Background
Juvenile idiopathic arthritis (JIA) is a potentially chronic disease that comprises 7 categories of childhood arthritis of unknown cause, that persists for more than 6 weeks and starts before the age of 16 [
1]. Osteopenia, bony deformity, erosions, and cartilage loss in carpalia, resulting in carpus shortening, can be complications of inflammation in JIA patients [
2‐
5]. Previous studies have shown that early damage on conventional radiography is correlated with functional deterioration and radiographic progression after 5 years [
2,
5,
6], and also with smaller chances to achieve clinical remission [
7]. Monitoring of radiographic damage progression is therefore important to evaluate treatment effect and predict prognosis. Since joint damage is assumed to be the result of ongoing inflammation, reaching inactive disease as early as possible and thereby preventing structural joint damage and consequently limitations in physical functioning, should be the goal of treatment [
8]. This is facilitated by the availability of new effective disease modifying antirheumatic drugs (DMARDs) [
9]. In accordance, current JIA treatment recommendations focus on earlier introduction of DMARDs aiming to achieve remission or at least low disease activity [
8,
9].
We have recently performed a randomized clinical trial using the treatment-to-target approach in recent-onset JIA patients, comparing 3 strategy-arms with different initial and subsequent treatment steps, aiming at inactive disease, including tapering and stopping DMARD therapy [
10,
11]. In this population we studied radiographic wrist damage using the Poznanski-score, at baseline and evaluated whether damage occurred or recovered with the abrogation of inflammation in the 3 strategy-arms. In addition we used the BoneXpert-method to determine the Bone Age (BA) and Bone Mineral Density (BMD) as markers for joint damage [
12].
Discussion
Our study is the first to describe longitudinal radiological outcomes of a tightly controlled treat-to-target approach, aimed at inactive disease during 24 months of treatment, in recent onset poly- and oligoarticular JIA patients. Despite a symptom duration of mean (SD) 7.6 (4.9) months and a JADAS-10 of 18.7 (5.6), at baseline, we found no significant differences in Poznanski-score and BA (as measured by the BoneXpert-method) of wrist radiographs compared to healthy children. Only in arm 3 BMD as measured by the BoneXpert-method was significantly lower than the normal reference population. After 24 months of treatment, there was no deterioration in any of the scores and in arm 3 BMD had statistically significantly improved. Mean JADAS10-scores over time were not associated with any of the radiological outcomes in this analysis.
Combined with rapid suppression of symptoms of active arthritis, prevention of damage is an important treatment goal in JIA. Damage has been most notably found in patients with longstanding and/or seropositive polyarticular JIA, but may also occur in seronegative polyarticular JIA and oligoarticular JIA [
19‐
21]. As has been shown in rheumatoid arthritis, it is thought likely that, also in JIA, damage progression is driven by inflammatory processes. Assessing damage in patients who are in very different phases of joint development can be challenging. In growing children, cartilage thinning, delayed or accelerated growth and reduced bone mineral density rather than bony erosions and joint space narrowing may indicate damage. Decreased bone age often reflects delayed bony maturation in JIA [
22,
23] but also increased focal bone maturation can be a result of joint inflammation [
15].
Compared to older cohorts [
12,
24] or recent cohorts with longer disease duration [
17], we found little damage at baseline in this cohort with recent-onset disease. Since we did not include patients with RF-positive polyarticular JIA, this could be a mildly affected cohort although initial JADAS10 scores were similar to other cohorts [
25]. In addition we found no significant damage progression. This is possibly due to our strategy of tightly controlled treatment-to-target aiming at inactive disease in all 3 treatment arms, resulting in rapid suppression of inflammation in most patients, without significant differences between the strategy-arms after 24 months. Only in arm 3 there was an initial greater clinical improvement [
10]. We cannot rule out an additional positive effect of use of etanercept, in all patients in arm 3, and in many in arms 1 and 2 after they failed to achieve remission on initial treatment with methotrexate (with or without temporary prednisone). Previous studies suggest that treatment with methotrexate cannot prevent joint damage progression whereas use of biologic DMARDs (used as initial treatment in our arm 3) may be more successful, although data are limited [
26,
27]. Apart from strategy, we did not find an effect of mean JADAS10-score over time, possibly in all patients due to rapid suppression of inflammation, therefore inhibiting the disease to have time to create damage.
To score differences in potentially little damage, we needed a sensitive scoring method. Conventional radiography has proven to be a useful modality to monitor wrist damage of JIA patients [
2,
4,
5,
13,
28‐
31]. Several methods, like the Dijkstra-score [
31], modified Sharp van der Heijde-score [
5,
19] the modified Larsen-score [
4,
32] and the Steinbrocker-scale [
33,
34] have been developed to evaluate radiographic damage to the osteochondral structures of the wrist and hand. The Dijkstra composite-score is limited in the grading of changes for severity over time [
35]. We stopped using the modified Sharp van der Heijde for pediatric assessment of joint damage [
5] as it proved too difficult to uniformly score subtle changes in joint space narrowing, bony erosions and bone deformity, as was recognized previously [
35]. Magnetic resonance imaging (MRI) and ultrasound (US) are suitable for monitoring disease activity for evaluating treatment response, and may also detect damage [
36]. However, interpretation of MRI findings of the osteochondral domain in JIA patients is challenging due to characteristics of the growing skeleton, in particular in hand and wrist joints. Bone marrow edema and bony depressions are also frequently seen on MRI in wrists of healthy children [
37‐
40]. Until now, no optimal method has been found to differentiate pathological and standardized age-specific findings in healthy children on MRI and US which limits their use to accurately assess damage and damage progression in the wrist of JIA patients.
The Poznanski-score, which measures relative carpal length on radiographs of the wrist, is able to detect deviating growth in absence of distinct joint space narrowing or erosions [
13]. A disadvantage of the Poznanski-score is that it requires open growth plates, which caused ineligibility in 14 of our patients, and unreliability in case of carpometacarpal erosions which hampers discriminating bony ends, which did not occur in our cohort. In addition, we used the relatively new BoneXpert method to score Bone Age and BMD, which, compared to a healthy reference population, can indicate damage due to inflammation.
The BoneXpert method, based on digital X-ray radiogrammetry (DXR), allows to determine the Bone Age and BMD compared to a normal reference population, at lower costs and with lower radiation than manually comparing the hand radiograph with images in the atlas by Greulich and Pyle [
16] and than measuring BMD by Dual Energy X-ray Absorptiometry (DXA) [
41]. The BMD measurement by BoneXpert is corrected for the size of the cortical bones to compensate for the high variation in stature of growing children, in contrast to DXA. Previous studies have reported on delayed bone maturation as reflected by negative Z-scores for Bone Age [
12,
17]. These studies had included patients with more severe or longstanding active disease. However, Borzutzky and others have warned previously, that determining bone age can be challenging in JIA due to accelerated maturation [
15,
42].
In JIA patients, BMD is often reduced [
12,
17,
43‐
45]. BMD was significantly lower at baseline in arm 3 (− 1.1 SD, (− 1.48; − 0.68)). This could indicate longstanding or more severe disease. Indeed symptom duration in arm 3 was slightly longer than in arms 1 and 2, although JADAS-10 scores at baseline were similar in the 3 arms. Possibly as a result of rapid and sustained suppression of inflammation, BMD improved significantly over time in arm 3. A previous study also reported improvement of BMD after therapy [
45]. It is speculated that this improvement is due to the anti-inflammatory effect of DMARD treatment [
46], more specifically due to etanercept [
47,
48]. However, no comparison cohort is available to prove that the treatment-to-target approach is responsible for a better radiological outcome.
Future studies are needed to delineate the effect of the treatment-to-target concept on improving bone health as reflected by bone maturation and BMD in JIA.
Our study has some limitations. Although comparable with other studies [
2,
26,
49] in children with JIA, we had a relatively small sample size (
n = 60), and we may have lacked power to detect small differences. In previous studies, results were based on clinically inflamed wrists only. Since we have examined also 6 patients with wrist radiographs of unaffected wrists, in this study we may have underestimated damage, although sensitivity analyses excluding patients who never had any clinical wrist arthritis over 24 months showed similar results. It remains to be determined whether joint damage is mainly due to local inflammation or (also) to systemic inflammatory processes of JIA.
Due to our choice of scoring methods, patients were excluded who had closed growth plates. Also we disregarded results of radiographs made outside the selected time frame. Follow-up time was relatively short compared to previous cohorts. However, often radiographic damage is expected to occur within the first one or two years [
2]. Finally, determination of bone health by BoneXpert software needs further validation, including further comparison with existing methods for the determination of BMD in JIA patients [
41,
50,
51].
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.