Physician practices for withdrawal of medications in inactive systemic juvenile arthritis, Childhood Arthritis and Rheumatology Research Alliance (CARRA) survey

Systemic juvenile idiopathic arthritis (SJIA) comprises 10–15% of juvenile arthritis. SJIA is unique from other categories of JIA, due to its similarity to monogenic autoinflammatory diseases, systemic features and response to IL1 inhibitors. The introduction of anti-IL1 and anti-IL6 agents revolutionized the treatment of SJIA allowing the majority of patients to achieve inactive disease states [1].

SJIA can follow a monophasic (one episode of disease followed by remission), polyphasic (multiple flares of either systemic or arthritic features), or persistent (unremitting) course. Currently there are no genetic or immunologic clues that allow treating physicians to determine which of these three courses patients will follow over time. Hence, clinical decisions on when to continue, taper, or withdraw medications are often subjective and vary from physician to physician. Clinicians need to balance the risk of medication withdrawal causing disease flare versus the risk of continued medication exposure and related side effects.

Childhood Arthritis and Rheumatology Research Alliance (CARRA) is the largest pediatric rheumatology network across North America, committed to fostering research and innovation in pediatric rheumatologic diseases. This paper describes a CARRA survey conducted by the SJIA workgroup, to study current physician practices across North America, regarding the definition of inactive disease in SJIA, and withdrawal patterns of medications in SJIA once inactive disease is reached.

CARRA membership has grown exponentially since inception and as of June 2017, it has 555 members across 119 sites in United States and Canada. The JIA workgroup consists of 117 members. The SJIA workgroup consists of 62 pediatric rheumatologists.

At the 2015 CARRA SJIA workgroup meeting (Austin, Texas) there was a unanimous decision to conduct a survey examining current attitudes and trends amongst pediatric rheumatology physicians for withdrawal of medications in inactive disease in SJIA. REDCap is a secure web-based application that facilitates building and housing online surveys [2]. A cross-sectional anonymous electronic survey using REDCap was distributed to 100 randomly selected physician-voting members of the CARRA JIA workgroup. Seattle Children’s Hospital (STUDY00000532) institutional review board (IRB) and Hospital for Special Surgery (2017–0276) approved the study. The study was deemed exempt by the IRB and consent was consent was not required.

Additional file 1: At the 2016 CARRA SJIA workgroup meeting (Toronto, Ontario), face-to-face input from group members (n = 63, 2 were parent representatives including BE, coauthor) was obtained to develop several fields, questions and case scenarios for the survey, following which a smaller core group refined and finalized the survey. CARRA leadership and Protocol Evaluation Committee approvals were obtained prior to dissemination of the survey. An 80% response rate was required including opt out responses such as 1) I do not take care of patients; 2) I do not take care of patients with this condition and 3) other reason. Responders were tracked (with actual responses being anonymous) and non-responders were sent regular reminder e-mails until 80% response rate was achieved. Respondents were asked to consider patients with SJIA only, as other categories of JIA were excluded. The survey had three broad sections: a) demographic information (years of practice post-fellowship and number of hours/week spent in clinical work); b) physicians’ opinions on clinical inactive disease (CID) in SJIA and c) existing practices when withdrawing medications in SJIA during CID or clinical remission on medications (CRM). The American College of Rheumatology or Wallace criteria define CID as satisfying all of the following: absence of active arthritis, uveitis and fever/ rash/ serositis/ splenomegaly/ lymphadenopathy due to JIA, normal inflammatory markers, physician global assessment 0 and ≤ 15 min morning stiffness [3]. CRM is defined as presence of CID for at least six continuous months on medication [3].

Multiple-choice questions were used and branching logic was introduced for several questions to facilitate and tailor participant responses. For certain fields free text was allowed, for example if participants did not agree with Wallace definitions for CID in SJIA. When determining what factors play a role in the decision to withdraw/ taper therapy, participants were given a list of factors (generated at 2016 SJIA workgroup meeting) and asked to rank the top five factors and quantify the importance of each (very important, moderately important, somewhat important, less important, unimportant). The section on withdrawal of medications in the survey focused on decisions around individual medication withdrawal including prednisone, methotrexate, anakinra, canakinumab, rilonacept, tocilizumab and combination therapy withdrawal. Descriptive statistics including percentages were used to analyze the data.

While the introduction of cytokine-directed therapy targeting IL-1 and IL-6 has revolutionized the treatment of SJIA, little is known regarding the approach to tapering and stopping such therapy. This CARRA survey of pediatric rheumatology providers offers the first systematic description of clinical practice patterns for medication withdrawal. A first key finding of this survey is the demonstrated agreement and consensus amongst providers in using the existing Wallace criteria for determining inactive disease and remission states in children with SJIA [3]. With the exception of steroids and non-steroidal anti-inflammatory medication withdrawal, most providers wait for clinical remission on medications to start tapering or withdrawing other medications. Since publication of the CARRA SJIA CTP plans [4] an overwhelming majority of providers have adopted these in their current practices and not surprisingly, 71% did not use the glucocorticoid CTP likely reflecting the wide acceptance of biologic use in early presentations of SJIA [5, 6]. The hesitancy to use glucocorticoid CTP plan and 100% consensus on tapering glucocorticoids first when used likely also reflect an attempt by treating providers to limit use of steroids with their expected side effect profiles [7]. This is exemplified by the fact that 85% of respondents listed side effects from prednisone as an important factor in decision-making regarding tapering of prednisone. The lack of use of patient reported outcomes or imaging in 71 and 65% of physicians respectively might reflect the lack of current SJIA specific patient reported outcomes and the lack of readily available imaging, additional cost implications or variable expertise for use of musculoskeletal ultrasound across centers. Despite the lack of use of specific patient reported outcomes most physicians take into account patient/ family preferences via a process of shared decision making as implied by the fact that 61% of respondents ranked patient or family preference in withdrawal of medications as a very or moderately important factor in their decision making process.

The second key finding of this study is the considerable variability amongst pediatric rheumatology providers in decisions on which medication to taper next when patients are on combination therapy, how long to wait before tapering of other medications after discontinuing prednisone, and over what duration to taper medications. This likely is due to the paucity of current evidence or biomarkers to guide physicians on best strategies for withdrawal of medications. A randomized, prospective trial of 364 children with JIA treated with methotrexate (10% with SJIA) examined whether longer time in CRM prevented flares after withdrawal of medication [8]. Here, continuing therapy for 12 months versus 6 months after achieving inactive disease did not significantly improve the overall flare rate. A more recent retrospective study of 1514 patients with JIA (only 4% with SJIA) has challenged this, finding flare was indeed a function of time in CRM, and that patients with inactive disease > 12 months had significantly lower flare rates [9]. There is similarly substantial interest in biomarkers to stratify patients for risk of flare upon medication withdrawal. The above study by Foell et al. did find that lower serum S100A8/A9 (calprotectin; MRP8/14) levels were associated with lower risk of flare over the subsequent 3 months [8]. Further work specifically in SJIA patients found that S100A8/A9 levels correlated with disease activity, and that higher levels during CRM predicted disease flares [10].

SJIA itself is a heterogeneous disease with certain subsets of patients following a monophasic course and others following a more persistent or relapsing course. Clinical features at presentation or during disease course are unable to predict which path an individual patient will follow; hence physicians are left to their own judgements when it comes to withdrawing medications often needing to balance the side effects of continued exposure to medications with the risk of disease flare on withdrawal of these drugs. Some physicians may taper medications based on the half life of drugs used. Survey respondents noted that in patients on canakinumab, rilonacept and tocilizumab, most would prefer increasing the time between injections or infusions rather than reducing the dose likely reflecting the preference in patients for fewer injections/ infusions. Results from the long-term extension phase III trials of canakinumab reported that 25% (n = 44) were able to reduce the dose of canakinumab to 2 mg/kg and 59% (n = 26) of these were able to maintain this reduced doing schedule for 25 months. Eighteen patients eventually flared with the tapered schedule and required either increase in dose back to 4 mg/kg of canakinumab or additional medications [11]. More recent data from the same group examined the efficacy and safety of two different canakinumab tapering regimens (3-step dose taper of 2 mg/kg/q 4 weeks then 1 mg/kg/q 4 weeks then stop versus interval increase in dose 4 mg/kg q 8 weeks then 4 mg/kg/q 12 weeks then stop). Patients were eligible for taper if they were off methotrexate and corticosteroids in inactive disease for 24 months. While several patients were able to taper to an extent on both regimens (71% able to reduce dose to 2 mg/kg q 4 weeks, 84% at 4 mg/kg q 8 weeks) only 33% (n = 25/75)were able to discontinue canakinumab and maintain without flare for 24 weeks [12]. Data from the TENDER clinical trial (N = 112) studied tocilizumab taper in SJIA after 2 years of treatment and CID for 3 months by spacing the infusions from every 2 weeks to 3 weeks then 4 weeks then stopping infusions (35%, n = 39 were eligible for taper). At the time of last data review (May 2014), 7 patients were able to discontinue tocilizumab while 2 patients had to revert back to regular q 2 week doing schedule [13].

Simonini et al. studied remission rates after withdrawal of biologic agents in 135 JIA children (8% of who had SJIA) and found that likelihood of staying in remission was higher for SJIA patients [14]. Similarly, in the ReACCH- Out Canadian JIA cohort (N = 1104) about 47% of SJIA patients (n = 76) achieved remission in 5 years [6]. Shoop-Worrall et al. also noted in their systematic review that rates of CID or remission in SJIA varied widely ranging from 0 to 100% thus reflecting the lack of current knowledge in what drives and/ or sustains remission or CID in SJIA [15].

Strengths of the study include a high response rate of 86%, with a majority of participants being seasoned pediatric rheumatology practitioners (72% being > 5 years into practice of pediatric rheumatology). This suggests good representation across a large sample of the CARRA network. Limitations of the study include that results were self-reported thus introducing recall bias. Many questions in the survey were closed ended and had distinct options to choose from thus lowering the validity rate of the survey. We are limited by the heterogeneity of surveyed pediatric rheumatologists including the fact that 28% of respondents had < 5 years of pediatric rheumatology experience and differences between access to medication across different centers, especially with regard to biological drugs.

Most CARRA members are using existing consensus treatment plans for SJIA and agree with validated definitions of CID and CRM. There was also agreement to taper steroids first, but there was considerable variability for all other medications.