According to published data, symptoms and signs correlate poorly with the degree of cervical spine inflammation and possible subluxations. In adult population, cervical spine involvement is seen in 35–65% of patients with JIA [
3]. This wide range of reported cervical spine involvement in adult patients with juvenile-onset idiopathic arthritis could, to some extent, reflects recruitment bias since some studies were focused on clinical manifestations and others on radiographs or MRI changes [
3,
9,
21,
22]. In our study, all children had LROM, 80% had neck pain, two neck stiffness and only one torticollis. These findings are in line with the German study suggesting that the LROM in cervical spine and not pain is a leading clinical sign of cervical spine arthritis [
14]. The frequency of cervical spine involvement in our cohort of patients with JIA is estimated to be 4%. It is difficult to compare epidemiological data with other prevalence studies, because the duration of observation of patients in different studies varied or the studies were conducted only in selected JIA subtypes such as polyarticular and systemic forms of the disease. It is also important to note that different treatment modalities and protocols were used in the era before biologics, and more aggressive treatment, which is being used recently, may prevent some symptoms to develop. In addition, our study was conducted only in patients with cervical spine symptoms and some asymptomatic patients may have been missed.
A significant clinical observation in our study was that all children had peripheral joints involvement and in addition, 20% of them had co-existing unilateral or bilateral TMJ arthritis. The TMJ MRI examinations from our patients with TMJ involvement showed severe TMJ inflammation in all cases with intensive synovitis, joint fluid and significant morphological changes of mandibular condyles. In only few TMJ MRI examinations atlanto-axial joint was seen on T2 coronal sequence and we found in one patient signs of inflammation in atlanto-axial joint without clinical complains of pain and LROM. To identify patients with TMJ arthritis and possible clinically covert cervical spine involvement we suggest to include the atlanto-axial joints in coronal plane also on standard TMJ MRI examinations.
Imaging in patients with JIA is particularly important for joints that are difficult to assess clinically such as hip, sacroiliac joint and spine. In adults radiography of the cervical spine is accepted as mandatory in rheumatoid arthritis patients with neck pain and it is well standardized. On the other hand, studies reported low sensitivity of radiography in children because it provides morphological and structural information more typical for late phase of inflammation (dens deformation, erosive changes) and various forms of malalignment and ankylosis, which are rare in children as a first manifestation of cervical spine arthritis [
13,
23]. At least in the initial stage cervical spine radiography is not a routine imaging in children, but it can be useful in children with malalignment and persistent changes seen on MR to obtain functional information of cervical spine. The most frequently noted abnormalities seen on cervical radiography are apophyseal joint ankyloses or fusion at C2-C3, followed by aAAS, subaxial subluxation at levels between second and seventh cervical vertebrae, and erosions of the dens resulting in an “apple core” deformity [
9]. Atlantoaxial impaction (AAI) is another serious complication of longstanding rheumatoid arthritis, rarely seen in childhood. In our cohort radiography of cervical spine was performed in 20% of children to evaluate the malalignment and morphological changes of bones due to persistent LROM or signs of aAAS presented on MRI. Contrast-enhanced MRI of cervical spine has become the method of choice in evaluation of inflammation in this region [
11,
14]. Standardization of MRI protocol enables more reliable follow-up examinations and comparison of initial MRI with controls [
15,
24]. The main advantage of MRI is its ability to detect early, often subclinical disease by directly visualizing not only soft tissues changes in a form of synovial thickening/enhancement, joint effusion, facet joints inflammation, and ligament evaluation, but also bone changes in a form of edema, cortical thinning, morphological changes (especially dens deformation, evaluation of bone erosions), and malalignment. The presence of bone marrow edema was recognized as a strong predictor of future erosions and it is considered as pre-erosive abnormality in adults and indication for treatment initiation to prevent permanent joint damage [
25]. Early detection of joint pathology is of outmost importance with regards to the prompt therapeutic intervention. It is important to note that normal synovia in childhood can provide some degree of enhancement, and that there are no imaging criteria to clearly distinguish normal findings from early signs of arthritis. Moreover, postcontrast MRI should be performed during first 5–10 min of injection, because beyond this time diffusion of contrast material into the joints limits differentiation between enhancing synovia and adjunct joint fluid [
26]. Rheumatoid arthritis MR imaging score (RAMRIS) has been shown to have acceptable applicability to children in evaluation of the inflammation and response to treatment despite differences in the pediatric and adult skeletons, but was primarily designed for peripheral joint and not for assessment of cervical spine [
27,
28]. Synovitis of atlanto-axial joints was present on MRI in all our patients, and more than half of them had synovitis in other cervical joints compartments and facet joints. The degree of inflammation presented on MRI varied from mild (unilateral atlanto-axial inflammation) to moderate (bilateral inflammation) (Fig.
1) and severe (extensive pannus in atlanto-axial region)(Fig.
2). The early introduction of anti-TNFα in treatment of JIA patients with severe form of the disease provides an opportunity to delay or even prevent severe joint inflammation and subsequent destruction [
20,
29]. MRI as objective imaging method has important role in evaluation of treatment response. The evaluation of early treatment with biologics is usually done by comparing initial and follow-up MRI examinations in each child. No official scoring system for MRI evaluation of cervical spine arthritis in children is available [
15]. The only MRI based follow up study of cervical spine arthritis in children treated with anti-TNFα was done by Hospach et al. [
14]. The cohort of patients in this study was comparable to our patients and included 13 children, 12 of whom were treated with biologics, but with longer median disease duration of 1.7 years after the diagnosis of JIA (in our cohort 5.2 months) and shorter observation time. A variable duration of treatment with MTX and biologics has not been specified. Our study was in concordance with the abovementioned study in evaluation of inflammation, and showing a significant reduction of MRI signs of inflammation in cervical spine. The most significant difference between the study of Hospach et al. and our study was the evaluation of “chronic/late” changes. The follow-up MRI examinations in the study of Hospach et al. described more chronic changes in a form of malalignment in 3 patients after treatment, ankyloses in 3, erosion in 2, and narrowing of spinal canal in 3. In our study, half of the children with delayed biologic treatment had minor chronic changes compared to children with early introduction of biologics, where only one girl (#8) with severe and persistent inflammation developed more severe chronic changes, including dens deformation with bulging into spinal canal, thickened transverse ligament and persistent aAAS (Fig.
2). The low rate of long term consequences in our study could be due to earlier recognition with shorter disease duration and early aggressive treatment with biologics.
Limitations of our study include the retrospective design and small cohort of patients. Additional limitation is also the lack of a real gold standard in determination of the severity of neck arthritis. In spite of the limited number of patients, our results demonstrated that early and aggressive treatment of cervical spine arthritis with anti-TNFα could provide good results with elimination of inflammation (early response) and has a potential to reduce/deminish the chronic changes. According to our experience, in all children with persistent LROM MRI should be performed and if inflammation is confirmed, anti-TNFα treatment should be considered [
20]. The follow-up MRI examinations are individually addressed. However, according to our and others experiences [
14], we suggest the first follow-up MRI after 9 months, and if the clinical and laboratory responses are satisfactory, the second follow-up MRI should be performed 1 year later to evaluate possible development of chronic/late sequels. If the clinical signs of cervical arthritis persist, MRI should be performed earlier (after 6 months). Due to high incidence of co-existing TMJ inflammation (20% in our cohort), we suggest that the transversal plane during MR examination of cervical spine include TMJ (at least one of the fluid sensitive sequences). On the contrary, in children with MRI examination of TMJ atlanto-axial joint should be included in one of the fluid sensitive coronal sequences.