Conducted at the National Eye Institute with a Collaborative Research and Development Award from the Immunex Corporation.
A randomized, placebo-controlled, double-masked clinical trial of etanercept for the treatment of uveitis associated with juvenile idiopathic arthritis†
Article first published online: 4 FEB 2005
Copyright © 2005 by the American College of Rheumatology
Arthritis Care & Research
Volume 53, Issue 1, pages 18–23, 15 February 2005
How to Cite
Smith, J. A., Thompson, D. J. S., Whitcup, S. M., Suhler, E., Clarke, G., Smith, S., Robinson, M., Kim, J. and Barron, K. S. (2005), A randomized, placebo-controlled, double-masked clinical trial of etanercept for the treatment of uveitis associated with juvenile idiopathic arthritis. Arthritis & Rheumatism, 53: 18–23. doi: 10.1002/art.20904
- Issue published online: 4 FEB 2005
- Article first published online: 4 FEB 2005
- Manuscript Accepted: 11 JUL 2004
- Manuscript Received: 28 JAN 2004
- Juvenile idiopathic arthritis;
- Tumor necrosis factor α inhibitor
To investigate the safety and efficacy of etanercept in the treatment of uveitis associated with juvenile idiopathic arthritis (JIA).
Children who met the American College of Rheumatology diagnostic criteria for JIA with active uveitis, who had anterior chamber cells of ≥1+ or requiring topical corticosteroid ≥3 times daily, and who were on a stable regimen for arthritis treatment were eligible. Study participants received etanercept (0.4 mg/kg) or placebo administered subcutaneously twice weekly for 6 months. All participants received open-label etanercept for an additional 6 months.
Five patients received placebo and 7 received etanercept. Three of the 7 patients treated with etanercept and 2 of the 5 placebo-treated patients were considered ophthalmic successes (P = 1.0). One patient in each treatment group was considered a treatment failure. Three of the 7 etanercept-treated and 2 of the 5 placebo-treated patients were neither successes nor failures by our definition. There were no serious adverse events for any patient during the entire study period. Reports of minor infections were comparable in each treatment group, 71% for etanercept and 60% for placebo (P = 0.58).
In this small pilot study there was no apparent difference in the anterior segment inflammation between patients treated with etanercept and placebo. The stringent criteria used to measure ophthalmic success of treatment and the small patient population limit the implications of our findings.
Juvenile idiopathic arthritis (JIA) is an autoimmune disease that can cause both severe arthritis and sight-threatening uveitis. Diagnosis of JIA is based upon the presence of idiopathic joint inflammation lasting >6 weeks in a child 16 years of age or younger (1). Chronic, bilateral, nongranulomatous anterior uveitis is the hallmark ocular manifestation of this disease and is associated with an average age of onset of joint symptoms of 4–6.3 years. In contrast, some patients develop acute uveitis, and in these children the joint disease has an average age of onset of 7.5–9 years, preceding the ocular disease by months or years (1–3). The likelihood of the development of uveitis in JIA varies with the pattern of joint involvement. Although uveitis is rarely seen in children with systemic JIA, polyarticular-onset JIA is associated with a uveitis prevalence of 5% and pauciarticular-onset JIA is associated with the highest prevalence of uveitis at 20% (4). Young girls with pauciarticular arthritis and a positive antinuclear antibody test result are at the highest risk of developing chronic iridocyclitis (5).
The chronic anterior uveitis seen in JIA is characterized by anterior chamber (AC) cellular infiltration, significant aqueous flare, and nongranulomatous keratic precipitates in a white, quiet-appearing eye. Affected children often do not report pain or visual loss. Topical corticosteroids remain the mainstay of initial therapy of JIA-associated uveitis (6, 7). In a study of 210 patients with JIA, Chylack found that uveitis responded to short courses of topical corticosteroids in 39% of cases (6). However, 61% of patients did not respond to corticosteroids and required prolonged treatment. Repeated periocular injections of depot preparations of corticosteroids are needed to manage some cases. Short courses of oral corticosteroids can be useful, but long-term use of oral corticosteroids in children can cause overwhelming side effects on growth and bone formation. Appropriate judicious use of steroid-sparing agents is critical to the management of recalcitrant uveitis in JIA.
Longstanding ocular inflammation can lead to band keratopathy, posterior and peripheral anterior synechiae, cataract, hypotony, cyclitic membrane, glaucoma, and most importantly, severe vision loss. Although the ocular inflammatory disease associated with JIA primarily involves the anterior segment, vitritis, cystoid macular edema, and posterior segment lesions may occur. The etiology of visual loss is often multifactorial in children with JIA, in whom amblyopia can be an important contributing factor.
Recent data suggest that tumor necrosis factor (TNF) is important in the pathogenesis of JIA (8). TNF is predominantly produced by macrophages (9) and has a number of biologic effects, including activation of T cells, proliferation of both T and B cells, and the attraction of macrophages and granulocytes into areas of inflammation (10). TNF also induces macrophages and other inflammatory cells to release substances that sustain an immune response and lead to tissue destruction (11, 12). In studies of children with JIA, TNF has been shown to be an integral component in the development of the joint disease. In one study, TNF was detected in 45% of samples of synovial fluid from 44 children with JIA (13). Mangge et al have demonstrated that the soluble TNF receptor (sTNFR) levels were also elevated in certain types of JIA, and appeared to correlate with disease activity (10). Studies of patients with rheumatoid arthritis (RA) suggest that the overall quantity of both cell-surface TNFR and sTNFR is inadequate to consume all TNF produced (11, 14–16). Excess TNF combines with its receptors, producing a cascade of damaging inflammatory effects on joints (14, 15). Multiple controlled clinical trials have now demonstrated the efficacy of etanercept for the treatment of RA in patients unresponsive to disease-modifying antirheumatic drugs (DMARDs), when used as monotherapy and in conjunction with methotrexate. Furthermore, etanercept monotherapy has been shown to inhibit joint damage more than methotrexate in patients with early active RA (17).
Blocking TNF activity inhibits experimental autoimmune uveitis, an animal model of uveitis. Dick et al demonstrated that a p55-TNFR IgG fusion protein delayed the onset of inflammation in animals with experimental autoimmune uveitis (18). In addition, histologic examination of retinas showed sparing of damage to the rod outer segments in mice treated with TNFR. Sartani et al also showed that anti-TNF therapy suppresses the induction of experimental autoimmune uveitis (19). Although some investigators have shown that blocking TNF activity can increase ocular inflammation in an endotoxin-induced model of uveitis (20), the pathophysiology of this animal model is different from JIA. Furthermore, the aqueous humor and sera of patients with uveitis demonstrate higher levels of TNF than controls; however, the serum levels are significantly higher than in the aqueous humor in patients with uveitis (21).
The objective of this study was to investigate the safety and efficacy of etanercept in the treatment of uveitis associated with JIA. Based on a priori definitions of success and failure, the rate of success in participants receiving etanercept will be compared with the rate of success in participants receiving placebo.
Participants eligible for the study were children between 2 and 18 years of age who met the American College of Rheumatology criteria for juvenile rheumatoid arthritis (5), had active uveitis with AC cells of ≥1+ or required topical corticosteroids ≥3 times daily, with no change in arthritis regimen for at least 8 weeks prior to enrollment. Exclusion criteria included media opacities; periocular injection of corticosteroids within 2 months of baseline; DMARD therapy, with the exception of prednisone at a dosage ≤1.0 mg/kg/day or methotrexate at a dosage ≤1.0 mg/kg/week up to a maximum of 25 mg/week; or evidence of spondylarthropathy or enthesopathy. All participants were required to use reliable methods of birth control during the study if they were of child-bearing potential or sexually active. The legal guardians of all participants signed an informed consent and children signed an assent approved by the National Eye Institute (NEI) Institutional Review Board.
This 1-year study was composed of 2 phases each lasting 6 months. The first phase was a double-masked randomized study and the second phase was a single arm, open-label cohort study. Participants were randomized at baseline in a 2:1 fashion to receive etanercept or placebo. All participants continued to receive their antiinflammatory medications in addition to their randomized treatment. Participants were examined 2 weeks and 1, 2, 3, 4, and 6 months after initiating their randomized treatment. At all study visits, topical corticosteroids were adjusted according to a standardized treatment schedule based on the amount of AC cells present, unless a serious, sight-threatening, nonuveitic ocular condition of the participant warranted other treatment. Participants would discontinue the randomized treatment and either withdraw from the study or initiate open-label etanercept if they met a failure or safety outcome within the masked portion of the study. Patients who experienced either a safety outcome or an ophthalmic failure outcome were considered treatment failures, were unmasked, and were withdrawn from their randomized treatment. All participants reaching 6 months of followup were provided with open-label etanercept and returned at months 7, 9, and 12 for reassessment.
Outcomes were defined to measure efficacy in treating uveitis, efficacy in treating JIA, and systemic and ocular safety (Table 1). There were primary outcomes for both treatment success and failure. The primary outcomes for successful treatment of uveitis were a reduction of AC cells to 0 or trace while using corticosteroid drops <3 times a day or a 50% reduction in the number or dosage of other antiinflammatory medications without an increase in inflammation. Conversely, a failure of treatment of uveitis was defined as AC cells of >1+ while using corticosteroids ≥3 times daily after 4 months of treatment. There were also secondary efficacy outcomes for uveitis: a 10-letter change in best corrected visual acuity (BCVA), a 2-step change in AC cells, a 2-step change in AC flare, a 2-step change in vitreous haze, or the presence of cystoid macular edema. The primary safety outcomes were a 2-line or greater decrease in BCVA due to inflammation, serious infection or sepsis, and any severe adverse event at least possibly related to study therapy. Only participants who met an ophthalmic success were considered treatment successes.
|Primary ophthalmic efficacy outcomes|
|Reduction of AC cells to 0 or trace with topical corticosteroid <3 times/day|
|50% reduction in number or dose of other antiinflammatory medications with no increase in inflammation|
|AC cells of >1+ in either eye and topical corticosteroid ≥3 times/day after 4 months of therapy|
|Development of a sight-threatening inflammatory ophthalmic lesion requiring immediate increase in systemic antiinflammatory therapy or a periocular injection of corticosteroids|
|Secondary ophthalmic outcomes|
|10-letter (2-line) changes in BCVA|
|2-step changes in AC cells|
|2-step changes in vitreous haze|
|2-step changes in AC flare|
|Presence of CME|
|Primary safety outcomes|
|10-letter (2-line) or greater drop from baseline in BCVA due to inflammation|
|Occurrence of serious infection or sepsis|
|Any severe adverse event at least possibly related to study therapy|
During the randomized portion of the study, participants were provided with prefilled syringes of etanercept or placebo for twice weekly subcutaneous injections at a dose of 0.4 mg/kg to a maximum of 25 mg. Syringes were labeled with generic study information. Treatment codes were maintained at the study coordinating center. Injections were performed under supervision at the NEI for the first week and by the participant or participant's parent, guardian, or health care provider subsequently. Participants were monitored by weekly phone calls between visits for the first 4 months for the occurrence of adverse experiences. During the open-label portion of the study, participants were provided with lyophilized etanercept vials and bacteriostatic water to be used for reconstitution and injection. Participants with infections interrupted their study therapy until the infection was controlled. If a participant developed a fever (>38°C or >100.4°F), study therapy was interrupted until his or her temperature was below this level for at least 24 hours.
Differences in event rates between the 2 groups were tested using a 2-sided Fisher's exact test. The exact Poisson distribution was used to test for a difference in the number of infections per patient. The exact binomial distribution was used to calculate 95% confidence intervals (95% CIs). This study was powered to detect differences in success rates ≥70 percentage points with >80% power. The placebo arm was expected to have a success rate ≤20%. If no events were observed, this design does not rule out event rates as high as 40% in the etanercept treatment arm.
Demographics, baseline characteristics, and followup.
The protocol was activated on September 17, 1999 and enrollment was closed on September 28, 2001 as recommended by the NEI Data Safety and Monitoring Committee (DSMC). Twenty-four patients were screened and 12 were enrolled. Table 2 provides demographic details on the study participants. The age at enrollment ranged from 6 to 15 years and both the mean and median ages were 11 years. Nine of the 12 participants were female. Randomization resulted in 7 patients receiving etanercept and 5 receiving placebo. All participants completed the study per protocol and none missed an expected study visit. Clinic staff and study investigators remained masked to treatment assignments for the duration of the study. Through 18 months of enrollment, the study had not met its projected accrual goal of 15 participants. The DSMC recommended closing the study to accrual 2 years after activation unless the data provided some evidence to support continuing. In September 2001, the DSMC reviewed the interim data, determined there was a very small likelihood that completion of the accrual would provide a significant result, and the study was closed to further accrual.
|Patient no./sex/race/age, years||Treatment||Medications||Vision||Cells||Flare|
|1/F/wht Hisp/11||Etanercept||Prednisone 0.2 mg/kg||20/16, 20/50||T/T||1/2|
|MTX 0.1 mg/kg (5 mg/wk)|
|Prednisone bid OD, > qid OS|
|2/F/wht Hisp/13||Placebo||MTX 0.3 mg/kg (17.5 mg/wk)||20/20, 20/25||1/1||1/2|
|Prednisolone >qid OU|
|3/M/blk/11||Etanercept||Prednisolone >qid OD||20/80, 20/16||1/0||2/0|
|4/F/wht/13||Placebo||MTX 0.5 mg/kg (10 mg/wk)||20/25, 20/400||T/1||1/2|
|Prednisolone tid OD, qid OS|
|5/F/wht/6||Etanercept||MTX 0.2 mg/kg (5 mg/wk)||20/20, 20/20||1/T||2/1|
|Prednisolone >qid OD, qd OS|
|6/F/wht Hisp/10||Etanercept||Prednisolone bid OD||20/200, 20/16||1/0||2/0|
|7/F/wht Hisp/7||Placebo||Loteprednol >qid OU||20/20, 20/16||1, 1||1/1|
|8/F/Wht/15||Etanercept||Prednisone 0.06 mg/kg||20/16, 20/25||0/1||0/1|
|FML ointment qd OS|
|Prednisolone qd OS|
|9/M/Pac Isld/13||Etanercept||Prednisolone tid OD||20/25, 20/16||T/T||1/1|
|10/M/Wht/11||Etanercept||MTX 0.4 mg/kg (25 mg/wk)||20/25, 20/200||T/1||0/1|
|Loteprednol bid OD, tid OS|
|11/F/Wht/6||Placebo||Prednisone 0.09 mg/kg||20/25, 20/32||0/T||1/1|
|MTX 0.3 mg/kg (7.5 mg/wk)|
|Polymyxin B, neomycin, dexamethasone qod OU|
|Rimexolone 1% qid OU|
|12/F/Wht/7||Placebo||Prednisone 0.68 mg/kg||20/20, 20/20||T/T||1/1|
|MTX 0.7 mg/kg (12 mg/wk)|
A total of 21 infections were reported during the study; however, there were no serious or unexpected adverse events reported. Two participants met the safety outcome of a 10-letter loss in BCVA from baseline. The first participant with a 10-letter loss was participant 4 (placebo). At her week-4 study visit, she experienced an 11-letter drop in BCVA in her left eye (20 letters at baseline to 9 letters at 4 weeks; 20/400 to 20/600). This decrease was attributed to an increase in lens opacity exacerbated by persistent inflammation. She was switched to open-label etanercept and was followed through month 12. The second participant with a 10-letter loss was participant 6 (etanercept). She experienced a 16-letter drop in right eye visual acuity from a baseline score of 35 due to inflammation (20/150 at baseline to 20/400 at week 8). The participant was considered a treatment failure, was terminated from the protocol, and received other immunosuppressive therapy.
Comparing the 2 treatment arms during the masked portion of the study, the proportion of participants who reported infections relative to the number of participants exposed to each treatment was similar (5 of 7 = 0.71 for etanercept and 3 of 5 = 0.6 for placebo; P > 0.5). The number of infections per participant appeared higher in the etanercept group (9 infections in 7 participants = 1.29), but it was not significantly different from the placebo group (3 infections in 5 participants = 0.6) (P = 0.24). At the time of enrollment, all 12 participants were eligible to receive study drug for 6 months and the 7 participants who were randomized to the etanercept arm were eligible to receive etanercept for 12 months. Six of 12 (0.5; 95% CI 0.21–0.79) of the participants exposed for up to 6 months reported 11 infections (11 infections in 12 participants = 0.92) whereas 5 of 7 (0.71; 95% CI 0.30–0.96) of the participants exposed for up to 12 months reported 16 infections (16 infections in 7 participants = 2.29).
Masked therapy (randomized comparison).
Table 3 summarizes the successes and failures as well as the number of participants with sufficient followup to be considered eligible for each. Three of 7 (95% CI 0.10–0.82) participants (3, 5, and 9) in the etanercept group and 2 of 5 (95% CI 0.05–0.85) participants (2 and 12) in the placebo group were successes (P > 0.5). Participants 5 and 9 (etanercept) and patients 2 and 12 (placebo) had AC cell grades of “trace” or less while receiving topical corticosteroids at a frequency <3 times a day at month 6. Participant 3 was receiving 50% of the baseline dosage of immunosuppressive medications by month 6. Two of 5 participants in the placebo group and 3 of 7 participants in the etanercept group neither failed nor met the outcomes defined to be success at 6 months. Because this was a placebo-controlled trial and, at the time of the study, very limited information was available on the use of TNFα antagonists in uveitis, our study protocol specified no reduction in systemic immunosuppressive medication during the first 4 months of the study. At the 4-month study visit, all patients maintained at least trace AC cells and all continued to require topical corticosteroids; therefore no patients were eligible for taper of their systemic medications in the masked portion of the study. There was no effect of methotrexate dosage on study treatment outcome. The inclusion criteria limited the dosage of methotrexate because we had limited safety information on the concurrent use of high-dose methotrexate and etanercept at the time the study was started.
|Treatment group||No. of participants||Success||Neither||Failure|
|6 months on masked placebo only||5||2 (2, 12)||2 (7, 11)||1 (4)|
|6 months on masked etanercept only||7||3 (3, 5, 9)||3 (1, 8, 10)||1 (6)|
|6 months on open label etanercept only||5||3 (2, 7, 12)||2 (4, 11)|
|All participants on etanercept for 6 months||12||6 (2, 3, 5, 7, 9, 12)||5 (1, 4, 8, 10, 11)||1 (6)|
|All participants on etanercept for 12 months||7||4 (1, 3, 5, 9)||2 (8, 10)||1 (6)|
Open-label etanercept therapy (exposure to drug).
Six of the 12 participants exposed to etanercept for 6 months were considered successes, 1 was considered a failure, and 5 were neither successes nor failures. If we combine the masked portion with the open-label portion of the study and pool all participants' evaluation of success or failure after 6 months of exposure to etanercept, we find a success rate of 6 of 12 (0.5; 95% CI 0.21–0.79). Based on our patients with a 12-month exposure to etanercept, we observed a success rate of 4 of 7 (0.57; 95% CI 0.18–0.90). Thus, with our relatively short experience with etanercept in a small number of patients, it appears that ∼50% will achieve our goals of success, a rate similar to that observed with standard care.
Etanercept was first shown to be effective at reducing signs and symptoms and inhibiting progression of structural damage in patients with moderate to severely active rheumatoid arthritis. Etanercept has been shown to be safe and effective and is the only biologic agent that is Food and Drug Administration (FDA) approved for the treatment of polyarticular JIA in children older than 4 years (22). In children with JIA who failed to respond to methotrexate alone, three-quarters responded within 3 months of treatment with etanercept and had measurable improvement in symptoms, joint abnormalities, stiffness, and pain (23). This represents a dramatic improvement in the treatment of the potentially crippling joint disease of JIA.
JIA-associated uveitis unresponsive to corticosteroid treatment is commonly treated with methotrexate, and Samson et al have reported a 59% success rate of methotrexate as monotherapy for JIA-associated uveitis (24). Such uveitis is also treated with azathioprine, cyclosporine, chlorambucil, and cyclophosphamide; some of these agents are associated with serious increased risk of malignancy. Mycophenolate mofetil, a pyrimidine inhibitor recently FDA approved for the prevention of renal allogenic graft rejection, has also been used to treat uveitis with some success (25).
In this small placebo-controlled, randomized, double-masked, prospective clinical trial, a substantial therapeutic benefit of etanercept in treating JIA-associated uveitis was not found. This study design was not capable of detecting small or moderate treatment effects, but it is possible that they exist. This study was only powered to find a very large treatment effect. For example, the study had 80% power to find a difference if the observed rates were 10% success on placebo compared with 71% success on etanercept. Obviously, this is a large difference and further investigation will be necessary to more precisely estimate the true effect of etanercept. However, these results indicate a likely upper bound for the treatment effect and suggest that there is not a substantial safety concern in its use for eye disease.
To date there have been no published masked, randomized, placebo-controlled trials of etanercept for JIA-associated uveitis and there is limited controlled data about the use of TNF inhibitors in the treatment of uveitis in general. Reiff and coauthors prospectively evaluated 10 children with pauciarticular JIA-associated uveitis, but did not include a control group, and claimed successful treatment with etanercept added to preexisting immunosuppressive therapy (26). These authors, however, defined success as a single-step reduction in the degree of anterior segment cellular infiltration (Hogan scale) or a decrease in AC cells to <0.5 (25). Ten (63%) of 16 affected eyes showed “significant” improvement in AC cells within the first 12 weeks of treatment, whereas 6 (43%) eyes did not. Despite this improvement in AC cells by eye, 7 of 10 children had an “incomplete treatment response” and were treated with an increased dose of etanercept. Few children met the success criteria of decrease in AC cells to the trace level; in fact, those children who met this criteria in 1 eye had 1–3.5 or more AC cells in the fellow eye, arguing against a major role for etanercept in decreasing the inflammation. Foster and coauthors reported no significant efficacy of etanercept over placebo in preventing relapses of uveitis in a masked study of patients being tapered from methotrexate; however, only 1 of 20 patients had JIA-associate uveitis (27).
Limitations of our study included the small study size and the possible effect of systemic immunosuppression on disease activity. During the study, 7 of 12 children (3 receiving etanercept and 4 receiving placebo) received modest doses of methotrexate at dosages ranging from 5 to 25 mg/week; however, there was no effect on outcome based on methotrexate usage. It is possible that etanercept may offer an additive effect when combined with higher doses of methotrexate, but there is no clear evidence at present that there is a major ocular benefit of etanercept. This is the only randomized, controlled trial of a TNF inhibitor for the treatment of JIA-associated uveitis, and our results differ from the positive response reported in case series and retrospective studies, highlighting the importance of prospective, placebo-controlled trials. Although etanercept is clearly effective in the treatment of arthritis, it has not been found to be more effective than placebo in this small, randomized, masked pilot study for uveitis. Chronic anterior uveitis remains an important cause of ophthalmic morbidity in children with JIA. Expanded understanding of the pathogenesis of the ocular inflammatory response during active uveitis and how it differs from that of associated arthritis may facilitate the identification and evaluation of other immunosuppressive or immunomodulatory strategies for treatment of JIA-associated uveitis.
- 6The ocular manifestations of juvenile rheumatoid arthritis. Arthritis Rheum 1977; 20: 217–23..
- 12Interleukin-2 treatment potentiates induction of oral tolerance in a murine model of autoimmunity. J Clin Invest 1994; 157: 439–47., , , , , .
- 22Safety and efficacy of tumor necrosis factor receptor p75 Fc fusion protein (Etanercept: ENBREL™) in polyarticular course juvenile rheumatoid arthritis [abstract]. Arthritis Rheum 1998; 41(9 Suppl ): S130., , , , .