Conventional radiography in juvenile idiopathic arthritis: Joint recommendations from the French societies for rheumatology, radiology and paediatric rheumatology

Background Juvenile idiopathic arthritis (JIA) can cause structural damage. However, data on conventional radiography (CR) in JIA are scant. Objective To provide pragmatic guidelines on CR in each non-systemic JIA subtype. Methods A multidisciplinary task force of 16 French experts (rheumatologists, paediatricians, radiologists and one patient representative) formulated research questions on CR assessments in each non-systemic JIA subtype. A systematic literature review was conducted to identify studies providing detailed information on structural joint damage. Recommendations, based on the evidence found, were evaluated using two Delphi rounds and a review by an independent committee. Results 74 original articles were included. The task force developed four principles and 31 recommendations with grades ranging from B to D. The experts felt strongly that patients should be selected for CR based on the risk of structural damage, with routine CR of the hands and feet in rheumatoid factor-positive polyarticular JIA but not in oligoarticular non-extensive JIA. Conclusion These first pragmatic recommendations on CR in JIA rely chiefly on expert opinion, given the dearth of scientific evidence. CR deserves to be viewed as a valuable tool in many situations in patients with JIA. Key Points • CR is a valuable imaging technique in selected indications. • CR is routinely recommended for peripheral joints, when damage risk is high. • CR is recommended according to the damage risk, depending on JIA subtype. • CR is not the first-line technique for imaging of the axial skeleton. Electronic supplementary material The online version of this article (10.1007/s00330-018-5304-7) contains supplementary material, which is available to authorized users.


Introduction
Juvenile idiopathic arthritis (JIA) is a heterogeneous group of chronic inflammatory joint conditions that can cause structural damage [1]. Seven mutually exclusive subtypes of JIA are defined in the 2001 Edmonton classification developed by the International League Against Rheumatism (ILAR) [2]. This classification has been challenged and modifications suggested, such as exclusion of systemic-onset JIA (sJIA) due to its similarity to autoinflammatory diseases [3,4].
The prevalence of joint damage among patients with JIA has been estimated at 8-27 % in extended oligoarticular JIA (oJIA), 35-67 % in polyarticular JIA (pJIA) and up to 80 % in rheumatoid factor (RF)-positive pJIA [5,6]. The main treatment objectives in JIA are to control the pain and to prevent structural damage. Joint space narrowing (JSN), bone erosions and demineralization are radiographic findings shared between JIA and adult rheumatoid arthritis (RA). Changes specific to the paediatric population are early growth plate closure, epiphyseal deformity and growth asymmetry [7].
Conventional radiography (CR), magnetic resonance imaging (MRI) and ultrasound (US) are the imaging modalities most often used to evaluate joint inflammation or structural damage [8]. MRI and US hold considerable promise but are still under evaluation in JIA. CR remains the most readily available imaging technique for detecting and monitoring structural damage. However, potential limitations of CR in JIA include the risk of radiationinduced harm to the patient, interpretation difficulties raised by skeletal immaturity, and the delayed development of structural joint damage. Furthermore, because JIA is rare, little is known about the potential effects of synthetic or biological disease-modifying anti-rheumatic drugs (DMARDs) on structural joint damage [9][10][11]. Thus, whereas recommendations based on large studies are available for the radiographic assessment of chronic inflammatory joint disease in adults [12,13], no similar guidelines have been developed for JIA. A task force was recently convened by the European League Against Rheumatism (EULAR) -Paediatric Rheumatology European Society (PReS) to develop recommendations about imaging studies for diagnosing and managing JIA [14]. Although this undertaking acknowledged, for the first time, that an assessment of imaging studies in JIA was needed, the task force neither focussed on CR nor provided specific guidance for everyday practice.
We established a multidisciplinary task force to develop guidelines on the use of CR for the diagnosis and follow-up of each JIA subtype in everyday practice. Our project was supported by the French Society for Rheumatology (SFR), French Society for Paediatric Rheumatology and Internal Medicine (SOFREMIP), French Society for Paediatric and Prenatal Imaging (SFIPP), French Society for Radiology (SFR), and largest non-profit paediatric rheumatology patient organisation in France (KOURIR).

Field of research
We considered the following situations, at diagnosis and during follow-up, in each of the following five subtypes of JIA (oJIA, pJIA with and without RF and/or anti-citrullinated peptide antibody (ACPA), juvenile psoriatic arthritis (jPsA), and enthesitis-related arthritis (ERA)) Undifferentiated arthritis, as a heterogeneous subset related to one or several subtypes, and systemic JIA, having a peculiar articular course and structural prognosis, were left aside. Experts also focused on juvenile monoarthritis. Special attention was directed to the cervical spine, hip and temporo-mandibular joints (TMJs).

Recommendation development process
The task force comprised 16 JIA experts (eight rheumatologists, five paediatricians, two paediatric radiologists experienced in skeletal disease and one patient organisation representative). We used the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) method [15,16] for elaborating, evaluating, disseminating and implementing recommendations elaborated by the EULAR and the Outcome Measures in Rheumatology (OMERACT) group [17,18], and the Population, Intervention, Comparison, Outcome (PICO) process to frame the research questions.
We considered structural radiographic abnormalities: JSN, erosions, pseudo-joint space widening for sacro-iliac joint [19,20] and ankylosis [12]. A research fellow (PM) assisted by two experts in systematic review methodology (CGV, methodologist; and VDP, convenor) performed a systematic literature review by searching PubMed, Scopus/Elsevier, and the Cochrane Library. Original articles including clinical trials, retrospective cohort studies, other retrospective studies, and case-control studies published between 1980 and December 2016 were identified. The following indexing was used: 'juvenile idiopathic arthritis' OR 'juvenile rheumatoid arthritis' OR 'juvenile chronic arthritis' OR 'juvenile psoriatic arthritis' O R ' e n t h e s i t i s -r e l a t e d a r t h r i t i s ' O R ' j u v e n i l e spondyloarthritis' AND 'radiography' OR 'X-ray' (see Appendix 1 for details). The quality of evidence and grades of recommendation were determined according to the standards of the Oxford Centre for Evidence-Based Medicine [21]. Recommendations were graded A to D depending on the level of the underlying evidence (from 1A to 4) [18].
The task force debated and formulated a preliminary set of recommendations based on the systematic literature review supplemented, when necessary, by their expert opinion. This set was then evaluated by a panel of 14 independent Frenchspeaking experts. Modifications were debated by the task force. The final recommendations were then rated on a 10point scale by the task force and independent panel through a Delphi process.

Recommendations
The experts elaborated four overarching principles and 31 recommendations. Table 2 lists the recommendations.

Overarching principles
Radiation exposure was taken into account (principle B), according to French Society for Radiology recommendations [90] (Appendix 2). Much of the cartilage is still radiotransparent in children younger than 5 years of age. In this age group, the need for CR must be evaluated with great care (principle C) [91].
Other imaging modalities such as US and MRI are increasingly used in JIA. Although promising, they are not discussed herein. They will be the focus of specific recommendations (principle D).
Oligoarticular JIA (oJIA) 1. CR should not be performed routinely as a diagnostic investigation in oJIA. The literature review identified ten studies in which CR was performed, even in patients younger than 4 years. Among them, one focussed specifically on oJIA [35] and nine investigated several JIA subtypes but reported data separately for oJIA [6, 24, 27, 36-38, 40, 42, 43]. The usefulness of CR is limited by the incomplete ossification of the epiphyses, most notably in the youngest age groups [33]. Therefore, when the diagnosis is definitive, CR is not recommended.
2. and 3. During follow-up, CR should be performed on affected joint(s) that remain symptomatic after 3 months. By 'symptomatic joints'*, we mean painful and/or swollen joints and/or joints that are limited in motion. In patients with persistently symptomatic* joints, the reiteration of CR during follow-up is at the discretion of the physician. Several studies showed evidence of radiographic progression early in the natural history of oJIA [24,27,35,38]. 4. In patients with clinically inactive disease (CID), CR should not be performed routinely. The diagnosis of CID relies on physician judgement, aided by validated tools [92][93][94]. No data are available on radiographic disease progression in clinically silent joints in patients with oJIA. 5. In patients with extended oJIA, the recommendations for pJIA should be applied. The number of affected joints is strongly associated with structural damage in oJIA [35]. 6. In patients with structural damage, the selection and timing of specific imaging techniques to further assess the damaged joint during follow-up is guided by clinical considerations.
Joints with structural damage must undergo specific CR evaluations during the patient's growth.
Polyarticular JIA (pJIA) 7. and 8. Routine CR of the wrists, hands and forefeet is strongly recommended at the diagnosis of polyarticular JIA with positive RF/ACPA. CR of other joints than wrists, hands and forefeet, is recommended at the diagnosis for symptomatic joints*only. Prospective studies were reviewed, with special attention to early pJIA. Erosions and JSN occurred preferentially at the hands, wrists and feet [11,31,43,[48][49][50][51], joints that were sometimes asymptomatic [31] CR at the diagnosis provides a reference for assessing disease progression. It is supported by 'adult' recommendations [13] for rheumatoid arthritis, which has a similar structural evolution. 9. and 10. In new-onset RF/ACPA-negative pJIA with adverse prognostic factors, CR at diagnosis should be performed as for RF/ACPA-positive pJIA. Box 1 lists the factors of adverse prognostic significance in pJIA [31,44,50,51]. These factors are associated with a pattern of joint damage over time similar to that seen in RF/ACPA-positive pJIA [38].
Box 1: Factors of adverse prognostic significance in polyarticular juvenile idiopathic arthritis (pJIA) Early involvement of wrists Symmetric arthritis Distal, small-joint arthritis Elevated ESR/CRP Pre-existing radiographic abnormalities ESR, erthrocyte sedimentation rate; CRP, serum C-reactive protein level 11. In new-onset, RF/ACPA-negative pJIA without adverse prognostic factors, at diagnosis, CR should be confined to symptomatic* joints. This recommendation is based on expert opinion.
12. In RF/ACPA-positive pJIA, CR of the hands, wrists and forefeet is strongly recommended 1 year after disease onset, and when transitioning from paediatric to adult healthcare. At other time points, the use of CR during follow-up is at the discretion of the physician. Prospective studies found evidence of joint damage even in asymptomatic joints [31]. Patients with long-standing disease had high prevalences of joint erosions (30-70 % in historical studies) [5,28,38,40,44,48,54], close to those in adults with RA [48]. In RA, joint destruction at asymptomatic sites is a major predictor of adverse outcomes [13,95]. However, radiographic progression with erosions in asymptomatic joints is not well documented in JIA and may have been underestimated. In a study of 471 joints in 67 patients with polyarticular JIA, radiographs showed erosions at the hands and feet in 36 % and 39 % of   [31]. Our literature review identified some data on the best times for CR. One study suggested a higher risk of radiographic progression within the first year after disease onset [51]. The experts felt that CR contributed to ease the transition from paediatric to adult healthcare [96]. 13. Routine CR of other joints is not recommended. No data were found on which to base specific recommendations.
14. During the follow-up of RF/ACPA-negative pJIA with adverse prognostic factors, CR should be performed as for RF/ACPA-positive pJIA (see recommendation #12). 15. During the follow-up of RF/ACPA-negative pJIA without adverse prognostic factors, the use of CR is at the discretion of the physician. No scientific data were available on which to base specific recommendations. 16. and 17. CR can be repeated in patients who remain symptomatic* longer than 3 months. In patients with structural damage, the selection and timing of specific imaging techniques during follow-up is guided by clinical considerations. The experts emphasised the need for careful attention to joints with active disease. In prospective studies, the time interval separating CR assessments of the same joints ranged from 8 months to 24 years. The 3-month interval in this recommendation was based on expert opinion.

Enthesitis-related arthritis (ERA)
18. In patients with axial ERA, CR of the spine and hip joints should be performed only when needed for the differential diagnosis. Axial manifestations may arise at the spine, hips and sacro-iliac joints. A radiographic view specifically designed to assess the sacro-iliac joints is not recommended, as the results are not interpretable in skeletally immature patients and radiation exposure is significant [20]. In patients with axial inflammatory pain, MRI (for both sacro-iliac and hip joints) and US (for the hip joint) may be more relevant [67]. 19. During the follow-up of axial ERA, CR should be considered only for the hip joints, depending on the clinical course and availability of US and/or MRI. ERA is associated with a high prevalence of hip joint arthritis [30,56,[58][59][60]. MRI or US are non-irradiating methods capable of detecting hip joint effusion; in addition, MRI can detect bone oedema. Therefore, in the future, MRI and US may deserve consideration as first-line imaging techniques. CR, however, is appropriate for monitoring known structural damage and deformities.
20. and 21. CR is not recommended for multifocal enthesitis. In patients with isolated enthesitis, CR can be JIA juvenile idiopathic arthritis, CR conventional radiography, oJIA oligoarticular juvenile idiopathic arthritis, pJIA polyarticular juvenile idiopathic arthritis, RF rheumatoid factor, ACPA anti-citrullinated protein antibody, ERA enthesitis-related arthritis, TMJ temporo-mandibular joint *Symptomatic joints: swollen and/or painful joints, and/or joints with motion range limitation considered as a tool for establishing the differential diagnosis. When isolated enthesitis is suspected, CR may contribute to the differential diagnosis (e.g. with post-traumatic changes or osteochondritis); otherwise, CR is unhelpful for assessing peri-articular manifestations.

22.
No specific recommendation can be made about CR in juvenile psoriatic arthritis. Scientific data are scarce [62][63][64][65][66]68]. The definition of this entity is still debated [68]. Traditionally, two subtypes are described, an axial inflammatory disease resembling axial ERA and a peripheral joint disease resembling oJIA [66]. 23. Guidance may be taken from the recommendations above, depending on the clinical presentation, or from recommendations issued for adults.

Situations of specific interest
Monoarthritis 24. At the diagnosis of acute monoarthritis, CR of the involved joint should be performed, with two perpendicular views. The French Society for Radiology [90] strongly recommends CR of any site of focal bone pain in paediatric patients, with the goal of excluding a tumour, osteomyelitis, or a haematological malignancy [34,97]. 25. At the diagnosis of acute monoarthritis, comparative CR of the contralateral joint is unnecessary. Because cartilage thickness varies within individuals, comparison to the healthy contra-lateral joint is uninformative [26,33].
Cervical spine 26. In patients with persistent neck pain related to JIA, MRI is preferable over CR.
27. When MRI is unavailable, CR is recommended only for the cervical spine and should consist only of a lateral view.
28. In patients with JIA who have neurological symptoms of spinal cord compression and neck pain, cervical MRI must be performed, on an emergency basis.
In a cohort study of oJIA, 2.4 % of patients had cervical spine damage at the diagnosis [35]. Cervical spine erosions and ankylosis are common in advanced pJIA [42,71]. Evidence-based data are too scarce to recommend any specific pattern of radiological follow-up. Atlanto-axial diastasis may be normal in paediatric patients, and dynamic CR is therefore irrelevant. MRI is the most sensitive imaging technique, and is mandatory when spinal cord compression is suspected [98].
Temporomandibular joints 29. CR of the TMJs is not recommended when cross-sectional imaging is available.
TMJ damage is common in JIA, with the prevalence ranging across studies from 17 % to 87 % [73]. The TMJ cartilage is thin and condylar erosions therefore develop early. The panoramic radiograph is often normal at disease onset.
Cross-sectional imaging offers better diagnostic performance. Imaging of the TMJs is not usually performed on a routine basis but is required in the event of pain, mouth-opening limitation or audible cracking of the TMJs [74, 76-81, 83, 84]. MRI is considered the best imaging technique, although distinguishing the normal appearance from abnormal changes can be challenging [99,100]. Cone-beam computed tomography allows three-dimensional reconstructions [101]. The usefulness of US TMJ imaging is under debate [77,102].
Hip joint 30. Routine CR of the hip joint is not recommended in patients with pJIA.
31. When CR of a symptomatic hip joint is performed, a single view should be obtained, i.e. either an antero-posterior view or a frog leg view.
In RF/ACPA-positive pJIA, hip joint damage is common [48] but CR of the hip joint is associated with a high level of ionising radiation exposure, so the hip is not among the joints for which routine CR is recommended .When available, MRI should be performed instead of, or in addition to, CR. If CR is performed, either an antero-posterior or a frog leg view is recommended, to visualise both hip joints and to allow the detection of bone erosions and/or avascular necrosis.

Discussion
CR is the most widely available imaging procedure worldwide. In paediatric patients, this advantage should be weighed against the heightened risks of radiation exposure and difficulty in interpreting joint radiographs before skeletal maturity is achieved. In addition, in JIA, radiographically visible joint damage takes time to develop, limiting the usefulness of CR. Specific recommendations about CR in paediatric patients are therefore needed, a fact that prompted the present work.
Obstacles to the development of recommendations about CR in JIA included the paucity of strong evidence about structural disease progression in JIA and the pooling of JIA subtypes in many studies. The low incidence of JIA contributes to explain the dearth of data. To maximise the usefulness of our recommendations to all physicians caring for patients with JIA, we focussed on CR and separated the five non-systemic, nonundifferentiated subtypes of JIA. Importantly, these recommendations are based not only on recently published data, but also, in many cases, on expert opinion, due to the paucity of paediatric studies. As a result, many of our recommendations are low grade, and in some cases obtaining guidance from recommendations for adults would seem to be the only option. However, the level of agreement among the multidisciplinary experts sitting on our panel was high.
Structural damage requires evaluation in JIA, especially in pJIA and extended oJIA, which carry the highest risk of adverse outcomes. In the treatment plans for pJIA developed by the CARRA, CR changes are considered an important outcome and their yearly assessment is suggested [55]. However, the risk associated with exposure to ionising radiation during CR is of major concern, as pointed out by the representative of the patient organisation during our study. Little evidence is available on which to base an objective quantification of this risk. Our experts considered that the risk was substantial for CR of the pelvis and lumbar spine but was too small at peripheral sites to constitute an argument against using CR. To minimise radiation exposure, the experts recommended having CR performed at centres with expertise in paediatric radioprotection.
Research is needed in a broad range of areas to fill the knowledge gaps we identified when developing our recommendations (Box 2). More specifically, most paediatric clinical trials failed to assess potential treatment effects on structural damage. Also, data on structural damage just before the transition to adult healthcare are needed, since treatment recommendations for adults are based on structural damage.
Box 2: Research agenda We considered neither MRI nor US, both of which are under evaluation in JIA. Both are non-irradiating, and US is also widely available and inexpensive, although it requires specific training. US is now performed almost routinely in adults with joint disease. In paediatric patients, however, differentiating normal from abnormal findings by MRI and US can be challenging [100,103]. Furthermore, very few physicians are specifically trained in paediatric US. The OMERACT and Health-e-Child Radiology groups are currently working together to standardise MRI protocols and interpretation in JIA [104][105][106].
In conclusion, CR still appears relevant in many situations in patients with JIA. CR is a widely available and inexpensive investigation that has an acceptable safety profile and can provide essential information about the structural course of the disease. Until validation studies of other imaging techniques, such as MRI and US, are completed, CR will remain the investigation of reference for assessing structural joint damage in patients with JIA.
Statistics and biometry No complex statistical methods, were necessary for this paper.
Ethical approval Institutional Review Board approval was not required; the methodology entirely relies on literature review and expert opinion.
Informed consent Informed consent was not required because no human subjects were involved.