Imaging of the hip in juvenile idiopathic arthritis

Hip involvement is common and estimated to occur in approximately 35–63% of children with juvenile idiopathic arthritis (JIA). It is more prevalent in the aggressive systemic subtypes, with irreversible changes occurring as early as within 5 years of diagnosis. Whilst clinical parameters and joint examination can be useful for assessing disease severity, subclinical disease is known to exist and delayed treatment may herald a lifetime of disability and pain. Early recognition of JIA changes is therefore crucial in determining treatment options. Validated scoring systems in the radiologic assessment of the hip for clinical drug trials may inform treatment outcomes, although robust tools for analysis are still lacking. This review article details the modalities utilised for imaging the hip in children with JIA with particular efforts focused upon reliability and validity in their assessment of joint disease. We conclude with a short literature review on the potential future techniques being developed for hip joint imaging in JIA.

Even when the hip is not the primary joint affected at the time of diagnosis, a threefold increase in cases with hip involvement has been observed in a cohort of JIA patients over Commented [ØO1]: author: per 100,000 children?

Commented [ØO2]:
author: abstract says 35%-63%, please correct 2 a 5-year follow-up period, despite being on medical therapy [4]. In addition, it has been shown that destructive hip disease, in particular, is more frequent in patients with the aggressive "very early-onset systemic" subtype of JIA [5]. The majority of children with hip involvement can develop irreversible changes within 5 years of diagnosis [6], and approximately 26-44% will require a total hip replacement within the first 10 years of disease onset [7], at least in the prebiological era. Hip joint involvement is therefore a marker that heralds future disability with the potential to incite a reduced sense of personal independence and deterioration in overall health, partially from lack of exercise and subsequent depression [8].
Whilst clinical parameters and joint examination in cases of JIA are useful for assessing severity and extent of disease, hip involvement can be easily missed given the difficulty both in palpating the inflamed hip joint synovium [9] and in observing joint malformation due to effusion or synovial hypertrophy, which is relatively easier to identify in smaller joints, such as the wrist and knee. It is also now known that with improved therapeutic agents, joint destruction can be prevented especially when initiated promptly, underpinning the importance of detecting inflammation in subclinical disease [10].
With these factors in mind, imaging of the hip joint, particularly in early disease, holds great potential for accurately assessing disease progression, determining escalations for medical therapy and the need for targeted steroid injections and counseling with regard to future surgical management. It is worth noting that imaging of the hip also provides a role in determining differential diagnoses for hip joint pain such as leukemic involvement, avascular necrosis, trauma or underlying osseous malignancy [11].
In this review article, we highlight imaging modalities that can be used to assess the hip joint in JIA including radiography, ultrasound (US) and magnetic resonance imaging (MRI). Particular efforts have been made to focus on the feasibility, reliability, and construction of a clinical validity for published scoring systems in each imaging modality.

Progression of changes seen in juvenile idiopathic arthritis of the hip
Before determining the imaging modality and radiographic features to assess, it is important to understand the natural history of JIA disease's involvement of the hip joint.
Whilst papers focusing specifically on changes in the hip are scarce and longitudinal imaging studies are lacking, it is generally believed from long-term and short-term studies of various joints that destruction follows a reasonably predictable pattern.
The progression is thought to start primarily with inflammatory soft-tissue changes (synovitis, tendinitis, bursitis) and effusions before periarticular osseous changes (bone marrow oedema). Eventually, further disease progression may lead to growth disturbances (caused by reduction in bone mineral density, epiphyseal enlargement and early physeal closure) with destructive osteocartilagineous changes (cartilaginous thinning, joint space reduction and bone erosions) and degenerative features (ankylosis, joint malalignment) and growth disturbances leading to leg length discrepancy [12] (Fig.   1).
The order and presence of the features are not absolute or prerequisites for other destructive changes. The rapidity of development may be different between patients and dependent both on JIA subtype and on treatment response. In addition, the peculiarities of 4 the growing skeleton, including incomplete ossification and age-related variation in cartilage thickness, make the evaluation of children's joints a real challenge. These are all factors one must bear in mind whilst interpreting the imaging and devising scoring systems.

Conventional radiography
Conventional radiography of the hips is available in most health care centres, and is relatively inexpensive and quick to acquire. Traditionally, it has been the mainstay of imaging in JIA, although given the degree of cartilage in the developing hip joint, early erosive changes can be overlooked with this modality [13]. Whilst radiographic scoring systems for rheumatoid arthritis in adults have been used to define outcomes in clinical trials for therapies [14], they are not directly translatable to children given the growth disturbances and changes in anatomy with age (Fig. 2).
The only validated radiographic scoring system available for JIA of the hip is the Childhood Arthritis Radiographic Score of the Hip [15], developed by a panel of five paediatric rheumatologists and based on clinical experience and known radiographic features of hip disease in JIA. In this system, each hip is scored separately on a single frontal radiograph of the pelvis and awarded points based on the subjective assessment of joint space narrowing, erosive change, growth abnormalities, subchondral cysts, joint malalignment, sclerosis of the acetabulum and avascular necrosis of the femoral head. A maximum possible score of 16 for all variables can be assigned per hip joint (i.e. 32 per patient). In terms of repeatability scores, the intraclass correlation coefficients ranged between 0.76-0.98 for interobserver repeatability (in 2 observers scoring 381 hip 5 radiographs) and 0.82-0.96 for intraobserver repeatability (1 observer of a random selection of 37 patients, 3 months apart).
A test of clinical and construct validity between the Childhood Arthritis Radiographic Score of the Hip and clinical markers (such as the Childhood Health Assessment Questionnaire, physician and parent's global assessment of child's wellbeing) did not show any predictable correlation, although a change (not an absolute value) in the Childhood Arthritis Radiographic Score of the Hip between a baseline assessment and follow-up after approximately 1 year was able to give some measure of long-term destructive hip changes. As this scoring system was only tested by two experienced radiologists, further work is needed to determine how user-friendly the system can be and whether it is appropriate across a wider range of patient abnormalities. The increasing use of biological agents that prevent rapid destruction has stimulated the need for imaging modalities more sensitive in detecting inflammatory pre-erosive changes. Early softtissue changes seen at the onset of JIA in the hip are not well determined on radiographic imaging and the use of ionising radiation (particularly to the gonadal regions in a predominantly early pubertal age group) may limit the widespread usage of this technique.

Ultrasound
The lack of ionising radiation, low cost, ability to detect soft-tissue changes and dynamic real-time scanning of the hip joint makes US a useful tool for joint assessment and guidance of steroid injection therapy in the setting of JIA. Several joints can be assessed 6 at the same sitting, and previous studies have suggested that US may be more accurate at detecting synovitis than physical examination [16]. These factors together with the lack of sedation (which may be required in MRI for some children) make US examination an attractive modality for hip joint assessment (Fig. 3).
There are no standardized US imaging protocols specific to the hip in the setting of JIA; however, the European Society of Musculoskeletal Radiology provides an excellent free online guide [17] with anatomical correlation and sonographic features of the hip joint, detailing insertion sites of the surrounding musculature and probe location for optimal imaging of hip joint effusions.
Caution should, however, be taken when translating adult musculoskeletal sonographic findings to the paediatric hip, as age-related variations in thickness of cartilage, and the appearances of ossification centres and normal epiphyseal and metaphyseal vessels can mimic pathology [18]. Normal values for hip joint capsule and synovial cavity thickness have now been established in apparently healthy children by Zuber et al. [19], based on 816 US hip studies in 408 patients ages 0-18 years old. Notably, the data set was drawn from a cohort of children referred to a rheumatological centre, but without musculoskeletal disease being found. Neither images demonstrating the US appearances nor measurement sites were presented. The authors reported a reasonable inter-and intraobserver mean precision of approximately 1.8% and 12.5%, respectively, based on 3 sonographers assessing 34 randomly selected patients in this cohort.
Standardized US scoring systems for the hip joint in JIA have not been established and studies thus far have mostly detailed correlations of sonographic 7 appearances and clinical symptoms in small joints. In one retrospective study of 92 JIA patients, 31.5% of patients demonstrated abnormal hip US findings (i.e. the presence of effusion and/or synovial thickening), which were highly associated with clinical symptoms of limited joint movement, but not with disease activity or subtype [20].
Another study comparing bilateral hand, wrist, ankle, hip and elbow ultrasounds in 27 JIA patients in clinical remission against 36 healthy controls discovered that there was a significant difference in the proportion of children with subclinical synovitis in the JIA group (41.7% vs. 11.1%). Unfortunately, hip involvement was not common in this subgroup and the majority of the abnormalities were seen in the elbows and ankles [21].
Finally, it has also been shown that US assessment after intra-articular steroid injection can monitor a response to treatment by showing reduction in joint effusion, although this was based on only four hips in one study [22].
Despite these interesting observations, true population-based reference standards as well as papers relating to the reliability and reproducibility of abnormalities with respect to the hip joint in JIA are lacking. US technique relies heavily on operator experience, patient cooperation, the depth of joint for adequate sonographic penetration (which may be harder to assess in larger, potentially obese and immobile children) and possible reduced visualization of the entire joint if pain limits patient movement of the joint. These factors may limit repeatability of further studies.

Magnetic resonance imaging
MRI is the only modality that can assess both the soft-tissue and bone marrow changes seen in JIA. Nevertheless, caution must be applied during MR interpretation as the 8 developing skeleton in children can pose several pitfalls. One such area includes understanding the predictable marrow transformation pattern from haematopoietic (red) to fatty (yellow) marrow, which starts in the peripheries (fingers and toes) before moving centrally (axial skeleton). Within long bones, the epiphyses first undergo fatty transformation, followed by the diaphysis and lastly the proximal metaphyses [23]. This pattern can sometimes lead to erroneous overcalling of abnormal marrow signal in the proximal femora on pelvic MRI studies. In addition, a rich, branching vascular network of epiphyseal vascular canals is normal in younger children and can result in diffusion of intravenous MR contrast agents into the surrounding cartilage. This can persist for several minutes after injection and easily fool the inexperienced interpreter into overcalling synovial enhancement and hypertrophy. As children develop and epiphyseal ossification ensues, the canals become less numerous [24]. Interpretation of imaging in child athletes and gymnasts should also be broached with caution given the subtle differences in normal developmental appearances, which include increased incidence of coxa valga, tendinous hypertrophy and oedema within the greater trochanter [25]. Whilst it is acknowledged that joint space narrowing is a feature of JIA, and that a small amount of joint fluid may be within normal limits, studies that provide absolute quantification of these parameters on MRI are lacking and still require subjective judgment by the experienced radiologist.
Different hospital departments usually have their own protocol for imaging an inflamed joint, although in general, the assessment of the hip joint includes the whole pelvis to allow visualization of both hips. The patient is imaged supine with the legs straight and the feet together in a neutral position. At the very least, MRI sequences will usually include one T1-weighted sequence (not fat saturated) to assess for appropriate bone marrow fatty conversion, a STIR (short tau inversion recovery sequence) to assess for bone marrow oedema and joint effusion and at least one contrast enhanced T1weighted sequence to assess for synovial enhancement and thickening (Fig. 4). To ensure accurate comparison between previous and present examinations, timing of postcontrast images should be standardized. Based on the existing, but sparse literature, we would suggest an interval of 3-5 min [26]. To aid diagnostic interpretation, imaging in two planes of view (axial and coronal) is preferential [27]. The MRI protocol used at Great Ormond Street Hospital is given in Table 1, performed on a 1.5-T MRI system with gadolinium-based contrast agent administered intravenously at standard dose.
As yet, a validated and universally utilised MRI scoring system for the hips has

Kirkhus et al. [30] utilised MRI to identify differences in appearances of hip joint
in hip arthritis. Their scoring system was created to apply across all affected joints rather than be specific to the hip joint, and more descriptive in nature. This encompassed a score of 0 to 3 for the degree of joint fluid present (0=none to 3=distention of the joint capsule), presence of synovitis (enhancement or thickening of >2 mm), bone marrow oedema (none, less than or greater than one-third of the epiphysis), bone erosions (yes/no), softtissue oedema (none, trace, marked) and maximal short-axis diameter of the largest regional lymph node. In their prospective imaging review of 59 patients, it was found that four patients with infectious arthritis (two of whom had hip joint imaging) had abnormally low contrast enhancement in the hyaline cartilage. This finding was statistically significant compared to imaging features in children with JIA, although the overall numbers of participants in the different arthritis groups and those with hip imaging were very small overall to generalize the conclusions. Although it is interesting to note MRI differences between arthritis subgroups, the scoring for the above mentioned studies was only performed once (either by a single or group of radiologists together) and at one time meaning repeatability data for MRI imaging scoring systems are also lacking. The scoring systems also do not take into account all changes that can be seen in JIA (e.g., growth disturbances) and may be simplistic for clinical trial usage. Information on reliability, repeatability and correlation of imaging findings with treatment outcomes is still largely unknown.