Is it time to move to active comparator trials in juvenile idiopathic arthritis?: A review of current study designs

Over the last 10 years, the number of controlled clinical trials in juvenile idiopathic arthritis (JIA) has grown exponentially due to several factors. These include the availability of new immune response–modifying drugs, generically called biologic agents (1-5), and implementation, both in the US (6) and in Europe (7), of adequate pediatric legislation that essentially mandates companies that wish to register a new treatment (drug or biologic) for a given disease in adults to conduct studies in children if there is a pediatric counterpart of the illness. Two large not-for-profit networks, the Paediatric Rheumatology International Trials Organisation (PRINTO) (8) and the Pediatric Rheumatology Collaborative Study Group (PRCSG), which cover most of the pediatric rheumatology centers worldwide, have facilitated these trials. Moreover, adequate outcome measures have been identified and validated to assess response to therapy in JIA (9-15) and other pediatric rheumatic diseases (16, 17). All of the above are necessary prerequisites for proper trial implementation and successful completion within reasonable time frames, both with respect to investigator-initiated trials (18-20) and trials conducted as part of pharmaceutical company–sponsored clinical development programs (1-4, 21). In 1997, validated criteria to evaluate response to therapy in JIA were published. These criteria were adopted by the ACR and are now known as the ACR Pediatric 30 (9). According to the ACR Pediatric 30, patients are considered to be responders to a given therapy if they demonstrate at least 30% improvement from baseline in at least 3 of any 6 JIA core set variables, with no more than 1 of the remaining variables worsening by >30%. The ACR Pediatric 30 allows researchers and clinicians to dichotomize patients into responders or nonresponders. Commonly, patients are also evaluated for ACR Pediatric 50, 70, 90, and 100 levels of response (at least 50%, 70%, 90%, or 100% improvement, respectively, in at least 3 of any 6 JIA core set variables with no more than 1 of the remaining variables worsening by >30%). The following variables are included in this definition of improvement: 1) the number of joints with active arthritis, defined as a joint with swelling or with limitation of motion accompanied by either pain on motion or tenderness (range 0–73) (22); 2) the number of joints with limited range of motion (range 0–69); 3) the physician's global assessment of disease activity, usually assessed using a doubly anchored 10-cm visual analog scale (VAS) or a 21-circle VAS (23), with anchoring words of 0 = inactive disease and 10 = very active disease; 4) the parent's assessment of the child's overall well-being, usually assessed using a doubly anchored 10-cm VAS or a 21-circle VAS, with anchoring words of 0 = very well and 10 = very poor; 5) a validated measure of daily physical functional ability, usually measured by the disability index of the Childhood Health Assessment Questionnaire (14, 24), or by other instruments derived from the literature, such as the Juvenile Arthritis Functionality Scale (15); and 6) a laboratory measure of inflammation, either the erythrocyte sedimentation rate or C-reactive protein level. The ACR Pediatric 30 criteria are now accepted by both the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) for all phase III trials in JIA seeking drug registration (25). The Adapted ACR Pediatric 30 is currently in use in trials in children with systemic JIA. In addition to the 6 core variables, the absence of spiking fever (≤38°C during the week preceding the evaluation) is required. The 21-circle VAS (23) has been recommended for use in trials in adults with rheumatoid arthritis (RA) (25). Future trials in JIA may use this same system due to its simplicity, feasibility, and increased precision, particularly for the assessment of inactive disease status, where the 10-cm VAS has been shown to possess inherent measurement error (26). Assessment of structural joint damage is a key outcome required by the FDA for studies in chronic arthritis, but it has never been studied in JIA as part of a clinical development program. Because several reliable radiographic scoring systems are now available (27-32) to properly quantify joint erosions and joint space narrowing, this key outcome should be considered in all future JIA trials. Similarly, evaluation of inactive disease status/clinical remission (i.e., no arthritis, no systemic JIA signs/symptoms, no uveitis, normal index of inflammation, and normal physician's global assessment of disease activity) while receiving and while not receiving medication (11) is an important outcome to be reported in JIA clinical trials. When designing comparative studies, several possible control groups (33) can be considered, such as placebo, active comparator (positive), external (e.g., historical), multiple control groups (several active doses with or without placebo), and fixed-dose (dose-response) control groups. However, not all types of control groups are similarly suitable in terms of scientific acceptability, and not all types of control groups are ethically acceptable. In particular, the ethics are questionable of trials that require a proportion of the children with a chronic condition such as JIA to be treated with a placebo when an effective treatment is readily available. There is intense debate about placebo-controlled studies among ethics committees/institutional review boards, practitioners, and families, any of which may reject such studies due to the prospect of a child with active JIA being assigned to receive placebo for several weeks or months (34-36). While a multitude of study designs can be considered (33), the three major trial designs that are relevant to JIA are briefly discussed here (Table 1). There is little argument that the classic parallel RCT with placebo remains the gold standard for establishing the efficacy and short-term safety of an experimental agent, especially when the ultimate goal is approval from the regulatory agencies. Advantages include the potential for testing treatment arms with clinical equipoise, determination of assay sensitivity (i.e., the ability to demonstrate effectiveness in the experimental arm versus the active comparator/placebo arm), and straightforward statistical analysis, usually through assessment of effect size. Disadvantages include ethical concerns due to the use of placebo, clinical inefficiency introduced by treating patients who have active disease with either placebo or experimental agents with unknown efficacy when other effective treatments are available, and the difficulty of generalizing the results to routine clinical care. This study design has been used in several trials in children with JIA (2, 19, 20, 37). More recently, this design has been used for trials of new drugs for systemic JIA, such as tocilizumab and canakinumab, while a trial of rilonacept used a new design called randomized placebo phase study (38). (See www.clinicaltrials.gov [identifier NCT00534495].) The consideration of placebo-controlled trials in systemic JIA may be justified because the disease has few therapeutic options, the placebo effect and the required sample sizes are expected to be low, and the duration of placebo treatment is short (1–3 months). However, an essential prerequisite is the provision of liberal escape rules that allow children with uncontrolled fever to escape after only a few days of ineffective treatment (with placebo or experimental agent). This design has been used in clinical trials of children with JIA (39-41) and encompasses the substitution of placebo with an active comparator (e.g., standard therapy), usually in the format of a noninferiority, equivalence, or superiority study. Advantages of this design include comparison with current standard therapy, a more patient-friendly study (i.e., all subjects receive active treatment), the potential prospect of having at least some treatment arms with clinical equipoise, and straightforward analysis, usually by reporting confidence intervals. Among the disadvantages of this study approach are that assay sensitivity is difficult to evaluate, especially if the comparator drug is very effective. Double-dummy designs are often necessary in situations in which the active agent has a different dosing pattern than the comparator. Above all, very large sample sizes are needed, with the potential for exposing a larger number of subjects to either an inefficacious or dangerous treatment. This study design was proposed by two of the authors (EHG and DJL) for use in pediatric rheumatology studies (18). Eligible children are treated in an open-label manner with the experimental therapy to be tested in the trial for a few months, after which responders (typically defined as those demonstrating an ACR Pediatric 30 response) are randomized in a double-blind manner either to continue the experimental therapy or to switch to placebo. In this segment of the study, called the double-blind withdrawal phase, patients who demonstrate a predefined definition of disease worsening (e.g., a flare) are withdrawn from the double-blind withdrawal phase and usually re-treated with the experimental therapy in an open-label manner. Advantages of this study approach include a placebo-controlled segment (in the double-blind withdrawal phase); a patient- and physician-friendly design, since all subjects receive the experimental treatment; limitation of exposure to placebo to responders only; use of information from all subjects in the analysis of efficacy and safety, thereby producing an efficient study design and minimizing the number of patients exposed to a potentially ineffective or unsafe experimental treatment; and the fact that the proportion of response in the open-label phase of the study more closely approximates routine clinical care than an RCT. Among the disadvantages of the randomized withdrawal design are a bias toward responders (e.g., only responders are randomized); the requirement for estimation of the response rate during the initial open segment as well as estimation of the time to flare; the requirement of performing a complex evaluation of response during the study visit (though it has become standard practice to have the evaluation of the primary outcome, as well as flare [10], dosage tapers, and inactive disease [11], be performed independently by the PRINTO/PRCSG networks on the behalf of the sponsors [3]); limited patient-years on placebo, thus producing a potential for limiting the validity of safety comparison with the experimental agent; use of nontraditional outcomes (e.g., time to flare or proportion of patients who experience a flare in the double-blind withdrawal phase); and impracticality for use in treatments with a very long duration of biologic effect, such as B cell–depleting antibodies. The withdrawal design has proven to be very effective and has been used in nearly all recent trials of biologic agents in children with JIA (1, 3-5). Data gathered using this design have led to the approval of biologic agents for children with JIA by both the FDA and the EMA. Sample size calculations for parallel RCTs are based on expected ACR Pediatric 30 response rates among those in the test agent group versus those in the placebo or active comparator agent groups. Conversely, in randomized withdrawal designs, sample size estimations for the pivotal phase of the trial are based on the expected rate of flares among those in the experimental group compared with those in the placebo group. The definition of a flare currently used in JIA is essentially the inverse of the ACR Pediatric 30 criteria for response (10). This definition requires that there be at least a 30% worsening in 3 of the 6 JIA core set parameters, with no more than 1 parameter improving by more than 30%. Because this definition uses percentages, patients with low numbers of joints with active arthritis or with limitation of motion may have an artificially high percentage increase simply because the denominator for the calculation is so small. Thus, the definition has been modified in this situation. If either the number of joints with active arthritis or the number of joints with limitation of motion is used in the definition of a flare, there must be at least a 2-joint increase in the number of joints with active disease or the number with limited range of motion. Also, if the physician's global assessment of disease activity or the parent's assessment of the child's overall well-being is used in the definition of a flare, worsening ≥2 cm (on a 0–10-cm scale) is required. Additionally, for systemic JIA patients only, reappearance of spiking fever (suggested definition >38°C, lasting for at least 2 days in the week preceding the evaluation) not due to infections would signify that a flare has occurred. These flare criteria have been validated and shown to be effective in limiting potential damage in subjects randomized to placebo or among those who lose response to the experimental agent. Over the last 10 years, nearly all phase III trials conducted in JIA for registration purposes have used the RCT withdrawal design. An exception is the phase III trial of infliximab (2). With increasing knowledge of safety issues in adult RA, it has become apparent over time what changes should be made to the inclusion/exclusion criteria for studies in children. The etanercept (1) and infliximab (2) trials enrolled JIA patients who had responded inadequately to or were intolerant of methotrexate (MTX). The adalimumab trial enrolled two strata of JIA patients: those who had never received MTX and those whose disease did not respond to MTX (4). Patients enrolled in the abatacept trial had demonstrated an inadequate response or intolerance to MTX and/or other biologic agents (3). Candidates for entry into trials for treatments of systemic JIA are usually nonresponders to all available treatment (steroids, anti–tumor necrosis factor [anti-TNF], and anti–interleukin-1 [anti–IL-1]). Of note, for the infliximab trial (2) the 2 doses studied (3 mg/kg and 6 mg/kg every 8 weeks) were established based on data available in adult RA; the trial demonstrated that while both doses were equally effective, the 6 mg/kg dose provided a more favorable risk/benefit profile in children. The number of biologic agents available for study has increased dramatically over the last 15 years. The issue of expanding the enrollment area became particularly necessary after etanercept, and subsequently other biologics, were approved for market for JIA worldwide. In order to expeditiously enroll patients in phase III trials, it became crucial to widen the geographic enrollment area, moving from North America (where an etanercept trial included 9 centers that enrolled 69 subjects) to Europe (with an adalimumab trial conducted at 31 centers enrolling 171 patients [88 in North America and 83 in Western and Eastern Europe]) to Latin America (with an infliximab trial conducted at 31 centers enrolling 123 patients [23 in North America, 72 in Europe, and 28 in Latin America] and an abatacept trial conducted at 43 centers enrolling 190 patients [27 in North America, 69 in Western Europe, and 94 in Latin America). Further, geographically broad recruitment of subjects is essential if trials are to be conducted within specific JIA categories (e.g., systemic JIA) or for drugs that are to be used in children whose disease to respond to at least biologic agent. issues to be considered with to the of a study in pediatric rheumatology include 1) the from sponsors to the drug to the it is for JIA in the specific or as long as it is to the is 2) the from sponsors to centers with adequate for (e.g., a for that into the specific and the of performing trials in pediatric 3) the of enrolling large numbers of children if the sample is to be by different JIA 4) the fact that in such as pediatric patients with rheumatic diseases is more than in adult pediatric rheumatology centers are and clinical assessment and requires and a for with children of different as well as with and 5) the low numbers of patients available for of new have that the ACR Pediatric 30 is low of a of improvement to as the primary efficacy outcome, especially when that current clinical practice mandates levels of response and overall control of JIA (e.g., ACR Pediatric or or the of an inactive disease The ACR Pediatric 30 criteria were to a of improvement to be for such as in order to be considered effective. the ACR Pediatric criteria were derived by the analysis of trial data for drugs, such as MTX (9). the time when the ACR Pediatric 30 criteria were biologic agents were concerns to the potential of the randomized withdrawal design are to the expected ACR Pediatric of response and/or to to parallel RCTs with of placebo, However, the is it is to conduct such trials with and the required sample sizes based on different using etanercept or MTX as the active comparator. 2 sample size calculations for a parallel randomized controlled active comparator trial using etanercept or MTX as the active comparator for different levels of ACR pediatric response 50, or Two possible are provided as in which the experimental drug is proposed to be or effective than the active comparator agent. to statistical if the in the response rate the active comparator (e.g., and the experimental agent is set at using the classic ACR Pediatric 30 response rate for etanercept, the overall number of subjects required to be enrolled in the two groups a is if the ACR Pediatric of response is the required number of JIA patients to be enrolled in the study is If were to the the experimental agent to have a response rate of to than the active comparator agent (e.g., at the ACR Pediatric 30 of the required sample size would be it is apparent that studies of biologic agents that use active MTX or study sample sizes that are simply not in JIA trials. In such large trials the of exposing larger numbers of children with JIA to experimental agents that are either ineffective or 3 the sample size estimations for randomized controlled superiority trials with active or In this the required number of patients to be enrolled for a trial more if the experimental therapy has a response rate of at least than that of the active comparator agent, and an ACR Pediatric 30 response is the required sample size is 190 for a trial in which etanercept is used as the active comparator. an ACR Pediatric response with MTX as active comparator and a to the required sample size is Sample sizes are larger for such as an ACR Pediatric response an ACR Pediatric response with a of overall sample sizes of or with the use of etanercept or respectively, as the active comparator would be required. The major with this of design is that it is it be or to future studies to be limited to agents that can be or than etanercept or the sample size estimations for parallel randomized controlled superiority trials with three treatment The three arms are the experimental agent, the active comparator or and placebo. In the with etanercept, the sample size is in the etanercept in the experimental and in the placebo arm using ACR Pediatric 30 as the primary If the active comparator is the sample size increase to These studies that include a placebo an active comparator of either etanercept or and an number of patients in the active and placebo with an ACR Pediatric 30 of sample size calculations are based on the following the trial can superiority of standard therapy or or experimental agent versus placebo, but the trial not be to demonstrate noninferiority, or of the experimental agent versus standard therapy or Of this of trial continue to the ethical of exposing children with active arthritis to placebo, the availability of effective The above are based on the that JIA is a while the disease is into different this all recent phase III trials have used as a that JIA patients with rheumatoid or and/or systemic JIA without active systemic in the 6 with systemic JIA (with and without active systemic are now considered with study designs parallel and/or withdrawal for agents seeking approval for use in systemic JIA (e.g., or necessary sample sizes be larger if there is a to by all or multiple different JIA in an to the of time that patients receive placebo, proposed a design for adult The trial with patients being randomized in the to receive either the experimental agent or placebo, with both groups to receive therapy (MTX). a standard of care treatment (e.g., is given to all patients who were to the placebo while patients in the experimental group continue to receive the experimental This design allows for the primary assessment of efficacy and safety of the experimental drug compared with placebo at 3 months, by of efficacy and safety, compared with standard of care. This proposed design, not a to the ethical of placebo which is particularly relevant when children. the current availability of several effective treatments for JIA, the issue of the of trial implementation in more This is especially for the drugs within the same medication (e.g., but with different types of or In these efficacy can be expected to be to that of tested agents in the same but safety may be and the proper dose in children with JIA must be If the of the has been established in JIA patients (i.e., therapy for and if adequate efficacy and safety data are available for the agent in adult RA, regulatory approval in JIA for that agent may only require studies in a limited sample of JIA This should be by adequate phase is for all approved to evaluate safety and clinical In for trials of drugs to a new medication the randomized withdrawal design remains the design in JIA since it the of scientific with the concerns of testing in a and systemic JIA the most important JIA categories for controlled The ACR Pediatric 30 to be the primary outcome measure for regulatory but outcomes should include more levels of response to assess the ability to more improvement in children with JIA, inactive Further, for drugs that to a of agents with proven in JIA, studies and to regulatory All authors were in the or it for important and all authors approved the to be published. had to all of the data in the study and for the of the data and the of the data Study and design. of and of