Imaging of the knee in juvenile idiopathic arthritis

For the application of US in paediatric rheumatology, both grey-scale B-mode and power/colour Doppler modes should be used in every examination. Also, standard scanning positions should be considered. For standard clinical paediatric ultrasound examination of the knee, the child is placed in supine position with a slightly flexed knee. The standard sagittal view includes the patella inferiorly, the quadriceps tendon, and the suprapatellar recess [9]. An axial view in the popliteal fossa can be considered to exclude a Baker cyst [9]. In the scope of research a more extensive ultrasound protocol of the knee can be considered (Table 1).

The main US features characterizing pathology in JIA at this stage of the development of a US-based outcome measure are synovial thickening and synovial effusion [10]. Synovial thickening (Fig. 1) is defined as abnormal, intra-articular, hypoechoic material that is non-displaceable. Synovial effusion (Fig. 1) is defined as an abnormal, intra-articular, anechoic or hypoechoic material that is displaceable. When looking for a synovial effusion, one should be aware that a change of patient position can influence the level of effusion detected. Moreover, a scanning technique avoiding excessive pressure is warranted because joint fluid can shift to another synovial recess when too much pressure is used [11].

Power/colour Doppler is useful to acquire information about tissue perfusion. It is important to clarify that synovitis can be detected on the basis of B-mode findings (synovial thickening or synovial effusion) alone. Nevertheless power and colour Doppler have proved to be useful in differentiating active (hypervascular) from fibrotic (hypovascular) pannus [11]. In growing children, the sole detection of mildly increased vascularity does not allow for the diagnosis of synovitis because of the different levels of joint vascularity according to a child’s age and level of physical activity of the joint [12]. It is important to recognize the presence of such normal vessels in paediatric joints from normal maturation and these should not be considered pathological [13].

In children with JIA, loss of cartilage thickness has been described. US has a strong correlation with MRI in the evaluation of cartilage thickness of the medial and lateral condyle and the intercondylar region, with a rho correlation of 0.70–0.86 [14].

We are currently developing a consensus-based US scoring system. The preliminary scoring system consists of three grades separated for synovitis and synovial perfusion (Table 2). Its reliability will be tested in a patient-based exercise.

Age-related US findings have been described in healthy children [13, 15]. There are considerable age-related variations in children’s joints, including the knee examined by US, where Doppler depicts vascularity, particularly within the epiphyseal cartilage of the children at a younger age. Also, age- and gender-related differences and standard reference values of the cartilage thickness of the knee in healthy children have been elaborated [16]. However, the number of studies focusing on normal ultrasound values of the knee in children is limited. Therefore the normal appearance of the synovial membrane, for instance, should be interpreted with care.

In children with JIA and knee involvement, the main imaging features include synovial thickening, joint effusion and bone marrow oedema. Moreover cartilage loss and bone erosions — although relatively rare in paediatric patients — can be observed. Tendinopathy, enthesopathy, inhomogeneity of the infra-patellar fat pad and bone cysts can be seen, as well, but these abnormalities are relatively uncommon in the knee joint. Moreover, the reliability of scoring these secondary features is unsatisfactory [18]. Table 3 shows an MRI protocol for evaluating the JIA knee joint.

Synovial inflammation leading to synovial thickening is the principal pathological process in JIA, and the presence of synovial thickening on knee MRI is associated with the clinical onset of JIA [19]. On MRI, the inflamed synovial membrane is thickened and irregular, and its outline may be wavy. In the knee, a synovial thickness of >2 mm is considered pathological. The central locations in the knee (around the cruciate ligaments and retropatellar and suprapatellar areas, Fig. 2) are most commonly affected [20]. The signal intensity of this thickened synovial membrane is low to intermediate on T1-weighted images and high on T2-weighted images, similar to joint effusion. T1-weighted images after administration of gadolinium-based contrast agents provide better differentiation between joint effusion and synovial thickening [21]. Omitting intravenous gadolinium-based contrast agents in the MRI assessment of joints in JIA is therefore not advised.

Although not specific for JIA, joint effusion (Fig. 2) can frequently be found in children with JIA [22]. On MRI, effusion demonstrates high signal intensity on fluid-sensitive images and low signal intensity on T1-weighted images. Joint effusions are predominantly located in the suprapatellar and central joint recesses.

MRI is the state-of-the-art imaging technique to visualize changes in bone marrow suggestive of bone marrow oedema (Fig. 3). Bone marrow oedema can be seen as lesions with poorly defined margins within trabecular bone. It is marked by high signal intensity on T2-weighted fat-saturated images and low signal intensity on T1-weighted images. In rheumatoid arthritis, longitudinal studies have shown that the presence of bone marrow oedema is a key predictor of early erosive joint damage in adults with rheumatoid arthritis [17]. Therefore bone marrow oedema and synovial thickening are considered to be the most sensitive MRI features for monitoring disease activity in rheumatoid arthritis [23‐25]. However to our knowledge there are a paucity of longitudinal studies, if any, focused on the prognostic value of bone marrow oedema in children with JIA.

MRI clearly shows the difference between changes in the articular cartilage and bone erosions. Bone erosions can be seen on T1-weighted images as loss of the normal low signal intensity of cortical bone and loss of the normal high signal intensity of trabecular bone (Fig. 3). Typically they have sharp margins. On T2-weighted images bone erosions appear as hypointense lesions, in contrast to subchondral cysts, which show as hyperintense signal on fluid-sensitive images.

Although MRI is the preferred imaging modality for detecting both inflammatory as well as destructive changes in JIA, international experiences with the use of MRI as an outcome measure in JIA are limited. Subsequently, this technique is underused in both clinical practice and research. In order to further internationally validate MRI as an outcome measure in JIA, a special interest group on Outcome Measures in Rheumatology on MRI was created in 2011 [26]. Within this special interest group on MRI in JIA, three working groups were initiated for further assessment of MRI scoring systems: one group focuses on temporomandibular joints, another on small joints and the third on large joints (primarily the knee joint).

In the last decade, two MRI scoring systems have been developed for the assessment of the paediatric knee joint: (1) the Juvenile Arthritis MRI Scoring system for use in JIA [27] and (2) the International Prophylaxis Study Group MRI Scoring system focusing on haemophilic arthropathy [28].

Recently the Outcome Measures in Rheumatology (OMERACT) special interest group on MRI in JIA conducted a reliability study evaluating both scoring methods [18]. This study showed good inter-observer reliability for MRI scores focusing on active disease (e.g., synovial thickening, joint effusion and bone marrow changes) and moderate to substantial inter-observer reliability for scores focusing on damage (e.g., cartilage lesions, bone erosions). Finally, the group proposed a combined juvenile arthritis MRI scoring system (JAMRIS) — shown in Table 4 — including items from both the original Juvenile Arthritis MRI Scoring and the International Prophylaxis Study Group systems [18].

Besides this international reliability study, the intra-observer and inter-observer reliability of the original Juvenile Arthritis MRI Scoring system has been evaluated, and both proved to be reliable [27]. These data are trustworthy, especially regarding MRI features focusing on active disease. Caution is needed with regard to MRI characteristics focusing on damage because these features are relatively uncommonly observed in JIA patients.

Although MRI has some practical limitations such as limited availability in some regions/countries and the challenge of maintaining the same position for a prolonged period of time, it has proved feasible to perform contrast-enhanced knee MRI in children with JIA as young as 5 years old without the use of anaesthesia or sedation [29]. Moreover the use of the original Juvenile Arthritis MRI Scoring system proved feasible because the scoring takes an acceptable median of 6.6 min per patient [21].

Within the last few years, two studies have focused on the sensitivity to changes in disease activity using the original Juvenile Arthritis MRI Scoring system [27, 30]. In both studies, with a follow-up period of 1 year, improvement of clinical JIA disease activity scores was associated with a significant decrease in MRI-based synovial thickening scores. These results represent good responsiveness of the Juvenile Arthritis MRI Scoring system, which is an important measure of validity. Moreover the Juvenile Arthritis MRI Scoring synovial thickening score proved useful in discriminating clinically active from clinically inactive JIA patients [31], indicating good discriminant validity.

Next to the responsiveness and discriminant validity, the clinical validity has also been evaluated. In a study focusing on whether clinical, laboratory or MRI measures were able to differentiate JIA with active arthritis from other causes of non-infectious arthritis in a group of patients with clinical signs of early arthritis, multivariate analysis showed that MRI-based synovial thickening was independently associated with JIA (odds ratio [OR] 6.58, 95% confidence interval 2.36–18.33) [19].

Because synovial inflammation is the hallmark of disease activity in JIA, it is important to determine the normal appearance and thickness of the synovial membrane in children. As stated, an intravenous injection of a gadolinium-based contrast agent is warranted for reliable evaluation of the synovial membrane. Therefore data on normal values for the synovial membrane in healthy children are sparse. Two studies that focused on the contrast-enhanced appearance of the synovial membrane in children have been performed: (1) A study by Nusman et al. [32] looked at the synovial membrane of the knee in children with inflammatory bowel disease who were clinically unaffected by arthritis; (2) Hemke et al. [33] determined the synovial membrane in knees of healthy children, using a standardized imaging protocol with post-contrast images obtained in the early phase (<5 min). The latter study showed that the normal synovial membrane thickness measures a maximum of 1.8 mm. The membrane was thickest around the cruciate ligaments, and retropatellar and suprapatellar recesses. The study by Nusman et al. showed a thickened synovial membrane (>2 mm) in more than half of patients with inflammatory bowel disease; however they did not observe any synovial thickness >4 mm. In all probability, the results from Hemke et al. are a more valid reflection of the true normal appearance of the synovial membrane because this study included healthy children (compared to patients with inflammatory bowel disease) and obtained post-contrast images in the early phase. Therefore, the Juvenile Arthritis MRI Scoring cut-off value of >2 mm for synovial thickening can be considered a valid measure.

Some joint fluid in the knee can be seen in the majority of healthy children. The largest pockets of normal joint fluid in healthy children are located in the central locations of the knee — around the cruciate ligaments and retropatellar region [33]. The mean diameter of the largest pocket of joint fluid in knees of healthy children is about 3 mm [15, 33].

Bone marrow changes suggestive of bone marrow oedema in knees of healthy children are relatively uncommon. In a study of 57 healthy children, bone marrow changes suggestive of bone marrow oedema were observed in 3 healthy children only. In all three children, the bone marrow changes were located in the apex patellae [33]. The presence of bone marrow changes suggestive of bone marrow oedema in the apex patellae in children with JIA should, therefore, be interpreted with care.

Moreover, zones of hematopoietic red bone marrow in the distal diaphysis and metaphysis of the femur might be seen as flame-like regions with signal characteristics fitting with bone marrow oedema. Typically, these marrow flames originate from the physis and have straight vertical margins [34]. These hematopoietic bone marrow flames are a normal finding in growing children and should not be mistaken for bone marrow oedema. So-called speckled bone marrow is another normal variant that might be mistaken for pathology. Small spots with bone marrow oedema signal characteristics are predominantly located in the feet and ankles of children younger than 15 years, although the spots can be seen in the tibia plateau and the distal epiphysis of the femur as well [35]. The speckled appearance might be caused by focal regions of residual hematopoietic bone marrow or physiological stress, possibly related to weight-bearing or altered biomechanics during normal growth.

The thickness of the normal articular cartilage in knees of growing children differs with age. As expected, the cartilage is normally thicker in younger children compared to older children. A study performed by Keshava et al. [15] clearly demonstrates these differences among different age groups in healthy boys. Moreover they showed that the normal thickness of the articular cartilage differs per location within the knee joint [15]. In the distal femur, ossification usually starts in the centre of the cartilaginous epiphysis. Because the newly formed ossification centre contains hematopoietic bone marrow, its signal intensity is the same as that of red marrow in the adjacent distal femoral metaphysis [36]. The ossification centre enlarges from endochondral bone development. Adjacent cartilage cells undergo hypertrophy during endochondral ossification, which results in increased signal intensity on T2-weighted images [34]. This area of normal high signal intensity is most obvious in the posterior part of the distal femoral epiphysis and can be quite discrete [37]. Furthermore, it is important to be aware of the high prevalence of ossification variants of the femoral condyle among boys ages 2–12 years and girls ages 2–10 years; this should not be considered pathological [37].