Objective To describe the prevalence of fibromyalgia (FM) in an axial spondyloarthritis (axSpA) population and to confirm that concomitant FM had a negative impact on tumour necrosis factor blockers’ (TNFb) response.
Design Prospective observational study with two visits 3 months apart.
Patients Adult patients with AxSpa initiating a TNFb.
Study groups FM was defined by the Fibromyalgia Rapid Screening Tool (FiRST) at baseline and also by a sustained positive FiRST (both visits) and by a fulfilment of the 1990 American College of Rheumatology criteria for FM.
Statistical analysis Prevalence of FM; evaluation of the impact of a concomitant FM on TNFb response (Bath Ankylosing Spondylitis Disease Activity Index (BASDAI 50) as primary endpoint), adjusted by factors known to have an impact on TNFb response.
Results Among the 508 patients included in the main analysis, 192 (37.8%) were screened at baseline as FM. Percentage of success after 12 weeks of treatment was lower in the FM group for most of the effectiveness endpoints (eg, BASDAI 50: 45.3% vs 54.1% in the FM/not FM groups according to the FiRST), except for the C reactive protein change endpoints which were not different across groups.
Conclusion This study confirms that FM coexists in patients with axSpA and that its presence seems to have a negative impact on TNFb response, which seems more related to the self-reported instruments used in its evaluation, rather than a different treatment effect of the molecule in this subgroup of patients.
- fibromyalgis/pain syndromes
Statistics from Altmetric.com
Fibromyalgia (FM) is a complex chronic condition of unknown aetiology and is considered as a pain amplification syndrome. Its hallmark symptoms are chronic widespread musculoskeletal pain and generalised tender points leading to significant physical disability and reduced quality of life.1 2 Its prevalence in the general population has been estimated around 2%–7%,1 but seems to be higher in patients with chronic rheumatic diseases such as rheumatoid arthritis (RA) and Sjögren syndrome, ranging from 7% to 30%.3–5
Patients with FM are known to report very high scores of pain, and this might lead to higher disease activity scores when these are based on patient-reported outcomes: that is, patients with RA and concomitant FM tend to present with higher Disease Activity Score 286 due to higher tender joint count and higher visual analogue scale-global score as compared with patients with RA without FM, making the interpretation of disease activity and treatment response difficult.7
Concomitant FM has also been reported in patients with axial spondyloarthritis (axSpA), with similar prevalences when compared with other chronic rheumatic diseases,5 8–10 and although only few data are available to date it has been reported that patients with SpA with concomitant FM tend to present with higher disease activity indices (ie, Bath Ankylosing Spondylitis Disease Activity Index (BASDAI)).9 11
The American College of Rheumatology (ACR) criteria (1990 ACR criteria and the ACR 2010 and modified 2010 criteria (2011))12–14 are the main classification criteria for FM. These criteria were mostly developed for research and classification purposes, and are difficult to apply in daily practice since they require some training to be implemented.15 Moreover, ACR 1990 classification criteria integrate tender points at physical exam that might reflect enthesitis, in patients with SpA. Therefore, with the aim of an easy and valid screening tool to facilitate the identification of patients with FM in clinical practice and research, the self-reported Fibromyalgia Rapid Screening Tool (FiRST) was developed, with preliminary high performances: a sensitivity of 90.5% and a specificity of 85.7% for the classification of FM, compared with a group of patients with chronic pain due to other rheumatic conditions (ie, RA, SpA, osteoarthritis).16
In this context, we decided to conduct a prospective study aiming (1) to describe the prevalence of concomitant FM in an axSpA population, and (2) to confirm that concomitant FM had a negative impact on tumour necrosis factor blockers’ (TNFb) treatment response.
Patients and methods
This is a prospective, longitudinal study (two visits: baseline visit and effectiveness visit occurring after 12 weeks of TNFb treatment) including 65 centres (64 French +1 Algerian centre) (ClinicalTrials.gov: NCT03039088).
The study included adult patients suffering from axSpA based on the opinion of the treating rheumatologist, in whom the treating rheumatologist made the decision of initiating or switching a TNFb because of axial symptoms of axSpA. Patients also had to be able to understand the questionnaire, and patients with an absolute contraindication to TNFb were excluded from this study.
No specific TNFb drug was evaluated in this study. Any TNFb could be prescribed, according to the rheumatologist’s choice.
FM was defined according to a score ≥5/6 on the FiRST questionnaire (online supplementary table 1).
Supplementary file 1
FM was also defined as a ‘sustained FiRST’ (ie, patients presenting a positive FiRST both at baseline and at the effectiveness visits) and also according to the 1990 ACR criteria (ie, patients with history of widespread pain and at least 11/18 tender points at the baseline visit).12
Data included demographics (age, gender, educational level, family situation, number of sick leave days, weight, height, smoking), data related to the classification criteria (all the items included in the Assessment in SpondyloArthritis international Society (ASAS) set of criteria for axSpA17 18), data related to concomitant FM (FM score (FiRST questionnaire), items included in the ACR 1990 criteria), data related to the investigations permitting to evaluate the degree of activity/severity of the disease (BASDAI,19 Ankylosing Spondylitis Disease Activity Score (ASDAS),20 Bath Ankylosing Spondylitis Functional Index (BASFI)),21 and data on treatments (ASAS-non-steroidal anti-inflammatory drug (NSAID) score22 (last week), conventional synthetic disease-modifying antirheumatic drugs, previous TNFb, but also analgesics, antidepressive agents and muscle relaxants).
At the efficacy visit
Data included ASAS-NSAIDs score (last 7 days), TNFb received during the study, BASDAI, ASDAS, BASFI, C reactive protein (CRP), and the decision to continue/switch/discontinue TNFb and the reason for switching or discontinuing this TNFb.
The primary effectiveness endpoint was the proportion of patients achieving a BASDAI response at the effectiveness visit (ie, after 12 weeks). A BASDAI response is defined by the reduction of at least 50% or two units compared with baseline.23
Secondary effectiveness endpoints were defined as either of the following:
An improvement between baseline and the effectiveness visit: ASAS 40,24 ASAS 20,24 ASDAS clinically important improvement and major improvement25; reduction of at least 50% in the ASAS-NSAID score26; reduction on the baseline CRP (mg/L).
A status after 12 weeks: ASDAS moderate disease activity (ASDAS MDA) and ASDAS inactive disease (ASDAS ID),25 respectively; an ASAS-NSAID score <10; and a CRP <6 mg/L at 12 weeks.
Description of the prevalence of FM in this axSpA population
Prevalence of FM in this axSpA population was determined by the percentage of patients with a FiRST questionnaire ≥5/6; the percentage of patients fulfilling the 1990 ACR criteria was calculated, and the agreement between both classification instruments was calculated by a prevalence-adjusted kappa (PABAK) statistic and its 95% CI.27 The agreement between ‘sustained FiRST’ and 1990 ACR criteria was also calculated by a PABAK (95% CI).
To evaluate the impact of the presence of FM on TNFb treatment response, the probability to present a BASDAI 50 response at the effectiveness visit was estimated by logistic regression, including in the model only FM (defined by the FiRST) as independent variable (univariable analysis). The adjusted multivariable model was a multivariable logistic regression estimating the probability to present a BASDAI 50 response at 12 weeks. The predefined variables (known factors influencing the TNFb response28–33) included in the multivariable model (along with FM) were age (below 40), gender (male), past or present X-ray sacroiliitis, past or present MRI sacroiliitis, abnormal CRP, smoking status (never), Human Leucocyte Antigen (HLA) B27 and absence of previous TNFb exposure.
The probability to present a treatment response at the effectiveness visit was also estimated for all the different secondary effectiveness endpoints, identically, by univariable and multivariable analyses, including in the model the same variables as in the main analysis.
The probability to present a treatment response at the effectiveness visit (defined by the BASDAI response but also by all other effectiveness endpoints) was also estimated, using the ACR 1990 criteria and ‘sustained FiRST’ definitions for FM.
Missing data handling and imputation
Patients with no data on BASDAI at baseline were excluded from the analysis. In case of missing data on the effectiveness endpoints at the effectiveness visit, among patients attending the effectiveness visit, non-responder imputation and baseline observation carried-forward imputation were performed for binary and continuous variables, respectively. If a patient discontinued the TNFb treatment before the effectiveness visit (regardless of whether he/she attended the efficacy visit), he/she was considered as a non-responder. For adjustment variables, missing values were considered as absent.
Figure 1 summarises the flow chart of the study. Among the 527 patients enrolled in the study, no baseline information was available for one patient, resulting in 526 patients with available data at baseline. In total 20 patients did not attend the effectiveness visit: 18 patients were lost to follow-up and 2 patients did not attend the visit due to a side effect and to a lack of efficacy, respectively.
The 506 patients attending the effectiveness visit and the 2 patients not attending the visit but for whom discontinuation of treatment was known were included in the main analysis, resulting in a total analysed sample for the effectiveness analyses of 508 patients.
Description of the prevalence of concomitant FM in this axSpA population
Among the 526 patients included at baseline, 202 patients (38.4%) were screened as FM by the FiRST questionnaire. Prevalence of FM according to the FiRST was similar in patients fulfilling each arm of the ASAS criteria for axSpA: among the 441/526 (83.8%) patients fulfilling the ASAS criteria for axSpA, 167 (37.9%) were screened as FM, and 151/395 (38.3%) and 16/46 (34.8%) in the imaging and clinical arms, respectively. Prevalence of FM in the subgroup of patients not fulfilling the ASAS criteria for axSpA was also comparable (35/85, 41.2%).
Among the 526 patients at baseline, 86 (16.3%) were classified as FM according to the 1990 ACR criteria for FM. The agreement between the two definitions was only fair, with a PABAK of 0.35 (0.27–0.43). According to this definition, prevalence of FM was slightly lower in the group of patients fulfilling the clinical arm of the ASAS criteria for axSpA: among the 441 patients fulfilling the criteria, 58 (13.2%) were screened as FM, and 53/395 (13.4%) and 5/46 (10.9%) in the imaging and clinical arms, respectively. Conversely, prevalence of FM in the subgroup of patients not fulfilling the ASAS criteria for axSpA was higher (28/85, 32.9%).
Among the 506 patients who attended the effectiveness visit, 194 had a positive FiRST screening at baseline, but only 94/506 (18.6%) were persistently classified as FM by the FiRST at the effectiveness visit. The agreement between the ‘sustained FIRST’ and the 1990 ACR criteria was moderate, with a PABAK of 0.55 (0.48–0.63). Prevalence of FM according to the FiRST was similar in patients fulfilling each arm of the ASAS criteria: among the 506 patients attending both visits, 423 (83.6%) fulfilled the ASAS criteria, 76 (18.0%) were screened as FM, and 66/377 (17.5%) and 10/46 (21.7%) in the imaging and clinical arms, respectively. Prevalence of FM in the subgroup of patients not fulfilling the ASAS criteria was also comparable (18/83, 21.7%).
Impact of the coexistence of FM on TNFb response
Patients and disease characteristics of the 508 patients included in the main analysis are presented in table 1. The mean age was 41.4 (±11.6), and 237 (46.7%) were women, with a mean disease duration of 6.1±8.5 years.
Among the 508 patients included in the main analysis, 192 (37.8%) were screened as FM by the FiRST questionnaire. Patients with FM were more frequently female (55.7% vs 41.1%), had less frequently a university degree (34.6% vs 51.6%), more frequently reported peripheral enthesitis (64.7% vs 47.8%), were less frequently HLA B27-positive (55.0% vs 70.5%), and reported higher BASDAI and ASDAS scores at baseline (6.5±1.6 vs 5.1±1.8 and 3.5±0.9 vs 3.2±0.9, respectively). Antidepressant intake was also more frequently reported in the FM group (26.8% vs 16.2%). No differences with regard to objective signs of inflammation (MRI sacroiliitis or abnormal CRP) or X-ray sacroiliitis were observed between the two groups.
Patients from the FM group presented less frequently a BASDAI response, 87/192 (45.3%) vs 171/316 (54.1%), but this difference did not reach a statistical significance neither in the univariable analysis nor in the multivariable analysis (adjusted OR (adjOR)=0.7 (95% CI 0.5 to 1.0)). Only X-ray sacroiliitis (adjOR=1.7 (95% CI 1.2 to 2.6)) and abnormal CRP (adjOR=1.7 (95% CI 1.2 to 2.5)) were retained as factors independently associated with a BASDAI 50 response (figure 2).
Almost all other effectiveness endpoints were significantly less frequent in the group of patients with FM, except for the CRP change endpoints, where there were no differences (table 2). This difference remained in the multivariable analysis, particularly when evaluating treatment response as a ‘state’ after 12 weeks of treatment (eg, ASDAS ID at the effectiveness visit), rather than an improvement on a score (table 2).
1990 ACR criteria
According to the 1990 ACR criteria, 82/508 (16.1%) patients were classified as FM. Patient and disease characteristics of patients fulfilling/not fulfilling the criteria are resumed in table 1. Identical to the main analysis, patients with FM according to the 1990 ACR criteria were more frequently female (58.5% vs 44.4%), presented more frequently with peripheral enthesitis (74.1% vs 50.35%) and were less frequently HLA B27-positive (37.1% vs 69.6%). Conversely, patients with FM according to the 1990 ACR criteria presented less frequently with X-ray and MRI sacroiliitis (42.0% vs 58.3% and 50.0% vs 70.9%, respectively). Mean BASDAI and ASDAS at baseline were also higher in the FM+ group (7.1 vs 5.4 and 3.8 and 3.3, respectively), and they also reported more frequently a history of antidepressant intake (31.7% vs 18.0%).
Sensitivity analysis using the 1990 ACR criteria to classify for FM did not find any significant difference in the percentage of responders in patients with/without FM (univariable nor multivariable), except for the ASDAS ‘states’ at the effectiveness visit, which were significantly less frequent in the group of patients with FM (24/82 (29.3%) vs 240/426 (56.3%), adjOR=0.4 (95% CI 0.2 to 0.6) and 5/82 (6.1%) vs 121/426 (28.4%), adjOR=0.2 (95% CI 0.1 to 0.5), for the ASDAS MDA and ASDAS ID, respectively) (online supplementary table 2).
Among the 506 patients who attended the effectiveness visit, a FiRST value was available at both visits for 504 patients. Among them, 94 (18.7%) patients were persistently classified as FM by the FiRST (see table 1).
According to this FM definition, and similarly to the main analysis, patients with FM were more frequently female (61.3% vs 42.1%), presented more frequently with peripheral enthesitis (68.1% vs 51.3%) and were less frequently HLA B27-positive (50.0% vs 68.3%). Mean BASDAI at baseline was also significantly higher in the FM+ group (6.9 vs 5.4) but no ASDAS (3.5 vs 3.3); they also reported more frequently a history of antidepressant intake (33.3% vs 17.4%) but also of third-ladder analgesics (26.7% vs 15.3%). Conversely, patients with FM according to this definition presented less frequently with X-ray sacroiliitis and abnormal CRP (39.8% vs 59.2% and 52.8% vs 64.0%, respectively).
Sensitivity analysis using the sustained FiRST to classify for FM found a lower percentage of responders in the FM group, regardless the effectiveness endpoint used, except for the change on CRP, which was similar in both groups (72/93 (77.4%) vs 316/411 (76.9%), adjOR=0.9 (95% CI 0.5 to 1.6)) (online supplementary table 1).
All adjORs for all effectiveness endpoints for all FM definitions are resumed in figure 3.
This study confirms that FM can coexist concomitantly and not concomitantly with axSpA, with a prevalence ranging from 16.1% to 37.8% depending on the definition used (1990 ACR criteria and FiRST, respectively); this confirms that concomitant FM is more frequent in patients with axSpA than in the general population1 2 but not more frequent than in other chronic rheumatic diseases.3–5 Our results also suggest that concomitant FM has an impact on TNFb treatment effect but only when this effect was evaluated by patient-reported outcomes, and this impact was even more important when treatment response is defined by a status to be reached after 12 weeks rather than a change. Indeed, regardless of the definition used for FM, change in CRP was never different with regard to the presence/absence of concomitant FM in this axSpA population.
Our findings also confirm that patients with axSpA with concomitant FM are more frequently female, with peripheral enthesitis history and were more frequently taking antidepressants.
The decreased frequency of HLA B27-positive, radiographic sacroiliitis and MRI inflammation at the sacroiliac joints (SIJ) in the FM+ is intriguing and might suggest that some patients participating in this trial might have been misdiagnosed and were in fact suffering from FM only. However, other clinical features suggestive of axSpA (eg, uveitis, psoriasis, inflammatory bowel disease) were equally distributed across the different subgroups. One potential explanation is the fact that this study was focused on patients with a high level of complaints that prompted their treating rheumatologists to initiate a TNFb treatment. These patients (ie, with high level of complaints) could have suffered from severe axSpA without FM or mild axSpA with concomitant FM. Since HLA B27 positivity, X-ray sacroiliitis and MRI sacroiliitis have been reported to be related with the severity of the disease, the observed results in our study are not surprising.
This study has some weaknesses but also some strengths. First, prevalence of FM in our study was higher than what has been reported in the literature.8 9 However, this was only true when using the FiRST questionnaire to define FM; when using either the 1990 ACR criteria or the ‘sustained FiRST’ definition, prevalence ranged between 17% and 19%, comparable to previous data. It is worth noticing that the FiRST questionnaire was developed as an easy-to-use screening tool for FM, that is, aiming for higher sensitivity rather than specificity (ie, in order to not miss any ‘true’ FM despite classifying some ‘false’ FM).34 Also, this study is a particular setting, that is, patients with axSpA eligible to initiate a TNFb treatment in clinical practice, with high levels of pain: it is possible that FiRST applied only at baseline captured some patients with ‘false’ FM that scored very high on the questions of the FiRST referring to pain and fatigue related to their axSpA symptoms, which faded after initiation of TNFb treatment, as reflected by the decrease in the number of patients consistently classified by the FiRST at 12 weeks. Indeed, agreement between 1990 ACR (which may be considered as the ‘gold standard’ for FM classification) and sustained FiRST was much better than with FiRST only at baseline, suggesting that this very easy questionnaire might better screen FM when applied twice (before treatment and at the visit aiming to evaluate treatment effect).
Second, there might be some concern with regard to the representativity of the sample of patients included in this study, that is, physicians being aware of the aim of the study would more likely include (only) patients with coexisting FM. However, neither the prevalence rates of FM (particularly according to the 1990 ACR criteria) nor the patients’ characteristics seem to support this. Furthermore, patients with/without FM according to the FiRST questionnaire were not different regarding the phenotype of the disease (ie, neither MRI nor X-ray sacroiliitis) and were equally fulfilling the different arms of the ASAS criteria. On the other hand, patients with FM were more likely female, presenting with enthesitic involvement pain, reporting higher BASDAI and ASDAS scores and with history of antidepressant intake. All these characteristics are consistent with previously reported data from patients with axSpA and concomitant FM.9 35
Third, we did not manage to confirm our main objective, that is, to confirm that coexisting FM in axSpA had a significant impact on TNFb treatment response assessed by the BASDAI response. However, this significant impact was observed for almost all the other effectiveness endpoints observed, and particularly when using a ‘state after 12 weeks’ endpoint. These results suggest that there is indeed an impact on the treatment response, but seems more related to the patient-reported outcomes used in the effectiveness endpoints, as suggested by the absence of difference across groups for the objective biological parameters (ie, CRP). Indeed, patients with concomitant FM respond to TNFb and improve their disease activity scores (ie, they felt ‘better’), but it seems difficult to achieve a remission or low disease activity state (ie, they did not feel ‘good’),36 since these states require very low patient-reported scores, which are unlikely to be achieved in patients with FM. This seems particularly important for the decision of the treatment target in patients with axSpA when applying a treat-to-target strategy37: for example, remission might not be a feasible target for these patients who will not likely reach ever this state, but should rather aim for a significant change or focus only on objective parameters (ie, CRP).
Finally, in this study we have applied the FiRST in an innovative way, that is, repeating the questionnaire at the effectiveness visit, and considering as FM only patients who were persistently classified as FM by the tool. Indeed, in a previous study, reliability of FiRST was reported to be good, but this was tested in a stable situation, where no medication change was performed.9 Here, FiRST was first scored in a situation where patients were very likely in high levels of pain (ie, since needing a TNFb treatment), and thereafter after 12 weeks of treatment. When tested after treatment, only half of patients presented a persistent positive FiRST. Interestingly, when using this definition, TNFb response rates were significantly lower in patients with FM, regardless of the endpoint used (but also more important for ‘status’ endpoints) except for CRP, suggesting again the important weight of patient reported outcomes in the evaluation of treatment response.
This study confirms that FM can coexist in patients with axSpA and that its presence has a negative impact on TNFb response, but that this impact seems related to the instruments used in its evaluation, rather than a different treatment effect of the molecule. Furthermore, this study suggests also that the presence of FM is not related to a specific phenotype of a patient with axSpA (ie, not related with the absence of imaging abnormalities).
These findings highlight the importance of evaluating the presence of coexisting FM in patients with axSpA when evaluating treatment response, but also when determining treatment targets. Further studies should be conducted to confirm these results and to evaluate the relevance of the implementation of the FiRST questionnaire in daily practice.
Handling editor Josef S Smolen
Contributors AMo, LG, SP, MD: conception, design, analysis and interpretation of data. AMo: drafting the article. All authors critically revised the manuscript for important intellectual content. All authors read and approved the final manuscript.
Funding This study was conducted thanks to an unrestricted grant from MSD.
Competing interests None declared.
Ethics approval Ethics approval was obtained according to local regulations and all patients gave their informed consent.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement The data sets generated and/or analysed during the current study are not publicly available due to consent restrictions. Programming codes used for statistical analysis during the current study are available from the corresponding author upon reasonable request.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.