Objectives To determine whether non-steroidal anti-inflammatory drugs (NSAIDs) have a significant effect on ultrasonographic (US) grey scale (GS) and power Doppler (PD) assessment of synovitis in rheumatoid arthritis (RA).
Methods Patients with RA taking NSAIDs were randomised to either stopping (for a minimum of 5 drug half-lives) or continuing the drug. All patients had a clinical assessment and US examination of both hands and wrists before and after stopping/continuing the NSAID. Changes at follow-up were compared between groups using Mann–Whitney U tests.
Results A total of 58 patients with RA were recruited. All the clinical assessment parameters (including disease activity, pain, general state of health and physician global visual analogue score and tender and swollen joints count) showed an increase in the group who stopped their NSAID treatment. The total GS and PD score showed median (first to third quartiles) increase of 9.5 (5.75 to 19.0) and 4.0 (2.0 to 6.0) per patient, respectively, in the patients who stopped their NSAID in comparison with 1.0 (–1.0 to 2.25) and 0.0 (–2.0 to 3.0), respectively, in the patients who continued their NSAID (p<0.001). There was an increase in the number of joints scoring >0 for GS and PD in the patients who stopped the NSAID. The inter- and intrareader agreement was good to excellent for the US examination.
Conclusion NSAID usage may mask the GS and PD signal and result in lower scoring despite continuing disease activity. Consideration should be given to the NSAID effect in designing clinical studies which use US to assess response to therapeutic.
Statistics from Altmetric.com
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.
Non-steroidal anti-inflammatory drugs (NSAIDs) are widely used to improve pain and stiffness in patients with rheumatoid arthritis (RA). Although useful for offering symptomatic relief in RA and reducing synovial inflammation, they do not appear to alter the course of the disease or prevent joint destruction. In inflamed joints, NSAIDs inhibit cyclo-oxygenase activity and consequently decrease prostaglandin (PG) production and its subsequent effects.1
Musculoskeletal ultrasound (US) offers a sensitive means of detecting and quantifying synovial inflammation, using grey scale (GS) and/or power Doppler (PD) changes.2 It has been shown to be a valid tool for assessing and monitoring response to therapeutics. Changes in the GS and PD scores in response to biological agents in RA have been reported by many trials.3,–,6
At first presentation to a rheumatologist, many patients referred from primary care are already taking prescription or over-the-counter NSAIDs. They are also used by many patients to control symptoms during times of flare while participating in clinical trials. The confounding effect of concurrent intra-articular and systemic steroids on synovitis assessment has been previously reported.7,–,10 Subsequently, the steroid effect is usually considered in the recruitment and/or analysis process of clinical trials. However, the effect NSAIDs on the GS and PD signal has not been reported.
The aim of this study was to investigate if NSAIDs affect the detection and grading of GS and PD synovitis in patients with RA.
Patients and methods
Patients fulfilling the American College of Rheumatology 1987 diagnostic criteria for RA11 were recruited. All patients had to be taking an NSAID as prescribed by their treating rheumatologist. Randomisation was stratified such that half of the patients had early RA (<3 years of diagnosis) and half had established RA (≥3 years of diagnosis), distributed equally between the randomised groups. All patients had to have been receiving stable doses of NSAID for at least 2 weeks and stable doses of disease-modifying antirheumatic drugs (DMARD) for at least 6 weeks before participating in the study. Patients did not have intramuscular, oral or intravenous steroids within 6 weeks of the first study visit or during the study. All patients were recruited from the outpatient rheumatology clinics at Chapel Allerton Hospital, Leeds from June 2009 to September 2010. They were randomised to either stopping or continuing NSAIDs for a minimum of five drug half-lives (eg, half-life of diclofenac sodium is 1–2 h and for celecoxib 11–16 h).12 13 Patients underwent clinical and US assessment at baseline and at follow-up. For the short half-life NSAIDs the drug was stopped for a period of 1–3 days before the follow-up assessment, while for longer half-life NSAIDs (eg, etoricoxib) a longer duration was required. The study was approved by the local ethics committee and informed written consent was obtained from all patients before study enrolment.
All patients were assessed clinically at baseline and at follow-up. The clinical assessments were performed by rheumatologists (MS and CW) who were blinded to the US assessment. The clinical assessment included 28 tender and swollen joint counts and physician global Visual Analogue Score (VAS). The patients were asked to complete a disease activity VAS, pain VAS and general state of health VAS.
Ultrasonography (GS and PD) was performed using an HDI Philips 5000 scanner (Phillips, Eindhoven, The Netherlands) with a multilinear 15-7 MHz hockey stick probe. An investigator (ASZ) trained in musculoskeletal US performed all examinations and was blinded to the patient's randomisation status. PD was assessed with the highest gain level without background noise, pulse repetition frequency of 750 Hz and medium wall filter.
Both wrists and hands were assessed in all patients. Eight joints in the hands and wrists were examined, including radiocarpal, intercarpal, ulnar-carpal joints and first to fifth metacarpophalangeal joints. Tendon sheaths were also assessed, including the extensor carpi ulnaris and second to fifth flexor digitorium tendon sheaths for the presence of tenosynovitis. The GS and PD were scored using the OMERACT definitions and proposed semiquantitative 0–3 scale.14,–,17 GS scoring was as follows: 0 = no synovial hypertrophy, 1 = mild hypertrophy, 2 = moderate hypertrophy and 3 = severe hypertrophy. The PD scoring was as follows: 0 = absence of signal, no intra-articular flow; 1 = mild hyperaemia, one or two vessels signal (including one confluent vessel); 2 = moderate hyperaemia, (>grade 1) and <50% of the GS area; 3 = marked hyperaemia, vessels signal in more than half of the synovial area. Tenosynovitis was defined according to the OMERACT criteria16 and the GS and PD signal scored using a locally devised semiquantitative 0–3 scale system (0 = normal, 1 = mild, 2 = moderate and 3 = severe). Currently, there is no validated scoring system for US-detected tenosynovitis.
To test the inter-reader reliability of US, 10 hands and wrists of 10 patients were rescanned immediately at the same visit by a second rheumatologist (JEF) experienced in US. The second reader was blinded to the readings of the first reader and clinical findings. Ten hands were also rescanned by the first reader (ASZ) within 48 h to test intrareader reliability. Patients were selected according to their willingness to have a re-scan.
Statistical analysis was performed using the SPSS statistical package, version 17.0 (SPSS, Chicago, Illinois, USA). To facilitate the analysis of different joint areas the scores were summed within the following areas: hand joints (five metacarpophalangeal joints, radiocarpal, intercarpal and ulnar-carpal joints) and tendon sheaths (second to fifth flexor digitorium tendons and extensor carpi ulnaris). For the clinical assessment, the changes in the disease activity VAS, global state of health VAS, pain and physician VAS and tender and swollen joints count were calculated by subtracting the VAS scores and joint numbers at follow-up from the VAS scores and joint numbers at baseline. To assess if NSAID intake affects scoring of the GS and PD, the changes in the score between baseline and follow-up were calculated by subtracting the total GS and PD scores (for each area) at follow-up from the scores at baseline. To assess if NSAID intake affects detection of the GS and PD signal in the joints, a comparison between the number of joints scoring more than 0 (>0) for the GS and PD per patient at baseline and follow-up was made. Changes in VAS scores, GS and PD scores and number of joints with GS and PD activity were summarised in each group using the median, first and third quartiles. Changes at follow-up were compared between groups using non-parametric Mann–Whitney U tests. The inter-reader and intrareader reliability were assessed using quadratic-weighted κ (κw) statistics. Values of κw <0.40 were considered to reflect poor agreement, between 0.40 and 0.75 fair to good agreement and >0.75 excellent agreement.18
A total of 58 patients were recruited. The mean age was 56, range 24–83; 71.2% female. The mean (SD) duration of symptoms was 96.34 (110.28) months. Four patients withdrew from the study before second scan and two were protocol violators—one by receiving a steroid injection and one by taking an alternative NSAID instead of stopping the NSAID treatment. The data from 52 patients who completed the study according to the protocol were analysed. For these patients, 26 stopped NSAIDs and 26 continued taking their NSAIDs. The duration of disease among patients was distributed as follows: 26 patients had early RA with mean (SD) duration of symptoms 10.57 (10.27) months and 26 patients had established RA with mean (SD) duration of symptoms 182.38 (96.43) months. The frequencies of NSAID preparations were as following; 26 patients were taking diclofenac, 12 celecoxib, 4 ibuprofen, 3 naproxen, 2 etoricoxib, 2 meloxicam, 1 ketoprofen, 1 etodolac, 1 mefenamic acid. There was no significant baseline difference in the parameters of the tender and swollen joints count or in the US assessment parameters between the patients groups (table 1).
The disease activity VAS, pain VAS and physician VAS showed a statistically significant increase in the group who stopped their NSAID treatment (table 2). There was a statistically significant increase in the number of swollen joints per patient in the NSAID stoppers in comparison with the non-stoppers (table 2). There was bigger change in the pain VAS, early morning stiffness and tender joint count in the groups with early RA than in the group with established RA when they stopped the NSAID treatment (table 3).
Both GS and PD scores showed a statistically significant increase in the patients who stopped NSAID in comparison with those who did not stop (figure 1–3). There was a median (first to third quartiles) increase in the number of joints in hands and wrist scoring >0 for GS by 4.0 (1.0 to 6.0) joints per patient in the group who stopped NSAID in comparison with 1.0 (0.25 to 1.0) joint in the patients who continued their NSAID (p<0.001). For PD, the number of joints in hands and wrist scoring >0 increased by 2.0 (0.75 to 3.0) joints per patient in the patient who stopped NSAID in comparison with no change in the patient who continued their NSAID (0.0 (–1.0 to 2.0), p = 0.005).
More joints showed an increase in the GS score per patient in the group of patients who stopped their NSAID than in those who did not stop (8.0 (5.7 to 11.25) joints and 2.0 (0.75 to 4.0) joints, respectively). Similarly, more joints showed an increase in PD score per patient in the group of patients who stopped NSAID than in those who did not stop (3.5 (2.75 to 5.25) joints and 0.5 (0.0 to 3.0) joints, respectively).
The number of joints that showed a decrease in GS score during the study was 0.0 (0.0 to 2.0) joints per patient in those who stopped their NSAID and 1.0 (0.75 to 3.0) joint per patient in those who continued their NSAID. The number of joints that showed a decrease in PD score during the study was 0.0 (0.0 to 1.0) joint per patient in those who stopped NSAID and 0.0 (0.0 to 2.0) joint per patient in those who continued NSAID. There was no significant difference in the change in the GS and PD scores between the patients with early RA and established RA when they stopped NSAID treatment (table 3).
A very small number of patients had tenosynovitis of the hand flexor tendons and extensor carpi ulnaris at baseline. There was no change in the GS and PD scores of the tendon sheaths between the two groups of patients. Stopping NSAID treatment did not lead to any increase or flare of GS or PD signal in the tendon sheaths.
Inter- and intrareader reliability
The inter-reader agreement was good to excellent for both GS and PD (κw for GS and PD scores were 0.72 and 0.88, respectively, p<0.001). The percentage exact agreement was 71% for GS and 81.3% for PD. The intrareader agreement was also excellent (κw for GS and PD were 0.96 and 0.90, respectively, p<0.001). The percentage exact agreement was 95% for GS and 88.1% for PD.
This study investigated the effect of NSAID intake on the detection and grading of US-detected synovitis. The results of this study demonstrated that stopping NSAIDs five half-lives before an US examination leads to an increase in clinical assessment parameters and US GS and PD scores. More joints showed a significant increase in GS and PD score per patient in the group of patients who stopped their NSAID than in those who continued. There was an increase in the disease activity VAS, global state of health VAS, pain and physician assessment VAS in the group of patients who stopped the NSAID. The number of tender and swollen joints also increased, but not to the same extent as that seen with sensitive US imaging. Less change was noted in the group of patients who did not stop their NSAID. The changes in the global health VASand tender joint count were not statistically significant. In patients who stopped NSAID there was an increase in total GS and PD score and the number of joints in hands and wrist scoring >0 for both GS and PD.
These results can be explained by a number of NSAID effects. NSAIDs may have an indirect effect on GS and PD by changing the synovial vessel tone through inhibition of the enzyme activity of cyclo-oxygenases (COX), which are involved in the formation of PGs.19 20 PGs play an important physiological role in blood vessel tone regulation. For example, PGD2 and PGI2, which are secreted in high levels in the process of inflammation by endothelial cells, enhance blood vessel dilatation19 21 22 and NSAIDs oppose this response. This effect on neovascularisation in the synovium may lead to a smaller PD signal detected by US. Another potential explanation is that NSAIDs may downregulate the vascular endothelial growth factor (VEGF) through COX inhibition. It is well known that VEGF plays a key role in the process of vasodilatation and endothelial cell proliferation and neovascularisation in the synovium.23 A recent study concluded that COX-2 may stimulate VEGF production.24
There have been no previously published data to illustrate the effects of NSAIDs on GS and PD of the synovium in inflammatory arthritis. A previous study tried to compare the effect of NSAIDs and paracetamol in reducing total effusion volume and synovial tissue volume in knee osteoarthritis.25 Patients had MRI after five half-lives washout of their NSAID and this was repeated 14 days after restarting the NSAID. After resumption of NSAID use there was a moderate decrease in the size of the effusion, relative to baseline. Patients noted a 49% decrease in pain relative to that at the time of the flare.
It was not the aim of our study to determine the efficacy of NSAID in RA as there have been a number of efficacy studies comparing NSAID with placebo.26,–,28 These studies all demonstrate a reduction in inflammatory parameters such as tender and swollen joint counts. Instead, the scope of our study was to determine the confounding effect of NSAID on US examination in an attempt to improve standardisation of image acquisition according to the goal of the OMERACT US Task Force goal.29 The acquisition of an US image is dependent on a number of machine-, operator- and patient-related factors. The effects of many of these factors have been underinvestigated, although there are data investigating the effects of transducer pressure,30 exercise, heating and cooling.31
This study has a number of limitations. Different NSAID preparations with different pharmacokinetics and COX-1 and COX-2 inhibition potency were collectively studied. That is because this study aimed to investigate the general confounding effect of NSAIDs rather than that of one particular agent. One potential criticism is that this study was single blinded as the patients were aware of the randomisation status. However, we considered that patient blinding status should not have direct effect on US assessment as this is dependent on the blinded assessor. The study did not also investigate the effect of restarting NSAIDs on the GS and PD. Time and feasibility issues were among the reasons for this as the patients had to attend an extra US examination for that purpose. We believe that the information about the effect of NSAIDs on GS and PD was gathered by investigating the flare up of synovitis induced by stopping the NSAIDs. A potential limitation of this study is the short duration for which NSAIDs were stopped; however, five half-lives has been reported to be sufficient to clear most of the NSAIDs from the body.32,–,34 It is possible that with a longer duration of washout a greater flare of synovitis might have occurred.
As far as we know, this is the first randomised controlled trial assessing the effect of NSAIDs on the scoring of joints in patients with RA using US. We have shown that stopping NSAIDs increases both the number of swollen joints and the scores of GS and PD synovitis, which might be important in the diagnosis and assessment of disease activity. Therefore NSAID use should be borne in mind at clinical assessment and during clinical trials, particularly when US is used as an outcome measure.
Competing interests None.
Provenance and peer review Not commissioned; internally peer reviewed.