Porphyromonas gingivalis and Disease-Related Autoantibodies in Individuals at Increased Risk of Rheumatoid Arthritis

Abstract

Purpose

To examine the relationship of Porphyromonas gingivalis (Pg) with the presence of autoantibodies in individuals at risk for rheumatoid arthritis (RA).

Methods

Participants included: 1) a cohort enriched with HLA-DR4 and 2) those at risk for RA by virtue of having a first-degree relative with RA. None satisfied 1987 ACR RA classification criteria. Autoantibodies measured included anti-citrullinated protein antibody (ACPA) and rheumatoid factor (RF; nephelometry, IgA, IgM, IgG). Individuals were considered autoantibody positive (n = 113) with ≥ 1 positive autoantibody with individuals further categorized as `high-risk' (n = 38; positive ACPA or ≥ 2 RF assays). Autoantibody negative individuals served as comparators (n = 171). Antibody to Pg, P. intermedia (Pi), and F. nucleatum (Fn) were measured. Associations of bacterial antibodies with group status were examined using logistic regression.

Results

Anti-Pg concentrations were higher in high-risk (p = 0.011) and autoantibody positive group (p = 0.010) than in the autoantibody negative group. There were no group differences in anti-Pi or anti-Fn concentrations. After multivariable adjustment, anti-Pg concentrations (but not anti-Pi or anti-Fn) were significantly associated with autoantibody positive and high-risk status (p < 0.05).

Conclusion

Immunity to Pg, but not Pi or Fn, is significantly associated with the presence of RA-related autoantibodies in individuals at risk for RA. These results support the hypothesis that infection with Pg may play a central role in the early loss of tolerance to self-antigens in RA pathogenesis.

Periodontitis (PD) has emerged as a risk factor for rheumatoid arthritis (RA). Affecting more than 20% of the general population (1), PD is an inflammatory disorder initiated by bacterial infection which detrimentally impacts the integrity of several different oral tissues including the gingiva, cementum, and periodontal ligament, ultimately leading to tooth loss. PD and RA share several common disease attributes including chronic tissue inflammation and, in severe cases, marked destruction of underlying bone. Similarities between PD and RA extend beyond shared histopathologic and inflammatory features. Both PD and RA share common risk factors for susceptibility, most notably HLA-DRB1 alleles and, importantly, cigarette smoking (2–9). Moreover, therapies used in RA treatment have been reported to ameliorate the signs and symptoms of PD (10–12).

Several cross-sectional case-control investigations have corroborated the association of PD with RA, although these findings were not replicated in two recent studies (13, 14). Compared to controls, RA patients experience more gingival bleeding, more missing teeth, twice as much loss of soft tissue attachment, and increased alveolar bone loss (15, 16). In one recent study, patients with RA were almost twice as likely as patients with osteoarthritis to have moderate to severe PD, an association that was independent of age, sex, race, and smoking history (17).

While most studies investigating the relationship of PD and RA have focused on shared inflammatory pathways, few have examined the associations of RA with the bacterial infections that initiate PD. A number of gram-negative oral pathogens have been implicated in PD and several have garnered attention. Chief among the organisms of interest is Porphyromonas gingivalis (P. gingivalis). P. gingivalis has been reported to be the only prokaryote known to express peptidylarginine deiminase (PAD) (18, 19), an enzyme responsible for the post-translational modification of arginine into citrulline. Given the predominant role of citrullinated proteins in RA pathogenesis, it has been speculated that infection with P. gingivalis could facilitate autoantigen presentation and tolerance loss in RA (19).

Investigations of P. gingivalis in RA have primarily involved studies examining RA patients with established disease. Based on these studies alone, it is not possible to know with certainty whether infection with P. gingivalis precedes RA onset or rather occurs subsequent to RA disease incidence. Therefore, in the present study, we sought to examine the association of P. gingivalis infection with the presence of RA-related autoantibodies among individuals at increased risk for the future development of RA, but who had not yet developed clinical RA. The existence of such an association in the absence of clinically-apparent inflammatory arthritis would strongly support the hypothesis that infection precedes disease and, therefore, is not simply a consequence of established RA or its treatments. The existence of such an association would also strongly support a central role of P. gingivalis in RA disease initiation.

Materials and Methods

Study population

Study subjects were participants in the ongoing longitudinal cohort study, Studies of the Etiology of Rheumatoid Arthritis (SERA). SERA is a multi-center prospective cohort study designed to investigate genetic and epidemiologic associations with RA-related autoimmunity during the pre-clinical period of RA development (20). SERA includes subjects at higher risk of developing RA, recruited from two populations: 1) a cohort enriched with the HLA-DR4 allele (the strongest genetic risk factor for RA), and 2) a cohort of first-degree relatives (FDRs; parent, full sibling, or offspring) of individuals with RA. Subjects were excluded from participation in SERA if they were less than 18 years of age, satisfied the 1987 American College of Rheumatology (ACR) classification criteria for RA (21), or were previously diagnosed with RA by a board certified rheumatologist. Individuals comprising the HLA-DR4 enriched cohort were parents of children enrolled in the Diabetes Autoimmunity Study in the Young (DAISY), a cohort of children with an increased risk of type 1 diabetes either through the presence of HLA-DR4 or a family history of type 1 diabetes (22). DAISY parents have a prevalence of DR4 positivity that approaches 45% (22), higher than background prevalence rates observed in populations of similar ancestry. FDRs of probands with RA were recruited by contact through the probands' rheumatologists from clinics at U.S. academic centers, Veterans' hospitals, and private and public sector rheumatology clinics based in New York, Chicago, Omaha (as the center for the Rheumatoid Arthritis Investigational Network), Denver, Seattle, and Los Angeles (20). This study was approved by Institutional Review Boards at all study sites and all SERA participants provided informed written consent prior to study enrollment.

In addition to undergoing a blood draw and systematic examinations for evidence of early inflammatory arthritis, SERA participants provided enrollment information relevant to: sociodemographics (including age, sex, race/ethnicity, and educational status) in addition to medical history and health-related behaviors (including comorbid diabetes and cigarette smoking status). Study participants were also asked about the presence of PD-related signs and symptoms at enrollment using the following questions: 1) Do your gums bleed after you brush your teeth? 2) Have you ever been told by a dentist or dental hygienist that you have gingivitis or gum disease? 3) Have you ever been told by a dentist or dental hygienist that you have deep gingival pockets? (23). Formal dental / periodontal examinations were not included in the SERA study; thus corresponding data were not available for these analyses.

Autoantibody testing and classification of subjects

The measurement of RA-related autoantibodies was completed at the University of Colorado Division of Rheumatology Clinical Research Laboratory (SERA Coordinating Center). Anti-citrullinated protein antibody (ACPA) was measured using a second-generation anti-cyclic citrullinated peptide (anti-CCP2) ELISA (Diastat, Axis-Shield, Dundee, Scotland, UK; positive test ≥ 5 U/ml). Rheumatoid factor (RF)-IgM, -IgG, and -IgA isotypes were measured by ELISA using QUANTA Lite™ kits with results reported in IU/ml. RF was also measured by nephelometry according to the manufacturer's specifications (Dade Behring, Newark, Delaware, USA, IU/ml). For all RF assays (ELISA and nephelometry), positivity was defined as a serum concentration exceeding that observed in 95% of healthy controls according to ACR RA criteria (21).

For this study, a subset of SERA participants was examined based on a strategy of oversampling autoantibody positive individuals with a random selection of visits from remaining autoantibody negative subjects. Analyses were limited to a single visit for each individual included in the study. SERA participants were classified as autoantibody positive if they were positive for at least 1 RA-related autoantibody with a subset of these patients further categorized as `high-risk' if they were positive for ACPA or at least two of the four RF assays done at a single visit. The high-risk definition is based on prior data suggesting that individuals with ACPA and/or RF have a substantially higher risk of subsequently developing future RA (24–26). Furthermore, this classification scheme was based specifically on data from 980 SERA RA case probands and 200 healthy randomly-selected blood donor controls, where positivity to ACPA or at least two of the four RF assays was 96% specific to RA. For participants with the high-risk autoantibody profile who had multiple study visits, we examined the first study visit at which the patient satisfied the high-risk profile. For autoantibody positive individuals not meeting the high-risk profile definition, we used the first study visit at which individuals were positive for a single RF test. Of note, in prior studies, a single RF isotype has yielded a specificity for future RA ranging exceeding 93% (24–26). For autoantibody negative individuals, a random single study visit was used. Autoantibody negative FDRs and DAISY parents without a positive autoantibody test at all available study visits served as the referent group for analyses.

HLA-DRB1 genotyping

Participants were genotyped for HLA-DRB1 shared epitope (SE) containing alleles as previously described (20). In brief, complete subtyping for HLA-DR4 alleles was done via a modification of a real-time PCR approach. HLA-DR4 subtypes considered SE positive included DRB1*0401, 0404, 0405, 0408, 0409, 0410, 0413, 0416, 0419, and 0421. A real-time low resolution PCR analysis was also performed to identify the presence of SE-containing DR1 alleles, including *0101, 0102, 0104, 0105, 0107, 0108, and 0111.

Antibody to oral pathogens

Antibody to P. gingivalis (IgG anti-Pg) was measured at the University of Nebraska Medical Center Experimental Immunology Laboratory (27), as previously described. In brief, strain 381 of P. gingivalis (American Type Culture Collection, Bethesda, MD) was grown in reducing broth (10 g of yeast extract, 30 g of trypticase soy broth, 1 g cysteine, 100 mg of dithiothreitol [DTT], 5 mg of hemin and 2.5 mg of menadione in a 1-liter volume) (28). The cells were grown with constant low-speed shaking (150 rpm) at 37°C for 24 hours to an optical density of 1.5 at 660 nm.

Outer membrane antigens from P. gingivalis were prepared as described (29). Briefly, cells were pelleted from the broth culture medium at 6,500 rpm for 20 minutes in 200 ml bottles. The supernatant was removed; the pellet re-suspended in 0.15 M NaCl containing protease inhibitor 3 and frozen in a minimal volume at −80°C overnight. The next day the suspension was thawed and washed twice more with 200 ml of 0.15 M NaCl containing protease inhibitor 3 by centrifugation at 6,500 rpm for 20 minutes. The pellet was suspended in 50 ml of a buffer containing 0.05 M sodium phosphate (with protease inhibitor), 0.15 M NaCl, and 0.01 M EDTA, pH 7.4. The suspension was sonicated on ice with a Misonix XL-2000 with a ¼ inch microtip (Newtown, CT) 5 times, 1 minute each time, at 70 watts, cooling the probe between sonications. The sonicate was centrifuged at 12,000 × g for 20 minutes using a 50Ti rotor in a Beckman Ultracentrifuge (Palo Alto, CA), the pellet discarded, supernatant collected and the protein concentration determined using the BioRad Protein Assay (Bio-Rad Laboratories, Inc., Hercules, CA).

An enzyme-linked immunosorbent assay (ELISA) was adapted from the procedure described by Engvall and Perlmann (30). P. gingivalis antigen was coated to ELISA plates using 100–200 ng/well of sonic extracts. Additionally, purified human IgG (Jackson Immunochemical, Avon, PA) was serially diluted down the appropriate ELISA plates starting at 1280 ng/ml, to construct a standard curve. Two-fold serum dilutions from patients (first dilution 1:100) were added to the plate and the bound human IgG detected with a peroxidase-conjugated, affinity-purified anti-human IgG, Fc-specific antibody (Jackson Immunochemical, Avon, PA) and then developed using TMB Substrate (Becton-Dickinson, Palo Alto, CA). Absorbance was detected after stopping the reaction using 2 N H₂SO₄ and reading at 450 nm using an MRX II Microplate Reader (Dynatech). Data were analyzed using Revelations Software from Dynatech and extrapolated to pg/ml using the purified human IgG as a standard curve. Prior studies have demonstrated that several strains of P. gingivalis, including strain 381, express common outer membrane proteins (31). Furthermore, case-control studies involving patients with PD have demonstrated higher mean antibody reactivity among cases compared to controls for a majority of outer membrane proteins identified (31, 32).

To examine potential issues of assay cross-reactivity and whether associations were unique to P. gingivalis, we also measured antibody to Prevotella intermedia (IgG anti-Pi) and Fusobacterium nucleatum using similar approaches. Similar to P. gingivalis, P. intermedia (formerly Bacteroides intermedius; American Type Culture Collection, Bethesda, MD) and F. nucleatum (American Type Culture Collection, Bethesda, MD) are gram negative anaerobes that act as major oral pathogens in PD (33, 34). P. intermedia is specifically known to frequently co-aggregate with P. gingivalis in PD-related bio-films (35). All bacterial antibody concentrations were determined in μg/ml extrapolated from a standard curve, and then log-transformed for analysis. Intra-assay coefficients of variation for the bacterial ELISAs ranged from 11% to 13%. There were no changes in measured bacterial antibody concentrations in a subset of 20 serum samples following RF depletion (performed using HeteroBlock, Omega Biologicals, Bozeman, Montana (36)).

Data analysis

Group characteristics (autoantibody positive and high-risk vs. autoantibody negative) and the prevalence of self-reported PD signs and symptoms were compared using the Chi-square test for categorical variables and the Student's t-test for continuous variables. Given skewed distributions, bacterial antibody concentrations were normalized using natural log transformation. Log-transformed bacterial serologies were compared by group using a Student's t-test. The correlation of transformed anti-bacterial antibody concentrations was examined using Spearman correlation.

Associations of bacterial antibody concentrations with autoantibody positive and high-risk status (vs. controls) were examined using logistic regression analysis, with autoantibody positive and high-risk status assessed in separate models. Associations were examined by generating odds ratios (ORs) and 95% confidence intervals (CIs). Subsequent multivariable regression analyses were conducted to adjust for confounding. In addition to both bacterial serologies (anti-Pg + anti-Pi and anti-Pg + anti-Fn modeled separately), variables examined in multivariable models included factors associated with RA and/or PD (37) (age, gender, race [Caucasian vs. other], ever smoking, HLA-DRB1 SE status [1 or 2 alleles vs. none], diabetes, and education [> high school vs. ≤ high school]). In sensitivity analyses, we also examined the associations of bacterial serologies with high-risk status, using an alternative definition of high-risk for future RA that included either ACPA positivity or positivity of two or more RF isotypes (eliminating RF-nephelometry results from the definition). All analyses were completed using STATA v10.1 (StataCorp, College Station, TX).

Results

There were 284 subjects included in the analyses, including 171 autoantibody negative and 113 autoantibody positive individuals. Of the 113 categorized as autoantibody positive, 38 were further categorized as high-risk based on the presence of a positive ACPA or positivity to two or more RF assays. When high-risk status was defined alternatively as positivity to ACPA or ≥ 2 of 3 RF isotypes (eliminating RF nephelometry from the definition), there were 33 individuals categorized as being high-risk. Group characteristics are summarized in Table 1. Compared to autoantibody negative subjects (44 ± 14 [S.D.] years), both autoantibody positive (48 ± 15 years; p = 0.044) and high-risk individuals (51 ± 16 years; p = 0.012) were older at enrollment. There were no other significant group differences noted in the autoantibody positive or high-risk groups compared to autoantibody negative subjects. Specifically, there were no group differences in the prevalence of known PD risk factors including HLA-DRB1 SE positivity, smoking, or diabetes (Table 1). Relative to autoantibody negative individuals, there were also no significant differences among autoantibody positive or high-risk individuals in the prevalence of self-reported signs and symptoms of PD including gum bleeding, a diagnosis of gum disease or gingivitis, or the presence of deep periodontal pockets.

Table 1.

Descriptive characteristics and the frequency of self-reported signs and symptoms of peridontitis (PD) among SERA participants^*

	Autoantibody Negative (n = 171)	Autoantibody Positive (n = 113)	High-Risk (n = 38)
Sociodemographics and RA Risk Factors
Age, yrs, mean (± S.D.)	44 (14)	48 (15)†	51 (16)†
Female, %	69	73	76
Caucasian race, %	77	82	82
Cigarette smoking, ever, %	37	31	29
Diabetes mellitus, %	5	4	5
> High school education, %	79	79	73
HLA-DRB1 shared epitope positive, %	55	55	58
Signs and symptoms of Periodontitis
Gum bleeding, %	22	26	19
Gum disease or gingivitis, %	21	23	19
Deep periodontal pockets, %	15	18	20

^*SERA = Studies of the Etiology of Rheumatoid Arthritis; autoantibody positive status was defined as subjects with at least one positive rheumatoid arthritis (RA)-related autoantibody; high-risk status was defined as individuals positive for anti-citrullinated protein antibody (ACPA) or positive for ≥ 2 rheumatoid factor (RF) assays (nephelometry, IgA, IgM, or IgG); autoantibody negative subjects defined as those negative for all RA-related autoantibodies examined; all individuals classified as high-risk also included in autoantibody positive group (all individuals positive for 1 or more RA-related autoantibody)

^†p < 0.05 vs. controls

Rates of positivity for different RA-related autoantibodies among autoantibody positive and high-risk individuals are summarized in Table 2. There were eight ACPA positive individuals, comprising 7% of those positive for at least one autoantibody and 21% of the high-risk group.

Table 2.

Frequency of rheumatoid arthritis (RA)-related autoantibody positivity in study participants classified as autoantibody positive or high-risk^*

	≥ 1 Autoantibody Positive (n = 113)	High-Risk (n = 38)
	Number (%)
Anti-citrullinated protein antibody (ACPA)	8 (7)	8 (21)
Rheumatoid Factor, nephelometry	38 (34)	18 (47)
Rheumatoid Factor, IgA	26 (23)	16 (42)
Rheumatoid Factor, IgM	36 (32)	25 (66)
Rheumatoid Factor, IgG	60 (53)	26 (68)

^*Autoantibody positive status defined as subjects with at least one positive rheumatoid arthritis (RA)-related autoantibody; high-risk subjects are defined as individuals positive for anti-citrullinated protein antibody (ACPA) or positive for ≥ 2 rheumatoid factor (RF) assays (nephelometry, IgA, IgM, or IgG); all individuals classified as high-risk also represented in autoantibody positive group (all individuals positive for 1 or more RA-related autoantibody); all patients classified as high-risk also included in autoantibody positive group

Circulating IgG antibody concentrations to P. gingivalis, P. intermedia, and F. nucleatum by study group (log-transformed) are shown in Figure 1. Anti-Pg antibody concentrations were significantly correlated with both anti-Pi (r = 0.60, p < 0.001) and anti-Fn (r = 0.45, p < 0.001). Log-transformed anti-Pg concentrations were higher in those with at least one positive autoantibody then in autoantibody negative subjects (4.89 ± 1.00 vs. 4.59 ± 0.88; p = 0.010), a difference that was numerically greater in high-risk individuals (5.03 ± 1.11 vs. 4.59 ± 0.88; p = 0.011). In contrast, there were no significant differences in anti-Pi concentrations among those with at least one positive autoantibody (5.46 ± 0.70; p = 0.564) or high-risk individuals (5.57 ± 0.48; p = 0.191) compared to autoantibody negative individuals (5.42 ± 0.66). Likewise, there was no difference in anti-Fn antibody concentrations among autoantibody positive individuals (4.49 ± 0.82, p = 0.59) or high-risk individuals (4.48 ± 0.90, p = 0.80) compared to autoantibody negative individuals (4.44 ± 0.84) (Figure 1). There were no significant differences for any of the bacterial serologies, including anti-Pg, when comparing ACPA positive subjects (n = 8) with autoantibody negative individuals (data not shown).

Figure 1.

Log-transformed antibody concentrations to P. gingivalis (upper left), P. intermedia (lower left), and F. nucleatum (upper right) based on autoantibody positive (Ab+; n = 113; positive for ≥ 1 RA-related autoantibody), high-risk (HR; n = 38; positive ACPA or positive for ≥ 2 RF assays), or autoantibody negative status (Ab−; n = 171). *p = 0.010; **p = 0.011; all other comparisons non-significant.

Univariate and adjusted associations of log-transformed anti-Pg, anti-Pi, and anti-Fn concentrations with the presence of at least one positive autoantibody and high-risk status (vs. autoantibody negative status) are summarized in Table 3. In unadjusted analyses, anti-Pg concentrations (per unit log unit increase) were significantly associated with autoantibody positive status (OR_unadj = 1.41; 95% CI 1.08, 1.85, p = 0.011) but anti-Pi (OR_unadj = 1.11; 95% CI 0.78, 1.59; p = 0.563) and anti-Fn (OR_unadj = 1.08; 95% CI 0.81, 1.45, p = 0.594) were not. Unadjusted associations of anti-Pg with high-risk status (OR_unadj = 1.68; 95% 1.12, 2.52; p = 0.012) were also significant while there were non-significant associations of anti-Pi with high-risk status (OR_unadj = 1.51; 95% CI 0.82, 2.80, p = 0.188). Anti-Fn antibody showed no association with high-risk status, results that did not change following multivariable adjustment. Results were not substantially changed following multivariable adjustment referent to anti-Pg concentrations, which remained significant. In contrast, associations of anti-Pi concentration with high-risk status were completely attenuated following multivariable adjustment (OR_adj = 0.97; 95% CI 0.42, 2.27; p = 0.953) (Table 3). In multivariable analyses that included all three bacterial serologies in addition to the aforementioned covariates, the association of anti-Pg with autoantibody positivity (OR_adj = 1.51; 95% CI 1.04, 2.20, p = 0.032) and high risk status (OR_adj = 1.64; 95% CI 0.94, 2.89, p = 0.083) did not change substantially although the association did not reach statistical significance for high-risk status. Of the covariates examined including age, gender, race, smoking, HLA-DRB1 SE status, diabetes, and education, only older age was significantly associated with autoantibody positive and high-risk status in multivariable models (data not shown). Associations of anti-Pg antibody with high-risk status were also unchanged and remained significant when an alternative definition of high-risk autoantibody was used consisting of positivity to either ACPA or 2 of 3 RF isotypes (data not shown).

Table 3.

Associations of antibodies to P. gingivalis (anti-Pg), P. intermedia (anti-Pi), and F. nucleatum (anti-Fn) with the presence of rheumatoid arthritis (RA)-related autoantibody among SERA participants^*

	Autoantibody Positive vs. Control O.R. (95% C.I.)			High Risk vs. Control O.R. (95% C.I.)
Bacterial IgG Ab (μg/ml, log-transformed)	Univariate	Multivariable Model A†	Multivariable Model B†	Univariate	Multivariable Model A†	Multivariable Model B†
Anti-Pg	1.41 (1.08 to 1.85)	1.56 (1.10 to 2.22)	1.39 (1.00 to 1.92)	1.68 (1.12 to 2.52)	1.72 (1.01 to 2.95)	1.70 (1.05 to 2.74)
	p = 0.011	p = 0.013	p = 0.047	p = 0.012	p = 0.046	p = 0.031
Anti-Pi	1.11 (0.78 to 1.59)	0.76 (0.47 to 1.23)	----	1.51 (0.82 to 2.80)	0.97 (0.42 to 2.27)	----
	p = 0.563	p = 0.268		p = 0.188	p = 0.953
Anti-Fn	1.08 (0.81 to 1.45)	----	0.98 (0.67 to 1.43)	1.06 (0.69 to 1.62)	----	1.01 (0.57 to 1.76)
	p = 0.594		p = 0.907	p = 0.800		p = 0.985

^*SERA = Studies of the Etiology of Rheumatoid Arthritis; autoantibody positive defined as subjects with at least one positive rheumatoid arthritis (RA)-related autoantibody; high-risk subjects defined as individuals positive for anti-citrullinated protein antibody (ACPA) or positive for ≥ 2 rheumatoid factor (RF) assays (nephelometry, IgA, IgM, or IgG); all individuals classified as high-risk also represented in autoantibody positive group (all individuals positive for 1 or more RA-related autoantibody)

^†Adjusted for age, gender, race, ever smoking, shared epitope status, diabetes, education, and alternative bacterial serology (Model A includes anti-Pg and anti-Pi; Model B includes anti-Pg and anti-Fn)

Discussion

Although associations between RA and PD have been noted for several decades (38), mechanisms underpinning this relationship have not been clearly elucidated. Recent studies, including one from our group (27, 39, 40), have suggested that infection with P. gingivalis is a co-factor in the development and progression of RA. We previously observed that antibodies to P. gingivalis are found in significantly higher concentrations among patients with RA than in controls (27). Furthermore, in patients with RA, anti-Pg antibody concentrations correlate with ACPA expression (27, 40). Of 11 different oral bacteria examined, P. gingivalis was recently shown to be the only organism capable of endogenously citrullinating both fibrinogen and enolase (41). This is important because autoantibodies directed against these antigens in their citrullinated form are both highly specific to RA and have been speculated to play a pathogenic role in disease progression (42). It has been hypothesized that the dual expression of gingipains (lysine- and arginine-specific proteases expressed by several oral bacteria) and PAD by P. gingivalis acts in concert in RA, the former producing a carboxy-terminal arginine residue that then serves as a target for bacterially expressed PAD (18). Citrullination by bacterial PAD appears to be distinct from mammalian PAD that more efficiently citrullinates internal arginine residues (43).

In addition to associations with ACPA, P. gingivalis along with other oral pathogens have also been speculated to play a role in RF expression. Patients with PD are more likely to be RF-seropositive than controls and RF has been identified in the gingival tissue and subgingival plaque from patients with PD (44). Because lysine and arginine residues exist in the Fc region of IgG (45), it has been suggested that modification of these domains by bacterially expressed gingipains leads to targeting and binding with RF (19). Whether the association of P. gingivalis with RA is directly operative through effects on RF or ACPA generation in the oral cavity or related lymphatic structures remains unknown, and needs further study.

The timing and development of RF and ACPA in relationship to each other in RA development is also unknown; some studies suggest that ACPAs are generated prior to RF (25, 26), although conversely, a study using the U.S. Department of Defense Serum Repository showed that RF positivity often precedes ACPA expression (46). This area of study is particularly relevant when trying to understand the mechanisms by which PD and infection with P. gingivalis may lead to the development of RA-related autoimmunity. For example, given the association of PD with RF elevations in the oral cavity, and the association in our study of elevated antibody to P. gingivalis in subjects that predominantly only had circulating elevations of RF, P. gingivalis infection may initially lead to RF development, followed by generation of ACPAs. Investigations into this area are planned in further study of longitudinal samples from the SERA project.

Hitchon and colleagues recently examined the association of P. gingivalis antibody with the presence of RA-related autoantibodies in the North American Native (NAN) population, a study that included patients with RA, unaffected relatives (FDRs), and controls (39). This study showed that antibodies specific to P. gingivalis were found in higher concentrations in both ACPA positive RA and among ACPA positive FDRs compared to ACPA negative RA cases and ACPA negative FDRs, respectively. In addition to important differences in target populations and techniques used for the measurement of bacterial serologies (a lipopolysaccharide [LPS]-based assay (39) vs. whole lysate-based approach in our study), there are other aspects that distinguish these efforts. ACPA positivity was relatively low in our study population (2.8%), precluding meaningful analyses of this subgroup. This rate of ACPA positivity is much lower than the prevalence of 19% reported among NAN FDRs. While this may relate to differences in the background prevalence of disease risk factors such as HLA-DRB1 SE (73% for NAN FDRs vs. 55% in our study population), this difference is largely attributable to variable definitions used for ACPA positivity. In the present study, individuals were considered to be ACPA positive based solely on results from the anti-CCP2 ELISA with prior reports demonstrating a substantially increased risk of developing RA with seropositivity to this assay (25, 26). In contrast, NAN RA cases and NAN FDRs were considered ACPA positive if seropositive for either the anti-CCP2 assay (5% positive) or if seropositive for any anti-CCP2 isotype (IgG_1–4, IgA, or IgM). As pointed out by these authors (39), the prognostic implications of anti-CCP2 isotype positivity among unaffected relatives remain unclear. In addition to ACPA positivity our high-risk group included individuals seropositive for two or more RF assays, a definition that was not operationalized in the prior NAN investigation but that has been shown to portend disease risk in other unaffected populations. For instance, positivity to two or more RF isotypes has been shown to have a specificity of 98% for the development of future RA (24).

Our results complement and extend prior reports in several meaningful ways. For each log-fold increase in anti-Pg antibody concentration, individuals in our study were 40% to 70% more likely to be seropositive for RA-related autoantibodies. These associations were independent of all other RA and PD risk factors examined. To our knowledge, ours is the first study to simultaneously examine associations of P. gingivalis and alternative oral pathogens with autoantibody expression in individuals at heightened risk for future RA. Recognizing that P. intermedia and other oral bacteria frequently co-aggregate with P. gingivalis in PD-related bio-films (35), the correlations observed between bacterial antibody concentrations were expected. Importantly, results of multivariable regression that simultaneously included antibody to P. gingivalis and an alternative oral pathogen (in addition to antibody to all three organisms examined) as independent variables showed that associations with the presence of RA-related autoantibodies were specific to P. gingivalis and did not extend to at least two other oral pathogens frequently implicated in PD. Similar rates of self-reported PD signs and symptoms reported across the study groups further support, but do not prove, the contention that infection with P. gingivalis drives the observed associations with RA-related autoantibody expression rather than disease-related autoantibody production being a consequence of non-specific periodontal inflammation. Recognizing only modest sensitivities and moderate predictive values for self-reported PD (23), it would have been optimal to have had results from periodontal examinations. However, results from standardized oral assessments were not part of this study although such assessments are planned for ongoing studies.

Additional studies with longer follow-up and larger sample sizes will be needed to more clearly define the epidemiologic links between P. gingivalis infection and RA onset. It will be essential that future efforts are designed to more precisely detail the temporal relationship of this infection with the immunologic responses that follow including the formation of RA-related autoantibodies. Although our study included simultaneous examinations of P. gingivalis and two other oral pathogens, additional efforts will be needed to examine the many other oral pathogens that have been identified but were not included in this study in addition to the complex interactions that are likely to exist between pathogens comprising oral and subgingival microbiomes. Regardless, these results demonstrate that associations of P.gingivalis infection with RA-related autoantibody expression exist among individuals without clinically apparent RA but with a higher background risk of future disease. Furthermore, based on these observations, it is unlikely that established RA serves as the initiating event in this relationship. Specifically, these results refute speculation that PD (and infection with P. gingivalis) simply represents a consequence of severe RA or an `opportunistic' product of immunosuppressive therapy. Importantly, these results provide insight into a potentially critical environmental trigger in RA pathogenesis, one that could be targeted in future interventions aimed at disease prevention.