Introduction
The etiology of rheumatoid arthritis (RA) is unknown, but both environmental and genetic factors are likely to play roles in its pathogenesis. Periodontal disease (PD), an inflammatory disease of tooth-supporting structures, may be an environmental trigger for RA. Compared with healthy controls, PD is more frequent in RA patients, both in those with new-onset and in those with long-standing disease, even when potential confounding factors such as smoking are taken into account [
1‐
5].
Furthermore, there is increasing evidence of a role for PD pathogens, particularly
Porphyromonas gingivalis (
Pg), in RA pathogenesis.
Pg is the only prokaryote known to possess a peptidylarginine deiminase (PAD), an enzyme that catalyzes the posttranslational modification of arginine residues to citrulline. Although citrullination may occur more generally in sites of inflammation, antibodies to citrullinated proteins (anti-cyclic citrullinated peptide (anti-CCP) antibodies) are specific for RA and are now valuable diagnostic markers for the disease [
6]. CCP antibodies are associated with a more aggressive course [
7] and may be detected prior to the onset of clinical disease [
8], suggesting a role in RA pathogenesis.
Pg, through its PAD enzyme, may citrullinate host or bacterial proteins [
9], altering their antigenicity and triggering autoimmunity and RA in predisposed individuals [
9,
10]. Further support for this hypothesis comes from animal models.
Pg enolase has been found to cause arthritis in DR4-IE-transgenic mice [
11], and
Pg infection has been shown to exacerbate collagen antibody-induced arthritis [
12].
Prior studies have demonstrated increased frequencies of antibody responses to
Pg in RA patients compared with healthy controls [
5,
13‐
16]. However, in these studies [
5,
13,
15,
16], patients generally had long-standing disease and were presumably receiving disease-modifying antirheumatic drugs (DMARDs), factors which may affect infection with PD pathogens and serum antibody responses. Moreover, clinical correlations with
Pg responses have been inconsistent. Some investigators have noted an association of
Pg antibodies with anti-CCP antibody levels, but not with RF values [
14,
15], whereas others found a correlation of
Pg immunoglobulin G (IgG) antibodies with RF levels, but not with CCP antibody values [
5,
16]. Additionally, some researchers found an association of
Pg antibodies and elevated C-reactive protein (CRP) levels [
13,
14], but others noted no correlation between these antibodies and Disease Activity Score based on 28-joint count (DAS28)-CRP values [
16].
Patients with early RA are an important group to study because they may benefit the most from treatment interventions for
Pg and PD. In the only previous study in which
Pg antibodies were examined in early RA patients, Scher
et al. [
4] found no significant difference in antibodies to a specific
Pg antigen (
Pg-specific chaperone protein HtpG) in patients with new-onset or chronic RA compared with control participants. However, antibody responses to whole-
Pg antigen preparations have not yet been reported in early RA patients. Furthermore,
Pg antibody responses in RA patients have not been compared to those of patients with other connective tissue diseases (CTDs), such as lupus, which are associated with broad immune stimulation.
In this study, we determined Pg antibody responses in patients with early RA and in comparison groups and correlated these results with standard RA biomarkers, disease activity scores and measures of function. We report here that both early and late RA patients had significantly more frequent and higher IgG antibody responses to whole-Pg sonicate antigens than healthy control participants and tended to have higher Pg antibody reactivity than patients with other CTDs. In early RA patients prior to DMARD therapy, Pg antibody responses correlated significantly with anti-CCP antibody reactivity and, to a lesser degree, with erythrocyte sedimentation rate (ESR) values. Moreover, there was a trend toward higher disease activity and more functional impairment in patients with Pg antibodies, and these trends remained during 12 months of DMARD therapy. Our observations are consistent with a role for Pg in disease pathogenesis in a subset of RA patients.
Discussion
In this study, 34% of patients with early RA, prior to DMARD therapy, had positive IgG antibody responses to
Pg. These antibody responses were of significantly greater frequency and magnitude than those of age-similar healthy hospital personnel or blood bank donors. Most of the RA patients were women, and the majority of hospital personnel were men. Although men may carry a greater risk of PD than women [
20], a highly significant difference was still shown between the RA patients and healthy participants. In addition,
Pg antibody levels in early RA patients were similar to or even higher than those in late RA patients, even though the latter group had more risk factors for PD, including older age and more frequent smoking history [
21]. Furthermore, the similar or higher
Pg antibody levels in DMARD-naïve early RA patients compared with DMARD-treated late RA patients suggest that immunosuppression from DMARD therapy did not alter or enhance
Pg antibody reactivity. Although 26% of the early RA patients at the time of enrollment were receiving steroids, which may potentially lower antibody levels,
Pg antibody levels tended to be higher in patients receiving steroids, suggesting that steroid treatment did not decrease
Pg antibody levels.
We also included a comparison group consisting of patients with other CTDs typically associated with systemic inflammation and autoantibody production, including lupus, mixed CTD, Sjögren syndrome or scleroderma. About 20% of these patients had positive antibody responses to Pg, which was a lesser percentage than RA patients (34%) but greater than age-similar healthy control participants (5%). We hypothesize that Pg reactivity in patients with other CTD may have been caused by PD, nonspecific immune stimulation or both. However, because almost all of the patients with other CTDs lacked anti-CCP antibodies, cross-reactivity with citrullinated proteins does not appear to account for increased Pg reactivity in the non-RA patient groups. Thus, we conclude that a subset of RA patients seen early in the illness prior to DMARD therapy have positive Pg IgG antibody responses, which were more frequent and of higher magnitude than those in individuals in the other groups we tested.
Determining which
Pg antigens to use for serologic tests is problematic. Using whole-
Pg lysates, a
Pg-specific lipopolysaccharide or recombinant citrullinated
Pg PAD, other investigators have previously demonstrated increased frequencies of
Pg antibody responses in patients with chronic RA compared with healthy control participants [
5,
13‐
16,
22], as we did using whole-sonicate antigens. In the only previous study of new-onset RA patients, antibody responses to
Pg specific chaperone protein HtpG were not significantly different from those in healthy control participants [
4]. However, antibody responses to this protein may protect against PD and may even be higher in healthy subjects [
23]. In our study, we chose to use whole-sonicate antigens because little is known about the frequencies of reactivity with specific
Pg antigens, even in patients with PD. Therefore, the use of whole-sonicate antigens, which encompass many
Pg proteins, would increase the possibility of detecting responses. Although such preparations would presumably also increase the potential for false-positive results because of cross-reacting antibodies to other organisms or cross-reactivity between bacterial and host citrullinated proteins [
24], we attempted to address this issue by testing serum samples from multiple comparison groups. Future serologic studies using multiple types of assays, such as ELISA and Western blot analysis, as well as different antigen preparations, including noncitrullinated
Pg proteins, will be important to define further the sensitivity and specificity of
Pg antibody responses in RA patients.
In our study, a limitation in our ability to interpret
Pg antibody responses is the lack of dental examinations in case and control participants. In our study, the cutoff value for a positive response was 2 SD above the mean value in the age-similar healthy control participants who reported no history of PD. Yet, it is possible that some of these individuals had PD, which might obscure a difference between RA patients and control participants. However, a highly significant difference was still shown between these groups. Others have previously noted an increased frequency of PD in RA patients [
1‐
5].
Pg antibody responses correlate well with the presence of PD [
25], and higher
Pg antibody titers and DAS28 scores have been noted in RA patients with more severe PD [
5]. However,
Pg antibodies may also be associated with a host protective response or carriage of
Pg as a commensal opportunistic pathogen [
23,
26]. Therefore, in future studies, it will be important to perform
Pg antibody studies along with dental examinations, as we have now initiated in our cohort.
In addition to RA, increased frequencies and severity of PD have been reported in several other rheumatic diseases, including ankylosing spondylitis [
27], scleroderma [
28] and psoriatic arthritis [
29]. Although patients with Sjögren syndrome have xerostomia, periodontitis does not seem to be more frequent in these patients [
30]. In addition to
Pg reactivity, antibody responses to several other PD pathogens have been found in RA patients, including
Prevotella intermedia,
Prevotella melaninogenica,
Actinobacillus actinomycetemcomitans and
Bacteroides forsythus [
13,
16]. Thus, among rheumatic diseases, increased frequencies of PD may not be unique to RA and increased frequencies of positive antibody responses to PD pathogens may include organisms other than
Pg.
What does seem to be unique in RA is the association between
Pg and anti-CCP antibody responses. Consistent with previous reports [
14,
15], our early RA patients with
Pg antibodies more often had anti-CCP antibody reactivity. Their anti-CCP responses were significantly higher, and the levels of anti-
Pg antibodies correlated directly with anti-CCP levels. In contrast, anti-CCP antibody responses were rarely found in patients with other CTDs, and, if present, the levels were very low. Moreover, these responses were not present in healthy control participants. Mikuls
et al. recently noted that
Pg antibodies may be present prior to the development of synovitis [
31], and, in our study, no patient who initially had negative results for
Pg antibodies developed these antibodies during 12 months of DMARD therapy. Taken together, these observations suggest that immunity to
Pg is one factor that may set the stage for autoimmunity and inflammatory synovitis in a subset of RA patients.
Citrullination of proteins may occur within the joint as well as at other sites of inflammation, such as the lung and periodontium [
32,
33]. In our study,
Pg-negative RA patients were more likely to be smokers, suggesting that the lung, rather than the gingiva, may have been the site of protein citrullination in these patients. Further studies which examine the lung and periodontium in RA patients are needed to evaluate this hypothesis. Although
Pg may be involved in RA pathogenesis through citrullination, there may be other important mechanisms by which
Pg contributes to disease activity in RA. For example, the organism may be more likely to skew CD4+ T-cell reactivity to a Th17 phenotype [
34], a response which has been implicated in autoimmunity [
35].
Despite specific mechanisms that may account for the association of
Pg immunity with RA, we found, as others have [
5,
14,
16], that RF, general markers of inflammation (ESR), scores of disease activity (DAS28 and CDAI) and functional impairment (HAQ) were also greater in patients with
Pg antibody responses. Moreover, the trends remained for the DAS28-ESR and ESR values during 12 months of DMARD therapy. Although the CDAI scores were also slightly higher in the
Pg antibody-positive group, the differences were not as great as those in the DAS28-ESR, suggesting that ESR is an important contributor to differences in disease activity scores between the groups. Thus,
Pg antibodies may simply be a marker for PD, a chronic inflammatory condition that may itself be associated with elevation of inflammatory markers. However, the trend toward more severe disease activity in
Pg antibody-positive patients using indices (HAQ and CDAI) that do not incorporate inflammatory markers suggests that
Pg itself may contribute to RA disease activity.
There are possible confounding issues in the association of
Pg antibody reactivity with greater disease activity in RA. For example, certain health behaviors, such as lack of routine dental care, might extend to noncompliance with RA treatment. Factors not studied here, such as the shared epitope (SE), may contribute to the severity of PD [
36] as well as RA, and the SE has been associated with periodontal bone destruction in RA patients [
37]. However, other studies in RA patients have not found a correlation between the SE and PD or
Pg antibodies [
5,
14]. Because of the heterogeneous environmental and host factors involved in RA, patient cohorts may vary, accounting for inconsistent results in clinical correlations among studies [
4,
5,
14‐
16]. However, consistent with our results, there is general agreement that
Pg immunity is associated with greater inflammation in RA patients.
In a preliminary study of patients with established RA, Ortiz
et al. reported that disease activity, as measured by DAS28 score and swollen joint counts, improved with nonsurgical treatment of PD [
38].
Pg antibodies as a marker for PD may prove to be useful in the identification of patients who would benefit from such treatments. It will be particularly important to conduct studies of therapies for PD and
Pg in patients with early RA because PD treatment may make the most difference in early stages of the disease.
Authors' contributions
SLA performed the clinical data collection, conducted the laboratory research experiments, performed data interpretation and analysis, contributed to the study design and was responsible for the writing of the manuscript. DSC, MCF, SU and GLC participated in the selection, follow-up and medical care of enrolled patients and reviewed the manuscript. GAM, TK and KS assisted with the design and conduct of the laboratory experiments and reviewed the manuscript. AS, as senior author, designed and coordinated the study and contributed to interpretation of the data and the intellectual input in drafting the manuscript. All authors read and approved the final manuscript.