Introduction

Juvenile idiopathic arthritis (JIA) is the most common chronic rheumatic disease in childhood with a prevalence that varies between 16 and 150 per 100,000 children [1]. JIA is not a single disease entity but an exclusion diagnosis which includes all forms of chronic arthritis that begin before the age of 16 years, persist for more than 6 weeks, and are of unknown origin. Indeed, JIA encompasses several disease categories, each with its own distinct clinical features and, in some cases, genetic background [2, 3]. Despite the heterogeneity, all forms are characterized by synovial inflammation, which if not adequately controlled may ultimately lead to cartilage loss and bone damage and consequent permanent disability [4]. Of note, JIA is one of the main causes of acquired disability in childhood [5].

Over the past decade, the management of JIA has dramatically changed, owing to the shift towards early aggressive interventions and the development of effective structure-modifying drugs, such as biological agents, that have considerably decreased the risk of permanent damage to joint structures [6–10]. As a consequence, it has become essential to identify patients who are more likely to develop unremitting illness carrying significant risk of joint destruction and who are suitable for aggressive treatment early in the disease course [11–13].

Imaging has a considerable role in the diagnosis of JIA, in assessing its severity and prognosis and in monitoring treatment efficacy. Conventional radiography (CR) is regarded as the mainstay of imaging evaluation of joint structural damage in children with JIA. However, the trend towards the early introduction of effective disease-modifying treatments has generated the need for alternative imaging modalities that are more sensitive in detecting pre-erosive inflammatory changes, in order to stratify patients for treatment and to monitor therapeutic efficacy more effectively. CR is inadequate in this setting, as it reveals late and largely irreversible consequences of synovial inflammation. Magnetic resonance imaging (MRI) and musculoskeletal ultrasound (MSUS) are therefore playing an expanding role in the assessment of arthritic joints. Despite the large amount of evidence that has accumulated on the value of these modern imaging modalities in predicting and monitoring treatment efficacy in adults with rheumatoid arthritis (RA) [14–20], this field has remained almost unexplored in JIA.

Unlike adults, imaging in children represents a real challenge because growing joints change anatomically over time. During childhood, skeletal maturation is marked by an orderly sequence of recognizable changes in the appearance of the skeleton. Such changes include timing and sequence of appearance of ossification centres, specific alterations in bone contours, and timing of the ultimate closure of the growth plates. All these growth-related physiological changes may be altered by chronic joint inflammation, resulting in radiographic changes, such as advancement of maturation and epiphyseal overgrowth secondary to hyperemia, or retardation of maturation due to damage, which are peculiar to JIA.

The present review aims to provide current evidence to support the use of MRI and MSUS for the assessment of patients with JIA, both in clinical and research settings. Furthermore, the specific aspects relevant to the evaluation of the growing skeleton will be discussed.

Magnetic Resonance Imaging

MR imaging, by providing multiplanar tomographic imaging with unparallel soft tissue contrast, allows the assessment of all joint structures involved in inflammatory arthritis. A thorough knowledge of the advantages and limitations of this imaging modality (Table 1) is crucial for identifying the clinical context in which MRI is most likely to be recommended. To diagnose JIA properly, a wide range of differential diagnoses needs to be excluded, many of which have recognizable features on imaging. CR is the first-step imaging modality in order to exclude other causes of joint pain and swelling in children, such as trauma, osteochondroses, osteoid tumors, or skeletal dysplasias [21–23]. In selected cases, however, MRI may be indicated to rule out other intra-articular disorders that mimic inflammatory arthritis such as hemangioma or pigmented villonodular synovitis [24–26]. In patients with known JIA, MRI is well suited to depict the involvement of particular joints that are not easy to assess clinically, such as the temporo-mandibular, hip, and sacroiliac joints [27–33]. Finally, MRI appears suitable for identifying odontoid lesions and assessing functional instability of the cervical spine [34].

Table 1 Advantages and limitations of MRI in the assessment of patients with JIA
Full size table

Over the last decade, however, along with breakthroughs in therapeutics, new expectations of MRI have emerged. By directly imaging synovitis, contrast-enhanced MRI has an intuitive advantage in assessing response to treatment over clinical surrogate measurements of inflammation (Fig. 1). Standardized and validated methods of evaluating MRI findings in JIA have progressed and validated pediatric scales are now available [35–38]. In a recent study comparing MRI and clinical criteria in 40 JIA patients who were starting a second-line therapy, the Rheumatoid Arthritis MRI synovitis score has proven to be a promising imaging biomarker for measuring therapeutic response [39•]. Notably, only the highest levels of clinical response were associated with significant decrease in synovitis and the halting of structural damage. These findings support the need to move towards more strict definitions of clinical response when assessing drug efficacy in JIA, and suggest the potential for MRI as a primary efficacy outcome in clinical studies. In this field of outcome assessment, prospective quantitative measurements of synovitis appear particular promising in evaluating treatment efficacy. Due to advances in imaging technologies, the rate and magnitude of synovial enhancement that reflect local tissue vascularity and capillary permeability can be reliably quantified using dynamic contrast-enhanced MRI (DCE-MRI). This technique has permitted the differentiation of active, hypervascular synovial membrane from the inactive, fibrotic pannus and provides a reliable assessment of disease activity in patients with JIA [40]. Furthermore, the use of pharmacokinetic modeling to quantify DCE-MRI, has provided an objective follow-up measure of therapeutic efficacy in JIA pilot studies [41, 42]. In addition, computerized measurements of wrist synovial volumes (Fig. 2), obtained using an automated segmentation program in a cohort of 56 JIA patients, were found to be more sensitive in detecting treatment efficacy and predicting progressive joint destruction compared to a semi-quantitative approach [43]. Quantitative methods are not typically part of standard daily practice, since they require post-processing analysis and dedicated software. Owing to their high sensitivity in detecting even the smallest treatment-related changes, they appear particularly promising in a clinical trial setting.

Fig. 1
figure 1

Wrist MRI in a 12-year-old girl with juvenile idiopathic arthritis. Contrast-enhanced coronal 3D Fast Field Echo T1-weighted MRI with fat saturation shows active synovitis in the radioulnar, radiocarpal and intercarpal joint recesses and diffuse bone marrow oedema involving several carpal bones and the metacarpal bases (a). Marked reduction of synovitis and bone marrow oedema after treatment with anti-tumour necrosis factor α (b)

Full size image
Fig. 2
figure 2

Contrast-enhanced wrist MRI of a patient with juvenile idiopathic arthritis. The automated segmentation of the voxel corresponding to inflamed synovial membrane (highlighted in yellow) provides a computerized measurement of the synovial volume

Full size image

Therapeutic advances have increased the expectations of treatment benefits with disease remission now becoming a realistic goal [44, 45]. Sustained synovitis detected by MRI was recently documented in a sizable proportion of JIA patients who satisfied clinically-defined remission criteria [39•, 43, 46]. However, no longitudinal studies have thus far investigated whether subclinical inflammation may end up in further joint damage and functional disability, as reported in adults with RA [47]. Overall, although MRI appears a promising outcome measure for the assessment of treatment efficacy and remission, further studies are required to establish whether targeting therapy to the measures obtained from imaging provides better outcomes compared to treating by using only clinical targets.

Early detection of patients who will develop erosive damage and consequent aggressive control of the disease is a highly desirable goal to reduce the chance of further disability. MRI is the only imaging modality able to visualize bone marrow edema (BMO), a key predictor of erosive joint damage and functional impairment in adults with RA [48–50]. Recent studies comparing MRI appearances with bone histology in patients with RA have revealed that BMO regions contain a vascularized, cell-rich inflammatory tissue in close apposition to activated osteoclasts, as revealed by extensive expression of RANKL [51]. These findings have suggested a new pathway for the development of bone erosions, which starts from an inflammatory process within BM leading to RANKL-induced osteoclastogenesis and bone resorption. Caution is, however, needed before considering BMO as a prognostic indicator in JIA, since longitudinal studies investigating whether the presence of BMO predates the development of bone erosions are still lacking. Furthermore, signal changes resembling BMO have recently been detected in a high percentage of carpal bones in healthy children of different ages [52••, 53]. Notably, the same phenomenon was not reported in similar studies including healthy adults [54, 55], thus highlighting the peculiarities of the growing skeleton and the need to include age-matched healthy subjects in MRI studies on JIA.

The availability of normative data is also pivotal for an accurate assessment of bone erosions. In line with studies in RA [56, 57], the higher sensitivity of MRI in detecting early erosive damage over the others imaging modalities has been demonstrated also in JIA [27, 28, 58]. However, a certain concern that the increased sensitivity of MRI may be at the cost of a reduced specificity has been raised after a MRI study on 88 healthy children in which bony depressions, mimicking erosive changes, were found in the carpal bones of all subjects [52••]. A sound knowledge of growth-related bony changes is of foremost value to establish whether the apparent changes on bone surface are pathological or part of normal development [59]. Unexpectedly, Malattia et al. did not find a clear advantage in the use of wrist MRI compared to CR when evaluating structural damage progression over 1 year [39•]. The fact that cartilage loss, whose evaluation is included in the radiographic scale but not in the MRI scoring system, is the most common form of damage throughout the disease course [60] might explain the higher sensitivity of CR in detecting structural changes. Additionally, in younger children, the radiographic changes in carpal bones are frequently observed and scored as deformity in shape rather than as discrete erosions [60]. This phenomenon is distinctive of JIA and is likely attributable to a combination of growth abnormalities, ossification of previous injured cartilage, and true bone erosions [61, 62]. Owing to the unique features of the growing skeleton, the MRI definition of damage developed for RA patients may not necessarily be applicable to pediatric patients. A more comprehensive MRI scale incorporating measurements of cartilage damage and bone deformity may increase the sensitivity of this imaging technique for testing the disease-modifying potential of antirheumatic drugs.

As previously mentioned, articular cartilage is a major target of the erosive process in chronic inflammatory arthritis, and the assessment of its integrity should constitute a key goal for any imaging modality used in children with JIA. MRI allows discrimination between different types of cartilage (articular, epiphyseal, and physeal) and the direct visualization of signs of cartilaginous damage such as thinning and erosions (Fig. 3) [63]. Of note, it is also possible, through the use of apposite MR techniques and software, to investigate biophysical properties and molecular changes in the composition of extracellular cartilage matrix that occur before morphologic changes can be detected by conventional imaging and when cartilage damage is potentially reversible [64–67].

Fig. 3
figure 3

Coronal T 1-weighted MR image of the hip of a patient with juvenile idiopatic arthritis shows bone erosions of both femoral heads; erosions in the roof of right acetabula are also present (arrow)

Full size image

Musculoskeletal Ultrasound

The use of MSUS is rapidly undergoing expansion into daily clinical practice for the diagnosis, monitoring, and managing of patients with JIA. MSUS is even more appealing for pediatric rheumatologists than other imaging modalities, as it does not entail sedation and ionizing radiation, allows a safe and multiplane assessment of several joints at one time, and is well accepted (Table 2) [68•]. The evolution of ultrasound equipment by manufacturers, with miniaturization and the availability of useful higher frequency transducers, have made joint evaluation considerably easier in the joints of children that are difficult to reach owing to their small size [69•]. Furthermore, the advent of Doppler modalities, as applied in the MSUS assessment, has allowed the identification of intra-synovial vascular signals, which can consistently help to differentiate active synovitis from inactive disease. Nevertheless, the current major limitation of the technique lies in its strong operator dependency and in the equipment used. Additionally, MSUS is not yet a properly validated tool for the assessment of synovitis in children with JIA [70•], as no definitions of common ultrasound abnormalities and standard reference values have been developed thus far.

Table 2 Advantages and limitations of MSUS in the assessment of patients with JIA
Full size table

The potential of MSUS to visualize most of the anatomical structures involved throughout JIA course is of considerable help in the successful management of disease. The integrity of cartilage can be appropriately detected by MSUS. Damage usually appears as cartilage loss with thinning or blurring of its anechoic structure. Recent studies have reported an acceptable inter- and intra-observer reliability [71] and a good agreement between MSUS and MRI for the measurement of cartilage thickness [72] in all the joints considered, except for the wrist. Furthermore, age- and sex-related normal standards for cartilage thickness for several joints have been established [73]. Recently, the same investigators described a decrease in cartilage thickness in JIA patients, when matched with similar joints of healthy subjects, with the loss being greatest in the knee of systemic and polyarticular JIA patients, which are commonly known to be the most clinically aggressive subtypes. Of interest, a decrease in joint cartilage thickness was also found in joints without previous arthritis involvement [74]. Overall, these findings highlight the potential of MSUS in the assessment of cartilage integrity, either in clinically active or in subclinical disease, and they may change the future approach in caring for patients with JIA by guiding very early and tailored treatment aimed at preventing the establishment of osteocartilaginous injury.

Indeed, structural damage is often a direct result of a chronic and uncontrolled synovitis and the occurrence of an early bone erosion has been associated with a poor long-term outcome in children with JIA [11]. The assessment of erosive changes in a growing skeleton is challenging, as some physiological bone irregularities in recently ossified bones can be misinterpreted as cortical erosions [69•]. Owing to the lack of studies on normal sonographic bone anatomy for all age groups, the use of MSUS may potentially lead both to an over- or under-estimation of structural damage in children with JIA, and this is the major current limitation for the application of the technique for such evaluation. In addition, MSUS performs at its best for assessing bone erosions only where accessibility is optimal and allows a perpendicular positioning of the probe over the examined bone structure. In this context, the anatomical areas which cannot be reached with a comprehensive 360° assessment, such as intercarpal and intertarsal bones and hip, are more complex to evaluate.

MSUS has proven to be more accurate than clinical evaluation in the assessment of joint inflammation [75–77]. As the current ILAR classification for JIA is based on the number of affected joints, the issue of MSUS detectable subclinical synovitis may be particular relevant in JIA. The detection of subclinical inflammation by MSUS may potentially lead to both reclassifying patients and shifting to a more aggressive treatment. Joint effusion and synovial hypertrophy are the most common abnormalities detected by gray-scale MSUS (Fig. 4), but they may not reflect ongoing active disease [47]. Color and power Doppler modalities are of particular benefit in the assessment of active joint inflammation in comparison with gray-scale alone (Fig. 5). However, the interpretation of juxta-articular Doppler signal is challenging in children, as it can be a sign of the increased synovial vascularization indicating inflammation, or expression of the physiologically enhanced blood flow of the well-vascularized epiphyseal cartilage [77]. Unlike adults with RA [78], the prognostic meaning of MSUS findings in JIA is still being debated, as abnormalities, including Doppler signal, have been documented not to predict a flare of disease [79••]. This shortcoming further highlights the need to define the normal sonographic anatomy throughout pediatric age groups before addressing the role of MSUS in children with JIA.

Fig. 4
figure 4

Longitudinal ultrasound scan of the hip of a 2-year-old JIA girl showing joint effusion (JE) and synovial hypertrophy (SH). GP growth plate; FH femoral head; FN femoral neck

Full size image
Fig. 5
figure 5

Longitudinal ultrasound scan of the supra-patellar pouch of the knee of a 12-year-old JIA boy showing joint effusion (JE) and synovial hypertrophy (SH). Red spots inside SH represent vascular signals consistent with active disease as detected by power Doppler modality

Full size image

Tendons are another anatomical compartment frequently involved throughout JIA course, and isolated tenosynovitis may be responsible for joint swelling and pain (Fig. 6). It is common knowledge that prominent adipose tissue and the small size of the joints create difficulties in interpretation even for the well-trained physician who attempts to identify clinically tenosynovitis in younger children. Recently, attention was centered around the assessment of joint and tendon disease using MSUS compared to clinical evaluation in patients with JIA and ankle involvement [80]. The poor agreement reported by the investigators suggests that clinical assessment of the ankle region is inadequate and supports the use of MSUS in identifying the real site of inflammation for joints with anatomical complexity and numerous adjacent tendons. Furthermore, MSUS may also help in the clinical setting to perform guided injections into joint recesses and peri-tendinous areas [81], enhancing both the efficacy of the procedure and minimizing the rate of side effects.

Fig. 6
figure 6

Tenosynovitis of medial tendons of the ankle in a 10-year-old girl with juvenile idiopathic arthritis. a Transverse ultrasound scan shows marked involvement of tibialis posterior (Tp) and flexor digitorum (Fd) tendons. b Longitudinal ultrasound scan of the tibialis posterior tendon

Full size image

In childhood, a clinical diagnosis of enthesitis, which is a common feature of the enthesitis-related arthritis (ERA) category of JIA, may be somewhat difficult to make, due to the soft tissue swelling around the bone anatomical landmarks at the enthesis insertion sites. MSUS coupled with PD has been used for detecting enthesis inflammation in children with JIA [82]. In that study, although clinical enthesitis was often associated with PD-MSUS enthesitis, PD signal without any clinical evidence of enthesitis was also documented. These findings raise the question of whether the presence of a PD signal may be an early sign of enthesis involvement in clinically-silent enthesitis. Of interest, in a recent study, MSUS has proven to be superior to a standardized dolorimeter examination in detecting enthesitis in children with ERA [83], further supporting the use of MSUS as an alternative and more objective measure of entheseal inflammation, both at diagnosis and follow-up time.

Conclusions

The imaging approach to JIA has radically changed over the last decade, and modern imaging modalities such as MR and MSUS are increasingly overtaking CR for diagnosis of JIA, assessing its severity and prognosis, monitoring disease course and treatment efficacy. These imaging modalities are still in their infancy and further research focusing on the validation and standardization of these techniques is warranted before considering their use in clinical practice. Due to the peculiarities of the growing skeleton, studies aiming to establish normative data for healthy children are of high priority.