Introduction
Juvenile idiopathic arthritis (JIA) is the most common rheumatic disease in childhood and an important cause of short- and long-term disability [
1]. Biologic disease-modifying antirheumatic drugs (bDMARDs) are proven effective in JIA patients and have been successfully implemented in the standard treatment regime [
2‐
5]. Still bDMARDs are costly and come with (dose-dependent) side effects [
2,
6‐
8]. Reducing or stopping bDMARDs (when disease activity is low) might therefore be a valuable strategy to reduce side effects (in particular infections) and costs. Dose reduction in adult population of rheumatoid arthritis (RA) patients has been proven to be successful, yet discontinuation is not recommended, as many adult RA patient flare after discontinuation [
9‐
13]. However, some JIA patients may recover spontaneously, with studies reporting over 50% of patients being in clinical remission off medication, 30 years after disease onset. Children therefore have more favorable outcomes than RA patients, this suggests that discontinuation might be a viable option in JIA patients [
14,
15]. However, the evidence about the success rate of discontinuation bDMARDs in JIA is not summarized yet. Furthermore when, how and in whom to discontinue bDMARDs in JIA is not known.
The aim of this review was to conduct a scoping review of all available evidence on relapse rates and relapse associated variables after bDMARD discontinuation in children with non-systemic JIA.
Methods
This review was guided by the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews Checklist, checklist and additional information can be found in supplement
1 [
16].
To better guide bDMARD discontinuation in children with non-systemic JIA, the following key questions were formulated;
1.
What is the relapse rate and time to relapse after discontinuation of a bDMARD?
2.
Which factors are associated with flares after treatment discontinuation;
a.
Are there differences in flare rate between age, type of bDMARD or JIA subgroups?
b.
Does disease severity (e.g., time in remission, treatment or disease duration, time to remission) affect flare rate or flare severity after bDMARD discontinuation?
c.
What is the difference between flare rate/flare severity when patients are tapered before bDMARD discontinuation compared to abrupt discontinuation?
d.
Does the use of multiple bDMARDs or concomitant therapy affect flare rate or flare severity, and when concomitant therapy is used, is there a preference in stopping one before the other?
Search strategy and eligibility of the studies
Eligible articles were identified by a systematic search of PubMed/MEDLINE database. The initial search was performed in April 2021, and was repeated in July 2022 to include articles published between April 2021 and July 2022. Different MeSH terms and synonyms for JIA, bDMARD, Abatacept, Adalimumab, Etanercept, Golimumab, Infliximab, and Tocilizumab were used. The complete list of used search terms and complete search strategy are set out in supplement
2.
Titles and abstracts were independently screened by two researchers. Any discrepancies were resolved by consensus agreement between the two investigators. After reading title and abstract, original articles written in English, Dutch, French or German reporting on bDMARD discontinuation in non-systemic JIA patients were selected. Review papers, guidelines, case reports and studies focusing on systemic onset JIA (soJIA), uveitis and psoriasis were excluded. Subsequently, after reading the full text, studies lacking data on previously mentioned key questions, studies with overlapping data and studies of which no full text article was available were excluded. Using references mentioned in relevant articles, a further manual search for additional articles was conducted.
Data selection
Data charting was primarily performed by one researcher. When data was ambiguous, interpretation of this data was verified by a second researcher. Data regarding study design, number of patients undergoing bDMARD discontinuation, mode of discontinuation, medication use, disease course, flare rate and response to therapy after flare were extracted from each eligible study. We defined patients as having flared when the study described the patient as having flared, relapsed, when therapy was restarted or when the patient was no longer defined as being in remission by the study. Additionally, patient demographics of the relevant study groups were noted, if provided. Statements about possible correlations to flare were obtained from text or extracted from presented tables or supplements when possible. After the outcome measures of the eligible studies were tabulated, they were analyzed with a purely descriptive approach.
Discussion
This review summarizes the current knowledge on relapse rate, time to relapse and possible flare associated variables after discontinuing bDMARDs in non-systemic JIA patients. Sixty to 78% of patients flared within one year after discontinuation, mean time to relapse was 2 to 8.4 months. None of the possible flare associated variables could definitively be linked to flares. Studies reporting a correlation were opposed with multiple studies finding no correlation, or correlations contradicting each other altogether. Comparison of data was further complicated by studies not displaying their data on possible flare associated variables in a verifiable and uniform manner.
A former review by Halyabar et al. also looked at treatment withdrawal in JIA but did not focus specifically on the non-systemic JIA group [
46]. As natural flare rates in soJIA are known to be significantly lower compared to other JIA subgroups [
47], it is important to look at these groups separately in relation to relapse rates following discontinuation of bDMARDs.
In the adult RA population relapse rates have also been studied. In this population, like in JIA, the treatment goal is to achieve remission soon after onset of RA, followed by the most optimal treatment that results in the lowest possible disease activity, the least adverse events and the lowest costs. This is achieved by tapering and when possible discontinuation of medication. In this adult population tapering of TNFi to a more optimal dose has been shown to be safe and feasible when low disease activity or remission is reached [
48,
49]. Furthermore, disease activity guided dose tapering seemed to be non-inferior to continuation of full dose TNFi [
50,
51]. This is in line with the studies of Cai and Mori reporting no loss of effectiveness after tapering ETN in a JIA population [
44,
45]. Still tapering can be accompanied by (temporary) flaring. Fautrel et al. found a relapse rate of 77% after 18 months in RA patients tapering ADA and/or ETN every 3 months followed by discontinuation, compared to 47% in RA patient continuing ADA and/or ETN in a standard dose [
52]. Unfortunately, similar to our findings in the JIA population, no markers for successful tapering have been found [
53] and even a multi-biomarker score could not predict successful tapering in adults [
54]. Nevertheless, protocolised tapering of TNFi seemed to be cost effective in the RA population [
55]. Studies have shown that discontinuation without prior tapering is inferior to full dose continuation in the adult RA population [
48]. It is therefore recommended to taper medication when low disease activity is reached in RA patients, thereby identifying a more optimal dose, as well as identifying patients who are able to discontinue their TNFi [
11]. It is not yet clear if this recommendation is applicable to the JIA population. One could argue that a more favourable disease course of JIA in general justifies a more liberal tapering policy.
When interpreting the data in this review there are some considerations to keep in mind. First, flares occur frequently in JIA patients in inactive disease, even when medication is not stopped. Guzman collected data of 1146 JIA patients in inactive disease (receiving different forms of therapy) and found 42.5% of patients developing a flare within one year of achieving inactive disease, with 26.6% of patients developing a significant flare, requiring treatment intensification [
47]. Therefore, some of the flares reported in the included studies in this review could also be due to the natural course of the disease and not directly related to stopping of medication.
Another aspect that one should consider, is the likely occurrence of selection bias: JIA patients receiving bDMARDs are probably of a more severe subclass and more flares can be expected in this subclass, either in general or after stoppage. This is again illustrated by Guzman, who showed that bDMARD use was associated with an increased risk of flare (HR 1.65).
This was especially true in the earlier years of bDMARD use, when these drugs were preserved for the most severe patients failing all other therapies. Since treatment strategies changed, bDMARDs are given earlier in the disease course and to less severely affected patients, which will likely result in fewer flares. Earlier studies in this review may therefore report higher flare rates than one would find today.
In addition to publication date, different discontinuation and treatment policies between studies should be kept in mind when interpreting data. For example,
time in clinical remission could be longer because of cautious discontinuation policies in general, or as a result of a decision to withhold discontinuation in a specific patient due to severity of the disease in this patient. These different reasons for a longer
time in clinical remission could very well explain the contradictory statements of correlation of
time in clinical remission and flare made by Prince, Simonini, Su and Lovell [
22,
27,
35,
42]. Likewise, selection bias could be an explanation for other correlations found in other flare associated variables. Health professionals could have been more inclined to prescribe concomitant MTX to more severe JIA patients and consequently, more flares could have been found by Lovell or Pratsidou-Gertsi [
22,
38]. Unfortunately, the articles did not present enough information to verify these potential explanations.
The most important limitation of this review is that more than half of the selected studies did not select flare rate as their primary outcome. Therefore data was not always sufficiently powered and statements on flare associated variables could not always be verified numerically. Furthermore, data regarding patients lost to follow up was missing in these studies. Additionally, even though we excluded soJIA in order to make the study group more homogeneous, JIA still is a heterogenous disease and general statements are therefore difficult to make. Finally, the follow up in the studies rarely exceeds one year, as only Klotsche reported flare rates up until 48 months after discontinuation [
20]. Long-term results of discontinuation are therefore still unclear. Despite these limitations the presented data in this review give an adequate representation of the current JIA population as data regarding relapse rates is comparable among the different studies.
Future studies are needed to show if
delayed remission is indeed a risk factor for development of flares post bDMARD stoppage, as suggested by Su et al. [
42]. A large multicentred cohort evaluating bDMARD discontinuation in JIA patients, using the same definitions for inactive disease, could provide more insight into this topic and might identify other possible flare related variables. We furthermore encourage future studies to display finding in a verifiable manner using uniform definitions and with a separate analysis of soJIA and non-systemic JIA subgroups.
In addition to patient characteristics and variables, biomarkers might also be helpful in predicting successful discontinuation [
18,
23,
56]. Furthermore, ultrasound guided discontinuation is suggested as a mode to reduce flares after discontinuation, however studies outcomes are conflicting [
57‐
60]. In summary, still more research needs to be conducted before a biomarker/ultrasound-based discontinuation strategy will be ready to be implemented.
In conclusion, this review showed that 60 to 78% of non-systemic JIA patients flare within one year after discontinuation, with a mean time to relapse of 2 to 8.4 months. Flares could not be reliably predicted by any predetermined variable at this point in time, mainly due to a lack of sufficient studies that primarily focussed on relapse rates and associated variables. This is the first study summarizing data on relapse rate and associated variables in non-systemic JIA patients after withdrawal of bDMARDs. Our current overview highlights the importance of future research and identifies several focus points for these studies. However this review shows that discontinuation of bDMARDs is feasible in JIA patients in general, and can be of assistance in daily practise in informing JIA patients and their parents on discontinuation of biologic therapy.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.