Article Text

Extended report
Rapid reduction in tenosynovitis of the wrist and fingers evaluated by MRI in patients with rheumatoid arthritis after treatment with etanercept
  1. M Pilar Lisbona1,
  2. Joan Maymó1,
  3. Javier Perich2,
  4. Miriam Almirall1,
  5. Jordi Carbonell1
  1. 1Department of Rheumatology, IMAS, Hospital del Mar, Universidad Autónoma de Barcelona, Barcelona, Spain
  2. 2Department of Radiology, IDIMAS-CRC, Hospital del Mar, Barcelona, Spain
  1. Correspondence to Dr Joan Maymó, Department of Rheumatology, IMAS, Hospital del Mar, Universidad Autónoma de Barcelona, Barcelona 08003, Spain; jmaymo{at}


Objective To assess the efficacy of etanercept in reducing tenosynovitis evaluated by MRI of the hand (h-MRI) in patients with active rheumatoid arthritis (RA) refractory to disease-modifying antirheumatic drug (DMARD) after 6 weeks of treatment.

Methods 31 patients with active RA defined by a disease activity score (DAS28) >3.2 and synovitis in the hands were randomised into two groups: 19 patients received 50 mg weekly subcutaneous etanercept added to previous DMARD treatment and 12 patients continued with previous DMARD therapy. Clinical evaluation, blood tests, functional capacity evaluation and h-MRI were performed at the start of the investigation and at week 6. Tenosynovitis was evaluated on T1-weighted sequences with fat suppression after gadolinium as the presence of a peritendinous signal enhancement on axial images using a new method including wrist and finger tendons. The reliability, sensitivity to change and responsiveness of this method were also evaluated.

Results Scores for tenosynovitis showed a significant reduction in the etanercept group compared with placebo (p=0.01) after 6 weeks of treatment. Adding MRI joint synovitis to tenosynovitis scores gave an even higher significant reduction in the etanercept group (p=0.007). A positive and statistically significant correlation between tenosynovitis and DAS28, erythrocyte sedimentation rate and C-reactive protein was found, but not with functional capacity. Responsiveness for tenosynovitis was small but was higher when joint synovitis scores were added.

Conclusion Addition of etanercept significantly reduced MRI tenosynovitis of the wrist and fingers in patients with active RA refractory to DMARD treatment. The method of scoring tenosynovitis showed good reliability and moderate responsiveness.

Statistics from


Rheumatoid arthritis (RA) is a chronic disease characterised by inflammation of synovial tissue of the joints and tendons. Tenosynovitis is an inflammatory process of the synovial tendon sheath that may result in degeneration and rupture of the tendons and may contribute to bone erosions, development of joint deformities and loss of functional capacity.1 Clinical evaluation of tenosynovitis is difficult and conventional radiography is not useful. MRI is a technique that allows good visualisation and assessment of inflammatory changes in the tendons and tendon sheaths.2 3

Scoring tenosynovitis was not included in the Rheumatoid Arthritis Magnetic Resonance Imaging Score (RAMRIS) described by Outcome Measures Rheumatoid Arthritis Clinical Trials (OMERACT) because of its low reliability.4 However, a novel method for tenosynovitis scoring has recently been described with good reliability and high responsiveness when evaluating treatment.5

Previous placebo-controlled studies have shown a significant reduction in joint synovitis and bone oedema on MRI after short periods of treatment with biological agents, but the effect of such treatment on tenosynovitis is not well known.6,,9

The aim of the present study was to evaluate the effect of etanercept on tenosynovitis by hand-MRI (h-MRI) using a semiquantitative assessment in patients with active RA refractory to disease-modifying antirheumatic drug (DMARD) therapy after 6 weeks of treatment. We also evaluated the correlation between changes in tenosynovitis and changes in the disease activity score (DAS28), erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) and functional capacity by Health Assessment Questionnaire (HAQ). In addition, reliability, sensitivity to change and responsiveness of the MRI tenosynovitis scoring method used was assessed.



Thirty-one patients with RA (fulfilling the 1987 American College of Rheumatology criteria for RA) were included.10 All patients had active disease defined by DAS28 >3.2 and synovitis of the hands (one or more joints involved) despite treatment with DMARDs.

Patients were randomised into two groups: 19 patients received a weekly subcutaneous dose of 50 mg etanercept added to previous DMARDs while 12 patients continued with previous DMARD treatment. All doses of DMARDs, corticosteroids and non-steroidal anti-inflammatory drugs were kept unchanged for the 6 weeks of the study. Paracetamol (maximum doses 3 g/day) was allowed in case of disease exacerbation. All patients fulfilled the recommendations of the Spanish Society for Rheumatology (SER) for biological therapy.11

Clinical evaluation, blood tests and MRI images were performed at baseline and after 6 weeks of treatment following a previously designed protocol. Clinical evaluation was performed by one rheumatologist (JM) and MR images were scored by two readers (JP, radiologist and MPL, rheumatologist), all blind to the patients' treatment.


MRI scanning of the dominant wrist and second to fifth metacarpophalangeal (MCP) joints was performed at baseline and at week 6 using a 1.5 Tesla Superconductive System (Signa Echo-speed Excite II; General Electric Medical Systems, Milwaukee, Wisconsin, USA) equipped with a dedicated wrist coil. Coronal and axial T1-weighted sequences (repetition time 460 ms; echo time 10 ms; slice thickness 3 mm; matrix 256 × 224; field of view 160 mm) and T2-weighted sequences with fat suppression (repetition time 4000 ms; echo time 94 ms; slice thickness 3 mm; matrix 256 × 224; field of view 160 mm) were performed. After intravenous injection of 0.2 ml/kg of the contrast agent Omniscan (GE Healthcare, Amersham, UK), coronal and axial T1-weighted sequences with fat suppression (repetition time 640 ms; echo time 15 ms; slice thickness 3 mm; matrix 256 × 224; field of view 160 mm) were performed. The same scanner and hand coil were used for both examinations. To score tenosynovitis, T1-weighted sequences with fat suppression before and after administration of gadolinium in the axial plane of the wrist and second to fifth MCP joints were used. The slices were taken consecutively starting from 1 cm proximally to the distal radioulnar joint to 1 cm distal to MCP joints, including 2–4 slices for each joint area or joint.

Tenosynovitis scoring system

MRI definitions

Synovitis, bone oedema and erosions were evaluated following RAMRIS definitions. Tenosynovitis was defined as above abnormal postgadolinium enhancement around a tendon (peritendinous halo) visible in two or more consecutive slices. Enhancement was defined as a signal intensity increase judged by comparison between T1-weighted images obtained before and after intravenous gadolinium contrast with fat suppression.

Tendons evaluated

The tendons of the wrist and fingers included for evaluation are shown in figure 1. Flexor carpi ulnaris, palmar longus and extensor tendons of the fingers were not included since they lack a synovial sheath.

Figure 1

Score sheet for tenosynovitis in a hand (wrist and metacarpophalangeal (MCP) joints) with rheumatoid arthritis.

Grading scale

The extent of tenosynovitis was defined as complete or incomplete halo. Incomplete halo was defined as enhancement surrounding less than two-thirds of a tendon or group of tendons, while complete halo was defined as enhancement surrounding more than two-thirds of the tendon or group of tendons. This definition was applicable even if only visualised in one slice. In total, 13 groups of tendons were evaluated with a semiquantitative score (0–2) for each group: 0, absence of enhancement; 1, presence of tenosynovitis with incomplete halo; and 2, presence of tenosynovitis with complete halo, giving a maximum score of 26 (see figures S1 and S2 in online supplement).

We computed the MRI inflammation score by combining the MRI synovitis and tenosynovitis score, with a possible maximum score of 47, and the MRI total inflammation score by adding MRI bone oedema to MRI synovitis and tenosynovitis scores with a maximum score of 116.

Two experienced readers in RAMRIS scoring (JP and MPL) read the images for tenosynovitis after 2 months of training with the new method. All sets of images were coded so readers were unaware of patient identity, treatment group or clinical data. The two readers read images independently and readings were performed directly on films. The two sets of MR images were read in chronological order and the reading time for each set was recorded. For evaluation of reliability, 15 pairs of h-MRI (baseline and at 6 weeks) were selected from the total of 31 patients. A second reading was completed after 1 week. Images were also scored measuring millimetres of maximum width of peritendinous enhancement of gadolinium following instructions and recommendations of the method described by Haavardsholm et al using commercial computer software (Centricity DICOM Viewer; GE Healthcare IT, Version 3.0).

Statistical analysis

To assess the reliability of the scoring method for tenosynovitis, the intraclass correlation coefficients (ICCs) for intrareader and inter-reader for a single measure and change were calculated using a two-way random effects model. Single-measure (SmICC) and average-measure (AvmICC) for status and change scores were calculated and were given as medians and ranges. Agreement was considered good if ICCs were >0.60 and very good if >0.80. The reliability of the RAMRIS method for this group of patients by the same readers has been reported in a recent paper.6

Sensitivity to change was assessed by calculating the smallest detectable difference (SDD) and expressed as minimal detectable change (MDC)—that is, as a percentage of the maximum score of the method. An MDC value of <20% is acceptable for detecting changes.12 The standardised response mean (SRM) was used to assess the responsiveness.13 Changes from baseline to week 6 were compared between the two groups using repeated measures analysis of variance and the Mann–Whitney U tests for non-normal distribution variables. The Spearman test of rank correlation was used for assessing correlation between clinical and laboratory parameters (DAS28, HAQ, ESR and CRP) with tenosynovitis, MRI inflammation and MRI total inflammation; p values ≤0.05 were considered to be significant.

SPSS Version 13.0 for Windows was used for the statistical analysis.


Both groups were homogenous for demographic, immunological, clinical and laboratory test parameters, although the etanercept group showed a slightly higher activity measured by DAS28. Synovitis by MRI was also slightly higher in the treatment group, but there were no differences in the tenosynovitis scores (table 1). The prevalence of tenosynovitis in our series of patients at baseline was 100% in both groups.

Table 1

Baseline demographic data, clinical, laboratory and MRI parameters between the two groups

At week 6, there was a median reduction of two points in tenosynovitis in the etanercept group compared with the control group, which was statistically significant (p=0.01). The rest of the parameters of MRI did not show statistically significant changes in median scores after treatment. Nevertheless, the etanercept group showed a marked reduction in the median score for synovitis from 13.0 to 9.0, but without reaching statistical significance (p=0.07). Combining tenosynovitis scores with synovitis scores (MRI inflammation score) showed a much higher statistical difference in reduction in the treatment group compared with the control group (p=0.007) than each parameter alone. Adding the bone oedema score (MRI total inflammation score) did not increase the statistical significance. A statistically significant reduction in activity variables DAS28, ESR, CRP and HAQ was found in the treatment group compared with the control group at week 6 (table 2).

Table 2

Clinical, laboratory and MRI parameters at baseline and at week 6 of the placebo and etanercept groups

In the etanercept group, a correlation was observed between the reduction in the parameters of MRI inflammation (tenosynovitis and tenosynovitis plus synovitis with or without bone oedema) with different parameters of activity (DAS28, ESR and CRP), but not with HAQ (table 3).

Table 3

Spearman correlation in etanercept group between changes in variables of clinical activity, functional capacity (HAQ) and changes in MRI tenosynovitis, MRI inflammation and total inflammation

The results of SRM with 95% CI are reported in table 4. MRI synovitis was highly responsive with an SRM=0.90, but the SRMs for the other RAMRIS parameters and for tenosynovitis were lower. Nevertheless, moderate to high responsiveness (SRM=0.71) was found for MRI inflammation and for MRI total inflammation scores.

Table 4

SRM with 95% CI for all measures at 6 weeks

The method for scoring tenosynovitis described in this study showed an acceptable reliability. The intrareader SmICC and AvmICC were high for both readers for status score (0.92–0.99) and slightly lower for change (0.78–0.97). The MDC was lower than 13% for both readers for status and change scores. Inter-reader ICC for status score was high (0.80–0.95) and moderate for change (0.62–0.76). Comparing our results with the novel method described recently by Haavardsholm et al, intrareader and inter-reader ICCs were similar for both methods. Also, when we used their method to read our patients, intrareader ICCs were similar for both methods but the inter-reader ICCs were slightly lower for status in the other method (table 5).

Table 5

Intrareader and inter-reader agreement on tenosynovitis for baseline and 6-week status and change in score of the two methods for scoring tenosynovitis (two-way random effect model, single measure ICCs, average measures ICCs, SDD and MDC)

The complete time for scoring a patient for tenosynovitis with our method (set of two MR images at baseline and at week 6) was 5 min (range 2–7) compared with 7 min (range 5–10 min) for the method used by Haavardsholm et al.


In this placebo-controlled study we found a significant reduction in tenosynovitis by h-MRI in patients with active RA after 6 weeks of treatment with etanercept. The efficacy of etanercept in rapidly reducing disease activity and slowing radiographic joint damage in RA has been shown in many studies.14,,16 MRI has also shown that tumour necrosis factor α (TNFα) blockers reduce synovitis and bone oedema over short periods of time,6,,8 but their effect on tendon inflammation in RA is not well known. MRI synovitis and bone oedema are predictors of progression of radiographic structural damage in the long term.17,,19 The relationship between tenosynovitis and structural damage or functional capacity is less well-known. However, tenosynovitis probably contributes greatly to pain, joint stiffness, bone erosions, joint subluxation and loss of capacity in RA.

Various studies have shown that the presence of tenosynovitis in active RA is frequent with a variable prevalence (53–96%) and is higher than in other inflammatory arthropathies.2 20,,22 The prevalence of 100% found in our series may be related to a high disease activity in patients with insufficient response to DMARD treatment and eligible for biological therapy.

Clinical evaluation of tenosynovitis of the hand is difficult and conventional radiography is useless. MRI has been shown to be more sensitive than physical examination in the evaluation of tendinous disorders of hands and feet in patients with RA.23 Moreover, extensive and early tendon involvement in RA correlates with a worse functional capacity outcome since it is a predictor for functional capacity of the hand and tendon ruptures in the long term.1 24 Evaluation of tenosynovitis and early treatment could therefore avoid invasive treatments such as prophylactic tenosynovectomy which was used in the last decade.25,,27

Our study is the first to show a statistically significant reduction in tenosynovitis after only 6 weeks between the etanercept group and the placebo group (see figure S3 in the online supplement).

Although there were differences in baseline values, especially in DAS28 and synovitis, there were no significant differences in baseline tenosynovitis values between the two groups. The number of patients was also small, but the reduction in tenosynovitis was statistically significant in the etanercept group. Most patients (73.7%) in the etanercept group showed some reduction compared with 16.7% in the control group (p=0.003).

Although synovitis and bone oedema were also reduced, these changes did not reach statistical significance. These results suggest that the response of tenosynovitis is more rapid than of synovitis or bone oedema after treatment with TNF blockers. Moreover, combining the scores for synovitis and tenosynovitis resulted in a more significant difference between the etanercept and control groups.

Tenosynovitis was not included for evaluation of MRI lesions in the RAMRIS method described by OMERACT because of the low reliability of previously reported methods to assess tendon changes by MRI. These methods have been described in different studies, which usually did not report data on reproducibility. They were also different in methodology of acquisition of images, types of tendon lesions evaluated and usually the scoring was dichotomic.2 3 20 21 23

Haavardsholm et al have described a novel method for scoring tenosynovitis by MRI with good results for reliability and sensitivity to change.5 These results are similar to the method used in our study. We developed our scoring system before the method described by Haavardsholm et al was published. Both methods use the standard RAMRIS recommendations for acquisition of images and defined tenosynovitis as enhancement after gadolinium independent of any thickness. Although synovitis-like changes in healthy subjects have been described, these changes were small, different methodology was used (T2 sequences instead of gadolinium, sagittal plane images) and the authors concluded that a slight peritendinous effusion or fluid collection could be observed in only a very few cases of healthy controls but no cases of ‘pannus’.3 28

The main difference between the method of Haavardsholm et al and our method was the use of fat suppression in T1-weighted sequences after gadolinium. We considered that, with fat suppression, images of tenosynovitis were more contrasted and therefore easier to read, as suggested by previous studies and by our expert radiologist. Another difference is that our method included flexor tendons at the level of the MCP joints as additional areas to score tenosynovitis in the hand. Standardised image acquisition of RAMRIS already includes these areas to score synovitis, erosions and bone oedema, so inclusion of these areas would not increase the time of image acquisition in a complete RAMRIS examination. Moreover, scoring the flexor tendons of the second to fourth fingers at the level of the respective MCP joints makes sense since these tendons have an independent synovial sheath. The flexor tendon sheaths of the first and fifth fingers were also included because tenosynovitis, as depicted by gadolinium, is not always homogeneous throughout the extension of these tendons. Despite the inclusion of flexor tendons at the level of the MCP joints, the time for scoring tenosynovitis was similar to the method described by Haavardsholm et al.

The use of the difference in complete or incomplete halo was chosen because we observed that, in those groups of tendons formed by more than one tendon with a common synovial sheath (abductor pollicis longus and extensor pollicis brevis (APL/EPB), extensor carpi radialis longus and brevis (ECRL/ECRB), extensor digitorum and indicis (ED), and flexor digitorum superficialis and profundus tendons (FD)), the enhancement after gadolinium frequently involved only one tendon or did not completely surround the entire group. In all these cases the presence of an incomplete halo was confirmed by the presence of tenosynovitis in coronal scans.

Haavardsholm et al have also reported a marked reduction in tenosynovitis after 3 months of treatment with different TNF blockers witha high responsiveness measured by SRM.29 In our series we found similar high responsiveness of RAMRIS synovitis and bone oedema and a somewhat lower response of tenosynovitis. This lower responsiveness of tenosynovitis could be related to a higher SD in the scoring of change, differences in time of outcome (only 6 weeks) or differences in the scoring method. Moreover, the median baseline scoring for tenosynovitis was also lower in our study than in this published study (8 vs 10), so the expected change might also be lower. As in the study by Haavardsholm et al, we found that synovitis and tenosynovitis showed the best results and higher SRM when combined together (MRI inflammation). However, adding bone oedema in our series did not give higher results of SRM, possibly because of a shorter period for outcome (6 weeks vs 3 months). Nevertheless, tenosynovitis and MRI inflammation scoring (combining tenosynovitis and synovitis) showed a clear statistically significant reduction after 6 weeks of etanercept treatment compared with the control group.

Different longitudinal studies have shown a statistically significant correlation between reduction in disease activity and reduction in MRI synovitis.6 18 19 30 In agreement with these results, we found that the reduction in tenosynovitis in the etanercept group correlated with a reduction in DAS28, ESR and CRP. We did not find a correlation with HAQ. Reduction in synovitis or bone oedema correlates with a reduction in HAQ.7 31 Tenosynovitis correlates with functional capacity and especially hand functional capacity in the long term.1 24 The lack of correlation between a reduction in tenosynovitis and HAQ in our study could be related to the short time of treatment, persistence of synovitis and to the fact that HAQ is a global measure of functional capacity.

In summary, in patients with active RA despite treatment, etanercept added to DMARDs rapidly and significantly reduced tenosynovitis measured by h-MRI. Scoring tenosynovitis should be included in MRI methods used to monitor the response to treatment in RA.


The authors thank Sergi Mojal and Josep Maria Manresa for their help in statistical analyses and Silvia Sanchez and Silvia Iniesta, investigational nurses from the Department of Rheumatology of Hospital del Mar, for their help.


View Abstract

Supplementary materials


  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the local ethics committee of Hospital del Mar, Barcelona and conducted under the principles of the Helsinki declaration. All patients signed an informed consent form for this study.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

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.