To investigate the validity, reproducibility, and responsiveness of a simplified power Doppler ultrasound (PDUS) assessment of joint inflammation compared with a comprehensive 44-joint PDUS assessment in patients with rheumatoid arthritis (RA) who started therapy with a biologic agent.
A total of 160 patients with active RA who started a biologic agent were prospectively recruited in 18 Spanish centers. The patients underwent clinical and laboratory assessment and blinded PDUS examination at baseline and 6 months. A PDUS examination of 128 synovial sites in 44 joints was performed. US synovitis and PD signal were semiquantitatively graded from 1 to 3 in all synovial sites. US count and index for synovitis and PD signal were obtained. PDUS intraobserver and interobserver reliability were evaluated. A process of data reduction based on the frequency of involvement of synovial sites by both synovitis and PD signal was conducted. Construct and discriminant validity of a simplified PDUS assessment was investigated.
A PDUS simplified assessment including 24 synovial sites from 12 joints detected 100% of patients with synovitis and 91% of patients with PD signal. There was a highly significant correlation between the 44-joint count and index for synovitis and PD signal and the 12-joint count and index for synovitis and PD signal at baseline and 6 months (r = 0.84–0.90, P < 0.0005). The smallest detectable difference was lower than the mean change in simplified PDUS variables.
A 12-joint PDUS assessment of RA joint inflammation may be a valid, feasible method for multicenter monitoring of therapeutic response to biologic agents.
For daily practice and clinical trials, the development of new, valid methods for assessing joint inflammation caused by rheumatoid arthritis (RA) is challenging but absolutely essential. Within the last decade, there has been an increasing number of studies on the criterion and construct validity of ultrasonography (US) with color Doppler or power Doppler (PD) techniques for evaluating inflammatory activity in selected joints affected by RA compared with histology (1–3), magnetic resonance imaging (4, 5), clinical and laboratory parameters (6–15), and radiographic outcome (11, 16). Doppler US is a noninvasive, relatively inexpensive bedside imaging modality that facilitates the scanning of all peripheral joints as many times as required at the time of consultation.
Synovitis can be detected in different synovial recesses accessible by US evaluation in peripheral joints (17). A comprehensive US assessment including multiple recesses of all accessible peripheral joints may be time consuming for the patient and doctor in daily practice and clinical trials. However, there is not enough evidence showing which joints and synovial recesses are better for detecting US synovitis. Therefore, for a feasible overall activity assessment of rheumatoid joints, it is necessary to determine which joints and which synovial sites should be evaluated.
The study by Scheel et al (18) was the first to determine which combination of synovial recesses of finger joints provides the most information on RA synovitis. Since then, there have been some cross-sectional data on simplified systems that correlated well with a comprehensive PDUS assessment of RA joint inflammation (19). However, to the best of our knowledge, there are no longitudinal studies demonstrating the validity, reliability, and sensitivity to change of a simplified US assessment compared with a comprehensive US assessment of joint inflammation in RA.
The purpose of the present study was to investigate the validity, reproducibility, responsiveness, and feasibility of a simplified PDUS assessment of joint inflammation as compared with a comprehensive 44-joint PDUS assessment in patients with RA who started treatment with a biologic agent.
PATIENTS AND METHODS
This study prospectively included 160 patients (125 women, 35 men) with RA according to the 1987 American College of Rheumatology (formerly the American Rheumatism Association) criteria for RA (20) who were recruited in the outpatient rheumatologic clinic of 18 Spanish centers from June 2005 to December 2006 and who started therapy with a biologic agent according to Spanish and international consenses on the use of biologic agents for the treatment of RA (21, 22). The mean ± SD age was 54.1 ± 12.4 years (range 19–84 years) and the mean ± SD disease duration was 111.9 ± 88.6 months (range 4–552 months). The study was conducted in accordance with the Declaration of Helsinki and was approved by the local ethics committees. Informed consent was obtained from all patients before study entry.
The patients underwent a clinical, laboratory, and PDUS evaluation at baseline (within 1 week before starting biologic therapy) and 6 months. Therapeutic decisions were made throughout the followup period depending on the RA clinical course without knowledge of the PDUS findings.
Clinical and laboratory assessment.
Clinical evaluation was performed by the same rheumatologist at each center, who was blinded to the PDUS findings. The following data were recorded for each patient at study entry: age; sex; symptom duration; nonsteroidal antiinflammatory drugs (NSAIDs), corticosteroids, disease-modifying antirheumatic drugs (DMARDs), and biologic agents received for RA before study entry; and presence of rheumatoid factor. Therapy received for RA from baseline was recorded at 6 months.
At baseline and 6 months, 28 joints were assessed for tenderness and swelling (23). Scores on a global pain intensity visual analog scale (0–100 mm) and a visual analog scale for the patient's overall assessment of disease activity (0–100 mm) were recorded. Erythrocyte sedimentation rate was also recorded from each patient's laboratory test obtained within 48 hours of each visit. Disease activity was estimated by calculating the 28-joint Disease Activity Score (DAS28) for all patients at baseline and 6 months (24).
Each patient underwent a PDUS assessment within 4 hours of each clinical evaluation by the same rheumatologist experienced in US (1 rheumatologist in 15 centers, 2 rheumatologists in 2 centers, and 3 rheumatologists in 1 center) who was blinded to the clinical and laboratory findings. The patients were asked not to talk about their clinical symptoms to the US examiner. The PDUS examination was carried out in a darkened room.
A systematic multiplanar gray-scale and PD examination of 44 joints was performed with the same real-time scanner in all centers (Logiq 5 PRO; General Electric Healthcare, Kyunnggi-do, Korea) using multifrequency linear array transducers (7–12 MHz). The PDUS assessment included bilateral shoulder, elbow, wrist, metacarpophalangeal (MCP), proximal interphalangeal (PIP) of the hands, hip, knee, ankle, tarsal, and metatarsophalangeal joints. The presence of synovitis and PD signal were investigated in different intraarticular and periarticular synovial anatomic sites at each joint. A total of 128 synovial sites were scanned in all patients at each visit (Table 1).
Table 1. Synovial intraarticular recesses and periarticular sites evaluated at each joint by power Doppler ultrasonography*
US scanning technique and pathology definitions were standardized among investigators prior to the study. This scanning method has been described in previous published studies (13,16,18,25–33). Synovitis was defined as the presence of abnormal hypoechoic material within joint recesses, tendon sheaths, or bursae according to the Outcome Measures in Rheumatology Clinical Trials (OMERACT) definitions and published descriptions (13, 16, 25, 27, 28, 34, 35). Intraarticular synovitis, tenosynovitis, and bursitis were graded semiquantitatively from 0 to 3 (where 0 = absence, 1 = mild, 2 = moderate, and 3 = marked) during the US examination.
Synovial blood flow was evaluated by PD in each of the intraarticular and periarticular synovial sites. Gray-scale and PD parameters were standardized among investigators. Pulse repetition frequency (PRF) was adjusted at the lowest permissible value to maximize sensitivity. This setting resulted in PRF from 500 Hz to 750 Hz depending on the anatomic area scanned. Flow was additionally demonstrated in 2 planes and confirmed by pulsed wave Doppler spectrum to exclude artifacts.
The intraarticular, tenosynovial, and intrabursal PD signals were graded on a semiquantitative scale from 0 to 3 (where 0 = absence, no synovial flow; 1 = mild, ≤3 isolated signals; 2 = moderate, >3 isolated signals or confluent signal in less than half of the synovial area; and 3 = marked, signals in more than half of the synovial area) during the US examination. A US count for joints with synovitis (44-USCS) and a US count for joints with PD signal (44-USCPD) in any synovial site were obtained at each visit.
Each joint was scored for synovitis and PD signal on a scale from 0 to 3. These scores corresponded to the maximum score for synovitis and PD signal, respectively, obtained from any of the synovial sites evaluated at each joint.
An overall US index for joints with synovitis (44-USIS) and an overall US index for joints with PD signal (44-USIPD; the sum of the synovitis and PD signal scores, respectively, obtained from each joint) were calculated at each US assessment.
The time spent on the PDUS examinations was recorded by the investigators. A representative image of PDUS findings is shown in Figure 1.
PDUS intraobserver reliability.
PDUS intraobserver reliability was assessed by recording representative images of the full baseline examination of the 160 patients included in the study. The stored images of each patient were blindly scored for synovitis and PD signal by the same investigator who performed the corresponding real-time PDUS examination a minimum of 3 months later.
PDUS interobserver reliability.
Interobserver reliability between US investigators was evaluated before patients' inclusion by scoring for synovitis and PD signal in 20 recorded images of the joints included in the PDUS assessment from 20 patients with active RA, randomly chosen by the investigator who coordinated the study (EN).
Statistical analysis was performed using SPSS statistical software, version 13.0 (SPSS, Chicago, IL). Quantitative variables (DAS28, PDUS parameters) were given as the mean ± SD and range.
As a first step, we undertook a process of data reduction based on the frequency of joint involvement by both synovitis and PD signal at baseline. The final visit was not used because there were fewer patients with PDUS pathologic findings than at baseline. A reduced PDUS assessment was selected from different joint combinations.
Once a reduced model was selected, we determined which synovial sites at each joint selected in the reduced model allowed us to detect >90% of the joints involved by synovitis and by PD signal. The final selected model was called simplified PDUS assessment. Afterwards, each investigator was asked to record the time taken to perform a simplified PDUS examination in 2 consecutive patients with active RA (DAS28 >3.2) attended at their outpatient rheumatologic clinic.
Validity of the simplified PDUS assessment was assessed for 2 characteristics. First, content and construct validity were evaluated by correlating the 44-joint PDUS count and index for synovitis and PD signal with the reduced and simplified PDUS parameters using Pearson's rank correlation coefficient. Correlations between the 44-joint PDUS, reduced PDUS, simplified PDUS variables, and DAS28 were obtained by Pearson's correlation coefficient.
Second, discriminant validity was investigated as follows. Intraobserver reliability was tested for the 44-joint PDUS assessment and the simplified PDUS assessment by calculating the intraclass correlation coefficient (ICC; 2-way mixed effects) for joint count and index for synovitis and PD signal. The smallest detectable difference (SDD) was obtained from agreement between the 2 assessments, 3 months apart, for the same images (36). Sensitivity to change of the PDUS variables was tested by comparing the SDD and the mean change in PDUS parameters from baseline to 6 months.
Interobserver reliability was evaluated by Kendall's W. A Kendall's W value <0.40 was considered poor, 0.40–0.50 moderate, 0.50–0.70 good, and 0.70–1 excellent.
Feasibility of the simplified PDUS assessment was estimated by comparing the time spent on the 44-joint PDUS examination and the simplified PDUS examination by the independent-samples t-test. A P value less than 0.05 was considered statistically significant.
Rheumatoid factor was positive in 116 patients (72.5%), negative in 40 patients (25%), and not available in 4 patients (2.5%). Before study entry, 25 patients (15.6%) had received 1 DMARD and 135 patients (84.4%) had received >1 DMARD. Twenty-three patients (17.5%) had received 1 previous biologic agent and 5 patients (3.1%) had received 2 previous biologic agents. These patients had switched to another biologic agent mainly because of ineffectiveness (60%) or adverse effects (8%).
At inclusion, 99 patients (61.9%) were taking methotrexate, 53 (33.1%) were taking leflunomide, and 8 (15%) were taking other DMARDs. Thirty-two patients (20%) were taking >1 DMARD. A total of 121 patients (75.6%) received prednisone and 135 (84.4%) received NSAIDs. A total of 123 patients (76.9%) started therapy with adalimumab, 18 (11.3%) started with infliximab, 18 (11.3%) started with etanercept, and 1 (0.6%) started with anakinra.
At 6 months, data were available for 133 of the 160 patients included in the study. After the 6-month followup, 127 patients received biologic therapy.
Disease activity and overall PDUS variables.
At study entry, the mean ± SD DAS28 score was 5.9 ± 1.1 (range 4–8.2). At baseline, the mean ± SD 44-USCS was 16.4 ± 10.9 (range 1–42), mean ± SD 44-USIS was 24.6 ± 19 (range 1–119), mean ± SD 44-USCPD was 7.8 ± 9.5 (range 0–40), and mean ± SD 44-USIPD was 9.6 ± 11.5 (range 0–57). Seventeen patients did not show PD signal in any location at baseline.
At 6 months, the mean ± SD DAS28 score was 4.1 ± 1.4 (range 0.1–7.5). At 6 months, the mean ± SD 44-USCS was 8.8 ± 7.8 (range 0–34), mean ± SD 44-USIS was 11.4 ± 11.4 (range 0–72), mean ± SD 44-USCPD was 3.1 ± 9.5 (range 0–26), and mean ± SD 44-USIPD was 3.3 ± 5.5 (range 0–34). At the final visit, synovitis was not detected in 5 patients and PD signal was not found in 52 patients.
Reduced PDUS assessment.
Different models of reduced PDUS assessment based on the most frequently involved joints at baseline are shown in Table 2. We selected a 12-joint model including bilateral elbow, wrist, second and third MCP, knee, and ankle joints for subsequent analysis. This reduced 12-joint PDUS assessment was able to detect 100% of patients with 44-USCS/44-USIS >0 and 94.4% of patients with 44-USCPD/44-USIPD >0. A reduced count for joints with synovitis (12-USCS) and PD signal (12-USCPD) along with a reduced index for synovitis (12-USIS) and PD signal (12-USIPD) were further calculated.
Table 2. Sensitivity for detecting synovitis and power Doppler signal of the reduced assessment models obtained from the most frequently involved joints at baseline*
PDUS = power Doppler ultrasound; USCS = ultrasonographic count for joints with synovitis; USIS = ultrasonographic index for joints with synovitis; USCPD = ultrasonographic count for joints with power Doppler signal; USIPD = ultrasonographic index for joints with power Doppler signal; WR = wrist; MCP = metacarpophalangeal; KN = knee; ANK = ankle; ELB = elbow; MTP = metatarsophalangeal.
We obtained a simplified 12-joint PDUS model (12S-joint PDUS) that consisted of a total of 24 synovial sites from the 12 selected joints (Table 3). This 12S-joint PDUS assessment detected 100% of patients with 44-USCS/44-USIS >0 and 90.9% of patients with 44-USCPD/44-USIPD >0. Then, a simplified count for joints with synovitis (12S-USCS) and PD signal (12S-USCPD) along with a simplified index for synovitis (12S-USIS) and PD signal (12S-USIPD) were obtained by the same method used for scoring the 44 joints.
Table 3. Sensitivity for detecting synovitis and power Doppler (PD) signal of the combination of synovial sites selected for the simplified ultrasonographic power Doppler assessment at each joint*
Synovial sites (bilateral)
Detected joints with synovitis, %
Detected joints with PD signal, %
Synovial sites scanned, no.
MCP = metacarpophalangeal.
Dorsal carpal recess
Lateral parapatellar recess
Anterior tibiotalar recess
Medial tendon sheaths
Lateral tendon sheaths
Correlation between the 44-joint PDUS assessment and the 12-joint and 12S-joint PDUS assessments.
The correlations between the 44-joint PDUS variables and the 12-joint and 12S-joint PDUS variables are shown in Table 4. There was a highly significant correlation between the 44-joint count and index for synovitis and PD signal and the corresponding 12-joint and 12S-joint counts and indices for synovitis and PD signal, both at baseline and 6 months.
Table 4. Correlations between the 44-joint power Doppler ultrasound (PDUS) parameters, 12-joint PDUS parameters, 12S-joint parameters, and DAS28 at baseline and 6 months*
Correlation between PDUS parameters and disease activity (DAS28).
All 44-joint, 12-joint, and 12S-joint PDUS variables demonstrated a significant correlation with the DAS28 at baseline and 6 months (Table 4). The differences between the correlation coefficients of the DAS28 with the 44-joint PDUS parameters, 12-joint PDUS parameters, and 12S-joint parameters at baseline and 6 months were not significant (Table 4).
Discriminant validity of simplified PDUS assessment intraobserver reliability and sensitivity to change.
Intraobserver reliability could be evaluated in 156 patients from all centers. The images from the baseline PDUS examination of 4 patients were not available. ICCs for the 44-joint PDUS, 12-joint PDUS, and 12S-joint PDUS variables are displayed in Table 5. The mean decrease in PDUS parameters is shown in Table 5. The SDD was lower than the mean change in all comprehensive, reduced, and simplified PDUS variables (Table 5).
Table 5. Intraobserver reliability and sensitivity to change of the comprehensive, reduced, and simplified power Doppler ultrasound parameters*
Intraobserver ICC (95% CI)
Mean decrease (95% CI)
ICC = intraclass correlation coefficient; 95% CI = 95% confidence interval; SDD = smallest detectable difference; S = simplified; see Table 2 for additional definitions.
Interobserver agreement was significant for both synovitis and PD signal (P < 0.0005). Kendall's W coefficient was 0.5 for synovitis and 0.8 for PD signal.
There was a significant difference between the mean time spent on the 44-joint PDUS examination (83.6 ± 25.8 minutes, range 47.5–120 minutes) and the mean time spent on the 12S-joint PDUS examination (22 ± 6.2 minutes, range 12.5–35 minutes; P < 0.0005).
Within the last decade, there has been an increasing use of musculoskeletal US in the diagnosis and monitoring of joint inflammatory activity in patients with RA (37–39) based mainly on the fact that it has a higher sensitivity for detecting synovitis than clinical examination (32,33,40,41). Several methods for evaluating RA joint inflammation have been described in previous studies conducted by individual research centers. Qualitative (4), semiquantitative (1–3, 8, 12–16, 41), and quantitative systems (2, 3, 5–11) have been used for assessing synovitis in any number of scanned joints ranging from 60 joints (13) to a reduced number of target RA joints such as wrist, hand, or toe joints (4–8, 11, 12). In addition, previous studies have not taken into account all of the synovial recesses accessible by US evaluation at each joint region. These variables have made comparison of study results difficult.
For this study, a large cohort of patients with RA who started biologic agents, which have been widely demonstrated to be effective (42–46), was investigated longitudinally. We chose a combination of gray-scale and PD multiplanar semiquantitative assessment of a large number of RA joints (44 joints, 128 synovial sites) as a comprehensive method for overall inflammatory activity. Other joints such as the acromioclavicular, sternoclavicular, PIP of the feet, and temporomandibular joints were not included because of the paucity of published data on US assessment of these joints in patients with RA. We also evaluated inflammatory PDUS findings in periarticular tendon sheaths and bursae in order to increase the comprehensiveness of the synovial inflammation assessment.
Although there were various joint combinations that yielded high sensitivity for detecting patients with synovitis and PD signal, we selected bilateral elbow, wrist, second and third MCP, knee, and ankle joints to maximize comprehensiveness and feasibility.
The selected count and index for synovitis and PD signal from 24 synovial sites at 12 joints showed a highly significant correlation with the comprehensive PDUS variables at baseline and after 6 months of biologic therapy. Thus, the simplified PDUS assessment of RA joint inflammation hereby developed has demonstrated content and construct validity.
There were similar significant correlations between the comprehensive and simplified variables and the DAS28 at baseline and 6 months. Other studies using fewer joints have found a weaker correlation between disease activity measured by US techniques and clinical and biochemical markers (5, 7, 8, 10).
Results of the discriminant validity, intraobserver reliability, and sensitivity to change of the simplified PDUS assessment were also comparable with the 44-joint PDUS assessment. We found a high intraobserver reliability. Throughout followup, the changes in the simplified counts and indices were higher than their SDD.
Despite the high number of US examiners, the interobserver reliability was good for synovitis and excellent for PD signal. It is likely that the strict standardization of scanning method and pathology definitions conducted before the study began contributed to reducing interobserver variability between investigators.
Some limitations of our study should be mentioned. With regard to intraobserver and interobserver reliability, image acquisition variability was not investigated. The assessment of selected PDUS images instead of the real-time examination of the joints obviously introduces bias into the study. However, testing intraobserver acquisition reliability would have been problematic for patients because of the time required to complete the comprehensive PDUS examination. Testing interobserver acquisition reliability was not considered due to the number of investigators involved in the study.
In conclusion, our results provide evidence that a simplified 24-recess, 12-joint PDUS assessment of synovitis may be valid, reliable, sensitive to change, and feasible for therapy monitoring in multicenter studies of established RA.
Dr. Naredo had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Analysis and interpretation of data. Naredo, Jesús Garrido.
Manuscript preparation. Naredo, Jesús Garrido.
Statistical analysis. Jesús Garrido.
ROLE OF THE STUDY SPONSOR
Abbott Laboratories granted the sonographic machines used for the study, the coordination, and the statistical analysis. Abbott Laboratories agreed to submit this manuscript for publication and approved its content.