Abstract
Objective. To develop and validate a magnetic resonance imaging (MRI) tenosynovitis (TS) score for tendons at the wrist and metacarpophalangeal (MCP) joint levels in patients with rheumatoid arthritis (RA).
Methods. Axial T1-weighted precontrast and postcontrast fat-saturated MR image sets of the hands of 43 patients with RA initiating rituximab therapy were obtained at baseline and after 14, 26, 38, or 52 weeks. The MR images were scored twice by 4 readers. Nine tendon compartments of the wrist and 4 flexor tendon compartments at the MCP joints were assessed. Tenosynovitis was scored as follows: 0: No; 1: < 1.5 mm; 2: ≥ 1.5 mm but < 3 mm; 3: ≥ 3 mm peritendinous effusion and/or postcontrast enhancement. Intrareader and interreader intraclass correlation coefficients (ICC), smallest detectable change (SDC), percentage of exact and close agreement (PEA/PCA), and standardized response mean (SRM) were calculated.
Results. Intrareader and interreader ICC for status and change scores were very good (≥ 0.80) for total scores for all readers. Intrareader SDC was ≤ 3.0 and interreader SDC was < 2.0. The overall PEA/PCA intrareader and interreader agreements for change scores in all tendons were 73.8%/97.6% and 47.9%/85.0%, respectively. Average SRM was moderate for total scores and 60.5% of the patients had a tenosynovitis change score ≥ SDC.
Conclusion. The TS score showed high intrareader and interreader agreement for wrist and finger tendons, with moderate responsiveness, and the majority of the patients showed a change above the SDC. This scoring system may be included as a component of the RAMRIS.
Tenosynovitis (TS) in the hand of patients with rheumatoid arthritis (RA) is a frequent and early occurring inflammatory feature that can cause tendon rupture and may be associated with subsequent bone erosion1,2,3,4,5. Magnetic resonance imaging (MRI) provides excellent visualization of bone and soft tissues in the hand, including TS.
The Outcome Measures in Rheumatology (OMERACT) MRI in Arthritis Working Group developed an RA MRI score (RAMRIS) for synovitis, osteitis, bone erosion, and recently, for joint space narrowing6,7. Adding TS to the RAMRIS may provide improved power to study and monitor the disease course of RA.
Haavardsholm, et al designed a TS score for the wrist that demonstrated good reliability and responsiveness8,9. In addition, the Working Group has suggested and validated a TS score for the fingers as part of the psoriatic arthritis MRI score (PsAMRIS)10,11,12. However, none of these scores covered the tendons at both wrist and metacarpophalangeal (MCP) joint level.
Our aim was to develop and validate an RA TS score for tendons at wrist and MCP joint levels. Here we present previously unpublished results that demonstrate the reliability of this novel score, enabling it to be included in the internationally renowned RAMRIS.
MATERIALS AND METHODS
Development of scoring system
In accordance with the processes in the OMERACT Handbook for technical tool development, previous MRI TS scores were identified through a thorough review of the literature by a fellow of the Working Group, which included rheumatologists, radiologists, methodologists, and industry representatives from 3 continents. The TS scores suggested by Haavardsholm, et al and the PsAMRIS appeared most thoroughly assessed methodologically and were tested in a pilot phase where the 2 TS scores were assessed at wrist and MCP joints by 2 readers at 2 timepoints8,9. The methodology by Haavardsholm, et al seemed more applicable to the wrist area and was therefore chosen for subsequent intrareader and interreader agreement analyses in a pre-exercise with 80 hands at 2 timepoints (baseline/1-year followup), in which the exact thickness of TS was also measured. The results were presented and discussed at a subsequent meeting of the Working Group, and modifications were implemented based on tendon sheath thickness distributions (Supplementary Figure 1, available with the online version of this article). The group then proceeded to perform the multireader exercise below.
Image selection
MR image sets were selected from a study13 in which patients with active RA (disease activity score > 3.2, rheumatoid factor-and/or anticyclic citrullinated peptide–positive, ≥ 1 previous disease-modifying antirheumatic drug) were treated with 2–3 doses of 1000 mg rituximab at baseline. MRI of the dominant hand was performed on a Siemens 1.5 Tesla MRI unit using a dedicated hand coil. Axial T1-weighted fat-saturated precontrast and postcontrast gradient echo MR image sets (slice thickness 0.45 mm, repetition time 30 ms, echo time 6.8 ms, field of view 150 mm) were obtained at baseline (n = 43) and after 14 (n = 5), 26 (n = 8), 38 (n = 15), or 52 (n = 15) weeks.
Scoring of images
Four readers (DG, FG, MØ, PB) with previous experience in MRI-assessment of TS participated in the exercise. The readers performed a calibration session the evening before the exercise. Forty-three paired MR image sets were blinded for patient data but not for chronology14, and were read twice on identical 23-inch screens over 2 days with reanonymizing and rerandomizing between the 2 reads.
Reader rules and scoring system
At the wrist, 6 extensor tendon compartments and 3 flexor tendon compartments were assessed between the radioulnar joint and the hook of hamate. At the level of the second to the fifth MCP joints, flexor tendons were assessed in an area from 1 cm proximal to 1 cm distal to each joint (Figure 1A).
Scoring sheet for the TS score. At the wrist, 6 extensor tendon compartments and 3 flexor tendon compartments are assessed, and at the level of the second to the fifth metacarpophalangeal (MCP) joints, the flexor tendons are assessed. The range of the scores is 0–27, 0–12, and 0–39 for the tendons at the wrist, MCP joints, and total score, respectively (A). Measuring the common tendon sheath perpendicularly from an individual tendon (B) will result in overestimation of the thickness of the tendon sheath. Therefore, the tendons within a common tendon sheath are assessed as 1 unit (C), illustrated by the dashed lines. The enhancing tendon sheath is measured perpendicularly to the unit at the thickest point. FCR: flexor carpi radialis; FPL: flexor pollicis longus.
TS was defined as peritendinous effusion and/or postcontrast enhancement of the tendon sheath seen on axial sequences over ≥ 3 consecutive slices. The maximum width of the effusion and/or enhancing tendon sheath was measured perpendicularly to the tendon, and the TS score was graded as follows: 0: No; 1: < 1.5 mm; 2: ≥ 1.5 mm but < 3 mm; 3: ≥ 3 mm peritendinous effusion and/or postcontrast enhancement (Figure 1A). Tendon sheaths of crossing tendons were measured proximally to the crossing point, and tendon bundles within a common tendon sheath were assessed as 1 unit (Figure 1B–C). Supplementary Figure 2 (available with the online version of this article) shows examples of TS.
Statistical analysis
Descriptive statistics and the Wilcoxon signed-rank test were used to assess changes in score over time.
Intrareader and interreader agreement were assessed using single and average measure intraclass correlation coefficient (smICC/avmICC), respectively. The smallest detectable change (SDC) was calculated for intrareader and interreader change scores15 and was also expressed as the percentage of maximum observed change (%MOC). The percentage of exact agreement (PEA, scores equal) and the percentages of close agreement (PCA, scores ≤ 1 different) between the 2 reads and the 4 readers were calculated for intrareader and interreader agreement, respectively.
Responsiveness was assessed by the standardized response mean (SRM). Ability to show change was also assessed by the percentage of patients with change score ≥ SDC.
RESULTS
Baseline and followup characteristics of the patients are presented in Supplementary Table 1 (available with the online version of this article). Median (range) change in total TS score was −1.0 (−1.0 to −2.5; p < 0.01).
Intrareader smICC for baseline scores were very good for all measures in all readers, except for MCP scores in 1 reader (smICC 0.79). Intrareader smICC for change scores were good to very good for all measures. All readers demonstrated very good intrareader smICC for total scores. Baseline/change interreader avmICC were > 0.90 (i.e., very good) for all measures. Median (range) intrareader SDC was 2.8 (2.1–3.0) for total scores. Interreader SDC for scores averaged over 4 readers were < 2.0 for wrist, MCP (< 1.0), and total scores. The percentage of patients with a total change score ≥ SDC was 39.5% to 54.7% for intrareader SDC. For interreader SDC, this percentage was 60.5%. The %MOC was below 20% for intrareader and interreader total scores (Table 1).
Intrareader and interreader agreement of the TS score.
The overall PEA/PCA intrareader and interreader agreements for change scores in all tendons were 73.8%/97.6% and 47.9%/85.0%, respectively (Table 1). PEA/PCA for individual tendons are presented in Supplementary Table 2 (available with the online version of this article). The inter-reader PEA/PCA for pairs of 2 readers were 72.4%/99.6% (baseline) and 73.3%/98.8% (change) for the best matched pair and 64.3%/98.3% (baseline) and 67.5%/97.0% (change) for the average pair when all tendons were considered.
The average SRM for total scores was moderate (Table 2). Overall, there were no differences in ICC, %MOC, PEA/PCA, and SRM between tendon sheaths at wrist versus MCP joint levels.
Responsiveness of the TS score.
DISCUSSION
This longitudinal multireader exercise of patients with active RA showed that the TS score had high intrareader and inter-reader agreement and moderate responsiveness. The results were similar at wrist and MCP joint level.
Intrareader and interreader ICC were very good, both for total baseline and change scores, indicating that the TS score is reliable and can monitor change over time.
The average intrareader PEA was high and the PCA was close to 100%. Interreader PEA was acceptable considering the increased difficulty of 4 assessors needing to reach exact agreement, as compared to 2 assessors. Because most clinical trials have 2 assessors, we also analyzed the PEA/PCA for paired readers, and found the average percentages close to the intrareader agreements and the percentages of the best paired assessors in line with the intrareader agreements.
The interreader SDC was ≤ 2.0 and the %MOC < 11%. Although the average SRM was only 0.50 (moderate), > 60% of the patients showed a change ≥ SDC. Variable treatment regimens (related to different timing of infusions) and different timing of followup MRI may have affected the responsiveness but would not be expected to alter the intrareader and interreader agreement. Future studies should assess the responsiveness in early RA cohorts and in placebo-controlled studies, where responsiveness can be compared between groups.
This TS score was developed as a modification from the score by Haavardsholm, et al after assessing this and the PsAMRIS TS score in a pilot phase8,9,10,11. A key modification of the methodology by Haavardsholm, et al was to narrow the intervals of tendon sheath thickness within each increment to potentially increase the ability to detect change. As proven by this exercise, the reliability of the current TS score remained high. The responsiveness was not increased compared to the results reported by Haavardsholm, et al9, but because therapies in the 2 cohorts were different, the results are not fully comparable. Other advances, compared to the score by Haavardsholm, et al, included clarifications on how to score tendon sheaths of crossing tendons and common tendon sheaths of tendon bundles (Figure 1B–C).
Other MRI TS scores have previously been used. Several have not assessed reliability and responsiveness1,16 or have scored TS qualitatively as absent/present2,4,17,18,19,20,21,22. Regarding the monitoring of change, qualitative scores may have less power to detect changes in TS. Semiquantitative TS scores have been suggested by McQueen, et al23 and Schirmer, et al24, but did not assess the performance in longitudinal settings nor the intrareader and interreader agreement. Lisbona, et al25 suggested a TS score for the hand (based on incomplete and complete halos of enhancing tendon sheath) that showed high intrareader and interreader ICC, but small SRM.
This TS score was developed as a potential addendum to the existing RAMRIS6,26. Therefore, we strived to design the score so that it covered the tendons at the joint regions included in the RAMRIS core set. Because TS is scored on the same MRI sequences and projections as synovitis, these pathologies may be scored simultaneously and therefore the addition of TS score will add only a small amount of time to assessing a hand when using the RAMRIS method. Based on this and the reliability and responsiveness data, we conclude that this TS score fulfills the OMERACT filter criteria concerning truth, discrimination, and feasibility27 and may be included as a component of the RAMRIS for assessing TS of the hand in RA clinical trials.
ONLINE SUPPLEMENT
Supplementary material accompanies the online version of this article.
Acknowledgment
We thank illustrator Axel Norén for help with Figure 1, and the Danish Rheumatism Association.
Footnotes
Supported by the Danish Rheumatism Association. The clinical trial from which these data were acquired was supported by a research grant and study drug from Roche. The research is supported by the NIHR Leeds Musculoskeletal Biomedical Research Unit. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR, or the UK Department of Health. Dr. Gandjbakhch has received consultation fees from Pfizer, AbbVie, Roche, and Janssen, and research support and grants from Pfizer and Roche. Dr. Haavardsholm has received consultation fees from Pfizer and MSD, and research support and grants from AbbVie, UCB, MSD, Pfizer, and Roche. Dr. Peterfy is a shareholder in Spire Sciences Inc. E.M. Vital has received research support and grants from Roche and AstraZeneca. P.G. Conaghan has been a consultant or on the speakers’ bureau for Abbvie, Lilly, Novartis, Pfizer, and Roche, and has received a research grant from BMS. Dr. Østergaard has received consultation fees from Abbvie, BMS, Boehringer-Ingelheim, Celgene, Eli-Lilly, Centocor, GSK, Hospira, Janssen, Merck, Mundipharma, Novartis, Novo, Orion, Pfizer, Regeneron, Schering-Plough, Roche, Takeda, UCB, and Wyeth, and has received research support and grants from Abbvie, BMS, Janssen, and Merck.
- Accepted for publication March 23, 2017.