Article Text
PDF
Abstract
Objective The aim of this study was to compare the efficacy of intramuscular versus ultrasound (US)-guided intratenosynovial glucocorticoid injection in providing disease control after 2, 4 and 12 weeks in patients with rheumatoid arthritis(RA) with tenosynovitis.
Methods Fifty patients with RA and tenosynovitis were randomised into two double-blind groups: (A) ‘intramuscular group’, receiving intramuscular injection of betamethasone and US-guided intratenosynovial isotonic saline injection and (B) ‘intratenosynovial group’ receiving saline intramuscularly and US-guided intratenosynovial betamethasone injection. All patients were in stable disease-modifying anti-rheumatic drug treatment prior to and during the study. Patients were excluded, and considered non-responders, if any treatments were altered during the follow-up period. ‘US tenosynovitis remission’, defined as US tenosynovitis grey-scale score ≤1 and colour Doppler score=0, was assessed at week 4 (primary outcome), and weeks 2 and 12, using non-responder imputation for missing data.
Results US tenosynovitis remission at week 4 was achieved in 25% (6/24) in the ‘intramuscular group’ versus 64% (16/25) in the ‘intratenosynovial group’, that is, a difference of −39 percentage point (pp) (CI −65pp to −13pp), Fisher exact test p=0.001. Corresponding values for the ‘intramuscular group’ versus the ‘intratenosynovial group’ at 2 and 12 weeks were 21% (5/24) versus 48% (13/25), that is, a difference of −27pp (CI −53pp to −2pp), p=0.072 and 8% (2/24) versus 44% (11/25), that is, difference of −36pp (−58pp to −13pp), p=0.003. Most US, clinical and patient-reported scores improved more in the ‘intratenosynovial group’ at all follow-up visits.
Conclusions In this randomised double-blind clinical trial, patients with RA and tenosynovitis responded significantly better to US-guided intratenosynovial glucocorticoid injection than to intramuscular glucocorticoid injection, both at 4 and 12 weeks follow-up.
Trial registration number EudraCT nr: 2013-003486-34.
- Synovitis
- Rheumatoid Arthritis
- Ultrasonography
- Corticosteroids
- Inflammation
Statistics from Altmetric.com
Background
Tenosynovitis (TS) is a frequent and important presentation of rheumatoid arthritis (RA), which besides the associated pain has severe potential consequences, as a recent study has shown that extensor carpi ulnaris TS is a strong predictor for erosive progression in early RA.1 However, clinically it is often difficult to distinguish TS from synovitis, as TS clinically may mimic joint involvement/swelling. Ultrasound (US) is an appropriate and often readily available tool for diagnosing TS and offers great capability for distinguishing between synovitis and TS.2
The response of TS to treatment with conventional synthetic disease-modifying anti-rheumatic drug (csDMARDs) and/or biological DMARDs (bDMARDs) is reported in one study to be similar to synovitis.2 However, it is common that patients have flares in joints or tendon sheaths, despite stable and effective DMARD treatment. The current treatment strategy to rapidly reobtain disease control during flares is glucocorticoid treatment, either orally for a limited period or as injection or optimisation of DMARD treatment. Several studies have focused on controlling joint flares,3–6 while few studies have investigated treatment of TS flares in RA.7 ,8
Intra-articular glucocorticoid injections are commonly used for a rapid disease control in early RA.9 ,10 In a trial of patients with RA, joints injected with glucocorticoid had a low frequency of flares after 2 years9 suggesting a role for intra-articular injections in a RA flare treatment strategy. Intramuscular injections (systemic treatment) versus US-guided local (ie, subacromial bursa) injections have to our knowledge only been compared once11 in a study investigating rotator cuff tears. However, no significant difference in overall shoulder pain and disability index score (primary outcome)11 was found after 6 weeks.
The aim of this study is to compare the efficacy of intramuscular versus US-guided intratenosynovial glucocorticoid injection to provide disease control after 2, 4 and 12 weeks in patients with RA and TS.
Patients and methods
Patients and study design
Patients with RA aged >18 years were eligible for inclusion if they had TS in the hand or ankle region, both on clinical examination, defined as pain on movement localised to the affected tendon sheath, and on US, according to the definition by the Outcome Measures in Rheumatology (OMERACT) US group.12 The following tendon sheaths were selected for possible intervention; abductor pollicis longus, extensor pollicis brevis, extensor carpi radialis longus, extensor carpi radialis brevis, extensor pollicis longus, extensor indicis and digitorum communis, extensor digiti minimi, extensor carpi ulnaris, flexor carpi radialis, flexor pollicis longus, flexores digitorum superficialis and profundus in the wrist/hand; tibialis anterior, extensor hallucis longus, extensor digitorum longus, peroneus longus, peroneus brevis, tibialis posterior, flexor digitorum longus and flexor hallucis longus in the ankle/foot.
All patients were recruited from the rheumatology outpatient clinic at Rigshospitalet, Denmark, from December 2013 to September 2015 by study independent physicians. All recruited patients were screened, that is, clinical and US examinations were performed of the above-mentioned selected tendon sheaths, by the study investigators. One or two tendon sheaths were chosen for intervention. If more than two tendons fulfilled the predefined definition for TS, the patients were not randomised due to ethical considerations. Patients were excluded if they had initiated csDMARDs and/or bDMARDs within the last 12 weeks prior to inclusion, changed dosage of csDMARDs and/or bDMARDs and/or glucocorticoid treatment within the last 6 weeks, or initiated non-steroidal anti-inflammatory drug therapy within the last 10 days prior to inclusion. During the follow-up period, patients were excluded if the DMARD treatment was changed.
Patients with RA and TS were randomised into two double-blind groups. An ‘intramuscular group’ receiving intramuscular injection of 14 mg (2 mL) of betamethasone dinatriumphosphate (BM) (eg, glucocorticoid) in the gluteal muscles and US-guided isotonic saline injection in up to two tendon sheaths (maximum 1 mL for each tendon sheath) and a ‘intratenosynovial group’ receiving 2 mL of intramuscular isotonic saline and US-guided BM injection in up to two tendon sheaths (maximum 7 mg (1 mL) for each tendon sheath). The equipotent prednisolone dose of 1 mg of BM is 8.33 mg. Randomisation was done in blocks of 10 by an independent ‘off-site’ person, controlled by the Danish Good Clinical Practice (GCP) unit. Follow-up was undertaken at 2, 4 and 12 weeks (±3 days) after injections.
Injection protocol
An independent research nurse was responsible for preparing the syringes. The syringes were blinded with self-adhesive labels covering the entire surface of the syringe, since BM dinatriumphosphat is slightly white and marked with intramuscular or tendon sheath. The label blinding was likewise controlled by the GCP unit. Three US-trained physicians performed the intervention, and were blinded for assessment and follow-up of the patients. The injections were performed with the needle point visualised by US inside the tendon sheath and outside the tendon, after disinfection and by ‘no touch technique’.9 US-guided injections were done after clinical and US assessment of the outcome measures. Patients were recommended to rest the injected hand/ankle for 48 hours after the injection.
US assessment
Two experts in musculoskeletal US, with a documented high intrareader and inter-reader agreement,13 blinded to the randomisation and intervention, performed the US assessment at baseline, 2, 4 and 12 weeks. A General Electric Logiq E9 (Milwaukee, Wisconsin, USA) US unit with a high frequency linear ML 6–15 probe was used. Colour Doppler (CD) was used as it is the most sensitive Doppler modality on this machine.14 The Doppler settings were made according to published recommendations,15 applying Doppler frequency of 7.5 MHz, pulse repetition frequency of 0.4 and gain just below the noise limit. The same Doppler settings were used for all examinations. TS was assessed by grey scale (GS) and CD using the semiquantitative scoring systems proposed by the OMERACT US group.12 This scoring system includes a four-grade semiquantitative scale for GS (ie, grade 0, normal; grade 1, minimal; grade 2, moderate; grade 3, severe) and Doppler (ie, grade 0, no intratenosynovial Doppler signal; grade 1, focal intratenosynovial Doppler activity; grade 2, multifocal intratenosynovial Doppler activity; grade 3, diffuse intratenosynovial Doppler activity). Furthermore, a quantitative Doppler evaluation was performed for each tendon sheath, using Q-analysis software programme on the Logic E9, calculating the fraction of colour pixels in the region of interest: pixel index (PI) 0–100. The PI was made on a transverse 5 s video clip of the most affected part of the tendon sheath with an average time consumption of 36 s. When two tendon sheaths were assessed, the mean US score/index was used for further analyses.
Clinical assessment
Four trained and calibrated physicians who were blinded to the randomisation and the selection of tendon sheath(s) for intervention and US assessment were chosen for clinical assessment of TS in the involved hand(s) or ankle(s) at baseline, 2, 4 and 12 weeks. A binary score for clinical tendon sheath involvement (absent/present) was performed in the involved hand(s) or ankle(s). Clinical signs of TS were scored as being present if there was pain on movement localised at the examined tendon sheath(s) and if the pain could be reproduced by resisted active movement of the affected tendons, with the forearm or lower leg stabilised. Further, findings such as crepitus, tenderness or swelling over the affected tendon sheaths supported the scoring, but were not mandatory.16 The mean clinical score was used for statistical analysis in case of two involved tendon sheaths.
All patients were blinded to the type of intervention and a patient-reported visual analogue scale (0–100 mm) for TS pain (VAS TS) was assessed at baseline, and after 2, 4 and 12 weeks. Furthermore, patients graded the effect of the treatment after 2, 4 and 12 weeks on a five-point scale (1: marked aggravated, 2: slightly aggravated, 3: unchanged, 4: slightly improved, 5: markedly improved).
Disease Activity Score for 28 joints using C reactive protein, patient global VAS (VAS Global) and Health Assessment Questionnaire were evaluated at baseline.
Outcome assessment
The primary outcome was the proportion of subjects in each group achieving US TS remission, defined as US TS GS score ≤1 and Doppler score=0, at week 4. The secondary outcome was obtaining US TS remission at week 12, whereas changes in clinical assessment, patient-reported outcomes and US parameters at weeks 2, 4 and 12 were exploratory outcomes. In the main statistical analysis, TS remission was only fulfilled in cases of two injected tendons sheaths, if both tendon sheets were fulfilling US remission definitions. A sensitivity analysis, in which US remission was registered if just one of the two tendons fulfilled the criteria, was also performed.
Ethical and legal considerations
The study was approved by the Danish National Ethical Committee on Health Research and the Danish Medical Agency. The study was assigned for the EU Clinical Trials Register (EudraCT nr: 2013-003486-34).
Statistical analysis
Statistical analyses were performed using SAS Enterprise Guide 7.1. Binary data (including the primary outcome) were analysed by Fisher's exact test, and relative risks were calculated between the groups at 2, 4 and 12 weeks. The 95% CI for the difference was computed by the Agresti-Caffo method. A non-responder imputation (NRI) was used for missing data in these analyses. Differences within and between groups were expressed as means (SD) and medians (25/75 percentiles). Data were analysed two sided using Wilcoxon's test for paired data and Mann–Whitney's test for unpaired data, with a p value<0.05 considered as significant. Further, a parametric sensitivity analysis of covariance (ANCOVA) was added. Group and baseline values were used as covariates in the model with a two-sided p value <0.05 considered as significant. Missing data were analysed as last observation carried forward in these analyses. Descriptive statistics were used to present the results of the patients' grading of treatment effect for weeks 2, 4 and 12.
Sample size considerations
We assumed that 40% of the intramuscular BM-treated and 80% of the intratenosynovial BM-treated patients would obtain the primary endpoint. With a statistical power of 0.7 and a probability of type 1 error of 0.05, 23 patients were required in each group. It was expected that approximately 10% would drop out of the study and therefore 50 patients were included in total.
Results
Study population and adverse events
Seventy-five patients with RA were screened from 18 December 2013 to 28 September 2015. Fifty patients were randomised and 49 received treatment. One patient did not receive allocated intervention, since a joint injection was needed as well. Last visit of the last patient was 21 December 2015. The patient disposition is illustrated in figure 1. Demographics and baseline characteristics of the patients are provided in table 1. No significant baseline differences between the groups were observed, except that patients receiving treatment for two inflamed tendon sheaths were more often observed in the ‘intramuscular group’ (10 patients) compared with ‘intratenosynovial group’ (3).
Demographics, clinical and laboratory characteristics of patients in each randomised group
Patient disposition. 1No data before new DMARD treatment was started, 2Visit out of window (before week 12). BM, betamethasone; DMARD, disease-modifying anti-rheumatic drug; NRI, non-responder imputation; LOC, last observation carried forward.
No serious adverse events were registered for any of the patients during the 12 weeks follow-up period and no tendon ruptures were registered in a safety follow-up analysis.
TS remission
US TS remission at week 4 was achieved in 25% (6/24) of the patients in the ‘intramuscular group’ and in 64% (16/25) in the ‘intratenosynovial group’. This difference was highly statistically significant, Fisher exact test p=0.001, that is, a difference of −39 percentage point (pp) (CI −65pp to −13pp).
At week 12, US TS remission was achieved in 8% (2/24) of the patients in the ‘intramuscular group’ and in 44% (11/25) of the patients in the ‘intratenosynovial group’, that is, a difference of −36pp (CI −58pp to −13pp), p=0.003. At 2 weeks, US TS remission was achieved in 21% (5/24) of the patients in the ‘intramuscular group’ and in 48% (13/25) of the patients in the ‘intratenosynovial group’, that is, a difference of −27pp (CI −53pp to −2pp), p=0.072 (figure 2).
Proportions of patients with US tenosynovitis remission at weeks 2, 4 and 12. BM, betamethasone; p, Fisher exact test (p); US, ultrasound; US tenosynovitis remission defined as US tenosynovitis grey-scale score ≤1 and colour Doppler score=0. The primary endpoint was at week 4.
A supplementary sensitivity analysis of TS remission, in which US remission was registered if just one of the two tendons fulfilled the criteria in cases where two tendons were injected, was also performed. No changes of week 4 data were found. However, there were minor changes for week 2 and 12. The week 2 values for ‘intramuscular group’ and the ‘intratenosynovial group’ were 29% (7/24) versus 52% (13/25), that is, a difference of −23pp (CI −50% to 4%), p=0.15 and for week 12 13% (3/24) versus 44% (11/25), that is, a difference of −31pp (CI −55% to −8%), p=0.026.
Changes in TS scores
Baseline values and the changes in these at 2, 4 and 12 weeks for GS, CD, PI, VAS TS and clinical assessment are presented in table 2. Furthermore, the exact values for weeks 2, 4 and 12 are illustrated in figure 3 and online supplementary table S1.
Tenosynovitis outcome values: at baseline, changes within patients receiving the intramuscular BM injection and patients receiving the US-guided intratenosynovial BM injection, and differences between groups, calculated as non-parametric statistic
Tenosynovitis exact values (mean) between intramuscular BM injections and US-guided intratenosynovial BM injections at baseline, and at weeks 2, 4 and 12. BM, betamethasone; CD, colour Doppler score; GS, grey-scale score; US, ultrasound; VAS tenosynovitis, patient-reported visual analogue scale (0–100 mm) for pain tenosynovitis.
supplementary tables
No baseline differences between the groups for GS score (Mann–Whitney U test; p=0.83), CD score (p=0.52), PI (p=0.66), VAS TS (p=0.45) and clinical assessment (p=0.07) values were found. Clinical blinded assessment was able to detect TS in 70% and 90% of the patients in ‘intramuscular group’ and ‘intratenosynovial group’, respectively.
In both groups, statistically significant decreases in all assessed US parameters (GS, CD and PI) were observed from baseline to 4 and 12 weeks, which is in contrast to VAS TS and clinical assessment where significant changes were found only in the ‘intratenosynovial group’. Figure 4 illustrates changes in TS after intratenosynovial BM treatment for all time points. In sensitivity analyses, using only observed data (no data imputation) there was a significant reduction for all assessed parameters in the ‘intratenosynovial group’, while only PI showed a significant decrease in the ‘intramuscular group’ (data not shown).
Ultrasound (US) assessed treatment response for tenosynovitis in the flexor tendon sheath of the second finger after US-guided intratenosynovial betamethasone injection. Treatment response for the second finger flexor tendon sheath (*) illustrated over 12 weeks. The US tenosynovitis grey-scale and colour Doppler scores were both 2 at baseline. At weeks 2, 4 and 12, the grey-scale score was 1 and the colour Doppler score 0. The second metacarpal joint is marked with two asterisks. The grey lines are areas without scan gel and the yellow boxes are the Doppler boxes.
Both CD score and PI decreased significantly within both groups (table 2): However, PI was only able to detect a statistically significant difference between the two groups after 12 weeks, whereas CD score found a statistically significant difference at all follow-up visits (2, 4 and 12 weeks).
In an ANCOVA sensitivity analyses of TS outcome values, the same levels of statistical significance for differences between the groups were found, except for PI at 4 weeks (p=0.037), see online supplementary table S2.
If change scores in the ‘intramuscular group’ and the ‘intratenosynovial group’ were compared at week 4, a significantly lower score was observed for all assessed parameters in the ‘intratenosynovial group’, while after 2 weeks a significant lower score was only found for GS and CD in the ‘intramuscular group’. Similar results were found when data were analysed without data imputation (data not shown).
The percentages of patients grading the treatment effect as markedly improved at weeks 2, 4 and 12 were 33%, 54% and 29%, respectively, in the ‘intramuscular group’, while 48%, 68% and 72%, respectively, in the ‘intratenosynovial group’.
The average dose of BM injected at baseline was 14 mg per patient in the ‘intramuscular group’, whereas the average BM dose in the ‘intratenosynovial group’ was 5.6 mg per patient.
Discussion
To our knowledge, this is the first study to compare systemic glucocorticoid injection with local US-guided glucocorticoid injection for the treatment of TS. Our study clearly showed that patients with RA with TS benefitted from local glucocorticoid injection compared with intramuscular administered glucocorticoid, as assessed 4 and 12 weeks after treatment.
Our primary outcome was defined as US TS GS score ≤1 and Doppler score=0 at week 4. This is a very strict definition for TS remission, since US GS grade 1 findings may be seen even in healthy individuals.17 ,18 There was a considerable difference between the groups after 4 weeks, and this was consistent at week 12. The patient-reported outcome showed similar difference between groups, illustrating that the difference in treatment effect is also clinically relevant for the patient. Thus, our study documents that US-guided intratenosynovial injections are superior to intramuscular injections of BM, assessed by imaging, and seen from the patient's point of view.
Interestingly, GS and CD US scores were the only parameters able to detect a significant intergroup difference after 2 weeks, while clinical parameters did not. This indicates that early US assessment may predict the clinical outcome and therefore is likely to be important in the monitoring of patients with RA. Thus, the data suggest that monitoring patients with US will allow quicker detection of changes in inflammatory activity, and thereby quicker intervention. This may allow a more efficient treat-to-target strategy through rapid disease control. Further studies are, however, needed to clarify this.
We expected that the use of PI in the present study would add a higher sensitivity to change, since PI includes all colour information in the grading of TS. However, the current study demonstrated no advantages of PI compared with the semiquantitative OMERACT scoring system.
It is noted that the administration route of BM has a high impact on the outcome. It is known that BM after intramuscular injection has a plasma half-life of about 81 hours,19 while the same values after intra-articular injections are approximately 151 hours.20 Even though the half-life within the joint or tendon sheaths after local injections have not been studied, these pharmacokinetic data indicate that BM is present longer after local injection than after intramuscular injection, which may explain the difference in treatment efficacy between groups.
Regarding the duration of effect it should be noted, that at week 12, only 2 of 11 patients in the ‘intramuscular group’ were in US remission, documenting a very limited long-term effect of intramuscular BM injection, in contrast to approximately half of the patients in the ‘intratenosynovial group’.
The changes in VAS-TS were similar in the two groups from baseline to 2 and 4 weeks, in contrast to clinical and US assessments. We have no certain explanation for this, but it seems likely that an expectation of improvement (placebo effect) by the patients in both groups has contributed.
A study limitation is the relatively small sample size, increasing the risk of type 2 error. Our study did however find a convincing, and statistically significant, difference between groups. Our primary analysis was done using data imputations (intention to treat; NRI or last observation carried forward), in a conservative way and we found similar results with and without the use of these imputations, supporting the reliability of our results. Overall, the randomisation provided comparable groups.
Both intramuscular and intratenosynovial injection(s) were well tolerated and no adverse events were observed in the study period. In the present study, we have used US-guided injections to ensure correct placement of the BM in the tendon sheath,21 ,22 but we have not evaluated whether US-guided injections are better than blinded local injections. However, a recent study of TS has indicated better short-term effect of US-guided intratenosynovial injections compared with blinded injection of corticosteroid8 and therefore we find it recommendable to use US-guided intratenosynovial injections for local treatment of TS in patients with RA.
The overall BM dose was lower in the ‘intratenosynovial group’ than in the ‘intramuscular group’. This further strengthens the advantage of intratenosynovial injections, since it may potentially lead to less side effects over time, as the cumulative dose of BM is likely to be lower if the patients are treated with local injections. Thus, we believe that our findings have the potential to improve the treatment of TS in patients with RA. In accordance with our findings, a previous study has shown long-lasting remission of injected joints and minimal side effects of intra-articular BM injections9 supporting the use of local BM injections in patients with RA. High overall disease remission rates have also previously been reported for treatment of active RA with local BM injections,23–25 but to our knowledge no double-blind comparisons with intramuscular glucocorticoid have yet been undertaken.
In conclusion, we found, in a randomised double-blind clinical trial, that patients with RA and TS responded significantly better clinically and by US when treated with US-guided intratenosynovial glucocorticoid injection compared with intramuscular glucocorticoid injection, both at 4 and 12 weeks.
Acknowledgments
The authors kindly thank Cecilie Cornelia Heegaard Brahe, Teresa Rozenfeld, Lise Hyldstrup, Brigitta Pedersen, Maj-Brit Thyme, Simon Krabbe and Signe Kaas Møller-Bisgaard (Copenhagen Center for Arthritis Research (COPECARE), Center for Rheumatology and Spine Diseases, Rigshospitalet, Glostrup, Denmark) for their great assistance with the study.
References
Footnotes
Handling editor Tore K Kvien
Contributors MAD and LT have made substantial contributions to study conception and design, acquisition of data, analysis and interpretation of data, drafting the article and final approval of the version of the article to be published. VF, DG, UMD and LMØ have made substantial contributions to acquisition of data, analysis and interpretation of data, revising the article critically for important intellectual content and final approval of the version of the article to be published. MØ and EN have made substantial contributions to study conception and design, analysis and interpretation of data, revising the article critically for important intellectual content and final approval of the version of the article to be published.
Competing interests None declared.
Patient consent Obtained.
Ethics approval The study was approved by the Danish National Ethical Committee on Health Research and the Danish Medical Agency. The study was assigned for the EU Clinical Trials Register (EudraCT nr: 2013-003486-34).
Provenance and peer review Not commissioned; externally peer reviewed.