Implications of evolving disease classification for drug approval in juvenile idiopathic arthritis

Abstract

The classification of inflammatory arthritis incorporates a sharp divide between diseases of childhood onset, grouped together as juvenile idiopathic arthritis, and diseases such as rheumatoid arthritis that begin by definition in adulthood. An important consequence of this divide is that regulatory authorities and many rheumatologists regard pediatric and adult arthritides as truly different, with the implication that drugs should be evaluated separately for each category. However, it is now clear that most forms of arthritis transcend the pediatric/adult boundary and that agents generally exhibit comparable success irrespective of age of onset, offering new opportunities in drug development and regulation focused on pharmacology and safety rather than efficacy. This paradigm shift will enable advances in arthritis treatment, originating either with adults or children, to translate more rapidly across the age spectrum.

1. Introduction

Among disease families, the inflammatory arthritides are unusual in that disease nomenclature incorporates an invariant cutoff by age of onset. All forms of chronic arthritis of unknown cause that begin before the 16^th birthday are grouped together as juvenile idiopathic arthritis (JIA), whereas all arthritides that begin after this date are considered distinct adult-only diseases such as rheumatoid arthritis (RA). While the names given to the pediatric and adult conditions are often similar, none overlap precisely, rendering it difficult to map arthritis cleanly across the age spectrum (Figure 1). As a result, pediatric and adult rheumatologists, insurers, and regulators understandably regard pediatric and adult arthritis as though they were distinct. For example, pathogenic findings and clinical trial results are extrapolated over the pediatric/adult boundary only with much reservation, and insurers commonly reject requests to use medications beyond the age category in which they were tested.

Figure 1: Overlap in Classification of Pediatric and Adult Arthritis Adapted from: [9, 11].

ILAR = International League of Associations for Rheumatology

PRINTO = Pediatric Rheumatology International Trials Organization

JIA = Juvenile idiopathic arthritis

Oligoarticular: less than or equal to 4 joints involved

ANA = Antinuclear antibody

RF = Rheumatoid factor

RA = Rheumatoid arthritis

In fact, both historical considerations and empiric evidence now render a sharp age divide within arthritis fully untenable. The age boundary of 16 years appeared first in the 1950s as an explicitly arbitrary cutoff reflecting the pattern of patient care in clinics and hospitals in Great Britain at the time, rather than because of age-restricted biologic or phenotypic features [1-3]. Given this origin, it would be a remarkable coincidence if age 16 were also pathophysiologically discriminatory, especially across the whole range of arthritides. A pediatric-restricted form of arthritis may indeed exist that is typically oligoarticular, often positive for antinuclear antibodies (ANA), and sometimes accompanied by chronic anterior uveitis, a complication rarely if ever observed in adult arthritis [4, 5]. However, aside from this disease – far from established as an independent entity, and in any case typically beginning before age 7 rather than 16 – all other forms of arthritis are continuous across the age spectrum both clinically and genetically (Figure 1) [5]. These include (1) seropositive polyarthritis (i.e. accompanied by rheumatoid factor or anti-citrullinated peptide antibodies), a consistent clinical phenotype that appears at appreciable frequency only in patients older than 8-10 years of age and shared genetic risk alleles irrespective of age of onset [6, 7]; (2) seronegative polyarthritis, a more heterogeneous syndrome that also exhibits shared genetic risk alleles in children and adults [7]; (3) diseases characterized by enthesitis, including ankylosing spondylitis and psoriatic arthritis, recognizable as spondyloarthropathies [8, 9]; and (4) systemic juvenile idiopathic arthritis JIA/adult-onset Still’s disease (AOSD), a highly inflammatory disease with a distinctive clinical phenotype including spiking fevers, biomarkers such as interleukin (IL)-18 elevation, drug response, and predilection for macrophage activation syndrome [9-11]. While the optimal boundaries between these phenotypes remain under investigation, there is now broad consensus that the existing International Leagues of Associations for Rheumatology (ILAR) criteria for JIA do not accurately delimit a set of arthritic diseases unique to childhood [5, 12].

2. Unique aspects of arthritis affecting children

Despite these similarities, children do differ from adults in important ways relevant to arthritis monitoring and treatment. The most evident of these is chronic anterior uveitis, a comorbidity that occurs in up to 30% of children in the highest-risk patients (early-onset ANA-positive arthritis) [4, 13]. Unlike acute anterior uveitis, which presents in both pediatric and adult spondylarthritis as eye redness and discomfort and is usually easily managed, chronic anterior uveitis is typically asymptomatic and yet highly destructive, necessitating systematic screening and aggressive disease-modifying therapy [14]. Additionally, childhood includes many important developmental processes sensitive to disease activity. For example, in growing children, uncontrolled arthritis can cause leg length discrepancy, micrognathia, and short stature [15]. Accumulation of bone is greatest in adolescence, such that systemic inflammatory or corticosteroid therapy can lead to low peak bone mass and early osteoporosis [16]. Less easily quantifiable but no less important, extended periods of debilitating symptoms during childhood can have a significant negative impact on a child’s mental health, education, and future job prospects [17]. As a result of these factors, pediatric rheumatologists tend to be more aggressive than adult rheumatologists with disease modifying anti-rheumatic drugs (DMARDs) to achieve remission as rapidly as possible and are usually unwilling to tolerate even low disease activity [18, 19].

Pharmacotherapy must also be approached differently in children. Compliance and quality of life require particular attention to use of liquids, flavorings, and infusions rather than injections. Drug absorption, distribution, metabolism, and excretion can differ between children and adults, requiring dosing adjustment for age, weight, and/or body surface area [20]. For example, tocilizumab and anakinra require higher per-kilogram dosing in younger children to adjust for shorter half-life [21-23]. Certain adverse drug effects may also be more prominent, as is commonly seen for nausea with methotrexate [24].

Finally, but critically, children differ from adults with respect to drug toxicity, including toxicity related to developmental stage. For example, corticosteroid exposure (together with high-grade systemic disease) contributed to the high rates of growth failure and osteoporosis historically associated with systemic JIA [16, 25, 26]. Growth impairment is at least a theoretical risk of Janus kinase (JAK) inhibitors because growth hormone receptor signals via JAK2 [27]. A new type of lung disease associated with systemic JIA, potentially related to IL-1 and -6 blockade, is observed primarily in those with onset in the first few years of life; however, not all affected patients have been exposed to these medications, so the pathophysiological connect remains unclear [28-30]. Children and adults often differ in exposures and comorbidities relevant to the toxicity profile of DMARDs, including use of alcohol and tobacco, pregnancy, type II diabetes, and cardiovascular disease. Since patients with JIA often require long-term therapy, there is the possibility for cumulative risk. For example, the alkylating agent cyclophosphamide (rarely used for JIA but exemplifying the principle) is associated with a several-fold increase in subsequent risk of malignancy [31-33]; if this effect were sustained in all years subsequent to exposure, the cumulative impact could be much higher in a child than in an adult. It remains unknown how this time-horizon effect is relevant for other toxicities observed in adults studies, including methotrexate as a risk factor for squamous cell skin cancer [34] and JAK inhibitors as a risk factor for lymphoma, lung cancer, and cardiovascular disease [35]. Importantly, since it is generally caregivers rather than the patients themselves who consent to therapeutic interventions, there is commonly less willingness to tolerate long-term risks in children than in adult patients who choose for themselves to assume such risks.

3. Clinical trials for arthritis in children and adults

Studies in children are vitally important to understanding how medications should be employed in pediatrics. However, standard randomized controlled trials (RCTs) can be complicated because JIA is relatively rare and because special ethical considerations apply to a population legally unable to provide independent informed consent [36]. To encourage pediatric drug labeling for off-patent and novel treatments respectively, the US Congress passed the Best Pharmaceuticals for Children Act (BPCA) in 2002 and the Pediatric Research Equity Act (PREA) in 2003, allowing the US Food and Drug Administration (FDA) to require studies for relevant pediatric subpopulations. Analogous initiatives were enacted by the European Medicines Agency. These policies have been effective [37], and there is in fact a substantial track record of successful RCTs in JIA [38, 39]. A review of pediatric trials submitted to the FDA through 2017 noted 20 studies in biologics (not all for JIA), making up 6.5% of all studies [37]. A list of approved biologics in pediatric and adult arthritis is outlined in Table 1, along with an enumeration of RCTs that were or were not successful in proving efficacy.

Table 1:

Approved Biologic Medications in Adult and Pediatric Arthritis

Drug class	Drug	Completed Pediatric RCT	Ongoing Pediatric RCT	FDA Approved for Pediatric Arthritis	FDA Approved for Adult Arthritis
Anti-CD20	Rituximab	0	0	No	RA
CTLA4-Immunoglobulin	Abatacept	1	0	pJIA ≥ 2 years old	RA
IL-1 inhibitor	Anakinra	2	0	No	RA
	Canakinumab	2	0	sJIA ≥ 2 years old	No
	Rilonacept	2	0	No	No
IL-6 inhibitor	Sarilumab	0	0	No	RA
	Tocilizumab	3	0	pJIA and sJIA ≥ 2 years old	RA
IL-12/23 inhibitor	Ustekinumab	0	0	No	PsA
IL-23 inhibitor	Guselkumab	0	0	No	PsA
	Risankizumab	0	0	No	No
	Tildrakizumab	0	0	No	No
IL-17 inhibitor	Brodalumab	0	0	No	No
	Ixekizumab	0	0	No	PsA
	Secukinumab	0	1	No	PsA, AS, nr-axSpA
TNF inhibitor	Adalimumab	5	1	pJIA ≥ 2 years old	RA, PsA, AS
	Etanercept	7	1	pJIA ≥ 2 years old	RA, PsA, AS
	Golimumab	1	0	pJIA and PsA ≥ 2 years old	RA, AS
	Infliximab	2	0	No	RA, AS, PsA
	Certolizumab pegol	0	0	No	RA, AS, PsA, nr-axSpA
JAK inhibitor	Tofacitinib	1	1	pJIA ≥ 2 years old	RA, AS, PsA
	Baricitinib	0	3	No	RA
	Upadacitinib	0	0	No	RA

IL = Interleukin

TNF = Tumor necrosis factor

JAK = Janus kinase

pJIA = Polyarticular JIA

sJIA = Systemic JIA

PsA = Psoriatic arthritis

RA = Rheumatoid arthritis

AS = Ankylosing spondylitis

nr-axSpA = Non-radiographic axial spondylarthritis

Despite this success, the traditional FDA approach requiring that drug efficacy must be demonstrated independently in JIA and adult arthritis has substantial disadvantages. First, since the more common adult diseases are targeted first, JIA patients experience significant delay in access to new agents. Such delay makes sense for new agents (especially first-in-class) where ensuring safety in adults is important before exposing children. However, it most cases the delay simply represents a barrier to optimal care, since all agents effective in adult arthritis have proven comparably efficacious in JIA, where tested in the comparable patient subset (Table 1) [40]. The rare exceptions, in which trials in JIA failed to confirm efficacy seen in adult studies, typically reflect suboptimal study design, different patient selection, or small group size. For example, the TNF inhibitors infliximab and golimumab both failed to meet their endpoints in JIA. However, the pediatric rheumatology community nevertheless regards these agents as effective, and indeed infliximab is one of the agents recommended by the American College of Rheumatology for JIA-associated chronic anterior uveitis [14]. The result of delayed drug approvals is evident in the US/Canadian Childhood Arthritis and Rheumatology Research Alliance (CARRA) Registry, where 15-19% of children with JIA had been prescribed at least one unapproved DMARD [41].

A second disadvantage, less obvious but equally important, is that RCTs targeting efficacy endpoints are not well suited to answer the real concerns of pediatric rheumatologists, patients, and families. To dose medications appropriately in patients of different ages and sizes, we need studies of pharmacokinetics and pharmacodynamics. To understand whether children are differentially susceptible to toxicity, including effects on growth and cancer risk, we need studies that are much larger and longer than those required to assess activity against arthritis, typically in the context of post-approval (Phase 4) registries [42, 43] or claims databases [44, 45]. Such observational data suffer the limitation that patients receiving different agents are often different in other ways as well, thus leading to confounding by indication and by access to care. Attribution of safety events is complicated by the fact that many patients are exposed to multiple agents sequentially or in combination. However, registries and claims data have the advantage that children are observed in a real-world setting wherein patients on one agent can be compared to patients on other drugs, better reflecting the actual choice a treating provider makes when selecting among treatments for JIA.

4. Future directions

If most forms of JIA represent recognizable entities spanning pediatric and adults age groups, and yet children differ in pharmacologically relevant ways, two consequential implications follow.

4.1. Where drug efficacy has been demonstrated in adults, approval in pediatrics should proceed based on pharmacokinetics and toxicity. Extrapolation of efficacy from adults to children can be a powerful tool in drug approval for JIA [46], and was the focus of a public FDA/MCERSI (University of Maryland Center of Excellence in Regulatory Science and Innovation) workshop on October 2, 2019, entitled "Accelerating Drug Development for Polyarticular Juvenile Idiopathic Arthritis (pJIA)", that delineated biological similarities between pediatric-onset and adult-onset arthritides and sought ways to use these similarities to improve pediatric drug access [47]. Extrapolation of efficacy is employed in most FDA submissions for pediatric labeling and associated with a higher rate of approval [36]. Indeed, regulatory authorities have a long history of approving medications for children in conditions expected to progress and respond to treatment in a similar way as in adults; examples include antimicrobials and pain medications [48]. A recent study of TNF and IL-6 inhibitors strongly supports such pharmacokinetic exposure-matching to extrapolate efficacy [40]. Conversely, drugs shown to be effective in children should also prima facie be considered effective in adults, for example in systemic JIA, where the findings of high-quality RCTs should be considered directly relevant for AOSD [49, 50]. This approach was employed successfully for golimumab, where despite an RCT that failed to show efficacy in polyarticular-course JIA, approval for use of the IV formulation was granted based on the similarity of this disease with adult arthritis together with a subsequent phase 3, open-label, single-arm clinical trial to test pharmacokinetics and safety in children [51]. Such a hybrid approach respects both the shared pathological features of arthritis across the age spectrum and potential differences in dosing and side effects in children.

Whether extrapolation from adult data can be extended to the pediatric phenotype of early-onset oligoarthritis is more complicated. Despite its phenotypic singularity, this condition shares HLA risk alleles with seronegative polyarticular JIA and seronegative adult RA, suggesting that it is biologically more similar than different [7, 10]. Further, many current agents target shared effector pathways in synovitis; for example, although seronegative and seropositive RA exhibit multiple well-established biological differences, most DMARDs exhibit subtle if any differences in clinical response [52]. Correspondingly, pediatric rheumatologists routinely use methotrexate and TNF inhibitors in refractory oligoarticular JIA, considering the efficacy if these agents to be comparable to those in other disease types. In our judgment, extrapolation of adult efficacy data to this pediatric subgroup is justified, although it remains especially important to remain vigilant for any differences in response.

4.2. Regulators should mandate that pharmaceutical companies support observational studies, including Phase 4 registries and other large databases, to identify toxicities unique to childhood that would not emerge during small, short-term RCTs. Such studies should compare newly approved drugs with relevant comparators and follow children for a decade or more after exposure, enabling early identification of long-term effects on outcomes such as cancer and fertility in the real-world setting.

5. Conclusions

The categorical separation of forms of arthritis by age of onset introduced a conceptual and regulatory hurdle in the management of children with arthritis. This division distanced juvenile-onset arthritis from the advances occurring in adult arthritis, such that therapies in routine adult use have often been available to children only with difficulty and after substantial delay. Historical and biological considerations show such a separation to be incorrect and the attendant delays unnecessary, opening up opportunities to simplify drug approval in pediatrics while focusing attention on pharmacokinetic differences and on the development of an infrastructure capable of monitoring for long-term toxicities. This revised approach will acknowledge both the ways in which children are little adults and ways in which they are not.

Key Points:

1. Pediatric and adult inflammatory arthritides belong to the same disease continuum.

2. In most cases, drug approval in juvenile idiopathic arthritis can reasonably proceed through extrapolation of efficacy from adult trials together with pediatric studies of pharmacokinetics and toxicity.

3. Registries of pediatric-onset arthritis are critical to defining drug safety in areas particularly relevant for children, including growth, fertility, and long-term side effects.