. 3F).

Correlation of the microbiota with periodontal disease and RA

Spearman correlation analysis (using FDR-corrected P values) was performed to assess associations among periodontal clinical parameters, RA-related parameters, and the 30 most abundant salivary genera. Several salivary genera were significantly positively correlated with periodontal parameters (Fig. 4A). The abundances of potential periodontal pathogens, including Porphyromonas and Prevotella, were positively correlated with clinical periodontal parameters and were also positively correlated with rheumatological parameters. In contrast, the abundances of taxa such as Saccharimonas, Selenomonas, Centipeda, and Leptotrichia were positively correlated with periodontal parameters but showed no statistically significant correlation with RF levels and ACPA titers after FDR correction. Additionally, Neisseria was negatively correlated with both periodontitis and RA-related parameters.

Fig. 4.

Heatmap of genus-level correlations between the oral and gut microbiota and clinical parameters. Spearman correlation analysis was performed to assess associations between (A) salivary, (B) subgingival plaque, and (C) fecal microbiota at the genus level and periodontal parameters (PLI, BOP, BI, and PD) and rheumatoid arthritis-related indices (ACPA, RF, CRP, and ESR). The color gradient indicates correlation coefficients (r): red for positive correlations and blue for negative correlations; significance levels: *P < 0.05, **P < 0.01, ***P < 0.001

Notably, subgingival plaque microbiota analysis indicated site-specific behavior (Fig. 4B). Porphyromonas abundance correlated positively with periodontal parameters but did not significantly differ from RA-related parameters. Conversely, the abundances of Neisseria and Corynebacterium were negatively correlated with periodontal parameters and tended to negatively correlate with RA-related parameters. Oral multisite analysis revealed that the abundances of genera such as Porphyromonas and Prevotella were positively correlated with periodontal parameters in both saliva and subgingival plaque. In saliva, these taxa were also significantly positively correlated with rheumatological parameters. However, this association was not significant for the subgingival plaque microbiota, highlighting niche-specific differences.

Further highlighting niche-specific influences, gut microbiota analysis indicated that Fusobacterium and Prevotella—although enriched in both compartments—exhibited opposite disease associations in the gut compared with the oral cavity: their abundances exhibited positive correlations with periodontal parameters but inverse associations with RA-related indices (Fig. 4C). This oral–gut divergence suggested that microbial behavior is strongly shaped by the anatomic niche, which may differentially influence host immune and inflammatory responses.

Functional profiling via PICRUSt2 analysis

Through PICRUSt2 functional annotation and LEfSe analysis (LDA score > 2.0), we found that the disease state of RA significantly impacted the functional characteristics of the oral microbiota (Fig. 5). In saliva samples, RA patients presented distinct metabolic signatures, which were enriched primarily in antibiotic synthesis pathways (e.g., ansamycin biosynthesis), energy acquisition pathways (e.g., photosynthesis), specific amino acid metabolism pathways (e.g., D-arginine and D-ornithine metabolism), and carbohydrate metabolic networks (including galactose, fructose, and mannose metabolism). In contrast, the salivary microbiota of healthy controls (N group) presented greater metabolic diversity, including multiple antibiotic synthesis pathways (e.g., vancomycin and streptomycin classes), complete energy metabolism networks (e.g., the tricarboxylic acid [TCA] cycle and oxidative phosphorylation), and complex lipid metabolism pathways (e.g., fatty acid biosynthesis and unsaturated fatty acid metabolism) (Fig. 5A).

Fig. 5.

Group-specific Kyoto Encyclopedia of Genes and Genomes (KEGG) functional profiles of the oral microbiota. KEGG pathway prediction via PICRUSt2 revealed differentially enriched pathways (linear discriminant analysis [LDA] score > 2.0/2.5; *P < 0.05) in the (A, B) salivary and (C, D) subgingival plaque microbiota between MP and SP subgroups (RA with non-RA controls)

Salivary microbiota functional analysis revealed that the RA-SP group presented additional pathways, such as those related to sphingolipid and riboflavin metabolism, which are linked to inflammation regulation. The N-SP group retained the main features of the healthy control group but included xenobiotic metabolism pathways such as naphthalene degradation (Fig. 5B).

Analysis of subgingival plaque samples highlighted more pronounced disease‒periodontal interactions: RA patients exhibited only activation of the pentose and glucuronate interconversions pathway, whereas the N group presented typical symbiotic metabolic features (e.g., fatty acid and LPS biosynthesis) (Fig. 5C). Notably, the N-SP group exhibited specific activation of the LPS biosynthesis pathway (LDA score = 2.54; P = 0.047) (Fig. 5D). In contrast, the functional profiles of the gut microbiota did not significantly differ between RA patients and healthy controls (LDA score < 2.0).

Discussion

This study reveals associations between periodontitis-related oral dysbiosis and rheumatoid arthritis. Using a multi-compartment microbiome analysis, we identified distinct microbial profiles in RA patients compared to healthy controls. Importantly, we found that specific oral bacteria, particularly in saliva, correlate with both periodontal and rheumatologic parameters. These findings enhance our understanding of oral-systemic connections in RA and provide potential targets for future diagnostic development.

In this study, significant differences in clinical indicators such as the PLI and tender joint count were identified between individuals with RA and healthy controls. RA patients exhibited higher plaque indices in both mild and severe periodontitis states (P = 0.014 and 0.004, respectively), suggesting potential deficiencies in oral hygiene management among RA patients [8, 27]. Furthermore, compared with N-SP patients, RA-SP patients exhibited a greater prevalence of tender joints (P = 0.001), reinforcing the association between periodontitis and RA disease activity [28]. These significant differences in clinical characteristics provide a critical context for subsequent microbiome analyses.

Significant alterations were observed in the oral microbial community of individuals with RA. While α-diversity metrics in saliva, subgingival plaque, and the fecal microbiota did not significantly differ between patients with RA and healthy controls across periodontal subgroups, in contrast to previous reports [29], β-diversity analysis of the salivary microbiota revealed distinct clustering in patients with RA, particularly under severe periodontitis conditions. These findings suggested that severe periodontitis may act that microbial dysbiosis in RA is not driven by changes in overall diversity but by shifts in the abundance of specific taxa. Notably, RA patients demonstrated marked enrichment of proinflammatory taxa, including Streptococcus, Prevotella, and Veillonella (LDA score > 3.5), in saliva and subgingival plaque, in contrast to health-associated Neisseria. This dysbiotic pattern aligns with observations in Chinese RA cohorts, where reduced Neisseria abundance is correlated with depletion of Defluviitaleaceae_UCG-011 and expansion of Prevotella_6 [30], collectively underscoring the immunomodulatory role of Neisseria in maintaining oral and systemic immune homeostasis. These findings are consistent with the “oral dysbiosis-autoimmunity” hypothesis [31, 32]. RA-associated oral dysbiosis drives disease progression through three interlinked mechanisms: local immune activation (toll-like receptor, TLR-mediated cytokine release), molecular mimicry (ACPA generation via bacterial citrullination), and cross-site inflammation (oral–gut–joint axis) [30, 33–35]. These mechanisms collectively highlight the critical role of oral dysbiosis in RA pathogenesis. Secreted proteases (e.g., gingipains) of Prevotella degrade oral mucosal tight junctions, facilitating bacterial translocation (e.g., lipopolysaccharide, LPS) and activating TLR2/4-mediated immune responses [36, 37]. Peptidylarginine deiminase (PAD) expressed by Prevotella catalyzes the citrullination of host proteins (e.g., fibrinogen), generating ACPA-targeted antigens that form immune complexes in the joint synovium [36, 37].

Healthy controls show significant enrichment of Neisseria, which confers protection against periodontitis in two different ways: (1) Ecological competition: its hydrogen peroxide production inhibits pathogen colonization (e.g., P. gingivalis) and maintains mucosal integrity [38], and (2) Metabolic regulation: its butyrate synthesis enhances intestinal epithelial tight junction proteins (e.g., ZO-1), reducing systemic inflammation [38].

Our stratified analysis suggests that severe periodontitis may act as a “second hit” that synergistically exacerbates oral microbiome dysregulation in RA. Stratified analysis revealed a dose-dependent relationship between periodontitis severity and changes in the RA microbiota. Saliva β diversity increased with periodontitis progression (MP group R²=0.05; SP group R²=0.077), which aligns with findings that periodontal pathogen load is correlated with RA activity [39]. Notably, subgingival plaque analysis revealed taxonomic shifts within the Actinomycetaceae family across RA groups, with genus-level Actinomyces enrichment in the broader RA cohort and family-level significance in the RA-SP subgroup. However, correlation analyses revealed negative associations between Actinomyces abundance and periodontal parameters (BI, BOP), with no significant correlation to RA-related parameters. This seemingly paradoxical pattern—enrichment in disease groups yet negative clinical correlations—suggests Actinomyces may represent a background dysbiotic element rather than a direct inflammatory driver. This ecological role differs fundamentally from established periodontal pathogens like Aggregatibacter actinomycetemcomitans (Aa), which promotes RA through specific mechanisms including neutrophil-mediated protein citrullination, molecular mimicry, and synergistic inflammation [40–42]. While both organisms are associated with oral dysbiosis, their relationships to disease parameters appear distinct: Aa demonstrates direct positive correlations with disease activity, whereas Actinomyces enrichment may reflect more complex ecological interactions within the dysbiotic biofilm community. The classic members of the periodontal pathogen “red complex”, Porphyromonas gingivalis, Treponema denticola, and Tannerella forsythia, demonstrated similar disease parameter correlations, suggesting a synergistic “pathogen alliance” in RA development [43].

Functional profiling via PICRUSt2 analysis revealed unique metabolic features in the saliva of patients with RA, including D-arginine and D-ornithine metabolism, potentially intensifying systemic inflammation and correlating with periodontitis severity. This aligns with microbial metabolic reprogramming theories [44, 45]. These findings expand upon clinical observations of a periodontitis–joint inflammation dose–response relationship [46]. RA patients exhibited significant niche-specific oral microbial changes. Integrated saliva and subgingival data revealed that RA patients had a “metabolic simplification” signature, reflecting a microecological imbalance from chronic inflammation. Periodontitis altered oral microbial function in RA patients, whereas the metabolic diversity and stability of microbial functional networks were stronger in healthy individuals.

Based on prior literature [47, 48], periodontal pathogens may induce gut dysbiosis and functional impairments that affect RA progression. The gut–joint axis in RA is critically mediated by proinflammatory gut taxa, which exacerbate systemic inflammation and autoimmunity through LPS-driven barrier disruption, Th17/Treg imbalance, molecular mimicry of citrullinated antigens, and dysregulation of tryptophan metabolism, synergistically amplifying joint destruction [49, 50]. In our study, salivary Prevotella abundance showed significant positive correlations with periodontal parameters (BI, BOP, and PD ≥ 4 mm) and a rheumatoid-related parameter (ESR). This correlation pattern is consistent with previous reports suggesting that Prevotella species may contribute to shared pathogenesis through mechanisms such as Th17 cell promotion [36]. Our findings revealed that Fusobacterium exhibited distinct disease associations depending on its anatomical niche: demonstrating no significant correlation with periodontal parameters in the oral cavity but a significant negative correlation with RF titers in the gut. This contrasting pattern reflects the organism’s well-documented, context-dependent ecological role. In the oral cavity, Fusobacterium is known as a keystone pathogen that promotes biofilm formation and facilitates the colonization of other pathogenic species through co-aggregation, thereby exacerbating local inflammation [51, 52]. However, its role in the gut is more complex and potentially beneficial in certain contexts. Gut-colonizing Fusobacterium may suppress systemic inflammation through several potential mechanisms: (1) competitive exclusion of more proinflammatory taxa, (2) enhancement of mucosal barrier function via stimulation of mucin production, and (3) facilitation of butyrate-mediated regulatory T-cell (Treg) differentiation, which can promote immune tolerance [52, 53]. These gut-specific immunomodulatory mechanisms provide a plausible explanation for the inverse association we observed between gut Fusobacterium abundance and RF titers, highlighting the complex role of this genus in shaping systemic immunity.

The observed divergence between salivary and subgingival plaque microbiomes in their association with systemic RA parameters may be attributed to several factors. Saliva serves as a pooled sample reflecting contributions from multiple oral niches (gingival crevicular fluid, mucosal surfaces, tongue coating, and plaque), thereby capturing a broader spectrum of oral microbes and their metabolic byproducts [54]. This integrative nature may render saliva more reflective of systemic immune and inflammatory status. In contrast, subgingival plaque represents a localized, site-specific biofilm community, which is more directly associated with periodontal tissue destruction and may have a more limited direct interaction with systemic immune pathways. Furthermore, the translocation of saliva—through routine swallowing or micro-aspiration—may facilitate more frequent and direct interaction with the gut immune system, thereby influencing systemic inflammation and autoimmunity in ways that are less pronounced for the more sequestered subgingival plaque microbiota. Based on our findings and the literature, we suggest that saliva may be the most appropriate oral microbiome source for investigating systemic conditions such as RA, due to its non-invasive collection, holistic representation of the oral microbial community, and stronger correlation with systemic inflammatory markers. However, subgingival plaque remains invaluable for studying localized periodontal pathogenesis and may complement saliva in comprehensive multi-niche studies.

In addition to the well-characterized periodontal pathogens, our LEfSe analysis revealed significant enrichment of other taxa in RA patients. For instance, Veillonella and Streptococcus were notably enriched in the saliva of RA patients, particularly in the severe periodontitis subgroup. Veillonella, a lactate-utilizing genus, may contribute to a dysbiotic oral environment by metabolic cross-feeding with acid-producing bacteria, thereby exacerbating local inflammation and potentially influencing systemic immune responses [55]. Similarly, Streptococcus species, while commensal, can exhibit pathobiont behavior under inflammatory conditions, promoting neutrophil recruitment and cytokine release [56]. In the gut microbiota, RA patients exhibited increased abundances of Gemmiger, Barnesiella, and Ligilactobacillus, which have been previously associated with altered short-chain fatty acid profiles and mucosal immunity [20]. The enrichment of these taxa may reflect a compensatory mechanism or a dysbiotic shift contributing to systemic inflammation, warranting further functional validation. These findings collectively underscore the complexity of microbiome-wide alterations in RA and highlight the potential roles of non-periodontal taxa in disease pathophysiology.

This study offers valuable insights into RA microbiota characteristics and periodontitis interactions but has several limitations. First, the cross-sectional, case-control design of our study precludes definitive causal inferences regarding the observed microbiome–disease associations. While our findings highlight significant correlations, prospective longitudinal studies are needed to establish temporal relationships and causality. Second, the small, all-female sample limits the generalizability of our findings. The exclusion of smokers, a necessary step to control for a major confounder, disproportionately affected the recruitment of male participants, in whom smoking prevalence is high. Therefore, validation in larger, mixed-gender cohorts is warranted. Third, we also acknowledge that the inclusion of a periodontally healthy subgroup would have been ideal for a more precise comparison. However, this was not feasible in the current study due to the exceptionally high prevalence of periodontal disease (over 90%) among Chinese adults [57], making the recruitment of periodontally healthy middle-aged individuals challenging. Future studies should make a concerted effort to include such subgroups to better differentiate disease-specific mechanisms. Fourth, we noted BMI differences between groups, a potential confounder given the established link between obesity, systemic inflammation, and gut dysbiosis. It is noteworthy that our RA patients exhibited lower BMI than controls, a finding that contrasts with meta-analyses linking higher BMI to increased RA risk [58, 59]. This discrepancy is likely because our study design, which excluded individuals with BMI ≥ 28 kg/m², resulted in a metabolically healthy cohort—over 90% of participants had a BMI < 25 (per WHO criteria). This restricted range substantially attenuated confounding by obesity. Nevertheless, residual confounding by BMI within this restricted range cannot be entirely ruled out and should be considered when interpreting the findings.

Conclusion

In summary, our integrated analysis suggests that periodontitis may modify the microbiome in rheumatoid arthritis through niche-specific dysregulation. Saliva emerged as a particularly informative niche. We observed alterations in multiple taxa, with Porphyromonas demonstrating the most consistent and dual positive correlations with both periodontal disease severity and key RA serological markers ACPA. This pattern was distinct from the site-specific associations of the subgingival plaque microbiota and the inverse correlations observed in the gut. The stronger systemic associations of the salivary microbiota highlight its clinical value as a holistic and non-invasive sample for investigating oral-systemic connections. Collectively, our findings provide a foundational reference for the field, highlighting salivary Porphyromonas as a candidate biomarker worthy of further investigation. They reinforce the importance of considering periodontal health in RA management and underscore the necessity of multi-niche microbiome analysis to fully unravel the oral-systemic connection in RA.

Abbreviations

ACPA

Anti-citrullinated protein antibody

BMI

Body mass index

BOP

Bleeding on probing

CAL

Clinical attachment level

DAS28

Disease activity score in 28 joints

MMP

Matrix metalloproteinase

NET

Neutrophil extracellular trap

PAD

Peptidylarginine deiminase

PD

Probing depth

PLI

Plaque index

RA

Rheumatoid arthritis

RF

Rheumatoid factor

Authors’ contributions

XW : Writing—original draft, conceptualization, methodology. TL: Writing—original draft, Formal Analysis, Investigation, Software. LS1: Writing—review & editing, Data curation, Investigation, Visualization. YH : Writing—review & editing, Software, Supervision. YZ: Writing—review & editing, Visualization. ZW: Writing—review & editing, Supervision. KY: Writing—review & editing, Data curation, Validation. FD: Writing—review & editing, Project administration, Resources. LS2 : Writing—review & editing, Funding acquisition, Supervision, Conceptualization. All the authors read and approved the final manuscript.

Funding

This work was financially supported by the National Natural Science Foundation of China (no. 82460196), the Key R&D Program of Jiangxi Province, China (no. 20252BCG330024), the Double Thousand Talents Project of Jiangxi Province (no. jxsq2023201045) and the High-level and High-skilled Leading Talent Training Project of Jiangxi Province (no. G/Y3034).

Data availability

The raw sequencing data were deposited at NGDC (the National Genomics Data Center) with project number PRJCA045624, and accession number CRA029936, https://ngdc.cncb.ac.cn/gsa/s/m4qcmviT.

Declarations

Ethics approval and consent to participate

The study protocol involving human participants was conducted in accordance with the principles of the Declaration of Helsinki. It was reviewed and approved by the Ethics Committee of the Second Affiliated Hospital of Nanchang University (Reference No. O-MedResEthRev No. (17); MR-36-24-041017). The trial was registered with the Chinese Clinical Trial Registry (Registration No. ChiCTR2500108690). All the individuals provided written informed consent prior to enrollment.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.