To assess the efficacy and safety of non-biological therapies in patients with axial spondyloarthritis (axSpA) to inform the update of the Assessment of SpondyloArthritis international Society (ASAS)/European League Against Rheumatism (EULAR) recommendations for the management of axSpA. A systematic literature review (2009–2016) of all non-pharmacological treatments, non-biological drugs (except targeted synthetic disease-modifying antirheumatic drugs (DMARDs)) and surgical therapies was performed. Randomised controlled trials (RCTs) and clinical controlled trials were assessed for efficacy and safety, while observational studies with a comparator were assessed for safety. All relevant efficacy and safety outcomes were included. Study heterogeneity precluded data pooling. If possible, Cohen's effect size was calculated for non-pharmacological treatments. In total, 45 papers and 2 abstracts were included. Studies on non-pharmacological treatments were very heterogeneous but overall confirmed a benefit for regular exercises, with small improvements in disease activity, function and spinal mobility. New studies on non-steroidal anti-inflammatory drugs (NSAIDs) confirmed their efficacy and new safety signals were not found. NSAIDs used continuously compared with on-demand did not reduce the modified Stoke Ankylosing Spondylitis Spinal Score (mSASSS) mean change over 2 years in patients with ankylosing spondylitis with normal C reactive protein (CRP; ≤5 mg/L) (1 ‘negative’ RCT (0.9 vs 0.8; p=0.62)), while for patients with high CRP, conflicting results were found (1 ‘positive’ RCT (0.2 vs 1.7; p=0.003), 1 ‘negative’ RCT (1.68 vs 0.96; p=0.28)). No new trials were found for conventional synthetic DMARDs (csDMARDs). Short-term high-dose systemic glucocorticoids showed limited efficacy. Regular exercises may improve several outcomes. Efficacy and safety of NSAIDs in axSpA are confirmed. Glucocorticoids are not proven to be effective in axSpA and new data on csDMARDs are lacking.
- Ankylosing Spondylitis
This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Statistics from Altmetric.com
Regular exercises may improve several outcomes.
Efficacy and safety of NSAIDs in axSpA are confirmed.
Glucocorticoids are not proven to be effective in axSpA.
No new data on csDMARDs in axSpA was found.
Treatment of axial spondyloarthritis (axSpA) can be a challenge due to a limited number of therapeutic alternatives.1 In the past decade, a plethora of non-pharmacological and pharmacological therapies have been applied, aiming to improve the patient’s quality of life, to reduce pain and physical impairment and to avoid work disability. Treatment with tumour necrosis factor α inhibitors (TNFi) is especially efficacious but because of drug cost treatment has been reserved for patients failing the so-called conventional compounds such as non-steroidal anti-inflammatory drugs (NSAIDs).2 Overall, a multidisciplinary approach with a combination of non-pharmacological and pharmacological treatment and, if needed, a surgical intervention comprises the full spectrum of the treatment of axSpA.2
A collaboration between the Assessment of SpondyloArthritis international Society (ASAS) and the European League Against Rheumatism (EULAR) has led to the first publication of the ASAS/EULAR recommendations for the management of ankylosing spondylitis (AS) in 2006,1 while an update had been published in 2010,2 based on evidence from systematic literature reviews (SLRs).3 ,4 In these recommendations, treatment was constrained to patients in later stages of axSpA (radiographic axSpA—r-axSpA—or AS). Another ASAS initiative issued recommendations for the use of TNFi in patients with axSpA, also taking the earlier, non-radiographic stages (nr-axSpA) into account.5 Still, no recommendations had yet covered the whole management spectrum (including non-pharmacological and pharmacological management) and the full spectrum of axSpA (including both nr-axSpA and r-axSpA). During the past years, accumulating evidence has shown that the disease is one continuum, including nr-axSpA and r-axSpA.6 This, together with the progress witnessed in the area of management of axSpA in the past years, justified an update of the recommendations for the management of axSpA.
The objective of the current SLR was to update the evidence on efficacy and safety of non-biological interventions (non-pharmacological treatment, non-biological drugs and surgical therapies). This SLR was performed together with another on biological and targeted synthetic disease-modifying antirheumatic drugs (tsDMARDs).7 ,8 Both SLRs aimed to inform the task force responsible for the update of the ASAS/EULAR recommendations for the management of axSpA.9
Search methodology and study selection
The systematic literature search was performed by using references from MEDLINE, EMBASE and Cochrane CENTRAL databases and as an update of the previous SLR conducted in 2009.4 The articles included in the present SLR had to be published between 1 January 2009 and 26 February 2016. In addition, abstracts from the annual conferences of EULAR and the American College of Rheumatology (ACR) 2014 and 2015 were included. The search strategy is presented in online supplementary text 1. Eligible study types for efficacy and safety assessment were randomised controlled trials (RCTs), clinical controlled trials (CCTs) and open-label long-term extension studies. Cohort studies or registries were considered for safety assessment but only if a comparator treatment was available, or if population-based incidence rates were reported and at least 50 participants per group were included. For surgical interventions, cohort studies with a comparator group, as well as case–control studies, were used to assess both efficacy and safety. SLRs were only considered appropriate to identify references from original studies, except for Cochrane reviews, which were included anyway. Research questions were reformulated according to the PICO (Participants, Interventions, Comparisons and Outcomes) method.10 Studies were selected with adult patients (age ≥18 years) and a diagnosis of axSpA. The interventions in the current SLR were defined as (1) non-pharmacological interventions (physiotherapy, exercise, balneotherapy, spa therapy, diet, education, self-education groups), (2) non-biological drugs, such as NSAIDs, local and systemic glucocorticoids, conventional synthetic disease-modifying antirheumatic drugs (csDMARDs) (methotrexate, leflunomide, sulfasalazine, hydroxychloroquine, azathioprine, cyclosporine, cyclophosphamide, auranofin, penicillamine or thalidomide), bisphosphonates, analgesics, opioids, opioid-like drugs, neuromodulators (antidepressants, anticonvulsants and muscle relaxants) and probiotics, and (3) surgical therapies. All doses, formulations, regimens (eg, on-demand, continuous) and treatment durations were assessed. Treatment comparators were defined as any non-pharmacological or surgical intervention, same non-biological interventions in different doses or regimens, other non-biological drugs, any combination therapy, placebo or none.
Outcomes considered for the assessment of treatment efficacy were the Bath AS Disease Activity Index (BASDAI11), Bath AS Functional Index (BASFI12), Bath AS Metrology Index (BASMI13), AS Disease Activity Score (ASDAS14 ,15) and ASDAS disease activity status,16 ASAS partial remission,17 patient’s global assessment of disease activity, pain levels, assessments of enthesitis, swollen and tender joint count. Outcomes considered for patient’s response to treatment were the ASAS response criteria17 (ASAS20, ASAS40 and ASAS5/6),18 ASDAS clinically important improvement (Δ≥1.1) and ASDAS major improvement (Δ≥2.0)16 and BASDAI response (improvement of ≥50% and/or ≥2 units). The AS Quality of Life (ASQoL19) index was considered to evaluate the Quality of Life. Additionally, work disability, work productivity, cost-efficacy and cost-effectiveness were assessed. Radiographic progression of the spine was assessed by the modified Stoke Ankylosing Spondylitis Spinal Score (mSASSS20). Inflammation on magnetic resonance imaging (MRI) was measured by the ASAS/Outcome Measures in Rheumatology (OMERACT) definition21 and the Spondyloarthritis Research Consortium of Canada (SPARCC) score (sacroiliac joints22 and spine23).
For safety outcomes, information was collected on withdrawals due to adverse events (AEs), serious AEs, infections, malignancies, cardiovascular disease, infusion/injection-site reactions, renal, gastrointestinal (GI) and hepatic effects, haematological abnormalities and demyelinating disease.
Data extraction and assessment of risk of bias (RoB)
Each article or abstract identified was assessed independently by two reviewers (AR and AS) for suitability according to the predefined inclusion criteria, followed by a full-text review. For every included study, relevant data were extracted. Additionally, the two reviewers evaluated the RoB of each study according to the ‘Cochrane tool’ for RCTs,24 the ‘Hayden-tool’ for cohort studies25 and the Newcastle-Ottawa Scale for case–control studies.26 Disagreements regarding the eligibility of the studies, data extraction and RoB assessment were resolved by discussion and consensus. In case of persistent disagreement, a third reviewer (SR) was involved.
Owing to the large heterogeneity of the studies, data could not be pooled and results are presented descriptively. As in the previous SLR,4 if possible, Cohen’s effect size (ES) (mean change in score divided by the baseline standard deviation (SD)) was calculated for non-pharmacological interventions, with Cohen’s ES<0 meaning worsening, 0–0.49 a small positive effect (ie, improvement), 0.5–0.79 a moderate effect and ≥0.8 a large effect. Additionally, if possible, the number needed to treat (NNT, number of patients who must be treated in order to obtain the benefit of interest in one additional patient) was presented.
Overall, the search yielded 11 649 articles (after de-duplication), of which 45 full-text articles and 2 abstracts were included in this SLR (flow chart in online supplementary figure S1 and online supplementary tables S1–S4; the articles on biological DMARDs and tsDMARDs are included in a separate SLR8). In total, 29 trials investigated benefits and harms of non-pharmacological therapies (28 papers;27–54 1 abstract55), 15 publications focused on non-biological drugs (13 papers;56–68 1 abstract;69 1 Cochrane review70), and 3 articles71–73 assessed the efficacy of surgical interventions.
No studies were found on csDMARDs, neuromodulators, diet or self-education groups.
Twenty-nine trials were identified assessing different non-pharmacological interventions in patients with axSpA (for details, see online supplementary tables S5–S9 (Exercises), S10–S14 (Education), and S15–S19 (Other non-pharmacological interventions)).27–55
Overall, the studies were heterogeneous (figure 1), differing mainly in the type and duration of intervention, group size and outcome parameters. The group size was often small: only four studies44–46 ,50 included more than 90 patients. One study51 enrolled patients with active axSpA defined according to the ASAS classification criteria and with a BASDAI≥3.5.6 All the remaining studies focused on patients with established r-axSpA according to the modified New York (mNY) criteria.
Nine studies28 ,29 ,34 ,42 ,44 ,45 ,49 ,51 ,53 had a low or unclear RoB and we have therefore focused on these, that is, we excluded the high RoB studies from in-depth analysis. An overview of these low/unclear RoB studies, together with Cohen's ES for BASDAI, BASFI, pain global and ASDAS, can be found in table 1. In summary, regular exercises can improve disease activity, pain, function and spinal mobility. However, the effects were usually small. Endurance combined with strength training, compared with ‘no exercises’, provided the largest effect on disease activity, both measured with the ASDAS (mean 2.3 vs 2.7 at start and 1.8 vs 2.6 at the end of observation, respectively) and BASDAI (mean 5.3/10 vs 5.3/10 units at start and 3.3 vs 5.2 at the end of observation, respectively).51 Laser therapy compared with placebo resulted in the largest effect on function as measured by the BASFI (mean 51.5/100 vs 48.6/100 at start and 37.4/100 vs 50.6/100 at the end of observation, respectively) and pain (mean 70.0/100 vs 67.5/100 at start and 33.1/100 vs 65.6/100 at the end of observation, respectively) (Cohen's ES of 0.84 for BASFI and 2.48 for pain, both for Laser therapy, respectively).49 Aquatic exercises compared with land-based exercises led to the best improvements in pain (mean 5.1/10 vs 4.9/10 at start and 2.6/10 vs 3.3/10 at the end of observation, respectively) (Cohen's ES of 0.96 for aquatic exercises), also with moderate improvements in BASDAI (at start mean 3.9/10 vs 4.0/10 and 2.6/10 vs 2.8/10 at the end of observation, respectively; Cohen’s ES 0.68).34
Five studies31 ,32 ,43 ,48 ,54 focused on a combination therapy of exercises and TNFi compared with treatment of TNFi only. However, none of these studies showed any additional effect on the function and spinal mobility of patients with axSpA by exercises added to TNFi therapy.
The main characteristics and efficacy data of the included studies on non-biological drugs are presented in tables 2 and 3; safety data from observational studies are shown in table 4. Additional data as well as the RoB assessment are presented in online supplementary tables S20–S24 for efficacy and S25–S27 for safety.
Non-steroidal anti-inflammatory drugs
A Cochrane review,70 comparing NSAIDs (traditional and cyclooxygenase (COX)-II inhibitors) to placebo as well as between them, included 39 studies (35 RCTs, 2 quasi-RCTs, 2 cohorts) up to June 2014. From the studies included in the Cochrane review, only two studies were published in or after 2009 (Poddubnyy et al,74 Kroon et al59), thus overlapping with the current SLR. Both focused on the effect of NSAIDs on radiographic progression. This Cochrane review showed that after 6 weeks of treatment, traditional NSAIDs and COX-II inhibitors were more efficacious than placebo (pain visual analogue scale (VAS) 0–100: −16.5 (95% confidence interval (CI) −20.8 to −12.2) with traditional NSAIDs (mean 44/100) versus placebo (mean 60.5), NNT 4 (range 3–6); −21.7 (95% CI −35.9 to −7.4), with COX-II inhibitors (mean 42.3) versus placebo (mean 64), NNT 3 (range 2–24)). Moreover, no measurable differences were seen between the different NSAIDs. No significant increase in AEs at 12 weeks were reported for NSAIDs.
In addition, five RCTs addressing NSAIDs were included in this SLR (tables 2 and 3), two of them focusing on the effect of NSAIDs on radiographic progression.58 ,59 Two studies comparing two NSAIDs (table 2, see online supplementary tables S20–S24) were included in the current SLR. Both studies56 ,69 confirmed the results of the aforementioned Cochrane review.70 The first study,69 at unclear RoB, showed that two different doses of etoricoxib (ETX) were as effective as naproxen (NPX) in improving the spinal pain intensity (SPI) score on a VAS (0–100) in patients with r-axSpA (SPI least square mean change from baseline at week 6: −29.0 for ETX 60 mg, −31.2 for ETX 90 mg, −30.6 for NPX 1000 mg). The second study,56 also at unclear RoB, demonstrated non-inferiority of celecoxib compared with diclofenac in decreasing the patient’s global assessment of pain intensity on a 0–100 scale (mean change at week 6: −23.7 celecoxib 200 mg, −26.7 diclofenac 75 mg).
The other two RCTs58 ,59 focused on radiographic progression. Sieper et al58 2016, at low RoB, evaluated the effect of diclofenac on spinal radiographic progression in patients with r-axSpA when taken continuously versus on-demand (table 3). No significant differences in the mSASSS mean change over 2 years were found, either in the whole group (1.28 vs 0.79; p=0.39, respectively) or in the subgroup with elevated C reactive protein (CRP) at baseline (1.68 vs 0.96; p=0.28). In contrast, Kroon et al59 (at low RoB) found a significant difference between continuous use of celecoxib versus on-demand in patients with r-axSpA with elevated CRP at baseline (0.2 vs 1.6; p=0.003; favouring continuous use). Study characteristics on both studies are provided in online supplementary table S20.
Two observational studies60 ,61 were identified assessing the safety of NSAIDs in axSpA (table 4 and online supplementary tables S25–S27). Only one study, Kristensen et al,60 at moderate RoB, focused on GI AEs. No differences in their incidence were found when comparing COX-II inhibitors with traditional NSAIDs. However, a significantly reduced risk of GI-AEs was identified in patients not using NSAIDs compared with patients on traditional NSAIDs.
Essers et al,61 at moderate RoB, reported a larger risk of ischaemic heart disease in patients with r-axSpA using NSAIDs (adjusted hazard ratio (aHR) (95% CI) 1.36 (1.00 to 1.85)) or COX-II inhibitors (aHR (95% CI) 3.03 (1.61 to 5.69)) compared with the general population. Kristensen et al60 also looked at atherosclerotic events and found no significant differences between traditional NSAIDs and COX-II inhibitors (at moderate RoB).
One RCT62 (at low RoB), Haibel et al (table 2) performed in patients with r-axSpA with active disease (BASDAI≥4) has shown no short-term differences between two different doses of prednisolone and placebo in the primary end point (BASDAI 50 week 2: 8% under placebo; 27% under prednisolone 20 mg, p value versus placebo=0.30; 33% under prednisolone 50 mg, p value versus placebo=0.16). However, there were significant effects observed on ASDAS-CRP (week 2 change scores: −0.34 for placebo; −1.16 for prednisolone 20 mg, p value versus placebo=0.004; −1.56 for prednisolone 50 mg, p value versus placebo=0.010) and CRP (week 2 change scores: −3.19 for placebo; −19.94 for prednisolone 20 mg, p value versus placebo=0.0016; −15.58 for prednisolone 50 mg, p value versus placebo=0.036). The number of AEs at 2 weeks was similar in the three-arm study (n=6 placebo; n=4 under prednisolone 20 mg; n=5 under prednisolone 50 mg).
One cohort study63 (at high RoB) assessed the safety of low-dose glucocorticoids and NSAIDS compared with NSAIDs alone in patients with r-axSpA (table 4). No significant differences were reported for serious infections and peptic ulcer disease. On the other hand, a higher incidence of dermatological AEs was found in patients receiving glucocorticoids (incidence rate/1000 patient-years 22.2 vs 6.6; p=0.003).
Other non-biological drugs
Five trials (four studies64 ,65 ,67 ,68 at high RoB and one study66 at low RoB) were identified assessing the efficacy and safety of other non-biological drugs such as probiotics and pamidronate (table 2 and online supplementary tables S20–S24). In summary, none of the studies provided convincing evidence that these therapeutic alternatives are effective.
Overall, three studies71–73 (all at high RoB) focusing on surgical interventions in patients with advanced r-axSpA were found. These studies suggested benefits for pedicle subtraction osteotomy and total hip replacement in patients with a fixed kyphotic deformity or advanced hip joint deformity, respectively (see online supplementary tables S28–S31).
This SLR summarises the current state of evidence for non-pharmacological treatments, non-biological drugs and surgical interventions in the treatment of axSpA, published after 2009.
The evidence favouring the efficacy of non-pharmacological interventions such as exercises, education and physiotherapy confirmed previous findings.4 Almost all studies that were analysed demonstrated that regular exercises may improve disease activity, function, spinal mobility and pain in patients with axSpA. However, since the trials were so heterogeneous, no data pooling could be performed. The absence of a meta-analysis makes it difficult to decide which type of exercise is preferable, also because improvements were often small, regardless of the type of intervention. Only one study51 focused on axSpA according to the ASAS criteria,6 including early and advanced stages of disease, and required a high disease activity (BASDAI≥3.5) for inclusion. All remaining trials included patients with advanced r-axSpA. Furthermore, data on the safety of the exercises were indecisive, since the information about possible AEs, such as vertebral fractures, was not available but may be relevant, particularly in advanced stages of the disease.
Another point of concern is the quality of the studies on non-pharmacological interventions. Although several studies were performed during the past years, the overall RoB was most often ‘high’. One main reason for this is the lack of blinding of the outcome assessors, which admittedly is challenging to achieve for some interventions such as physiotherapy or exercises. Still, designing blinded studies for such interventions is not without precedent: previous trials, for example, with a sham intervention as a comparator, have shown that interventions broadly considered to be effective may fail to demonstrate superiority when tested against a formal comparator, and when blinding is ensured.75
As already shown in the previous SLRs informing ASAS-EULAR recommendations,4 ,76 NSAIDs are effective for the treatment of axSpA with no difference in efficacy between different classes. Compared with the last SLR, new evidence regarding the effect of NSAIDs on radiographic progression has been published but the results are not consistent. Until now, there is no evidence that NSAIDs reduce spinal radiographic progression in patients with r-axSpA with normal CRP, while contradicting evidence towards less radiographic progression is available for patients with elevated CRP and continuous NSAIDs intake.58 ,59 In addition, one cohort study showed an inhibitory effect of continuous NSAID use on radiographic progression in patients with r-axSpA and elevated CRP.74 Taken all together, the potential inhibitory effect of NSAIDs on spinal radiographic progression is still an open question and warrants more research to draw definite conclusions.
In comparison to the previous SLR,4 no new findings on the safety of NSAIDs were obtained. Overall, only four studies with moderate quality could be analysed on this topic, all together confirming the previous data at least for patients with established r-axSpA, while eligible observational studies focusing on safety in patients with nr-axSpA were not available.
In contrast to the previous SLRs, a low RoB RCT62 has addressed the short-term efficacy of high doses of systemic glucocorticoids. This study failed to formally demonstrate superior efficacy of glucocorticoids for patients with r-axSpA with active disease (BASDAI≥4), since it did not meet its primary end point (BASDAI 50). However, significant differences in secondary outcomes were found, such as in ASDAS-CRP and CRP levels for prednisolone 20 mg and 50 mg and in the BASDAI levels for prednisolone 50 mg as compared with placebo. This proof-of-concept 2-week trial with high doses of glucocorticoids in axSpA has shown only very modest efficacy.
Finally, similar to the studies identified in the last SLR,4 the level of evidence for surgical interventions remained low. Only three small studies testing surgical interventions to a comparator were found. The remaining captured (but not included) studies were case series or cohort studies without a comparator, thus hampering the assessment of possible treatment effects. The limited data suggest that patients with advanced r-axSpA may benefit from spinal corrective osteotomy or total hip arthroplasty when indicated.
In summary, in the latest SLR on non-biological treatment in axSpA, the evidence on efficacy and safety of NSAIDs was confirmed, while no new data were found on treatment with csDMARDs. Thus far, oral glucocorticoids did not demonstrate efficacy in axSpA. Regular exercises may improve outcomes, but with modest effect sizes. This SLR has informed the 2016 update of the ASAS-EULAR recommendations for the management of axSpA.
Correction notice This article has been corrected since it was first published online.
Funding The study was funded by the European League Against Rheumatism (EULAR), and the Assessment of SpondyloArthritis international Society (ASAS).
Competing interests AS: Fundação para a Ciência e Tecnologia (grant number: SFRH/BD/108246/2015). XB: AbbVie, BMS, Boehringer Ingelheim, Celgene, Centocor, Chugai, Janssen Biologics, Novartis, Pfizer, UCB. DvdH: AbbVie, Amgen, Astellas, AstraZeneca, Bristol Myers Squibb, Boeringer Ingelheim, Celgene, Daiichi, Eli-Lilly, Galapagos, Janssen, Merk, Novartis, Pfizer, Roche, Sanofi-Aventis, UCB, Imaging Rheumatology BV. JB: Abbott, Bristol Myers Squibb, Celgene, Celltrion, Chugai, Johnson & Johnson, MSD, Novartis, Pfizer, Roche, UCB Pharma. RL: Abbott/AbbVie, Ablynx, Amgen, AstraZeneca, BMS, Centocor, Janssen (formerly Centocor), GSK, Merck, Novo-Nordisk, Novartis, Pfizer, Roche, Schering-Plough, TiGenics UCB, Wyeth, Director of Rheumatology Consultancy BV. FdVB: AbbVie, Celgene, Janssen, Merck, Novartis, Pizer and UCB.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement No additional data are available.
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.