Abstract
Objective. A pilot study testing novel ultrasound (US) joint-selection methods in rheumatoid arthritis.
Methods. Responsiveness of novel [individualized US (IUS) and individualized composite US (ICUS)] methods were compared with existing US methods and the Disease Activity Score at 28 joints (DAS28) for 12 patients followed for 3 months. IUS selected up to 7 and 12 most ultrasonographically inflamed joints, while ICUS additionally incorporated clinically symptomatic joints.
Results. The existing, IUS, and ICUS methods’ standardized response means were −0.39, −1.08, and −1.11, respectively, for 7 joints; −0.49, −1.00, and −1.16, respectively, for 12 joints; and −0.94 for DAS28.
Conclusion. Novel methods effectively demonstrate inflammatory improvement when compared with existing methods and DAS28.
Ultrasonography (US) is increasingly used to monitor joint inflammation in rheumatoid arthritis (RA)1. The optimal method of selecting reduced joint counts for US monitoring is not established. Various criteria have been used for reduced joint selection (e.g., frequency of involvement, feasibility, representativeness of joints, logistic regression models)2. Scanning fewer joints saves time3 and yet can be representative of the inflammatory changes using extended scanning4. A systematic review studying US joint counts in RA2 highlighted 2 existing methods5,6 with good validity: 1 using a predefined 7-joint count5 and another using a 12-joint count derived out of using US reduction from the frequency of inflammatory involvement from 44 joints6. Because the extent and distribution of affected joints differ between individuals, such methods do not ensure selection of the most affected joints or the greatest number of affected joints per individual for US monitoring.
In our pilot study, novel individualized joint selection methods are designed to improve the number of affected joints per patient for US scanning (limited by a target joint number). The novel individualized US (IUS) method selects up to 7 and 12 most inflamed joints detected on US, while the novel individualized composite US (ICUS) method additionally incorporates clinically symptomatic joints. The key feature is joint selection dictated by severity, beginning with the most severely affected joint and continuing until the target count is reached. Target joint limits were set at 7 and 12 to maintain consistency with joint counts specified by existing methods5,6. We hypothesized that the novel methods would improve experimental efficacy, resulting in greater sensitivity for detecting change when compared with existing methods and the Disease Activity Score at 28 joints (DAS28).
MATERIALS AND METHODS
Patients
Seropositive patients with RA (DAS28 > 3.2) with ≥ 5 tender and/or swollen joints starting or escalating disease-modifying antirheumatic drugs (DMARD) and corticosteroid therapy in the rheumatology unit at the Singapore General Hospital were enrolled from March 2013 to May 2014 and followed up for 3 months. The local institutional review board approved this pilot study. Patients gave written informed consent. Patients were excluded if they had connective tissue diseases, other inflammatory arthritides, pregnancy, Hepatitis B/C, previous joint replacements, or a limb amputation.
At 0 and 3 months, a 44-joint clinical assessment (by a metrologist) and US were performed on the same day. A rheumatologist experienced in musculoskeletal US (blinded to the metrologist’s findings) acquired and scored the US images.
Clinical assessment
Joints assessed for tenderness and pain included bilateral shoulders, elbows, wrists, metacarpophalangeal joints (MCPJ) 1–5, proximal interphalangeal joints 2–5, thumb interphalangeal joint, hips, knees, ankles, midtarsal, and metatarsophalangeal joints 1–5. Of these 44 joints, 40 were assessed for swelling (hips and midtarsals excluded using the approach of DAS447). Joint tenderness and swelling were scored as 1 = yes and 0 = no. Joint pain was scored as 0 = none, 1 = mild, 2 = moderate, and 3 = severe. These scores form the clinical subscores used in the ICUS method (further described under the section “Individualized methods”).
US evaluation
US was performed with General Electric Healthcare LOGIQe machine with a multifrequency linear array transducer (5–13 MHz) or Philips Medical Systems EPIQ 5G machine with a multifrequency linear array transducer (5–17 MHz). US outpatient facility, machine, probe, and settings were kept the same for each patient. Standardized scanning was based on the European League Against Rheumatism (EULAR) guidelines8. The joint/tendon sites scanned are listed in Table 1. US pathology definitions from the EULAR OMERACT (Outcome Measures in Rheumatology) US workgroup were used1. Greyscale synovial hypertrophy (SH) and power Doppler (PD) vascularity were scored semiquantitatively (0 = none, 1 = mild, 2 = moderate, 3 = severe) based on US scoring definitions used by Backhaus, et al5. This method was used at the hand and feet joints (because these joints were included in the study by Backhaus, et al) and extrapolated for use in other joints (including the elbow, knee, ankle, midtarsal, shoulder, and hip). However, at certain medium-sized joints (e.g., elbow, knee, and ankle) whereby relevant images from a scoring atlas9 are available, semiquantitative scoring (0 to 3) was performed based on the relevant images from the scoring atlas. The shoulder and hip joints were not included in the original study by Backhaus, et al or the atlas, so where applicable, a score of 0 (normal) was taken to be greyscale SH less than the mean plus 2 SD of normal range10. Tenosynovitis was scored as 1 = yes and 0 = no for greyscale and PD findings. These scores form the US subscores used in the existing and individualized methods (further described under the sections “Existing methods” and “Individualized methods”).
Joint and tendon sites used in the individualized and existing methods.
Existing methods
The existing methods include the 7 and 12 joints used by Backhaus, et al5 and Naredo, et al6, respectively (Table 1). The individual joint score (IJS) per joint was calculated as the sum of the US subscores divided by the maximum possible score at that joint, so as to equalize score weights across the joints. The maximum number of affected joints selected by the existing 7- and 12-joint methods are therefore 7 and 12, respectively.
Individualized methods
In the 7-joint approach, the individualized methods selected up to a maximum of 7 most affected joints for monitoring. In the 12-joint approach, the individualized methods selected up to a maximum of 12 most affected joints for monitoring. For the IUS method, the IJS at each joint was calculated as the sum of the US subscores divided by the maximum possible score at that joint. For the ICUS method, the IJS at each joint was calculated as the sum of the clinical and US subscores divided by the maximum possible score at that joint. An affected joint for the existing and individualized methods was defined as IJS > 0.
The joint selection process (Figure 1) was as follows: the IJS from the 44 joints were ranked from largest to smallest score. The target joint count was set at 7 and 12. Joints with the highest IJS were identified. Joint selection progressed from the small to medium and larger joints as described in Figure 1. This process was repeated using joints with decreasing IJS until the target joint count was reached. The rationale for this joint selection process was (1) RA frequently involves the small joints, (2) scanning small joints is often easier, and (3) the semiquantitative scoring method was developed using smaller joints11.
Individualized methods of the joint selection process.
Statistical analysis
For the above methods, the IJS from the selected joints were summed to obtain a total inflammatory score (TIS) per patient. Patients’ mean TIS were calculated at 0 and 3 months and were used for deriving the standardized response mean (SRM), calculated as the mean change in the TIS score divided by the SD of the change in the TIS score. The threshold values from Cohen for effect size (ES) are often used for interpretation, i.e., trivial [ES < 0.20, small (0.20 ≤ ES < 0.50), moderate (0.50 ≤ ES < 0.80), and large (ES ≥ 0.80)]12,13,14,15. The average number of affected joints at baseline by these methods was reported. The SRM (at 3 mos) were calculated for these methods and the DAS28.
RESULTS
Patient characteristics
Twelve patients with RA [mean (SD) age 57.6 (6.5) yrs, 83.3% women, 83.3% Chinese, 8.3% Indian, and 8.3% other ethnic groups] with mean (SD) disease duration at baseline of 55.8 (71.2) months completed the study. All patients were started or escalated with DMARD and corticosteroid therapy prior to the baseline US scans. Within 3 months prior to recruitment, 8 patients (66.7%) were receiving oral DMARD (which included methotrexate, sulfasalazine, hydroxychloroquine, and azathioprine) while 11 patients (91.7%) were receiving prednisolone. The mean DAS28 at baseline and 3 months was 5.21 and 4.32, respectively. The SRM for DAS28 was −0.94.
Seven-joint and 12-joint approaches
Using 7 joints, the affected joints (average number) for the existing, IUS, and ICUS methods were 3, 7, and 7, respectively, with corresponding SRM of −0.39, −1.08, and −1.11.
Using 12 joints, the affected joints (average number) for the existing, IUS, and ICUS methods were 7, 11, and 12, respectively. The SRM were −0.49, −1.00, and −1.16, respectively (Table 2).
Results of the individualized and existing methods.
DISCUSSION
Our novel methods using a reduced joint number effectively demonstrate inflammatory improvement when compared with existing methods and the DAS28. This is unsurprising, given our emphasis on selecting affected joints (i.e., selecting as many affected joints, beginning with the most severely affected ones, until the target joint limit is reached) for followup scanning.
The SRM of the existing methods was consistent with the SRM reported in 2 RA studies12,13 using US monitoring. The SRM was −0.2595 at 5 months in 1 study12 using US of bilateral MCPJ 1–5 and −0.46 at 3 months in another study13 using US at the dominant wrist.
Our individualized methods required a 44-joint US assessment at baseline to select the target joints. While this means additional time at the baseline scan, followup scans require less time when compared with the baseline scan because only the selected joints need rescanning.
Our pilot results will need to be confirmed in larger cohorts. Future studies incorporating control group(s) for comparison, as well as correlation with other patient outcomes (e.g., disease remission, structural alteration, functional prognosis) would be necessary. The number of reduced joints to assess during followup could also be explored in larger studies.
Acknowledgment
We thank our colleagues from the Singapore General Hospital (SGH), Department of Rheumatology and Immunology, who referred patients for our study and also the staff of the Autoimmunity and Rheumatology Centre of the SGH where this study was performed. We would also like to acknowledge funding support from the Estate of Tan Sri Khoo Teck Puat (Khoo Clinical Scholars Pilot Award).
Footnotes
Grant support from the Khoo Clinical Scholars Pilot Award.
- Accepted for publication September 30, 2015.
