Ultrasound findings on patients with juvenile idiopathic arthritis in clinical remission




To assess whether children with juvenile idiopathic arthritis (JIA) in clinical remission show pathologic findings on either gray-scale or power Doppler ultrasound of their joints.


Children with JIA were eligible if they were in clinical remission for at least 3 months, as defined by the absence of clinically active joints and serologic markers of inflammation. Gray-scale as well as power Doppler ultrasonography of the wrist, knee, and ankle were carried out on previously affected joints and unaffected contralateral joints. Images were read by 2 independent readers. Findings were categorized as 1) structural abnormalities in the case of synovial thickening or increased joint fluid on gray-scale ultrasound or 2) power Doppler positive in the case of an abnormal power Doppler signal.


The study cohort consisted of 28 patients. Eight of 14 patients with previous wrist involvement had pathologic gray-scale findings, and 3 of these 14 patients also had pathologic Doppler findings in the wrist. None of the 20 patients with past knee involvement had pathologic gray-scale or Doppler findings in the knee. Six of 15 patients with previous ankle involvement had pathologic gray-scale findings and 1 of the 15 patients had pathologic Doppler findings in the tibiotalar joint.


This study demonstrates that some patients who meet clinical criteria for remission continue to show ongoing pathology on joint ultrasound, which may be suggestive of persistent inflammation.


The diagnosis of active arthritis in the context of juvenile idiopathic arthritis (JIA) is mostly clinical, both initially as well as in followup. Joints with swelling as well as joints with joint line tenderness or pain on motion and limited range of motion are considered active joints. In daily clinical practice, it is often difficult to assess whether a joint is truly swollen secondary to an arthritic effusion or whether the perceived joint swelling is due to other factors, including subcutaneous fat, soft tissue edema, or paratenonitis. Similarly, pain and limitation of motion in a joint are not always due to synovitis, and this might introduce significant inaccuracies into the diagnosis and assessment of treatment response (1–3).

The exact assessment of joint disease activity has become even more important with advances in treatment over the past years. The induction of permanent remission is possible for an increasing percentage of children but cannot always be reliably demonstrated on clinical examination alone (4, 5). In addition, the exact determination of remission status is also important for the decision to taper medication, thereby preventing side effects from long-term use.

Imaging modalities can be used to complement the clinical examination during joint assessments, and these include radiographs, magnetic resonance imaging (MRI), and ultrasound (US) (6–10). Whereas radiographs are important in the detection and followup of secondary changes of bone (erosions), they are not useful in the detection of joint inflammation. Assessment of synovial inflammatory activity by MRI has shown a close correlation with histologic findings (11). However, MRI is time consuming, expensive, and not widely available for routine clinical use in many countries. In contrast, US is relatively cheap, fast, and virtually free of side effects and would be especially useful for the joint assessment in children. In addition, all peripheral joints can be examined as many times as required at the time of consultation, improving the accuracy of clinical evaluation. Two modalities of US are routinely used for the assessment of joints: gray-scale and Doppler US. Structural abnormalities found on gray-scale US such as joint effusions or synovial thickening could be residual findings and may not necessarily represent ongoing active disease. Both color Doppler and power Doppler US techniques detect synovial blood flow, which is a sign of increased synovial vascularization (12). Doppler US is therefore considered superior in distinguishing active synovitis from inactive intraarticular synovial thickening compared to the gray-scale technique alone (13–18). Adding a US examination to the clinical examination might produce a more accurate assessment of disease activity than the clinical examination alone (1, 2, 19).

Data regarding the US assessment of children with JIA are limited at this point. Available studies have mostly assessed children with active disease (1, 2, 19), and only a minority (19) have used both gray-scale as well as power Doppler assessments. The current study was therefore designed to determine whether patients in clinical remission show either pathologic gray-scale or power Doppler findings on US examination indicating persistent structural abnormalities (gray scale) and/or evidence of active inflammation (pathologic power Doppler signals).

Significance & Innovations

  • In juvenile idiopathic arthritis (JIA), musculoskeletal ultrasound can be an important addition to the clinical examination in the precise determination of remission status.

  • A number of patients with JIA in clinical remission show ongoing signs of inflammation on ultrasound.

  • Careful analysis and pediatric-specific criteria for ultrasound interpretation are important to avoid false-positives, especially for the Doppler assessment.



Patients with JIA according to the International League of Associations for Rheumatology (ILAR) criteria (20), with the exception of the enthesitis-related and “other” subtypes, were eligible for the study if they had a minimum age of 2 years and had been in remission for a minimum of 3 months at the time of enrollment. Applying previously published criteria (21), remission was defined as the absence of any active joints on clinical examination; no evidence of uveitis on slit-lamp examination; normal serologic markers of inflammation; no fever, rash, serositis, or splenomegaly; and a physician's global assessment showing no disease activity either on or off medication. An active joint was defined as either a joint with swelling, or if no swelling was present or detectable (for example, in the hip), with limitation of motion, accompanied by pain on motion, tenderness, or both. Consecutive patients who attended the pediatric rheumatology clinic at the Children's Hospital of Eastern Ontario between April 2009 and September 2009, met the eligibility criteria, and were seen on a day the sonographer was available were asked to participate in the study. The study was approved by the Children's Hospital of Eastern Ontario Research Ethics Board. Informed consent was obtained from the parents and assent was obtained from the children, where applicable.

For the purpose of this study, the following clinical data were extracted from patient files: age, sex, ILAR subtype of JIA, disease duration, duration of remission, and medication. All extracted data were verified at the study visit.

US assessment.

Equipment and settings.

A General Electric Logiq e US device equipped with a multifrequency linear probe at a maximum frequency of 13 MHz for gray-scale sonography and 6.7 MHz for power Doppler sonography was used. The regions of interest for Doppler US included the bony margins, articular space, and a view of the surrounding tissues. The pulse repetition frequency was set between 500 and 750 Hz, depending on the joint scanned. Low wall filters were used. Color gain was set just below the level at which color noise appeared in the underlying bone.

Assessed joints.

The following joints were assessed: the radiocarpal and midcarpal joints, the knee joint, and the tibiotalar joint. In general, each anatomic region (wrist, knee, and ankle) was only assessed in a given patient if there had been previous clinical joint involvement of that region. If only one of the two joints in each region had been previously affected clinically, the contralateral joint was also assessed.

Recording and interpretation of US scans.

All US scans were recorded by a single rheumatologist experienced in musculoskeletal US in children (JR). All images were read independently by two readers (JR and MR-P). In the case of disagreement, the two readers met and resolved the discrepancy. Pathologic findings were categorized as either structural abnormalities on gray-scale US or a positive power Doppler signal.

Definition of structural abnormalities.

A joint effusion was defined as an abnormal, compressible, anechoic joint space, whereas synovial hypertrophy was defined as an abnormal, partially compressible, hypoechoic joint space according to the Outcome Measures in Rheumatology Clinical Trials definitions (22). No grading of the gray-scale changes was done, as these have not yet been established for children. Joint effusion or synovial thickening in the wrist was defined as the presence of any anechoic or hypoechoic synovial recess (23). For the knee joint, the cutoff was set as an anteroposterior diameter of >2.7 mm, representing more than 2 SDs of the normal joint space according to published normative data (23). For the tibiotalar joint, the threshold was a synovial space with a sagittal diameter of >1.5 mm (Roth J, et al: unpublished observations) according to unpublished data in 20 healthy children across an age range of 1–16 years.

Coding of power Doppler signal.

Presence of Doppler signals in the cartilage and within the joint space can be observed in adults (24, 25) as well as in children (26). In adults they are usually single vessels and not confluent. In order to not interpret physiologic blood flow as abnormal Doppler signals, only signals observed in an area of synovial hypertrophy were considered. The signals were categorized as follows: 0 = absent, 1 = presence of single-vessel dots, 2 = presence of confluent vessel dots in less than half of the synovial area, and 3 = presence of confluent vessel dots in more than half of the synovial area. This categorization, which has not been generally validated in pediatrics, was done for the sole purpose of setting a threshold of grade 2 signals and above that would be considered abnormal.

Intraobserver reliability was determined by blinded rescoring of the archived images 3 months after the initial assessment.

Scan positions.

The positioning of the probe for the various scans was as follows: for the wrist the patient was in a sitting position with the hands palm-side down in a neutral position on an examination table. The probe was positioned in the longitudinal plane over the dorsal aspect of the lunate and capitate bone and the scan was extended across the carpus (Figure 1A). Both the radiocarpal as well as the midcarpal joints were evaluated by determining the anteroposterior diameter of the synovial space measured from bone or cartilage (whichever was present) at the dorsum of the respective wrist bones. Pathologic findings were confirmed in a second plane. For the knee the patient was in a supine position with the lower extremities parallel to each other. A measurement was taken with the knee joint in neutral position. The probe was positioned longitudinally over the suprapatellar recess, cranial to the superior edge of the patella (Figure 1B). The maximum anteroposterior distance of the joint space was measured. The scan was extended medially and laterally as small effusions can sometimes be seen, especially in the lateral part of the suprapatellar recess. Pathologic findings were confirmed in a second plane.

Figure 1.

A, Standard position of the probe for the wrist. Dorsal longitudinal scan starting in the midline visualizing the lunate and capitate bone. A schematic depiction of the positioning of the ultrasound (US) probe is given on the left and the corresponding gray-scale US image is shown on the right. The dark area surrounding the lunate and capitate bone shows cartilage and not fluid, as can be seen by the cartilage interface sign (thin white line on top of the cartilage). The white stars indicate the sites where the presence of a joint effusion or synovial hypertrophy was assessed (radiocarpal joint left star, midcarpal joint right star). The scan was extended medially and laterally to detect any pathologic findings. B, Standard position of the probe for the knee. For the knee joint, the suprapatellar pouch was visualized in the midline with the knee in a neutral position. The probe was positioned sagitally over the suprapatellar recess cranial to the superior edge of the patella (p). Note that the left part of the patella still consists of cartilage extending up to the suprapatellar fat pad. The scan was extended medially and laterally. The scan position is schematically shown on the left and in a gray-scale US image on the right indicating where the measurement was taken (*) in case of the presence of detectable synovial fluid or hypertrophy. C, Standard position of the probe for the tibiotalar joint. For the tibiotalar joint, the patient was in a supine position with the knee in 90° flexion and the foot resting on the surface of the examination table, resulting in plantar flexion. The scan was started in a sagittal midline plane along the tibiotalar joint and then extended both medially and laterally. The probe position is shown schematically on the left side. On the gray-scale US image on the right side, the measurement location is shown (∗) at 50% of the convexity of the talar dome, sagittal to the bone and cartilage surface.

For the tibiotalar joint the patient was in a supine position with the knee in 90 degrees flexion and the foot resting on the surface of the examination table, resulting in plantar flexion of the tibiotalar joint. A longitudinal scan over the midline of the tibiotalar joint was done and extended to both sides. The synovial space was measured at 50% of the talar dome convexity (Figure 1C). Pathologic findings were confirmed in a second plane.

Statistical analysis.

Numbers of joints as well as patients that were found to be abnormal on gray-scale as well as Doppler US are given in relation to the number of patients/joints scanned in total. Confidence intervals for the percentage of abnormal joints were calculated. Intraobserver reproducibility was assessed by computing the percentage of exact agreement and by means of kappa statistics.


General characteristics of patients.

Twenty-eight children, 16 girls and 12 boys, were included in our study. The characterization of the disease subtypes and medication use can be found in Table 1.

Table 1. Disease characteristics of the patients*
  • *

    Values are the percentage unless otherwise indicated. RF = rheumatoid factor; US = ultrasound; anti-TNF = anti–tumor necrosis factor; C-HAQ = Childhood Health Assessment Questionnaire.

Antinuclear antibody positive41
HLA–B27 positive7
RF positive4
Methotrexate, ever/at US67/52
Anti-TNF, ever/at US15/15
Glucocorticoids, ever/at US26/4
Arthritis type 
 Oligoarticular persistent33
 Oligoarticular extended11
 Polyarticular RF positive4
 Polyarticular RF negative22
C-HAQ score, mean ± SD0.02 ± 0.09

The mean ± SD age at diagnosis was 6.2 ± 4.6 years. The mean ± SD time since diagnosis was 51.2 ± 37.8 months. All of the patients participating in our study had been in clinical remission for at least 3 months, with a mean ± SD duration of 14.4 ± 16.7 months. The detailed information on disease/remission duration is given in Table 2.

Table 2. Disease and remission duration
 Mean ± SD
Age at diagnosis, years6.2 ± 4.6
Time since diagnosis, months51.2 ± 37.8
Duration of active disease for each joint, months 
 Wrist11.1 ± 13.0
 Knee11.8 ± 12.5
 Ankle11.8 ± 16.6
Duration of remission for each joint, months 
 Wrist20.4 ± 15.9
 Knee21.9 ± 19.5
 Ankle19.9 ± 15.1

Intraobserver reliability.

The results of intraobserver reliability with regard to the presence or absence of gray-scale abnormalities were 91.7% (95% confidence interval [95% CI] 74.2–97.7%) for exact agreement, with a kappa statistic of 0.813 (95% CI 0.564–1). For the power Doppler US findings, the level of agreement was 95.8% (95% CI 79.8–99.3%), with a kappa statistic of 0.833 (95% CI 0.517–1).

Results of the US assessments according to joint.

Table 3 shows the US findings of the wrist joint. Fourteen patients had a total of 24 wrists that were affected previously. As the radiocarpal and midcarpal joints were assessed separately, this led to a total number of 48 wrist joints that were analyzed. Four patients had a total of 4 wrists (8 joints) that were previously unaffected. Among the previously affected wrists, 8 of 14 patients had abnormal gray-scale findings and 3 of 14 had abnormal Doppler findings. There were subsets of patients who had isolated abnormalities of either the proximal or midcarpal joint, but there were no patients with pathologic Doppler signals without corresponding gray-scale abnormalities. Half of the previously unaffected wrist joints did show abnormal findings on gray-scale US, but none had pathologic Doppler signals.

Table 3. Ultrasound results for the wrist joint*
WristPreviously (clinically) affectedPreviously (clinically) unaffected
  • *

    Results for the wrists are given per patient and per wrist joint. Because the proximal (radiocarpal) and midcarpal wrist joints were analyzed separately, the number of wrist joints is double the number of wrists. While results for the overall analysis are given according to patient, these results are further detailed according to radiocarpal and midcarpal joint as well as gray-scale and Doppler findings. All abnormal Doppler findings were shown in patients having gray-scale changes, i.e., no patient had only pathologic Doppler signals without corresponding gray-scale changes. 95% CI = 95% confidence interval.

Analysis per patient  
 Total no. of patients144
 Gray scale  
  No. of patients pathology8/142/4
  Percentage pathologic (95% CI)57.1 (28.9–82.3)50.0 (6.8–93.2)
  No. of patients pathology3/140/4
  Percentage pathologic (95% CI)21.4 (4.7–50.8)0 (0–60.2)
Analysis per joint  
 Gray scale  
  No. of pathologic joints  
   Proximal only4/242/4
   Midcarpal only2/240/4
  No. of pathologic joints  
   Proximal only3/240/4
   Midcarpal only0/240/4

US findings for the knee and the tibiotalar joint are shown in Table 4. In the case of the knee joint, the US examination was always concordant with the clinical examination, with no structural abnormalities or positive Doppler signal found in any patient (Table 4). In the tibiotalar joint, 7 of the 22 previously affected joints showed gray-scale abnormalities, with 1 joint showing abnormal Doppler signals as well as gray-scale changes.

Table 4. US results for the knee and tibiotalar joints*
 Knee: previously affectedKnee: previously unaffectedTibiotalar: previously affectedTibiotalar: previously unaffected
  • *

    Ultrasound (US) findings for the knee and the tibiotalar joint are shown for gray-scale as well as Doppler US. 95% CI = 95% confidence interval.

Analysis per patient    
 Total no. of patients208158
 Gray scale    
  No. of patients pathology0/200/86/151/8
  Percentage pathologic (95% CI)0 (0–60.2)0 (0–36.9)40 (16.3–67.7)12.5 (0.3–52.7)
  No. of patients pathology0/200/81/150/8
  Percentage pathologic (95% CI)0 (0–60.2)0 (0–36.9)6.7 (0.2–32.0)0 (0–36.9)
Analysis per joint    
 Total no. of joints328228
 Gray scale    
  No. of pathologic joints0/320/87/221/8
  No. of pathologic joints0/320/81/220/8


The aim of our study was to assess whether patients with JIA in clinical remission show abnormalities on either gray-scale or power Doppler US. To our knowledge, this is the first study assessing the value of US in children with JIA in clinical remission.

The results of this study demonstrate that a clinically important number of patients in remission continue to have abnormal US findings in the wrist and ankle joints, including positive power Doppler signals. On the other hand, there was complete concordance of the clinical and US assessment for the knee joint. The wrist and ankle might therefore be joint regions that would especially benefit from an additional assessment with US.

This is further illustrated by the fact that with US a given joint can sometimes be analyzed in more detail than with clinical examination. In the wrist, gray-scale midcarpal involvement was frequent and often coincided with radiocarpal joint involvement, but was also found as the sole manifestation of wrist joint abnormalities in 2 of 24 patients. It is also important to assess tendons separately from joints, and this cannot always be done reliably on clinical examination alone either. For the ankle joint it may be very useful to assess the tibiotalar and subtalar joints as well as the various tendons (2). However, this detailed assessment was not performed in our study, as the past joint involvement was determined only clinically and normative data were only available for a limited set of joints; for example, in the case of the ankle no normative data were available for the subtalar joint or the tendons surrounding the ankle. We therefore restricted the analysis to the joints shown in this study that are also among the most commonly involved joints in JIA (4).

Our findings differ from some previous pediatric publications where a significant dissociation of clinical and US findings has been described mainly for the knee joint, although this was partly in the context of active disease and not necessarily patients in remission (1, 27). There is also a small chance that by not doing lateral and medial scans to assess the lateral and medial recesses, we might have missed some pathologic US findings. With regard to the wrist, other groups did find significant dissociations between clinical and US assessments as well (19).

In contrast to some studies assessing patients with active disease (19), we did not scan all of the joints in all patients as the comparison was to be made to the clinical assessment of joints that were previously active and in remission at the time of the study. We did nevertheless scan contralateral joints of previously affected joints even if they had not been found to be active in the past.

For the wrist, half of the previously unaffected wrists in patients with a history of unilateral wrist involvement did show abnormal gray-scale findings. The question arises whether gray-scale abnormalities were overcalled, i.e., whether healthy volunteers would show some degree of gray-scale abnormalities as well, especially given the fact that none of the previously unaffected but gray-scale–positive joints showed Doppler signals. Published normative data (23) nevertheless suggest that none of these gray-scale findings should be present in healthy children, and our patients might have had subclinical contralateral arthritis in the past.

From a clinical point of view, the proportion of patients with polyarticular and systemic disease in our study population was higher than expected in JIA in general and the duration of active disease was also quite long in some patients, indicating a more complicated disease course and therefore a higher potential for residual changes.

We have tried to avoid overcalling active synovitis by differentiating between gray-scale abnormalities and Doppler-positive joints. Inflammation may lead to chronic changes such as synovial hypertrophy that will often persist despite the resolution of the acute inflammation. Doppler signals have been shown to be more accurate in the assessment of active synovitis than gray-scale abnormalities alone (13–18).

In our study, the presence of significant Doppler signals was required in order to consider a joint power Doppler positive, as some degree of normal intrasynovial blood flow can be expected, especially in the wrist (24). To consider any Doppler signal a sign of synovitis, as has been done in recent studies (19), might be problematic, especially in children where significant physiologic blood flow can be expected (26). We also did not attempt to grade the gray-scale findings, as the semiquantitative grading system that has been established in adult patients has not been validated in pediatrics. The limitations of our study are the relatively small number of patients and the use of a US system that is not among the high-end machines currently available. This might especially impact the Doppler sensitivity but would, if anything, mean that we have missed rather than overcalled pathologic Doppler signals. Finally, the intraobserver reliability was assessed in children in remission only, as this was the subject of the study, but might be different in children with active disease and more pathology.

Significant differences between the clinical examination of joints and the US assessment have been found in adult rheumatology. Active arthritis was found in 20% less joints by clinical examination than US examination in patients with rheumatoid arthritis (28). Other studies have shown an even higher percentage, with 73.3% of patients in complete clinical remission showing the presence of gray-scale synovial hypertrophy and 43.3% showing associated increased power Doppler flow (29, 30). The discrepancies between US and clinical examination might well explain structural deterioration despite clinical remission. Furthermore, it has been proposed that identification of subclinical disease may change patient classification (19) and lead the clinicians toward more aggressive treatment (7) and better outcomes.

In the future, US will likely play an important role in the assessment of joint disease activity in the pediatric population, as is the case for assessment of rheumatoid arthritis in adults. However, in order to be able to do so, large normative databases including the various joints and tendons need to be established that can then provide a basis for scoring systems in pediatric rheumatology. In addition, prospective trials are needed to evaluate the dynamics and evolution of US findings throughout the disease course.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Roth 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.

Study conception and design. Rebollo-Polo, Koujok, Bruns, Roth.

Acquisition of data. Rebollo-Polo, Weisser, Roth.

Analysis and interpretation of data. Rebollo-Polo, Jurencak, Roth.