Introduction
Rheumatoid arthritis (RA) is a common chronic systemic inflammatory disease that causes pain and disability. Pain in active RA may reflect synovitis, but not all people with RA respond to antiinflammatory treatment, and other pain mechanisms may be important. Noninflammatory pain mechanisms in RA may include variations in central pain processing, or coincidental pathology. Pain phenotypes in patients with RA that include widespread pain, mood disturbance, and fatigue overlap with those in fibromyalgia (FM). Fibromyalgia may represent either comorbidity or a continuous phenotypic spectrum associated with variations in central pain processing [
1]. Noninflammatory pain mechanisms in RA may share some common characteristics with those found in other arthropathies, such as osteoarthritis (OA), in which greater sensitivity to pain at nonjoint sites, as well as at sites affected by arthritis, may indicate augmented central pain processing [
2]. Previous studies have identified and partially characterized people with RA who fulfill classification criteria for FM [
3]. In general, people with RA and FM show higher pain, higher measures of disease activity, and poorer mental health, but also less structural damage [
4,
5]. Psychological distress may be associated with greater pain sensitivity [
6].
The 28-joint disease activity score (DAS28) is commonly used in RA as a clinical measure of inflammatory disease activity and to guide treatment decisions [
7,
8]. It appears that higher pain levels contribute to higher DAS28 scores [
9,
10]. DAS28 combines four discrete components into an overall measure of RA inflammatory disease activity; a 28-swollen joint count (SJC), 28 tender-joint count (TJC), acute-phase response (commonly the erythrocyte sedimentation rate (ESR)), and a general health assessment using a visual analogue score (VAS-GH). Of these components, SJC and acute-phase response are observed by the assessor, and VAS and TJC are reported by the patient. DAS28-P is a derived index for the contribution of patient-reported components to the DAS28, which, we have proposed, may provide a quantitative assessment of noninflammatory contributions to pain in early RA. A high DAS28-P was associated with worse pain and disability in early RA [
11], and DAS28-P may be high in people with FM [
12]. Tenderness may be associated with augmented pain processing and other features of FM in RA, and DAS28 scores may overestimate inflammatory disease activity in people with a high TJC [
12,
13]. The DAS28-P index is currently a research tool and has not been validated previously as a measure of augmented pain in people with long-standing RA.
Quantitative sensory testing can be used to assess pain mechanisms [
14]. Pressure pain-detection thresholds (PPTs) measure the intensity required for blunt pressure stimuli to produce pain. Differences in PPTs between individuals at sites distant to tissue injury may indicate augmented central pain processing [
2]. In this study, we hypothesized that, in people with RA on stable disease-modifying treatment, pain sensitivity, measured by lower PPTs, is associated with a high contribution of patient-reported factors to DAS28 and with features of FM. Associations between higher DAS28-P indices and pain sensitivity or FM would support interpretation of DAS28-P as a measure of augmented pain processing. We also aimed to explore, within a single patient group, associations between pain sensitivity, FM classification, mental health, and their effects on disease-activity assessment.
Discussion
We provide evidence that augmented pain processing is associated with pain and distress in people with established RA, and confounds assessment of inflammatory disease activity when using DAS28. We also show that DAS28-P has potential to be used as a marker for augmented pain processing in people with established RA during stable treatment with disease-modifying agents.
Low PPTs (more sensitive) were associated with greater reported pain in patients with established RA. Low PPTs have been associated with higher reported pain in other chronically painful musculoskeletal conditions, including osteoarthritis [
14] and FM [
26]. Low PPTs are also associated with neuropathic pain characteristics, as determined by questionnaires [
27,
28], and we confirmed this association in people with RA by using the neuropathic scale of the McGill Pain Questionnaire. However, similar associations were found with other pain characteristics, suggesting that low PPTs may be a feature of chronic musculoskeletal pain in general, rather than indicating specifically neuropathology.
We found that PPTs over joints were correlated with PPTs at sites distant to joints, and associations between PPTs and reported pain, mental health, and fatigue were equally strong at both articular and nonarticular sites. These findings suggest that PPTs reflect general patient-level characteristics, rather than only peripheral sensitization in inflamed joints. Similarly, widespread reductions in PPTs in people with OA [
14] or FM [
26] have been taken to indicate augmented central pain processing, and sensitivity to pressure has been associated with increased activity in several brain regions during functional MRI studies [
29,
30]. PPTs are lower in people with RA than in controls [
31] and generalized hyperalgesia and allodynia occur in RA [
32]. These and our data indicate that low PPTs may provide evidence of augmented central pain processing in people with RA.
We found that a high proportion (48%) of patients attending secondary care follow-up with established RA satisfied FM-classification criteria. Others have also found a higher prevalence of FM in people with RA than would be expected in the normal population [
33]. In one large U.S. longitudinal cohort study, approximately 10% of participants with RA satisfied FM criteria at one time point, and close to 20% satisfied FM criteria at some point during follow-up [
3]. Our participants were selected on the basis of DAS28 > 3.1 and current pain, despite receiving stable treatment with traditional or biologic DMARDs, and were largely attending secondary care appointments for DMARD monitoring or annual review. People with worse pain may be more likely to volunteer for studies of pain. The prevalence of FM in our study may well reflect its specific recruitment context. Overlap between features of FM (fatigue and somatic symptoms) and RA may lead to overclassification of FM in people with RA, although these overlapping features may also reflect common underlying mechanisms, or a subgroup of people with established RA that is prone to augmented pain.
FM is itself a condition with augmented pain processing, identified by quantitative sensory testing and functional MRI [
34]. Consistent with this, we found that participants with RA who satisfied FM criteria also exhibited lower PPTs than did those who did not. Associations were found between low PPTs and worse scores on questionnaires addressing a range of FM characteristics, including widespread pain, somatic symptoms, fatigue, and low mood. Each of these questionnaires produced data that did not significantly diverge from normal distributions, and the WPI and SSS, plus their summated total, were each associated with PPTs and other fibromyalgic features. These data are consistent with the proposal that fibromyalgia represents a continuous trait rather than a dichotomous comorbidity in people with RA, and that low PPT is a component of fibromyalgia in RA.
One sixth of our participants displayed DAS28 > 5.1, suggesting high disease activity, and the remainder satisfied DAS28 criteria for active RA. High DAS28 scores in this patient group were largely determined by high VAS-GH and TJC, and high TJCs can lead to overestimation of inflammatory disease activity in RA [
10]. Lower PPTs were associated with high VAS-GH and TJC, but not with ESR or SJC, indicating that augmented pain processing may be associated with inflated DAS28 in people with well-controlled inflammatory disease. This confounding of DAS28 scores was closely associated with other features of FM.
We have previously shown that DAS28-P, the proportion of DAS28 attributable to the patient-reported components VAS-GH and TJC, is associated with current pain, and predicts poor pain prognosis in people with early RA [
11]. The current study further indicates associations between DAS28-P and pain in RA of more than 2 years’ duration, and supports our hypothesis that DAS28-P may reflect augmented central pain processing in people with RA.
Participants in this study were attending for annual review or DMARD monitoring, and the low observed prevalence of swollen joints and elevated ESR was consistent with well-controlled inflammation. Where swollen joint counts and ESR contribute little to DAS28, variations in tender-joint counts and VAS-GH may have similar effects on both DAS28 and DAS28-P. In early RA, where inflammation was often incompletely controlled, we demonstrated only weak associations between DAS28-P and DAS28 [
11], whereas in the current study, DAS28-P was strongly associated with DAS28 (r = 0.73). Analysis of DAS28 components confirmed that associations of DAS28 with low PPTs and fibromyalgia were attributable to the patient-reported DAS28 components. Our data indicate that, for people with RA in whom inflammatory disease is well controlled, DAS28 itself may predominantly reflect augmented pain processing, and should be used with great caution as a measure of inflammatory disease activity.
Joint inflammation may itself drive augmented central pain processing [
35], and RA therefore may contribute to the development of FM [
3]. However, associations of FM with poorer mental health and with fatigue, in both RA and non-RA populations, indicate that factors in addition to synovitis may be associated with the development of augmented pain processing and fibromyalgia in RA. Our data indicate that changes in pain processing may persist after synovitis has settled, but further research is required to determine possible contributions of central processing to pain in people with active synovitis. The lack of significant associations between SJC or ESR and PPT suggests that ongoing inflammation was not the predominant drive for pain sensitivity in these patients, but interventional studies would be required to determine whether more intensive suppression of inflammation in early arthritis could reduce the contribution of augmented processing to pain even after inflammation is controlled.
Our research has several limitations, being a cross-sectional study of an RA population on stable treatment within a secondary-care context. Fibromyalgia can affect many aspects of life quality on which data were not collected in the current study, including sleep disturbance and loss of concentration/memory. Additional factors not anticipated in our study may have confounded some of our findings.
The use of 28 swollen-joint counts and ESR to measure inflammation in RA also has limitations. ESR might be within the normal range in people with active RA [
36], and DAS28 does not address feet, which might be a site of synovitis. Future studies could use ultrasound or magnetic resonance imaging to determine whether subclinical inflammation might contribute to pain or reduced pain-detection thresholds in people with RA, and whether more-intensive immunosuppression might further improve pain outcomes. Participants were using a range of analgesic medications, which may have affected pain reporting and PPTs, although
post hoc analyses did not reveal any association between analgesic use and PPTs in our study group. Our findings may be limited to the specific PPT protocol used, including sequence of test sites, and further research would be required to determine whether conclusions can be generalized to other quantitative sensory testing modalities, such as thermal pain thresholds.
It is unclear from our cross-sectional study whether augmented pain processing exacerbates pain in RA, whether chronic pain, as experienced in RA, leads to changes in pain processing, or whether confounding factors contribute to the observed associations. It is likely that altered pain processing may both contribute to and be a consequence of joint pathology. Associations of pain and PPT with FM measurement, both in the current and in previous [
10,
13] studies, additionally support the concept that comorbidity may contribute to both pain severity and reduced pain thresholds in RA. Further prospective research is merited to explore whether and which PPT methodologies could be used to identify a subgroup of people with RA who may benefit from treatments that target pain processing. Aspects of procedure, equipment, and cut-off PPT values may each affect the validity of PPT in defining patient subgroups.
Pain remains a major problem for people with RA, and we have shown that pain remains problematic despite inflammatory disease control with traditional and biologic DMARDs [
11]. Our study confirms interrelations that have been inferred from previous studies between augmented pain, mental health, fibromyalgic features, and measures of RA disease activity [
4,
5,
13,
37]. Augmented pain processing may be associated with poor pain prognosis in people treated with traditional antiinflammatory strategies, especially in people with high DAS28-P [
11]. Further escalation of traditional or biologic DMARD is unlikely to control the pain and distress associated with augmented central pain processing, but may result in important adverse events. Other pharmacologic, psychological, and physiotherapeutic approaches may be helpful for people with augmented pain processing associated with other forms of chronic musculoskeletal pain, such as FM or low back pain [
38-
40], and should continue to be investigated in people with RA [
41].
There is potential for DAS28-P, individual DAS28 components, or PPTs to offer stratification tools for selecting patients for specific treatments, but their potential should first be examined in larger, generalizable populations and validated in prospective studies. At present, clinicians should consider both inflammatory disease activity and augmented central pain processing as potential mediators of pain, and plan treatment based on holistic assessment in partnership with patients with RA.
Competing interests
NJ, IP, DW, and MW declare that they have no competing interests. DFM reports grants from Pfizer UK, outside the submitted work; DAW reports grants from Pfizer Ltd, personal fees from Pfizer Ltd, outside the submitted work.
Authors’ contributions
NJ, MW, and DW carried out the patient recruitment and data collection. DAW, IP, and NJ were involved in study conception. DAW, DFM, IP, NJ, MW, and DW were involved in the study design. DFM, DAW, and NJ performed data analysis. NJ, DFM, DW, MW, IP, and DAW were all involved in writing, editing, appraising, and approving the final manuscript that was submitted.