The subgingival microbiome in patients with established rheumatoid arthritis

The subgingival microbiome in patients with established rheumatoid arthritis Abstract Objectives To profile and compare the subgingival microbiome of RA patients with OA controls. Methods RA (n = 260) and OA (n = 296) patients underwent full-mouth examination and subgingival samples were collected. Bacterial DNA was profiled using 16 S rRNA Illumina sequencing. Following data filtering and normalization, hierarchical clustering analysis was used to group samples. Multivariable regression was used to examine associations of patient factors with membership in the two largest clusters. Differential abundance between RA and OA was examined using voom method and linear modelling with empirical Bayes moderation (Linear Models for Microarray Analysis, limma), accounting for the effects of periodontitis, race, marital status and smoking. Results Alpha diversity indices were similar in RA and OA after accounting for periodontitis. After filtering, 286 taxa were available for analysis. Samples grouped into one of seven clusters with membership sizes of 324, 223, 3, 2, 2, 1 and 1 patients, respectively. RA-OA status was not associated with cluster membership. Factors associated with cluster 1 (vs 2) membership included periodontitis, smoking, marital status and Caucasian race. Accounting for periodontitis, 10 taxa (3.5% of those examined) were in lower abundance in RA than OA. There were no associations between lower abundance taxa or other select taxa examined with RA autoantibody concentrations. Conclusion Leveraging data from a large case–control study and accounting for multiple factors known to influence oral health status, results from this study failed to identify a subgingival microbial fingerprint that could reliably discriminate RA from OA patients. periodontitis, subgingival microbiome, rheumatoid arthritis, osteoarthritis Rheumatology key messages Patients with established RA and OA appear to harbour similar subgingival microbial communities. Factors most closely associated with subgingival microbial composition in this study included periodontitis, smoking, marital status and race. Introduction Previous reports have implicated periodontitis in the aetiopathogenesis of RA [1–11]. Based on the high prevalence of periodontal dysbiosis in periodontitis, there has been interest in elucidating the role that specific microorganisms could play in the link between periodontitis and RA risk. Porphyromonas gingivalis, a Gram-negative anaerobe that is strongly associated with periodontitis, has perhaps garnered the most attention. Unique among prokaryotes, P. gingivalis expresses a peptidyl-arginine deiminase that catalyses protein citrullination [12]. This organism has further demonstrated the capacity to citrullinate proteins in vitro, generating antigens implicated in RA pathogenesis [13]. Recently, investigators have shown robust associations of antibody recognizing Aggregatibacter actinomycetemcomitans to RA risk, in addition to demonstrating that this organism induces hypercitrullination in host neutrophils generating a citrullinome paralleling that of the RA joint [14]. In another study, the subgingival presence of Prevotella and Leptotrichia species were overrepresented in treatment-naïve RA patients compared with healthy controls, irrespective of underlying periodontitis status [10]. To date, most studies examining the role of periodontal pathogens in RA have been highly selective, examining either one or a limited number of pre-specified microbes [1, 3, 8]. Given inherent difficulties in culturing these organisms, most studies have also used serologic testing as surrogates of bacterial exposure. Thus, data generated using non-biased approaches to fully characterize the oral microbiome in RA are limited [10, 15]. In one such study, investigators characterized the subgingival microbiome samples from 65 RA patients [10], while another used dental plaque to characterize the microbiome of 77 cases [15]. Using healthy controls for comparison, both studies demonstrated unique, but non-overlapping, microbiome characteristics specific to RA. To what degree other confounders such as smoking or sociodemographics might have contributed to these findings remains unclear. Leveraging data and subgingival samples from a large, well-characterized case–control study, we characterized the microbiome of patients with established RA and diseased controls with OA [8]. We sought to identify whether RA patients harboured unique subgingival microbial signatures after accounting for the effects of other factors that could influence the oral microbiota, including periodontitis status. We hypothesized that patients with established RA would demonstrate a unique subgingival microbiome when compared with OA controls and that this difference would be independent of periodontitis status. If true, these results would inform future research strategies of RA treatment and/or prevention that could conceivably target specific oral pathogens. Methods Study participants and clinical assessments Study subjects included participants in a multicentre case–control investigation examining the relationship of periodontitis with RA [8]. There were 617 study participants (287 RA cases and 330 OA controls) enrolled from from five participating centres. All cases satisfied the 1987 ACR classification criteria for RA with an age of disease onset >18 years [16]. OA in controls was based on medical documentation from a corresponding healthcare provider or imaging results consistent with degenerative arthritis. With demographics similar to cases, we used OA patients as comparators to minimize differences between RA cases and healthy control populations employed in other studies [10, 15]. The study was approved by the Institutional Review Board at each centre and all participants provided written informed consent prior to study enrolment. Clinical assessments and sample collection Cases and controls underwent a standardized and calibrated full mouth periodontal evaluation with periodontitis defined [17]. Subgingival plaque was collected from up to four mesiobuccal sites (one per quadrant) from posterior teeth from all but one participant (n = 616). Following removal of visible supragingival plaque, subgingival plaque samples were collected using a single sterile endodontic paper point for each site [18]. Up to four samples were pooled for each patient and frozen at −70°C until analysis. Serum was collected and RA cases were assessed for circulating concentrations of ACPA using a second-generation ELISA (Diastat, Axis-Shield; positivity ⩾5 U/ml). RF (positive ⩾15 IU/ml) and high-sensitivity CRP (mg/l) were determined by nephelometry (Siemens Healthcare Diagnostics, Tarrytown, NJ, USA). Disease activity in RA patients was quantified through the measurement of tender and swollen joint counts (0–28), and provider and patient global well-being scores (0–100 mm visual analogue scales). A DAS-28-CRP was calculated [19]. Subgingival microbiome identification Genomic DNA was extracted from each subgingival sample using the Mobio Ultraclean Microbial DNA extraction kit (Carlsbad, CA, USA). For each sample, bacterial 16 S rRNA V1–V3 gene segments were amplified in duplicate using composite 27 F (5′ AGAGTTTGATCCTGGCTCAG 3′) and 534 R (5′ ATTACCGCGGCTGCTGG 3′) primers. The forward and the reverse primers each contained unique barcodes to allow multiplex deep sequencing. Each 20 µl PCR contained 2 µl of the purified DNA template, 1× Accuprime PCR buffer II, 0.2 µM of the forward primer, 0.2 µM of the reverse primer and 1.0 U of Accuprime Taq High Fidelity polymerase. PCR amplification was performed as follows: denaturation at 95°C for 2 min, followed by 25 cycles of denaturation at 95°C for 20 s, annealing at 56°C for 30 s and extension at 72°C for 5 min. The replicate barcoded PCR products for each sample were pooled and analysed on 1.5% SYBR Safe (Invitrogen, Carlsbad, CA, USA) agarose gel. Gel slices containing the amplicons of expected size (∼600 bp) were excised and purified using the Qiagen gel extraction kit (Qiagen, Valencia, CA, USA). Purified PCR products were quantified using Qubit HS DNA quantification kit (Invitrogen, Carlsbad, CA, USA) and pooled with equal molar concentration. The use of barcodes allowed multiplexing and bidirectional sequencing on the Illumina MiSeq platform (Illumina, San Diego, CA, USA) in four pools, generating 16 145 880 raw reads. After trimming primer and barcode sequences and quality control, 11 039 494 V1–V3 16 S rRNA gene sequence reads were available for analysis (averaging 19 855 ± 6809 reads per sample and an amplicon length of 487 nt). Good’s coverage estimates were >0.95 for all samples. Rarefaction analysis indicates that sampling effort was sufficient for the majority of the samples, and further sampling would yield few additional taxa. Paired-end raw Illumina sequence reads were joined using Fast Length Adjustment of Short reads (https://ccb.jhu.edu/software/FLASH/) and filtered with custom scripts written in R (https://www.r-project.org/) using the following quality criteria: minimum overlap of 10 bp from both paired ends, exact match to barcode and primer, no ambiguous bases and an average quality score of 30. Barcodes were variable lengths (4–8 bp) for each paired end and formed a unique combination for identification in each pool. In downstream analysis, the barcodes and primers (27 F and 534 R) were trimmed. Trimmed reads were then searched against the Human Oral Microbiome Database (HOMD; non-extended version 10.1) [20] and an HOMD sequence (full-length) was assigned for each read based on the best hit for that read. The tag matching of the V1–V3 reads to full length HOMD sequences formed the basis of reference operational taxonomic unit (OTU), where the full-length HOMD sequences and taxonomy information were used in the analysis. Reads with <80% alignable coverage and <97% identity were excluded from additional analysis. Statistical analysis Descriptive statistics including relative frequency for categorical variables and mean (s.d.) for continuous variables were calculated. χ2 and non-parametric Wilcoxon rank sum tests were used to compare group characteristics. Sequencing data were filtered to exclude taxa for which >75% of patients in all four groups defined by RA-OA and periodontitis status had zero values. Alpha diversity indices were computed using the filtered data, and compared by group (RA-periodontitis, RA-no periodontitis, OA-periodontitis and OA-no periodontitis) via calculation of: total number of OTUs; Chao index, reflecting the total number of species (microbial richness); evenness, reflecting the relative abundance of species; Shannon diversity index, based on both evenness and richness (larger values reflecting more diverse microbiota); and Good’s coverage estimator, reflecting the total percent of species represented. The Shapiro-Wilk test was used to assess normality. Based on the normality testing, Shannon index, evenness and Good’s estimator were compared using the non-parametric Kruskal-Wallis test with Bonferroni’s correction for pairwise comparisons. The number of OTUs and Chao index were compared using one-way analysis of variance with a Tukey-Kramer post hoc test used for pairwise comparisons. Additional analyses were undertaken to examine associations of RA-related factors including disease activity (DAS-28-CRP), disease duration and select medications (prednisone, MTX or biologics) with measures of microbial diversity. Using data limited to patients with RA, Pearson or Spearman correlations were calculated (as appropriate based on normality) to examine associations of the aforementioned alpha diversity indices with DAS-28-CRP and disease duration. Additionally, two-sample t-tests or Wilcoxon rank sum tests were used as appropriate to compare indices among RA patients based on the use of prednisone, MTX or any biologic disease-modifying agent. Filtered raw count sequencing data were normalized with the method of trimmed mean of M-values and converted to log2-counts per million per sample by voom function built in the Linear Models for Microarray Analysis (limma) package in Bioconductor [20–22]. Two approaches were used to compare subgingival microbial communities by group. An unsupervised cluster analysis was completed and logistic regression models were generated to first examine prevailing microbiome patterns and associations with RA-OA status [23]. Specifically, hierarchical cluster analysis with averaging-linkage agglomeration on microbiome profiles was used to classify samples into clusters. Stepwise logistic regression was then used to identify variables associated with membership in clusters 1 vs 2, the two largest cluster memberships (or patient groups). The significance level for entry and removal of variables were set to be 0.05. RA vs OA disease status was retained in the model regardless of P-values. In addition to the presence of periodontitis, other factors examined included age, sex, self-reported race/ethnicity (Caucasian vs other), smoking status (ever vs never), BMI (kg/m2), self-reported diabetes, education and marital status. A heat-map was generated to illustrate OTUs with at least a 2-fold difference in abundance between clusters 1 and 2 using Genesis 1.8.1 (Graz University of Technology, Graz, Austria). Using a second approach, differential abundance analysis was conducted using the limma package [20–22]. The analysis was conducted using two linear models: the first fitted with RA vs OA status, periodontitis status and their interaction (base model); and the second fitted with the same factors as in the base model in addition to race, marital status and smoking (fully adjusted model). The Bejamini-Hochberg method was used to control the false discovery rate to be no >0.05 for multiple hypotheses correction [24]. Principal component analysis plots were generated as an additional means of examining the associations of patient characteristics with subgingival micobiome composition. Microbial complexes (red, orange, yellow and purple) were defined according to Socransky et al. [25], with red and orange complexes most closely linked to periodontitis while yellow and purple complexes have been more closely associated with oral health. With reports of differential abundance of Haemophilus, Prevotella, Leptotrichia and possibly Aggregatibacter species in RA patients vs controls [10, 14, 15], additional analyses explored differential abundance of P. gingivalis and other species at the genus level. In addition to P. gingivalis (OTU 619), OTUs including Haemophilus, Prevotella, Leptotrichia and Aggregatibacter were identified in filtered normalized data and median normalized counts were calculated for each of the four groups. A linear model was fitted with RA-OA disease status, periodontitis and the interaction of these, in addition to race, marital status and smoking. Spearman correlations assessed the relationship between autoantibody (RF and anti-CCP) concentrations and normalized OTUs. To limit the possibility of type I error, autoantibody correlations were examined for OTUs based on prior reports suggesting associations of specific oral bacteria (P. gingivalis, Haemophilus and Aggregatibacter species, in addition to Anaeroglobus geminatus) with autoantibody expression [10, 14, 15, 26, 27], as well as OTUs found in lower abundance between RA cases and OA controls. Results Of the 616 patients providing samples, 556 (90%) had sufficient DNA for sequencing. Characteristics for RA (n = 260) and OA (n = 296) patients are summarized in Table 1. Compared with those with OA, patients with RA were more likely to be smokers, had lower BMIs and were less likely to have diabetes. As previously reported in the full cohort [10], RA patients included in this analysis were more likely to have periodontitis than OA patients (38% vs 27%, P = 0.007). Table 1 RA and OA patient characteristics Characteristics RA (n = 260) OA (n = 296) P-value Demographics and comorbidity Age, mean (s.d.), years 59 (12) 60 (11) 0.68 Men, % 65 61 0.35 Race, % 0.25     Caucasian 77 72     African American 17 23     Other 6 5 Married, % 69 61 0.06 Ever smoking, % 62 46 <0.001 Periodontitis, % 38 27 0.007 BMI, mean (s.d.), kg/m2 29.8 (6.7) 31.7 (6.7) <0.001 Diabetes mellitus, % 18 25 0.04 RA disease characteristics     RF positive, % 78 – –     Anti-CCP antibody positive, % 85 – –     DAS-28-CRP, mean (s.d.) 3.3 (1.3) – –     MTX use, % 63 – –     Prednisone use, % 29 – –     Current biologic use, % 32 – – Characteristics RA (n = 260) OA (n = 296) P-value Demographics and comorbidity Age, mean (s.d.), years 59 (12) 60 (11) 0.68 Men, % 65 61 0.35 Race, % 0.25     Caucasian 77 72     African American 17 23     Other 6 5 Married, % 69 61 0.06 Ever smoking, % 62 46 <0.001 Periodontitis, % 38 27 0.007 BMI, mean (s.d.), kg/m2 29.8 (6.7) 31.7 (6.7) <0.001 Diabetes mellitus, % 18 25 0.04 RA disease characteristics     RF positive, % 78 – –     Anti-CCP antibody positive, % 85 – –     DAS-28-CRP, mean (s.d.) 3.3 (1.3) – –     MTX use, % 63 – –     Prednisone use, % 29 – –     Current biologic use, % 32 – – Table 1 RA and OA patient characteristics Characteristics RA (n = 260) OA (n = 296) P-value Demographics and comorbidity Age, mean (s.d.), years 59 (12) 60 (11) 0.68 Men, % 65 61 0.35 Race, % 0.25     Caucasian 77 72     African American 17 23     Other 6 5 Married, % 69 61 0.06 Ever smoking, % 62 46 <0.001 Periodontitis, % 38 27 0.007 BMI, mean (s.d.), kg/m2 29.8 (6.7) 31.7 (6.7) <0.001 Diabetes mellitus, % 18 25 0.04 RA disease characteristics     RF positive, % 78 – –     Anti-CCP antibody positive, % 85 – –     DAS-28-CRP, mean (s.d.) 3.3 (1.3) – –     MTX use, % 63 – –     Prednisone use, % 29 – –     Current biologic use, % 32 – – Characteristics RA (n = 260) OA (n = 296) P-value Demographics and comorbidity Age, mean (s.d.), years 59 (12) 60 (11) 0.68 Men, % 65 61 0.35 Race, % 0.25     Caucasian 77 72     African American 17 23     Other 6 5 Married, % 69 61 0.06 Ever smoking, % 62 46 <0.001 Periodontitis, % 38 27 0.007 BMI, mean (s.d.), kg/m2 29.8 (6.7) 31.7 (6.7) <0.001 Diabetes mellitus, % 18 25 0.04 RA disease characteristics     RF positive, % 78 – –     Anti-CCP antibody positive, % 85 – –     DAS-28-CRP, mean (s.d.) 3.3 (1.3) – –     MTX use, % 63 – –     Prednisone use, % 29 – –     Current biologic use, % 32 – – Alpha diversity indices across the four groups defined by RA-OA and periodontitis case status are shown in Table 2. Although both OTU numbers and the Chao index were higher in those with periodontitis than those without, these were not significantly impacted by RA-OA status. There were no significant differences by group for the Shannon index, evenness or Good’s coverage estimator. In additional analyses limited to RA patients, there were no correlations between the alpha diversity indices and DAS-28-CRP or disease duration, with the exception of a weak, but statistically significant, positive correlation between DAS-28-CRP and evenness (r = 0.15, P = 0.02). Likewise, we found no evidence of an association in RA patients between treatments received (prednisone, MTX or biologic) and microbial diversity (data not shown). Table 2 Alpha diversity measures in patients with RA and OA with and without evidence of periodontitis Measure RA-periodontitis (n = 99) RA-no periodontitis (n = 161) OA-periodontitis (n = 81) OA-no periodontitis (n = 215) P-valuea Number OTUsa, mean (s.d.) 151.9 (38.5) 138.6 (38.4) 158.5 (41.3) 140.8 (37.8) 0.0002 Chao indexa, mean (s.d.) 169.3 (41.6) 155.6 (41.8) 177.3 (45.5) 157.2 (40.1) 0.0002 Shannon indexb, median (range) 3.37 (1.95–4.45) 3.42 (1.42–4.45) 3.43 (1.54–4.26) 3.32 (1.74–4.26) 0.53 Evennessb, median (range) 0.67 (0.42–0.82) 0.70 (0.37–0.82) 0.68 (0.31–0.80) 0.68 (0.38–0.80) 0.11 Good’sb, median (range) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 0.45 Measure RA-periodontitis (n = 99) RA-no periodontitis (n = 161) OA-periodontitis (n = 81) OA-no periodontitis (n = 215) P-valuea Number OTUsa, mean (s.d.) 151.9 (38.5) 138.6 (38.4) 158.5 (41.3) 140.8 (37.8) 0.0002 Chao indexa, mean (s.d.) 169.3 (41.6) 155.6 (41.8) 177.3 (45.5) 157.2 (40.1) 0.0002 Shannon indexb, median (range) 3.37 (1.95–4.45) 3.42 (1.42–4.45) 3.43 (1.54–4.26) 3.32 (1.74–4.26) 0.53 Evennessb, median (range) 0.67 (0.42–0.82) 0.70 (0.37–0.82) 0.68 (0.31–0.80) 0.68 (0.38–0.80) 0.11 Good’sb, median (range) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 0.45 a Values satisfying normality assumption compared by group using ANOVA with Tukey-Kramer correction. Significant pairwise comparisons for the number of OTUs were: RA-periodontitis vs RA-no periodontitis (P = 0.04), OA-periodontitis vs RA-no periodontitis (P = 0.001) and OA-periodontitis vs OA-no periodontitis (P = 0.003); all other comparisons non-significant. Significant pairwise comparisons for the Chao index were: RA-periodontitis vs RA-no periodontitis (P = 0.05), OA-periodontitis vs RA-no periodontitis (P = 0.0008) and OA-periodontitis vs OA-no periodontitis (P = 0.001). b Values not satisfying normality assumption compared using Kruskal-Wallis test. OTU: operational taxonomic unit. Table 2 Alpha diversity measures in patients with RA and OA with and without evidence of periodontitis Measure RA-periodontitis (n = 99) RA-no periodontitis (n = 161) OA-periodontitis (n = 81) OA-no periodontitis (n = 215) P-valuea Number OTUsa, mean (s.d.) 151.9 (38.5) 138.6 (38.4) 158.5 (41.3) 140.8 (37.8) 0.0002 Chao indexa, mean (s.d.) 169.3 (41.6) 155.6 (41.8) 177.3 (45.5) 157.2 (40.1) 0.0002 Shannon indexb, median (range) 3.37 (1.95–4.45) 3.42 (1.42–4.45) 3.43 (1.54–4.26) 3.32 (1.74–4.26) 0.53 Evennessb, median (range) 0.67 (0.42–0.82) 0.70 (0.37–0.82) 0.68 (0.31–0.80) 0.68 (0.38–0.80) 0.11 Good’sb, median (range) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 0.45 Measure RA-periodontitis (n = 99) RA-no periodontitis (n = 161) OA-periodontitis (n = 81) OA-no periodontitis (n = 215) P-valuea Number OTUsa, mean (s.d.) 151.9 (38.5) 138.6 (38.4) 158.5 (41.3) 140.8 (37.8) 0.0002 Chao indexa, mean (s.d.) 169.3 (41.6) 155.6 (41.8) 177.3 (45.5) 157.2 (40.1) 0.0002 Shannon indexb, median (range) 3.37 (1.95–4.45) 3.42 (1.42–4.45) 3.43 (1.54–4.26) 3.32 (1.74–4.26) 0.53 Evennessb, median (range) 0.67 (0.42–0.82) 0.70 (0.37–0.82) 0.68 (0.31–0.80) 0.68 (0.38–0.80) 0.11 Good’sb, median (range) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 0.45 a Values satisfying normality assumption compared by group using ANOVA with Tukey-Kramer correction. Significant pairwise comparisons for the number of OTUs were: RA-periodontitis vs RA-no periodontitis (P = 0.04), OA-periodontitis vs RA-no periodontitis (P = 0.001) and OA-periodontitis vs OA-no periodontitis (P = 0.003); all other comparisons non-significant. Significant pairwise comparisons for the Chao index were: RA-periodontitis vs RA-no periodontitis (P = 0.05), OA-periodontitis vs RA-no periodontitis (P = 0.0008) and OA-periodontitis vs OA-no periodontitis (P = 0.001). b Values not satisfying normality assumption compared using Kruskal-Wallis test. OTU: operational taxonomic unit. Hierarchical cluster analysis with averaging-linkage agglomeration on microbiome profiles was used to classify samples into one of seven unique clusters. The number of memberships of clusters 1–7 was 324, 223, 3, 2, 2, 1 and 1, respectively. With only nine patients with membership, clusters 3–7 were excluded from the further analysis. OTUs with >2-fold difference in abundance between clusters are shown (Fig. 1). Bacteria comprising red and orange complexes [25], thought to represent periodontitis pathogens, were generally found in greater abundance in clusters 2 vs 1. All three red complex bacteria (P. gingivalis, Tannerella forsythia and Treponema denticola) were found in ⩾2-fold abundance in cluster 2 compared with cluster 1, as were two of three orange complex bacteria (Eubacterium nodatum and Streptococcus constellatus). Bacteria from the yellow and purple complexes, reflecting early colonizers and a healthier periodontium, were universally found in greater abundance in cluster 1 vs 2. Fig. 1 View largeDownload slide View largeDownload slide View largeDownload slide Heatmap demonstrating clusters 1 and 2 taxa Heatmap of operational taxonomic units (OTUs) demonstrating more than a 2-fold difference in abundance between cluster 1 (c1) and cluster 2 (c2). Panels shown are from single contiguous heatmap. Fig. 1 View largeDownload slide View largeDownload slide View largeDownload slide Heatmap demonstrating clusters 1 and 2 taxa Heatmap of operational taxonomic units (OTUs) demonstrating more than a 2-fold difference in abundance between cluster 1 (c1) and cluster 2 (c2). Panels shown are from single contiguous heatmap. Using stepwise logistic regression, RA case status was not significantly associated with membership in clusters 1 vs 2 [odds ratio (OR) = 1.18; 95% CI: 0.80, 1.73] (Table 3). Being married (OR = 1.57; 95% CI: 1.06, 2.33) and Caucasian race (OR = 1.94; 95% CI: 1.26, 2.99) were associated with an increased odds of membership in cluster 1 vs 2. In contrast, periodontitis (OR = 0.22; 95% CI: 0.15, 0.33) and ever smoking (OR = 0.65; 95% CI: 0.44, 0.96) were associated with a lower likelihood of membership in cluster 1. Table 3 Associations of patient factors with cluster membership (cluster 1 vs 2) based on subgingival microbiome composition Patient factor Odds ratio (95% CI) P-value RA vs OA 1.18 (0.80, 1.73) 0.40 Periodontitis vs no periodontitis 0.22 (0.15, 0.33) <0.001 Ever vs never smoking 0.65 (0.44, 0.96) 0.03 Married vs not married 1.57 (1.06, 2.33) 0.03 Caucasian vs other 1.94 (1.26, 2.99) 0.003 Patient factor Odds ratio (95% CI) P-value RA vs OA 1.18 (0.80, 1.73) 0.40 Periodontitis vs no periodontitis 0.22 (0.15, 0.33) <0.001 Ever vs never smoking 0.65 (0.44, 0.96) 0.03 Married vs not married 1.57 (1.06, 2.33) 0.03 Caucasian vs other 1.94 (1.26, 2.99) 0.003 Table 3 Associations of patient factors with cluster membership (cluster 1 vs 2) based on subgingival microbiome composition Patient factor Odds ratio (95% CI) P-value RA vs OA 1.18 (0.80, 1.73) 0.40 Periodontitis vs no periodontitis 0.22 (0.15, 0.33) <0.001 Ever vs never smoking 0.65 (0.44, 0.96) 0.03 Married vs not married 1.57 (1.06, 2.33) 0.03 Caucasian vs other 1.94 (1.26, 2.99) 0.003 Patient factor Odds ratio (95% CI) P-value RA vs OA 1.18 (0.80, 1.73) 0.40 Periodontitis vs no periodontitis 0.22 (0.15, 0.33) <0.001 Ever vs never smoking 0.65 (0.44, 0.96) 0.03 Married vs not married 1.57 (1.06, 2.33) 0.03 Caucasian vs other 1.94 (1.26, 2.99) 0.003 Two principal components, PC1 (12.3%) and PC2 (8.9%), together explained 21.2% of the bacterial variation. Principal component analysis plots did not demonstrate evidence of clustering based on RA-OA case status, but showed variable degrees of clustering based on marital status, race, periodontitis and cigarette smoking (Fig. 2). Fig. 2 View largeDownload slide Principal component analysis plots Two principal components (PC1 and PC2) explained most of the bacterial variation observed in subgingival samples. PCA plots examining the relationship of RA vs OA case status, marital status, race, periodontitis and smoking with microbiome composition are shown. Plots did not demonstrate evidence of clustering based on RA-OA case status, but showed variable degrees of clustering based on marital status, race, periodontitis and cigarette smoking. PCA: principal component analysis. Fig. 2 View largeDownload slide Principal component analysis plots Two principal components (PC1 and PC2) explained most of the bacterial variation observed in subgingival samples. PCA plots examining the relationship of RA vs OA case status, marital status, race, periodontitis and smoking with microbiome composition are shown. Plots did not demonstrate evidence of clustering based on RA-OA case status, but showed variable degrees of clustering based on marital status, race, periodontitis and cigarette smoking. PCA: principal component analysis. Excluding taxa for which >75% of patients in all four groups had zero values, there were 286 OTUs available for analysis. Among those with periodontitis, there were 10 OTUs (3.5% of OTUs examined) in lower abundance in RA vs OA samples in the base model with false discovery rate-adjusted P-values. These taxa included Catonella sp. (OTU 451), Clostridiales sp. (OTU 85), Lachnospiraceae sp. (OTU 96), Peptostreptococcaceae sp. (OTU 495), Porphyromonas sp. (OTU 285), Prevotella multiformis (OTU 685), Prevotella sp. (OTU 443), Selenomaonas sp. (OTU 479) and Treponema sp. (OTUs 230 and 236) (Table 4). Results were unchanged following full adjustment, with the exception that differential expression of Selenomaonas sp. was no longer significant. Although we observed evidence to suggest the under-abundance of Porphyromonas sp. in RA patients with periodontitis compared with OA patients with periodontitis, there was not similar evidence of differential expression for P. gingivalis (data not shown). In subjects without evidence of periodontitis, only one OTU was found in lower abundance in RA samples and in the fully adjusted model after accounting for false discovery rate, Streptococcus sp. (OTU 486), a difference that did not achieve statistical significance in the base model. There were no OTUs identified in the filtered data consistent with A. actinomycetemcomitans. Table 4 Differential expression of OTUs between RA and OA patients based on periodontitis status Base modela Fully adjusted modela OTUs with differential expression Fold difference (RA vs OA) P-value (FDR adjusted) Fold difference (RA vs OA) P-value (FDR adjusted) Patients with periodontitis  Catonella sp. (OTU 451) 0.424 0.002 0.466 0.005     Clostridiales sp. (OTU 85) 0.486 0.002 0.511 0.005  Lachnospiraceae sp. (OTU 96) 0.507 0.010 0.484 0.005  Peptostreptococcaceae sp. (OTU 495) 0.590 0.012 0.596 0.012  Porphyromonas sp. (OTU 285) 0.520 0.019 0.535 0.026  Prevotella multiformis (OTU 685) 0.456 0.009 0.469 0.005  Prevotella sp. (OTU 443) 0.523 0.019 0.540 0.022  Selenomaonas sp. (OTU 479) 0.589 0.039 0.617 0.081  Treponema sp. (OTU 230) 0.467 0.002 0.500 0.005  Treponema sp. (OTU 236) 0.429 0.019 0.465 0.035 Patients without periodontitis  Streptococcus sp. (OTU 486) 0.705 0.086 0.689 0.042 Base modela Fully adjusted modela OTUs with differential expression Fold difference (RA vs OA) P-value (FDR adjusted) Fold difference (RA vs OA) P-value (FDR adjusted) Patients with periodontitis  Catonella sp. (OTU 451) 0.424 0.002 0.466 0.005     Clostridiales sp. (OTU 85) 0.486 0.002 0.511 0.005  Lachnospiraceae sp. (OTU 96) 0.507 0.010 0.484 0.005  Peptostreptococcaceae sp. (OTU 495) 0.590 0.012 0.596 0.012  Porphyromonas sp. (OTU 285) 0.520 0.019 0.535 0.026  Prevotella multiformis (OTU 685) 0.456 0.009 0.469 0.005  Prevotella sp. (OTU 443) 0.523 0.019 0.540 0.022  Selenomaonas sp. (OTU 479) 0.589 0.039 0.617 0.081  Treponema sp. (OTU 230) 0.467 0.002 0.500 0.005  Treponema sp. (OTU 236) 0.429 0.019 0.465 0.035 Patients without periodontitis  Streptococcus sp. (OTU 486) 0.705 0.086 0.689 0.042 a Differential expression analysis conducted using two separate linear models: the first fitted with RA vs OA status, periodontitis status and the interaction between these factors (base model); and the second fitted with the same factors as the base model in addition to race, marital status and smoking (fully adjusted model). OTU: operational taxonomic unit; FDR: false discovery rate. Table 4 Differential expression of OTUs between RA and OA patients based on periodontitis status Base modela Fully adjusted modela OTUs with differential expression Fold difference (RA vs OA) P-value (FDR adjusted) Fold difference (RA vs OA) P-value (FDR adjusted) Patients with periodontitis  Catonella sp. (OTU 451) 0.424 0.002 0.466 0.005     Clostridiales sp. (OTU 85) 0.486 0.002 0.511 0.005  Lachnospiraceae sp. (OTU 96) 0.507 0.010 0.484 0.005  Peptostreptococcaceae sp. (OTU 495) 0.590 0.012 0.596 0.012  Porphyromonas sp. (OTU 285) 0.520 0.019 0.535 0.026  Prevotella multiformis (OTU 685) 0.456 0.009 0.469 0.005  Prevotella sp. (OTU 443) 0.523 0.019 0.540 0.022  Selenomaonas sp. (OTU 479) 0.589 0.039 0.617 0.081  Treponema sp. (OTU 230) 0.467 0.002 0.500 0.005  Treponema sp. (OTU 236) 0.429 0.019 0.465 0.035 Patients without periodontitis  Streptococcus sp. (OTU 486) 0.705 0.086 0.689 0.042 Base modela Fully adjusted modela OTUs with differential expression Fold difference (RA vs OA) P-value (FDR adjusted) Fold difference (RA vs OA) P-value (FDR adjusted) Patients with periodontitis  Catonella sp. (OTU 451) 0.424 0.002 0.466 0.005     Clostridiales sp. (OTU 85) 0.486 0.002 0.511 0.005  Lachnospiraceae sp. (OTU 96) 0.507 0.010 0.484 0.005  Peptostreptococcaceae sp. (OTU 495) 0.590 0.012 0.596 0.012  Porphyromonas sp. (OTU 285) 0.520 0.019 0.535 0.026  Prevotella multiformis (OTU 685) 0.456 0.009 0.469 0.005  Prevotella sp. (OTU 443) 0.523 0.019 0.540 0.022  Selenomaonas sp. (OTU 479) 0.589 0.039 0.617 0.081  Treponema sp. (OTU 230) 0.467 0.002 0.500 0.005  Treponema sp. (OTU 236) 0.429 0.019 0.465 0.035 Patients without periodontitis  Streptococcus sp. (OTU 486) 0.705 0.086 0.689 0.042 a Differential expression analysis conducted using two separate linear models: the first fitted with RA vs OA status, periodontitis status and the interaction between these factors (base model); and the second fitted with the same factors as the base model in addition to race, marital status and smoking (fully adjusted model). OTU: operational taxonomic unit; FDR: false discovery rate. Based on recent reports, we examined associations of all OTUs including Haemophilus, Prevotella, Leptotrichia and Aggregatibacter with RA. There were 4 OTUs including Haemophilus, 39 including Prevotella, 15 including Leptotrichia and 4 including Aggregatibacter identified in the filtered normalized data. Among those with periodontitis, only OTUs including Prevotella were differentially expressed in RA vs OA patients (0.68-fold difference, P = 0.006). With reports also showing associations of select pathogens with autoantibody expression in RA, we subsequently examined correlations of related OTUs with RF and anti-CCP antibody concentrations in RA patients (all patients and restricted to seropositive patients). We observed no significant correlations of A. geminatus or P. gingivalis, or with Haemophilus or Aggregatibacter containing OTUs with RF or anti-CCP antibody concentrations (data not shown). There were no associations of OTUs found in lower abundance in RA, shown in Table 4, with autoantibody concentrations (data not shown). Discussion Using samples and data from a large case–control study, we explored links between oral bacteria and RA by profiling both the abundance and diversity of the subgingival microbiome in a large group of patients with established disease, using OA patients for comparison. Results from this study failed to support the existence of a robust subgingival microbial fingerprint that could reliably be used to discriminate RA patients from those with OA. Scher et al. [10] were among the first to deploy 16 S rRNA pyrosequencing to characterize subgingival microbial composition in RA. Using samples from treatment-naïve patients (n = 31), patients with established RA (n = 34) and a smaller number of healthy controls (n = 18), the authors similarly observed no meaningful group differences referent to either microbial diversity or richness. The presence of periodontitis, more common in RA patients, was associated in our study with a higher number of OTUs and a higher Chao index (reflecting greater microbial richness). Similarly, Scher and colleagues found no evidence of microbial clustering based on RA case status, although the authors observed clustering based on the presence of moderate-to-severe periodontitis. In addition to replicating these results, we have extended these observations further by showing that other factors beyond the presence or absence of periodontitis appear to influence subgingival microbial composition. Subgingival samples from patients with a history of smoking, non-Caucasian race and those not married were more likely to segregate into a cluster characterized by an over-abundance of known pathogens. These results emphasize the importance of accounting for these factors in future research exploring the relationship between oral microbes and RA. Although others have also identified ethnic/racial background as a strong predictor of human microbiome composition [28], mechanisms underpinning this association (as well as associations with other socioeconomic factors such as marital status observed in the present study) are unknown and warrant further study [29]. Demonstrating similar microbial composition across groups, Scher et al. [10] identified Prevotella and Leptotrichia taxa in greater abundance in patients with untreated RA vs healthy controls independent of periodontitis. Using metagenomic shotgun sequencing of dental plaque samples, Zhang et al. [15] observed a decreased abundance of Haemophilus species and a simultaneous over-abundance of Lactobacillus salivarius in early RA patients vs controls. In contrast, we found a decreased subgingival abundance of OTUs including Prevotella in RA cases compared with controls. Results from these previous reports contrast with those from our study, which failed to identify an over-abundance of any taxa but identified an under-abundance of 10 different OTUs in RA patients relative to OA patients after accounting for periodontitis status and other factors. Of note, several of the OTUs identified to be in lower abundance have been implicated in periodontitis risk including Peptostreptococcus, Porphyromonas, Prevotella and Treponema species. Differences in the findings across studies may relate to the varied methods used (metagenomic shotgun sequencing vs 16 S rRNA 454 pyrosequencing vs Illumina sequencing) as well as the populations sampled. Although we failed to observe any significant associations of RA disease duration with microbial diversity in our study, marked differences in RA disease duration across studies could help to explain discrepancies in findings, with RA disease durations varying from a mean of just a few months [10] to >12 years in our study [8]. This difference may be particularly relevant as microbial dysbiosis reported to characterize untreated RA appears may be partially reversed with disease-modifying therapies [15]. Indeed, others found that red complex bacteria (T. forsythia and P. gingivalis) were present in greater abundance in the periodontium of early RA patients than in patients with established disease [10]. The authors speculated that this difference might have related to the effects of RA treatments, particularly agents shown to exert antimicrobial properties [30, 31]. Limited by its cross-sectional design, results from the present study do not support a major impact from treatments used, as we observed no associations of microbial diversity in RA patients with the use of prednisone, MTX or biologic therapies. Another possibility might be that RA patients mount more vigorous immune responses to select subgingival microbes, eventually leading to an under-abundance as observed in our study. Whether OTUs containing Prevotella or other OTUs found in greater abundance among controls in this study could somehow be protective of RA development is unknown. The association of select oral microbes with RA risk has been hypothesized to be driven by bacterially mediated citrullination leading to the local expression of ACPA. Previous studies have shown associations, for instance, of circulating antibodies to P. gingivalis and, more recently, to A. actinomycetemcomitans [14] with ACPA. Using bacterial 16 S rRNA sequencing as a direct measure of subgingival bacterial exposure, we did not detect OTUs corresponding to the presence of A. actinomycetemcomitans, a pathogen that has been most closely associated with localized aggressive forms of periodontitis in younger individuals [32], nor did we identify associations of P. gingivalis with ACPA. Separate studies performed in patients with early-onset disease yielded associations of alternative microbes, both Haemophilus species [15] and A. geminatus [10], with RA-related autoantibodies. Neither of these associations was replicated in the current study. It is possible that bacterially mediated ACPA production, if present, is more relevant in early disease, becoming less relevant over time as autoantibody production is established in other articular or extra-articular sites. To date, several reports have examined the concept of targeting periodontitis as a means of treating RA. With its cross-sectional and observational study design, this investigation alone does not fully address questions about the relationship of subgingival microbiome composition with RA disease progression, recognizing the need for additional studies involving RA patients earlier in the disease process. Similarities in subgingival bacterial communities between established RA and OA in the present study as well as the lack of any robust associations between RA disease activity and measures of microbial diversity, however, do not support the concept of an intervention targeting specific subgingival pathogens in patients with established RA. In summary, these results demonstrate that subgingival microbiome profiles in patients with established RA are similar to those observed in OA patients, and appear to be primarily influenced by other factors impacting oral health status. Funding: This work was supported by a Veterans Affairs Merit Award (to T.R.M.) [grant number CX000896] and a grant from the Rheumatology Research Foundation (to T.R.M.) [no grant number assigned]. Disclosure statement: The authors have declared no conflicts of interest. References 1 Bello-Gualtero JM , Lafaurie GI , Hoyos LX et al. Periodontal disease in individuals with a genetic risk of developing arthritis and early rheumatoid arthritis: a cross-sectional study . J Periodontol 2016 ; 87 : 346 – 56 . Google Scholar CrossRef Search ADS PubMed 2 de Pablo P , Dietrich T , McAlindon TE. Association of periodontal disease and tooth loss with rheumatoid arthritis in the US population . J Rheumatol 2008 ; 35 : 70 – 6 . Google Scholar PubMed 3 de Smit MD , Westra J , Vissink A et al. Periodontitis in established rheumatoid arthritis patients: a cross-sectional clinical, microbiological and serological study . Arthritis Res Ther 2012 ; 14 : R222 . Google Scholar CrossRef Search ADS PubMed 4 Dissick A , Redman RS , Jones M et al. Association of periodontitis with rheumatoid arthritis: a pilot study . J Periodontol 2010 ; 81 : 223 – 30 . Google Scholar CrossRef Search ADS PubMed 5 Kässer UR , Gleissner C , Dehne F et al. Risk for periodontal disease in patients with longstanding rheumatoid arthritis . Arthritis Rheum 1997 ; 40 : 2248 – 51 . Google Scholar CrossRef Search ADS PubMed 6 Malmstrom M , Calonius PE. Teeth loss and the inflammation of teeth-supporting tissues in rheumatoid disease . Scand J Rheumatol 1975 ; 4 : 49 – 55 . Google Scholar CrossRef Search ADS PubMed 7 Mercado FB , Marshall RI , Klestov AC , Bartold PM. Relationship between rheumatoid arthritis and periodontitis . J Periodontol 2001 ; 72 : 779 – 87 . Google Scholar CrossRef Search ADS PubMed 8 Mikuls TR , Payne JB , Yu F et al. Periodontitis and Porphyromonas gingivalis in patients with rheumatoid arthritis . Arthritis Rheumatol 2014 ; 66 : 1090 – 100 . Google Scholar CrossRef Search ADS PubMed 9 Pischon N , Pischon T , Kröger J et al. Association among rheumatoid arthritis, oral hygiene, and periodontitis . J Periodontol 2008 ; 79 : 979 – 86 . Google Scholar CrossRef Search ADS PubMed 10 Scher JU , Ubeda C , Equinda M et al. Periodontal disease and the oral microbiota in new-onset rheumatoid arthritis . Arthritis Rheum 2012 ; 64 : 3083 – 94 . Google Scholar CrossRef Search ADS PubMed 11 Tolo K , Jorkjend L. Serum antibodies and loss of periodontal bone in patients with rheumatoid arthritis . J Clin Periodontol 1990 ; 17 : 288 – 91 . Google Scholar CrossRef Search ADS PubMed 12 Rosenstein ED , Greenwald RA , Kushner LJ , Weissmann G. Hypothesis: the humoral immune response to oral bacteria provides a stimulus for the development of rheumatoid arthritis . Inflammation 2004 ; 28 : 311 – 8 . Google Scholar CrossRef Search ADS PubMed 13 Wegner N , Wait R , Sroka A et al. Peptidylarginine deiminase from Porphyromonas gingivalis citrullinates human fibrinogen and alpha-enolase: implications for autoimmunity in rheumatoid arthritis . Arthritis Rheum 2010 ; 62 : 2662 – 72 . Google Scholar CrossRef Search ADS PubMed 14 Konig MF , Abusleme L , Reinholdt J et al. Aggregatibacter actinomycetemcomitans-induced hypercitrullination links periodontal infection to autoimmunity in rheumatoid arthritis . Sci Transl Med 2016 ; 8 : 369ra176 . Google Scholar CrossRef Search ADS PubMed 15 Zhang X , Zhang D , Jia H et al. The oral and gut microbiomes are perturbed in rheumatoid arthritis and partly normalized after treatment . Nat Med 2015 ; 21 : 895 – 905 . Google Scholar CrossRef Search ADS PubMed 16 Arnett F , Edworthy S , Bloch D et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis . Arthritis Rheum 1988 ; 31 : 315 – 24 . Google Scholar CrossRef Search ADS PubMed 17 Machtei EE , Christersson LA , Grossi SG et al. Clinical criteria for the definition of “established periodontitis”. J Periodontol 1992 ; 63 : 206 – 14 . Google Scholar CrossRef Search ADS PubMed 18 Walker C , Puumala S , Golub LM et al. Subantimicrobial dose doxycycline effects on osteopenic bone loss: microbiologic results . J Periodontol 2007 ; 78 : 1590 – 601 . Google Scholar CrossRef Search ADS PubMed 19 Prevoo ML , van’t Hof MA , Kuper HH et al. Modified disease activity scores that include twenty-eight-joint counts. Development and validation in a prospective longitudinal study of patients with rheumatoid arthritis . Arthritis Rheum 1995 ; 38 : 44 – 8 . Google Scholar CrossRef Search ADS PubMed 20 Law CW , Chen Y , Shi W , Smyth GK. voom: precision weights unlock linear model analysis tools for RNA-seq read counts . Genome Biol 2014 ; 15 : R29 . Google Scholar CrossRef Search ADS PubMed 21 Ritchie ME , Phipson B , Wu D et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies . Nucleic Acids Res 2015 ; 43 : e47 . Google Scholar CrossRef Search ADS PubMed 22 Smyth GK. Linear models and empirical bayes methods for assessing differential expression in microarray experiments . Stat Appl Genet Mol Biol 2004 ; 3 :Article3. 23 Ge X , Rodriguez R , Trinh M , Gunsolley J , Xu P. Oral microbiome of deep and shallow dental pockets in chronic periodontitis . PLoS One 2013 ; 8 : e65520 . Google Scholar CrossRef Search ADS PubMed 24 Benjamini Y , Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing . J Roy Stat Soc Series B Stat Methodol 1995 ; 57 : 289 – 300 . 25 Socransky SS , Haffajee AD , Cugini MA , Smith C , Kent RL. Microbial complexes in subgingival plaque . J Clin Periodontol 1998 ; 25 : 134 – 44 . Google Scholar CrossRef Search ADS PubMed 26 Hitchon CA , Chandad F , Ferucci ED et al. Antibodies to Porphyromonas gingivalis are associated with anti-citrullinated protein antibodies in rheumatoid arthritis patients and their relatives . J Rheumatol 2010 ; 37 : 1105 – 12 . Google Scholar CrossRef Search ADS PubMed 27 Mikuls TR , Payne JB , Reinhardt RA et al. Antibody responses to Porphyromonas gingivalis (P. gingivalis) in subjects with rheumatoid arthritis and periodontitis . Int Immunopharmacol 2009 ; 9 : 38 – 42 . Google Scholar CrossRef Search ADS PubMed 28 Human Microbiome Project C . Structure, function and diversity of the healthy human microbiome . Nature 2012 ; 486 : 207 – 14 . CrossRef Search ADS PubMed 29 Fortenberry JD. The uses of race and ethnicity in human microbiome research . Trends Microbiol 2013 ; 21 : 165 – 6 . Google Scholar CrossRef Search ADS PubMed 30 Greenstein RJ , Su L , Haroutunian V , Shahidi A , Brown ST. On the action of methotrexate and 6-mercaptopurine on M. avium subspecies paratuberculosis . PLoS One 2007 ; 2 : e161 . Google Scholar CrossRef Search ADS PubMed 31 Rolain JM , Colson P , Raoult D. Recycling of chloroquine and its hydroxyl analogue to face bacterial, fungal and viral infections in the 21st century . Int J Antimicrob Agents 2007 ; 30 : 297 – 308 . Google Scholar CrossRef Search ADS PubMed 32 Shaddox LM , Huang H , Lin T et al. Microbiological characterization in children with aggressive periodontitis . J Dent Res 2012 ; 91 : 927 – 33 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Rheumatology Oxford University Press

The subgingival microbiome in patients with established rheumatoid arthritis

Loading next page...
 
/lp/ou_press/the-subgingival-microbiome-in-patients-with-established-rheumatoid-Wi2WD6llPJ
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For permissions, please email: journals.permissions@oup.com
ISSN
1462-0324
eISSN
1462-0332
D.O.I.
10.1093/rheumatology/key052
Publisher site
See Article on Publisher Site

Abstract

Abstract Objectives To profile and compare the subgingival microbiome of RA patients with OA controls. Methods RA (n = 260) and OA (n = 296) patients underwent full-mouth examination and subgingival samples were collected. Bacterial DNA was profiled using 16 S rRNA Illumina sequencing. Following data filtering and normalization, hierarchical clustering analysis was used to group samples. Multivariable regression was used to examine associations of patient factors with membership in the two largest clusters. Differential abundance between RA and OA was examined using voom method and linear modelling with empirical Bayes moderation (Linear Models for Microarray Analysis, limma), accounting for the effects of periodontitis, race, marital status and smoking. Results Alpha diversity indices were similar in RA and OA after accounting for periodontitis. After filtering, 286 taxa were available for analysis. Samples grouped into one of seven clusters with membership sizes of 324, 223, 3, 2, 2, 1 and 1 patients, respectively. RA-OA status was not associated with cluster membership. Factors associated with cluster 1 (vs 2) membership included periodontitis, smoking, marital status and Caucasian race. Accounting for periodontitis, 10 taxa (3.5% of those examined) were in lower abundance in RA than OA. There were no associations between lower abundance taxa or other select taxa examined with RA autoantibody concentrations. Conclusion Leveraging data from a large case–control study and accounting for multiple factors known to influence oral health status, results from this study failed to identify a subgingival microbial fingerprint that could reliably discriminate RA from OA patients. periodontitis, subgingival microbiome, rheumatoid arthritis, osteoarthritis Rheumatology key messages Patients with established RA and OA appear to harbour similar subgingival microbial communities. Factors most closely associated with subgingival microbial composition in this study included periodontitis, smoking, marital status and race. Introduction Previous reports have implicated periodontitis in the aetiopathogenesis of RA [1–11]. Based on the high prevalence of periodontal dysbiosis in periodontitis, there has been interest in elucidating the role that specific microorganisms could play in the link between periodontitis and RA risk. Porphyromonas gingivalis, a Gram-negative anaerobe that is strongly associated with periodontitis, has perhaps garnered the most attention. Unique among prokaryotes, P. gingivalis expresses a peptidyl-arginine deiminase that catalyses protein citrullination [12]. This organism has further demonstrated the capacity to citrullinate proteins in vitro, generating antigens implicated in RA pathogenesis [13]. Recently, investigators have shown robust associations of antibody recognizing Aggregatibacter actinomycetemcomitans to RA risk, in addition to demonstrating that this organism induces hypercitrullination in host neutrophils generating a citrullinome paralleling that of the RA joint [14]. In another study, the subgingival presence of Prevotella and Leptotrichia species were overrepresented in treatment-naïve RA patients compared with healthy controls, irrespective of underlying periodontitis status [10]. To date, most studies examining the role of periodontal pathogens in RA have been highly selective, examining either one or a limited number of pre-specified microbes [1, 3, 8]. Given inherent difficulties in culturing these organisms, most studies have also used serologic testing as surrogates of bacterial exposure. Thus, data generated using non-biased approaches to fully characterize the oral microbiome in RA are limited [10, 15]. In one such study, investigators characterized the subgingival microbiome samples from 65 RA patients [10], while another used dental plaque to characterize the microbiome of 77 cases [15]. Using healthy controls for comparison, both studies demonstrated unique, but non-overlapping, microbiome characteristics specific to RA. To what degree other confounders such as smoking or sociodemographics might have contributed to these findings remains unclear. Leveraging data and subgingival samples from a large, well-characterized case–control study, we characterized the microbiome of patients with established RA and diseased controls with OA [8]. We sought to identify whether RA patients harboured unique subgingival microbial signatures after accounting for the effects of other factors that could influence the oral microbiota, including periodontitis status. We hypothesized that patients with established RA would demonstrate a unique subgingival microbiome when compared with OA controls and that this difference would be independent of periodontitis status. If true, these results would inform future research strategies of RA treatment and/or prevention that could conceivably target specific oral pathogens. Methods Study participants and clinical assessments Study subjects included participants in a multicentre case–control investigation examining the relationship of periodontitis with RA [8]. There were 617 study participants (287 RA cases and 330 OA controls) enrolled from from five participating centres. All cases satisfied the 1987 ACR classification criteria for RA with an age of disease onset >18 years [16]. OA in controls was based on medical documentation from a corresponding healthcare provider or imaging results consistent with degenerative arthritis. With demographics similar to cases, we used OA patients as comparators to minimize differences between RA cases and healthy control populations employed in other studies [10, 15]. The study was approved by the Institutional Review Board at each centre and all participants provided written informed consent prior to study enrolment. Clinical assessments and sample collection Cases and controls underwent a standardized and calibrated full mouth periodontal evaluation with periodontitis defined [17]. Subgingival plaque was collected from up to four mesiobuccal sites (one per quadrant) from posterior teeth from all but one participant (n = 616). Following removal of visible supragingival plaque, subgingival plaque samples were collected using a single sterile endodontic paper point for each site [18]. Up to four samples were pooled for each patient and frozen at −70°C until analysis. Serum was collected and RA cases were assessed for circulating concentrations of ACPA using a second-generation ELISA (Diastat, Axis-Shield; positivity ⩾5 U/ml). RF (positive ⩾15 IU/ml) and high-sensitivity CRP (mg/l) were determined by nephelometry (Siemens Healthcare Diagnostics, Tarrytown, NJ, USA). Disease activity in RA patients was quantified through the measurement of tender and swollen joint counts (0–28), and provider and patient global well-being scores (0–100 mm visual analogue scales). A DAS-28-CRP was calculated [19]. Subgingival microbiome identification Genomic DNA was extracted from each subgingival sample using the Mobio Ultraclean Microbial DNA extraction kit (Carlsbad, CA, USA). For each sample, bacterial 16 S rRNA V1–V3 gene segments were amplified in duplicate using composite 27 F (5′ AGAGTTTGATCCTGGCTCAG 3′) and 534 R (5′ ATTACCGCGGCTGCTGG 3′) primers. The forward and the reverse primers each contained unique barcodes to allow multiplex deep sequencing. Each 20 µl PCR contained 2 µl of the purified DNA template, 1× Accuprime PCR buffer II, 0.2 µM of the forward primer, 0.2 µM of the reverse primer and 1.0 U of Accuprime Taq High Fidelity polymerase. PCR amplification was performed as follows: denaturation at 95°C for 2 min, followed by 25 cycles of denaturation at 95°C for 20 s, annealing at 56°C for 30 s and extension at 72°C for 5 min. The replicate barcoded PCR products for each sample were pooled and analysed on 1.5% SYBR Safe (Invitrogen, Carlsbad, CA, USA) agarose gel. Gel slices containing the amplicons of expected size (∼600 bp) were excised and purified using the Qiagen gel extraction kit (Qiagen, Valencia, CA, USA). Purified PCR products were quantified using Qubit HS DNA quantification kit (Invitrogen, Carlsbad, CA, USA) and pooled with equal molar concentration. The use of barcodes allowed multiplexing and bidirectional sequencing on the Illumina MiSeq platform (Illumina, San Diego, CA, USA) in four pools, generating 16 145 880 raw reads. After trimming primer and barcode sequences and quality control, 11 039 494 V1–V3 16 S rRNA gene sequence reads were available for analysis (averaging 19 855 ± 6809 reads per sample and an amplicon length of 487 nt). Good’s coverage estimates were >0.95 for all samples. Rarefaction analysis indicates that sampling effort was sufficient for the majority of the samples, and further sampling would yield few additional taxa. Paired-end raw Illumina sequence reads were joined using Fast Length Adjustment of Short reads (https://ccb.jhu.edu/software/FLASH/) and filtered with custom scripts written in R (https://www.r-project.org/) using the following quality criteria: minimum overlap of 10 bp from both paired ends, exact match to barcode and primer, no ambiguous bases and an average quality score of 30. Barcodes were variable lengths (4–8 bp) for each paired end and formed a unique combination for identification in each pool. In downstream analysis, the barcodes and primers (27 F and 534 R) were trimmed. Trimmed reads were then searched against the Human Oral Microbiome Database (HOMD; non-extended version 10.1) [20] and an HOMD sequence (full-length) was assigned for each read based on the best hit for that read. The tag matching of the V1–V3 reads to full length HOMD sequences formed the basis of reference operational taxonomic unit (OTU), where the full-length HOMD sequences and taxonomy information were used in the analysis. Reads with <80% alignable coverage and <97% identity were excluded from additional analysis. Statistical analysis Descriptive statistics including relative frequency for categorical variables and mean (s.d.) for continuous variables were calculated. χ2 and non-parametric Wilcoxon rank sum tests were used to compare group characteristics. Sequencing data were filtered to exclude taxa for which >75% of patients in all four groups defined by RA-OA and periodontitis status had zero values. Alpha diversity indices were computed using the filtered data, and compared by group (RA-periodontitis, RA-no periodontitis, OA-periodontitis and OA-no periodontitis) via calculation of: total number of OTUs; Chao index, reflecting the total number of species (microbial richness); evenness, reflecting the relative abundance of species; Shannon diversity index, based on both evenness and richness (larger values reflecting more diverse microbiota); and Good’s coverage estimator, reflecting the total percent of species represented. The Shapiro-Wilk test was used to assess normality. Based on the normality testing, Shannon index, evenness and Good’s estimator were compared using the non-parametric Kruskal-Wallis test with Bonferroni’s correction for pairwise comparisons. The number of OTUs and Chao index were compared using one-way analysis of variance with a Tukey-Kramer post hoc test used for pairwise comparisons. Additional analyses were undertaken to examine associations of RA-related factors including disease activity (DAS-28-CRP), disease duration and select medications (prednisone, MTX or biologics) with measures of microbial diversity. Using data limited to patients with RA, Pearson or Spearman correlations were calculated (as appropriate based on normality) to examine associations of the aforementioned alpha diversity indices with DAS-28-CRP and disease duration. Additionally, two-sample t-tests or Wilcoxon rank sum tests were used as appropriate to compare indices among RA patients based on the use of prednisone, MTX or any biologic disease-modifying agent. Filtered raw count sequencing data were normalized with the method of trimmed mean of M-values and converted to log2-counts per million per sample by voom function built in the Linear Models for Microarray Analysis (limma) package in Bioconductor [20–22]. Two approaches were used to compare subgingival microbial communities by group. An unsupervised cluster analysis was completed and logistic regression models were generated to first examine prevailing microbiome patterns and associations with RA-OA status [23]. Specifically, hierarchical cluster analysis with averaging-linkage agglomeration on microbiome profiles was used to classify samples into clusters. Stepwise logistic regression was then used to identify variables associated with membership in clusters 1 vs 2, the two largest cluster memberships (or patient groups). The significance level for entry and removal of variables were set to be 0.05. RA vs OA disease status was retained in the model regardless of P-values. In addition to the presence of periodontitis, other factors examined included age, sex, self-reported race/ethnicity (Caucasian vs other), smoking status (ever vs never), BMI (kg/m2), self-reported diabetes, education and marital status. A heat-map was generated to illustrate OTUs with at least a 2-fold difference in abundance between clusters 1 and 2 using Genesis 1.8.1 (Graz University of Technology, Graz, Austria). Using a second approach, differential abundance analysis was conducted using the limma package [20–22]. The analysis was conducted using two linear models: the first fitted with RA vs OA status, periodontitis status and their interaction (base model); and the second fitted with the same factors as in the base model in addition to race, marital status and smoking (fully adjusted model). The Bejamini-Hochberg method was used to control the false discovery rate to be no >0.05 for multiple hypotheses correction [24]. Principal component analysis plots were generated as an additional means of examining the associations of patient characteristics with subgingival micobiome composition. Microbial complexes (red, orange, yellow and purple) were defined according to Socransky et al. [25], with red and orange complexes most closely linked to periodontitis while yellow and purple complexes have been more closely associated with oral health. With reports of differential abundance of Haemophilus, Prevotella, Leptotrichia and possibly Aggregatibacter species in RA patients vs controls [10, 14, 15], additional analyses explored differential abundance of P. gingivalis and other species at the genus level. In addition to P. gingivalis (OTU 619), OTUs including Haemophilus, Prevotella, Leptotrichia and Aggregatibacter were identified in filtered normalized data and median normalized counts were calculated for each of the four groups. A linear model was fitted with RA-OA disease status, periodontitis and the interaction of these, in addition to race, marital status and smoking. Spearman correlations assessed the relationship between autoantibody (RF and anti-CCP) concentrations and normalized OTUs. To limit the possibility of type I error, autoantibody correlations were examined for OTUs based on prior reports suggesting associations of specific oral bacteria (P. gingivalis, Haemophilus and Aggregatibacter species, in addition to Anaeroglobus geminatus) with autoantibody expression [10, 14, 15, 26, 27], as well as OTUs found in lower abundance between RA cases and OA controls. Results Of the 616 patients providing samples, 556 (90%) had sufficient DNA for sequencing. Characteristics for RA (n = 260) and OA (n = 296) patients are summarized in Table 1. Compared with those with OA, patients with RA were more likely to be smokers, had lower BMIs and were less likely to have diabetes. As previously reported in the full cohort [10], RA patients included in this analysis were more likely to have periodontitis than OA patients (38% vs 27%, P = 0.007). Table 1 RA and OA patient characteristics Characteristics RA (n = 260) OA (n = 296) P-value Demographics and comorbidity Age, mean (s.d.), years 59 (12) 60 (11) 0.68 Men, % 65 61 0.35 Race, % 0.25     Caucasian 77 72     African American 17 23     Other 6 5 Married, % 69 61 0.06 Ever smoking, % 62 46 <0.001 Periodontitis, % 38 27 0.007 BMI, mean (s.d.), kg/m2 29.8 (6.7) 31.7 (6.7) <0.001 Diabetes mellitus, % 18 25 0.04 RA disease characteristics     RF positive, % 78 – –     Anti-CCP antibody positive, % 85 – –     DAS-28-CRP, mean (s.d.) 3.3 (1.3) – –     MTX use, % 63 – –     Prednisone use, % 29 – –     Current biologic use, % 32 – – Characteristics RA (n = 260) OA (n = 296) P-value Demographics and comorbidity Age, mean (s.d.), years 59 (12) 60 (11) 0.68 Men, % 65 61 0.35 Race, % 0.25     Caucasian 77 72     African American 17 23     Other 6 5 Married, % 69 61 0.06 Ever smoking, % 62 46 <0.001 Periodontitis, % 38 27 0.007 BMI, mean (s.d.), kg/m2 29.8 (6.7) 31.7 (6.7) <0.001 Diabetes mellitus, % 18 25 0.04 RA disease characteristics     RF positive, % 78 – –     Anti-CCP antibody positive, % 85 – –     DAS-28-CRP, mean (s.d.) 3.3 (1.3) – –     MTX use, % 63 – –     Prednisone use, % 29 – –     Current biologic use, % 32 – – Table 1 RA and OA patient characteristics Characteristics RA (n = 260) OA (n = 296) P-value Demographics and comorbidity Age, mean (s.d.), years 59 (12) 60 (11) 0.68 Men, % 65 61 0.35 Race, % 0.25     Caucasian 77 72     African American 17 23     Other 6 5 Married, % 69 61 0.06 Ever smoking, % 62 46 <0.001 Periodontitis, % 38 27 0.007 BMI, mean (s.d.), kg/m2 29.8 (6.7) 31.7 (6.7) <0.001 Diabetes mellitus, % 18 25 0.04 RA disease characteristics     RF positive, % 78 – –     Anti-CCP antibody positive, % 85 – –     DAS-28-CRP, mean (s.d.) 3.3 (1.3) – –     MTX use, % 63 – –     Prednisone use, % 29 – –     Current biologic use, % 32 – – Characteristics RA (n = 260) OA (n = 296) P-value Demographics and comorbidity Age, mean (s.d.), years 59 (12) 60 (11) 0.68 Men, % 65 61 0.35 Race, % 0.25     Caucasian 77 72     African American 17 23     Other 6 5 Married, % 69 61 0.06 Ever smoking, % 62 46 <0.001 Periodontitis, % 38 27 0.007 BMI, mean (s.d.), kg/m2 29.8 (6.7) 31.7 (6.7) <0.001 Diabetes mellitus, % 18 25 0.04 RA disease characteristics     RF positive, % 78 – –     Anti-CCP antibody positive, % 85 – –     DAS-28-CRP, mean (s.d.) 3.3 (1.3) – –     MTX use, % 63 – –     Prednisone use, % 29 – –     Current biologic use, % 32 – – Alpha diversity indices across the four groups defined by RA-OA and periodontitis case status are shown in Table 2. Although both OTU numbers and the Chao index were higher in those with periodontitis than those without, these were not significantly impacted by RA-OA status. There were no significant differences by group for the Shannon index, evenness or Good’s coverage estimator. In additional analyses limited to RA patients, there were no correlations between the alpha diversity indices and DAS-28-CRP or disease duration, with the exception of a weak, but statistically significant, positive correlation between DAS-28-CRP and evenness (r = 0.15, P = 0.02). Likewise, we found no evidence of an association in RA patients between treatments received (prednisone, MTX or biologic) and microbial diversity (data not shown). Table 2 Alpha diversity measures in patients with RA and OA with and without evidence of periodontitis Measure RA-periodontitis (n = 99) RA-no periodontitis (n = 161) OA-periodontitis (n = 81) OA-no periodontitis (n = 215) P-valuea Number OTUsa, mean (s.d.) 151.9 (38.5) 138.6 (38.4) 158.5 (41.3) 140.8 (37.8) 0.0002 Chao indexa, mean (s.d.) 169.3 (41.6) 155.6 (41.8) 177.3 (45.5) 157.2 (40.1) 0.0002 Shannon indexb, median (range) 3.37 (1.95–4.45) 3.42 (1.42–4.45) 3.43 (1.54–4.26) 3.32 (1.74–4.26) 0.53 Evennessb, median (range) 0.67 (0.42–0.82) 0.70 (0.37–0.82) 0.68 (0.31–0.80) 0.68 (0.38–0.80) 0.11 Good’sb, median (range) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 0.45 Measure RA-periodontitis (n = 99) RA-no periodontitis (n = 161) OA-periodontitis (n = 81) OA-no periodontitis (n = 215) P-valuea Number OTUsa, mean (s.d.) 151.9 (38.5) 138.6 (38.4) 158.5 (41.3) 140.8 (37.8) 0.0002 Chao indexa, mean (s.d.) 169.3 (41.6) 155.6 (41.8) 177.3 (45.5) 157.2 (40.1) 0.0002 Shannon indexb, median (range) 3.37 (1.95–4.45) 3.42 (1.42–4.45) 3.43 (1.54–4.26) 3.32 (1.74–4.26) 0.53 Evennessb, median (range) 0.67 (0.42–0.82) 0.70 (0.37–0.82) 0.68 (0.31–0.80) 0.68 (0.38–0.80) 0.11 Good’sb, median (range) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 0.45 a Values satisfying normality assumption compared by group using ANOVA with Tukey-Kramer correction. Significant pairwise comparisons for the number of OTUs were: RA-periodontitis vs RA-no periodontitis (P = 0.04), OA-periodontitis vs RA-no periodontitis (P = 0.001) and OA-periodontitis vs OA-no periodontitis (P = 0.003); all other comparisons non-significant. Significant pairwise comparisons for the Chao index were: RA-periodontitis vs RA-no periodontitis (P = 0.05), OA-periodontitis vs RA-no periodontitis (P = 0.0008) and OA-periodontitis vs OA-no periodontitis (P = 0.001). b Values not satisfying normality assumption compared using Kruskal-Wallis test. OTU: operational taxonomic unit. Table 2 Alpha diversity measures in patients with RA and OA with and without evidence of periodontitis Measure RA-periodontitis (n = 99) RA-no periodontitis (n = 161) OA-periodontitis (n = 81) OA-no periodontitis (n = 215) P-valuea Number OTUsa, mean (s.d.) 151.9 (38.5) 138.6 (38.4) 158.5 (41.3) 140.8 (37.8) 0.0002 Chao indexa, mean (s.d.) 169.3 (41.6) 155.6 (41.8) 177.3 (45.5) 157.2 (40.1) 0.0002 Shannon indexb, median (range) 3.37 (1.95–4.45) 3.42 (1.42–4.45) 3.43 (1.54–4.26) 3.32 (1.74–4.26) 0.53 Evennessb, median (range) 0.67 (0.42–0.82) 0.70 (0.37–0.82) 0.68 (0.31–0.80) 0.68 (0.38–0.80) 0.11 Good’sb, median (range) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 0.45 Measure RA-periodontitis (n = 99) RA-no periodontitis (n = 161) OA-periodontitis (n = 81) OA-no periodontitis (n = 215) P-valuea Number OTUsa, mean (s.d.) 151.9 (38.5) 138.6 (38.4) 158.5 (41.3) 140.8 (37.8) 0.0002 Chao indexa, mean (s.d.) 169.3 (41.6) 155.6 (41.8) 177.3 (45.5) 157.2 (40.1) 0.0002 Shannon indexb, median (range) 3.37 (1.95–4.45) 3.42 (1.42–4.45) 3.43 (1.54–4.26) 3.32 (1.74–4.26) 0.53 Evennessb, median (range) 0.67 (0.42–0.82) 0.70 (0.37–0.82) 0.68 (0.31–0.80) 0.68 (0.38–0.80) 0.11 Good’sb, median (range) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (0.99–1.00) 0.45 a Values satisfying normality assumption compared by group using ANOVA with Tukey-Kramer correction. Significant pairwise comparisons for the number of OTUs were: RA-periodontitis vs RA-no periodontitis (P = 0.04), OA-periodontitis vs RA-no periodontitis (P = 0.001) and OA-periodontitis vs OA-no periodontitis (P = 0.003); all other comparisons non-significant. Significant pairwise comparisons for the Chao index were: RA-periodontitis vs RA-no periodontitis (P = 0.05), OA-periodontitis vs RA-no periodontitis (P = 0.0008) and OA-periodontitis vs OA-no periodontitis (P = 0.001). b Values not satisfying normality assumption compared using Kruskal-Wallis test. OTU: operational taxonomic unit. Hierarchical cluster analysis with averaging-linkage agglomeration on microbiome profiles was used to classify samples into one of seven unique clusters. The number of memberships of clusters 1–7 was 324, 223, 3, 2, 2, 1 and 1, respectively. With only nine patients with membership, clusters 3–7 were excluded from the further analysis. OTUs with >2-fold difference in abundance between clusters are shown (Fig. 1). Bacteria comprising red and orange complexes [25], thought to represent periodontitis pathogens, were generally found in greater abundance in clusters 2 vs 1. All three red complex bacteria (P. gingivalis, Tannerella forsythia and Treponema denticola) were found in ⩾2-fold abundance in cluster 2 compared with cluster 1, as were two of three orange complex bacteria (Eubacterium nodatum and Streptococcus constellatus). Bacteria from the yellow and purple complexes, reflecting early colonizers and a healthier periodontium, were universally found in greater abundance in cluster 1 vs 2. Fig. 1 View largeDownload slide View largeDownload slide View largeDownload slide Heatmap demonstrating clusters 1 and 2 taxa Heatmap of operational taxonomic units (OTUs) demonstrating more than a 2-fold difference in abundance between cluster 1 (c1) and cluster 2 (c2). Panels shown are from single contiguous heatmap. Fig. 1 View largeDownload slide View largeDownload slide View largeDownload slide Heatmap demonstrating clusters 1 and 2 taxa Heatmap of operational taxonomic units (OTUs) demonstrating more than a 2-fold difference in abundance between cluster 1 (c1) and cluster 2 (c2). Panels shown are from single contiguous heatmap. Using stepwise logistic regression, RA case status was not significantly associated with membership in clusters 1 vs 2 [odds ratio (OR) = 1.18; 95% CI: 0.80, 1.73] (Table 3). Being married (OR = 1.57; 95% CI: 1.06, 2.33) and Caucasian race (OR = 1.94; 95% CI: 1.26, 2.99) were associated with an increased odds of membership in cluster 1 vs 2. In contrast, periodontitis (OR = 0.22; 95% CI: 0.15, 0.33) and ever smoking (OR = 0.65; 95% CI: 0.44, 0.96) were associated with a lower likelihood of membership in cluster 1. Table 3 Associations of patient factors with cluster membership (cluster 1 vs 2) based on subgingival microbiome composition Patient factor Odds ratio (95% CI) P-value RA vs OA 1.18 (0.80, 1.73) 0.40 Periodontitis vs no periodontitis 0.22 (0.15, 0.33) <0.001 Ever vs never smoking 0.65 (0.44, 0.96) 0.03 Married vs not married 1.57 (1.06, 2.33) 0.03 Caucasian vs other 1.94 (1.26, 2.99) 0.003 Patient factor Odds ratio (95% CI) P-value RA vs OA 1.18 (0.80, 1.73) 0.40 Periodontitis vs no periodontitis 0.22 (0.15, 0.33) <0.001 Ever vs never smoking 0.65 (0.44, 0.96) 0.03 Married vs not married 1.57 (1.06, 2.33) 0.03 Caucasian vs other 1.94 (1.26, 2.99) 0.003 Table 3 Associations of patient factors with cluster membership (cluster 1 vs 2) based on subgingival microbiome composition Patient factor Odds ratio (95% CI) P-value RA vs OA 1.18 (0.80, 1.73) 0.40 Periodontitis vs no periodontitis 0.22 (0.15, 0.33) <0.001 Ever vs never smoking 0.65 (0.44, 0.96) 0.03 Married vs not married 1.57 (1.06, 2.33) 0.03 Caucasian vs other 1.94 (1.26, 2.99) 0.003 Patient factor Odds ratio (95% CI) P-value RA vs OA 1.18 (0.80, 1.73) 0.40 Periodontitis vs no periodontitis 0.22 (0.15, 0.33) <0.001 Ever vs never smoking 0.65 (0.44, 0.96) 0.03 Married vs not married 1.57 (1.06, 2.33) 0.03 Caucasian vs other 1.94 (1.26, 2.99) 0.003 Two principal components, PC1 (12.3%) and PC2 (8.9%), together explained 21.2% of the bacterial variation. Principal component analysis plots did not demonstrate evidence of clustering based on RA-OA case status, but showed variable degrees of clustering based on marital status, race, periodontitis and cigarette smoking (Fig. 2). Fig. 2 View largeDownload slide Principal component analysis plots Two principal components (PC1 and PC2) explained most of the bacterial variation observed in subgingival samples. PCA plots examining the relationship of RA vs OA case status, marital status, race, periodontitis and smoking with microbiome composition are shown. Plots did not demonstrate evidence of clustering based on RA-OA case status, but showed variable degrees of clustering based on marital status, race, periodontitis and cigarette smoking. PCA: principal component analysis. Fig. 2 View largeDownload slide Principal component analysis plots Two principal components (PC1 and PC2) explained most of the bacterial variation observed in subgingival samples. PCA plots examining the relationship of RA vs OA case status, marital status, race, periodontitis and smoking with microbiome composition are shown. Plots did not demonstrate evidence of clustering based on RA-OA case status, but showed variable degrees of clustering based on marital status, race, periodontitis and cigarette smoking. PCA: principal component analysis. Excluding taxa for which >75% of patients in all four groups had zero values, there were 286 OTUs available for analysis. Among those with periodontitis, there were 10 OTUs (3.5% of OTUs examined) in lower abundance in RA vs OA samples in the base model with false discovery rate-adjusted P-values. These taxa included Catonella sp. (OTU 451), Clostridiales sp. (OTU 85), Lachnospiraceae sp. (OTU 96), Peptostreptococcaceae sp. (OTU 495), Porphyromonas sp. (OTU 285), Prevotella multiformis (OTU 685), Prevotella sp. (OTU 443), Selenomaonas sp. (OTU 479) and Treponema sp. (OTUs 230 and 236) (Table 4). Results were unchanged following full adjustment, with the exception that differential expression of Selenomaonas sp. was no longer significant. Although we observed evidence to suggest the under-abundance of Porphyromonas sp. in RA patients with periodontitis compared with OA patients with periodontitis, there was not similar evidence of differential expression for P. gingivalis (data not shown). In subjects without evidence of periodontitis, only one OTU was found in lower abundance in RA samples and in the fully adjusted model after accounting for false discovery rate, Streptococcus sp. (OTU 486), a difference that did not achieve statistical significance in the base model. There were no OTUs identified in the filtered data consistent with A. actinomycetemcomitans. Table 4 Differential expression of OTUs between RA and OA patients based on periodontitis status Base modela Fully adjusted modela OTUs with differential expression Fold difference (RA vs OA) P-value (FDR adjusted) Fold difference (RA vs OA) P-value (FDR adjusted) Patients with periodontitis  Catonella sp. (OTU 451) 0.424 0.002 0.466 0.005     Clostridiales sp. (OTU 85) 0.486 0.002 0.511 0.005  Lachnospiraceae sp. (OTU 96) 0.507 0.010 0.484 0.005  Peptostreptococcaceae sp. (OTU 495) 0.590 0.012 0.596 0.012  Porphyromonas sp. (OTU 285) 0.520 0.019 0.535 0.026  Prevotella multiformis (OTU 685) 0.456 0.009 0.469 0.005  Prevotella sp. (OTU 443) 0.523 0.019 0.540 0.022  Selenomaonas sp. (OTU 479) 0.589 0.039 0.617 0.081  Treponema sp. (OTU 230) 0.467 0.002 0.500 0.005  Treponema sp. (OTU 236) 0.429 0.019 0.465 0.035 Patients without periodontitis  Streptococcus sp. (OTU 486) 0.705 0.086 0.689 0.042 Base modela Fully adjusted modela OTUs with differential expression Fold difference (RA vs OA) P-value (FDR adjusted) Fold difference (RA vs OA) P-value (FDR adjusted) Patients with periodontitis  Catonella sp. (OTU 451) 0.424 0.002 0.466 0.005     Clostridiales sp. (OTU 85) 0.486 0.002 0.511 0.005  Lachnospiraceae sp. (OTU 96) 0.507 0.010 0.484 0.005  Peptostreptococcaceae sp. (OTU 495) 0.590 0.012 0.596 0.012  Porphyromonas sp. (OTU 285) 0.520 0.019 0.535 0.026  Prevotella multiformis (OTU 685) 0.456 0.009 0.469 0.005  Prevotella sp. (OTU 443) 0.523 0.019 0.540 0.022  Selenomaonas sp. (OTU 479) 0.589 0.039 0.617 0.081  Treponema sp. (OTU 230) 0.467 0.002 0.500 0.005  Treponema sp. (OTU 236) 0.429 0.019 0.465 0.035 Patients without periodontitis  Streptococcus sp. (OTU 486) 0.705 0.086 0.689 0.042 a Differential expression analysis conducted using two separate linear models: the first fitted with RA vs OA status, periodontitis status and the interaction between these factors (base model); and the second fitted with the same factors as the base model in addition to race, marital status and smoking (fully adjusted model). OTU: operational taxonomic unit; FDR: false discovery rate. Table 4 Differential expression of OTUs between RA and OA patients based on periodontitis status Base modela Fully adjusted modela OTUs with differential expression Fold difference (RA vs OA) P-value (FDR adjusted) Fold difference (RA vs OA) P-value (FDR adjusted) Patients with periodontitis  Catonella sp. (OTU 451) 0.424 0.002 0.466 0.005     Clostridiales sp. (OTU 85) 0.486 0.002 0.511 0.005  Lachnospiraceae sp. (OTU 96) 0.507 0.010 0.484 0.005  Peptostreptococcaceae sp. (OTU 495) 0.590 0.012 0.596 0.012  Porphyromonas sp. (OTU 285) 0.520 0.019 0.535 0.026  Prevotella multiformis (OTU 685) 0.456 0.009 0.469 0.005  Prevotella sp. (OTU 443) 0.523 0.019 0.540 0.022  Selenomaonas sp. (OTU 479) 0.589 0.039 0.617 0.081  Treponema sp. (OTU 230) 0.467 0.002 0.500 0.005  Treponema sp. (OTU 236) 0.429 0.019 0.465 0.035 Patients without periodontitis  Streptococcus sp. (OTU 486) 0.705 0.086 0.689 0.042 Base modela Fully adjusted modela OTUs with differential expression Fold difference (RA vs OA) P-value (FDR adjusted) Fold difference (RA vs OA) P-value (FDR adjusted) Patients with periodontitis  Catonella sp. (OTU 451) 0.424 0.002 0.466 0.005     Clostridiales sp. (OTU 85) 0.486 0.002 0.511 0.005  Lachnospiraceae sp. (OTU 96) 0.507 0.010 0.484 0.005  Peptostreptococcaceae sp. (OTU 495) 0.590 0.012 0.596 0.012  Porphyromonas sp. (OTU 285) 0.520 0.019 0.535 0.026  Prevotella multiformis (OTU 685) 0.456 0.009 0.469 0.005  Prevotella sp. (OTU 443) 0.523 0.019 0.540 0.022  Selenomaonas sp. (OTU 479) 0.589 0.039 0.617 0.081  Treponema sp. (OTU 230) 0.467 0.002 0.500 0.005  Treponema sp. (OTU 236) 0.429 0.019 0.465 0.035 Patients without periodontitis  Streptococcus sp. (OTU 486) 0.705 0.086 0.689 0.042 a Differential expression analysis conducted using two separate linear models: the first fitted with RA vs OA status, periodontitis status and the interaction between these factors (base model); and the second fitted with the same factors as the base model in addition to race, marital status and smoking (fully adjusted model). OTU: operational taxonomic unit; FDR: false discovery rate. Based on recent reports, we examined associations of all OTUs including Haemophilus, Prevotella, Leptotrichia and Aggregatibacter with RA. There were 4 OTUs including Haemophilus, 39 including Prevotella, 15 including Leptotrichia and 4 including Aggregatibacter identified in the filtered normalized data. Among those with periodontitis, only OTUs including Prevotella were differentially expressed in RA vs OA patients (0.68-fold difference, P = 0.006). With reports also showing associations of select pathogens with autoantibody expression in RA, we subsequently examined correlations of related OTUs with RF and anti-CCP antibody concentrations in RA patients (all patients and restricted to seropositive patients). We observed no significant correlations of A. geminatus or P. gingivalis, or with Haemophilus or Aggregatibacter containing OTUs with RF or anti-CCP antibody concentrations (data not shown). There were no associations of OTUs found in lower abundance in RA, shown in Table 4, with autoantibody concentrations (data not shown). Discussion Using samples and data from a large case–control study, we explored links between oral bacteria and RA by profiling both the abundance and diversity of the subgingival microbiome in a large group of patients with established disease, using OA patients for comparison. Results from this study failed to support the existence of a robust subgingival microbial fingerprint that could reliably be used to discriminate RA patients from those with OA. Scher et al. [10] were among the first to deploy 16 S rRNA pyrosequencing to characterize subgingival microbial composition in RA. Using samples from treatment-naïve patients (n = 31), patients with established RA (n = 34) and a smaller number of healthy controls (n = 18), the authors similarly observed no meaningful group differences referent to either microbial diversity or richness. The presence of periodontitis, more common in RA patients, was associated in our study with a higher number of OTUs and a higher Chao index (reflecting greater microbial richness). Similarly, Scher and colleagues found no evidence of microbial clustering based on RA case status, although the authors observed clustering based on the presence of moderate-to-severe periodontitis. In addition to replicating these results, we have extended these observations further by showing that other factors beyond the presence or absence of periodontitis appear to influence subgingival microbial composition. Subgingival samples from patients with a history of smoking, non-Caucasian race and those not married were more likely to segregate into a cluster characterized by an over-abundance of known pathogens. These results emphasize the importance of accounting for these factors in future research exploring the relationship between oral microbes and RA. Although others have also identified ethnic/racial background as a strong predictor of human microbiome composition [28], mechanisms underpinning this association (as well as associations with other socioeconomic factors such as marital status observed in the present study) are unknown and warrant further study [29]. Demonstrating similar microbial composition across groups, Scher et al. [10] identified Prevotella and Leptotrichia taxa in greater abundance in patients with untreated RA vs healthy controls independent of periodontitis. Using metagenomic shotgun sequencing of dental plaque samples, Zhang et al. [15] observed a decreased abundance of Haemophilus species and a simultaneous over-abundance of Lactobacillus salivarius in early RA patients vs controls. In contrast, we found a decreased subgingival abundance of OTUs including Prevotella in RA cases compared with controls. Results from these previous reports contrast with those from our study, which failed to identify an over-abundance of any taxa but identified an under-abundance of 10 different OTUs in RA patients relative to OA patients after accounting for periodontitis status and other factors. Of note, several of the OTUs identified to be in lower abundance have been implicated in periodontitis risk including Peptostreptococcus, Porphyromonas, Prevotella and Treponema species. Differences in the findings across studies may relate to the varied methods used (metagenomic shotgun sequencing vs 16 S rRNA 454 pyrosequencing vs Illumina sequencing) as well as the populations sampled. Although we failed to observe any significant associations of RA disease duration with microbial diversity in our study, marked differences in RA disease duration across studies could help to explain discrepancies in findings, with RA disease durations varying from a mean of just a few months [10] to >12 years in our study [8]. This difference may be particularly relevant as microbial dysbiosis reported to characterize untreated RA appears may be partially reversed with disease-modifying therapies [15]. Indeed, others found that red complex bacteria (T. forsythia and P. gingivalis) were present in greater abundance in the periodontium of early RA patients than in patients with established disease [10]. The authors speculated that this difference might have related to the effects of RA treatments, particularly agents shown to exert antimicrobial properties [30, 31]. Limited by its cross-sectional design, results from the present study do not support a major impact from treatments used, as we observed no associations of microbial diversity in RA patients with the use of prednisone, MTX or biologic therapies. Another possibility might be that RA patients mount more vigorous immune responses to select subgingival microbes, eventually leading to an under-abundance as observed in our study. Whether OTUs containing Prevotella or other OTUs found in greater abundance among controls in this study could somehow be protective of RA development is unknown. The association of select oral microbes with RA risk has been hypothesized to be driven by bacterially mediated citrullination leading to the local expression of ACPA. Previous studies have shown associations, for instance, of circulating antibodies to P. gingivalis and, more recently, to A. actinomycetemcomitans [14] with ACPA. Using bacterial 16 S rRNA sequencing as a direct measure of subgingival bacterial exposure, we did not detect OTUs corresponding to the presence of A. actinomycetemcomitans, a pathogen that has been most closely associated with localized aggressive forms of periodontitis in younger individuals [32], nor did we identify associations of P. gingivalis with ACPA. Separate studies performed in patients with early-onset disease yielded associations of alternative microbes, both Haemophilus species [15] and A. geminatus [10], with RA-related autoantibodies. Neither of these associations was replicated in the current study. It is possible that bacterially mediated ACPA production, if present, is more relevant in early disease, becoming less relevant over time as autoantibody production is established in other articular or extra-articular sites. To date, several reports have examined the concept of targeting periodontitis as a means of treating RA. With its cross-sectional and observational study design, this investigation alone does not fully address questions about the relationship of subgingival microbiome composition with RA disease progression, recognizing the need for additional studies involving RA patients earlier in the disease process. Similarities in subgingival bacterial communities between established RA and OA in the present study as well as the lack of any robust associations between RA disease activity and measures of microbial diversity, however, do not support the concept of an intervention targeting specific subgingival pathogens in patients with established RA. In summary, these results demonstrate that subgingival microbiome profiles in patients with established RA are similar to those observed in OA patients, and appear to be primarily influenced by other factors impacting oral health status. Funding: This work was supported by a Veterans Affairs Merit Award (to T.R.M.) [grant number CX000896] and a grant from the Rheumatology Research Foundation (to T.R.M.) [no grant number assigned]. Disclosure statement: The authors have declared no conflicts of interest. References 1 Bello-Gualtero JM , Lafaurie GI , Hoyos LX et al. Periodontal disease in individuals with a genetic risk of developing arthritis and early rheumatoid arthritis: a cross-sectional study . J Periodontol 2016 ; 87 : 346 – 56 . Google Scholar CrossRef Search ADS PubMed 2 de Pablo P , Dietrich T , McAlindon TE. Association of periodontal disease and tooth loss with rheumatoid arthritis in the US population . J Rheumatol 2008 ; 35 : 70 – 6 . Google Scholar PubMed 3 de Smit MD , Westra J , Vissink A et al. Periodontitis in established rheumatoid arthritis patients: a cross-sectional clinical, microbiological and serological study . Arthritis Res Ther 2012 ; 14 : R222 . Google Scholar CrossRef Search ADS PubMed 4 Dissick A , Redman RS , Jones M et al. Association of periodontitis with rheumatoid arthritis: a pilot study . J Periodontol 2010 ; 81 : 223 – 30 . Google Scholar CrossRef Search ADS PubMed 5 Kässer UR , Gleissner C , Dehne F et al. Risk for periodontal disease in patients with longstanding rheumatoid arthritis . Arthritis Rheum 1997 ; 40 : 2248 – 51 . Google Scholar CrossRef Search ADS PubMed 6 Malmstrom M , Calonius PE. Teeth loss and the inflammation of teeth-supporting tissues in rheumatoid disease . Scand J Rheumatol 1975 ; 4 : 49 – 55 . Google Scholar CrossRef Search ADS PubMed 7 Mercado FB , Marshall RI , Klestov AC , Bartold PM. Relationship between rheumatoid arthritis and periodontitis . J Periodontol 2001 ; 72 : 779 – 87 . Google Scholar CrossRef Search ADS PubMed 8 Mikuls TR , Payne JB , Yu F et al. Periodontitis and Porphyromonas gingivalis in patients with rheumatoid arthritis . Arthritis Rheumatol 2014 ; 66 : 1090 – 100 . Google Scholar CrossRef Search ADS PubMed 9 Pischon N , Pischon T , Kröger J et al. Association among rheumatoid arthritis, oral hygiene, and periodontitis . J Periodontol 2008 ; 79 : 979 – 86 . Google Scholar CrossRef Search ADS PubMed 10 Scher JU , Ubeda C , Equinda M et al. Periodontal disease and the oral microbiota in new-onset rheumatoid arthritis . Arthritis Rheum 2012 ; 64 : 3083 – 94 . Google Scholar CrossRef Search ADS PubMed 11 Tolo K , Jorkjend L. Serum antibodies and loss of periodontal bone in patients with rheumatoid arthritis . J Clin Periodontol 1990 ; 17 : 288 – 91 . Google Scholar CrossRef Search ADS PubMed 12 Rosenstein ED , Greenwald RA , Kushner LJ , Weissmann G. Hypothesis: the humoral immune response to oral bacteria provides a stimulus for the development of rheumatoid arthritis . Inflammation 2004 ; 28 : 311 – 8 . Google Scholar CrossRef Search ADS PubMed 13 Wegner N , Wait R , Sroka A et al. Peptidylarginine deiminase from Porphyromonas gingivalis citrullinates human fibrinogen and alpha-enolase: implications for autoimmunity in rheumatoid arthritis . Arthritis Rheum 2010 ; 62 : 2662 – 72 . Google Scholar CrossRef Search ADS PubMed 14 Konig MF , Abusleme L , Reinholdt J et al. Aggregatibacter actinomycetemcomitans-induced hypercitrullination links periodontal infection to autoimmunity in rheumatoid arthritis . Sci Transl Med 2016 ; 8 : 369ra176 . Google Scholar CrossRef Search ADS PubMed 15 Zhang X , Zhang D , Jia H et al. The oral and gut microbiomes are perturbed in rheumatoid arthritis and partly normalized after treatment . Nat Med 2015 ; 21 : 895 – 905 . Google Scholar CrossRef Search ADS PubMed 16 Arnett F , Edworthy S , Bloch D et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis . Arthritis Rheum 1988 ; 31 : 315 – 24 . Google Scholar CrossRef Search ADS PubMed 17 Machtei EE , Christersson LA , Grossi SG et al. Clinical criteria for the definition of “established periodontitis”. J Periodontol 1992 ; 63 : 206 – 14 . Google Scholar CrossRef Search ADS PubMed 18 Walker C , Puumala S , Golub LM et al. Subantimicrobial dose doxycycline effects on osteopenic bone loss: microbiologic results . J Periodontol 2007 ; 78 : 1590 – 601 . Google Scholar CrossRef Search ADS PubMed 19 Prevoo ML , van’t Hof MA , Kuper HH et al. Modified disease activity scores that include twenty-eight-joint counts. Development and validation in a prospective longitudinal study of patients with rheumatoid arthritis . Arthritis Rheum 1995 ; 38 : 44 – 8 . Google Scholar CrossRef Search ADS PubMed 20 Law CW , Chen Y , Shi W , Smyth GK. voom: precision weights unlock linear model analysis tools for RNA-seq read counts . Genome Biol 2014 ; 15 : R29 . Google Scholar CrossRef Search ADS PubMed 21 Ritchie ME , Phipson B , Wu D et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies . Nucleic Acids Res 2015 ; 43 : e47 . Google Scholar CrossRef Search ADS PubMed 22 Smyth GK. Linear models and empirical bayes methods for assessing differential expression in microarray experiments . Stat Appl Genet Mol Biol 2004 ; 3 :Article3. 23 Ge X , Rodriguez R , Trinh M , Gunsolley J , Xu P. Oral microbiome of deep and shallow dental pockets in chronic periodontitis . PLoS One 2013 ; 8 : e65520 . Google Scholar CrossRef Search ADS PubMed 24 Benjamini Y , Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing . J Roy Stat Soc Series B Stat Methodol 1995 ; 57 : 289 – 300 . 25 Socransky SS , Haffajee AD , Cugini MA , Smith C , Kent RL. Microbial complexes in subgingival plaque . J Clin Periodontol 1998 ; 25 : 134 – 44 . Google Scholar CrossRef Search ADS PubMed 26 Hitchon CA , Chandad F , Ferucci ED et al. Antibodies to Porphyromonas gingivalis are associated with anti-citrullinated protein antibodies in rheumatoid arthritis patients and their relatives . J Rheumatol 2010 ; 37 : 1105 – 12 . Google Scholar CrossRef Search ADS PubMed 27 Mikuls TR , Payne JB , Reinhardt RA et al. Antibody responses to Porphyromonas gingivalis (P. gingivalis) in subjects with rheumatoid arthritis and periodontitis . Int Immunopharmacol 2009 ; 9 : 38 – 42 . Google Scholar CrossRef Search ADS PubMed 28 Human Microbiome Project C . Structure, function and diversity of the healthy human microbiome . Nature 2012 ; 486 : 207 – 14 . CrossRef Search ADS PubMed 29 Fortenberry JD. The uses of race and ethnicity in human microbiome research . Trends Microbiol 2013 ; 21 : 165 – 6 . Google Scholar CrossRef Search ADS PubMed 30 Greenstein RJ , Su L , Haroutunian V , Shahidi A , Brown ST. On the action of methotrexate and 6-mercaptopurine on M. avium subspecies paratuberculosis . PLoS One 2007 ; 2 : e161 . Google Scholar CrossRef Search ADS PubMed 31 Rolain JM , Colson P , Raoult D. Recycling of chloroquine and its hydroxyl analogue to face bacterial, fungal and viral infections in the 21st century . Int J Antimicrob Agents 2007 ; 30 : 297 – 308 . Google Scholar CrossRef Search ADS PubMed 32 Shaddox LM , Huang H , Lin T et al. Microbiological characterization in children with aggressive periodontitis . J Dent Res 2012 ; 91 : 927 – 33 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

RheumatologyOxford University Press

Published: Mar 19, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off