Methods

Studies included in the analysis are listed in table 1. The phase 2 study reported by Gorman et al19 had a 3-month double-blind period followed by a 6-month open-label period. Two7 10 of the remaining three placebo-controlled studies7 10 17 had open-label extensions.16 20

Adverse events that coded to the Coding Symbols for a Thesaurus of Adverse Reaction Terms (COSTART) preferred term uveitis or iritis (iridocyclitis, a verbatim term, codes to uveitis) were included in the analysis. For simplicity, uveitis, iritis, iridocyclitis are henceforth referred to as uveitis.

History of uveitis was not specifically collected on the case report form but would have been captured as part of the general medical history. Subjects were considered to have a history at baseline if they had reported uveitis at baseline medical evaluation, or if the subject or doctor reported uveal inflammation on a baseline extraspinal involvement questionnaire.

Events were classified as either new onset in subjects with no history or flare in those with a history or prior event during trials. Subjects receiving placebo who had an event during the double-blind study were included in the placebo group; if they had an event after they switched to etanercept they were included in the etanercept group. Severity (mild/moderate/severe), duration of symptoms (where available) and treatment were recorded.

Rate of uveitis events per 100 subject years were calculated for each arm of the double-blind studies and pooled rates for etanercept and placebo were also calculated. A long-term rate per 100 subject years was calculated for etanercept that included all the events occurring in both the double-blind and open-label extension studies.

Statistical analysis

Baseline demographics and disease characteristics for subjects included in this analysis were pooled and calculated using descriptive statistics. Between-group differences with confidence intervals (CIs) were calculated for each study and for the pooled population using the method of Sahai and Kurshid.21

Results

Mean disease duration was longer in subjects who reported uveitis events than in their respective treatment groups (table 2). The subgroups of subjects who experienced an event were 91–100% HLA-B27 positive compared with the 79–82% seen in the pooled groups.

Of the total 1323 subjects, 78 reported at least one uveitis event during the study; 25 experienced new events and 69 reported one or more flares (table 3). Of the 249 subjects who received placebo in double-blind trials, 187 continued into the open-label extensions and received etanercept. Eight subjects who had at least one event when receiving placebo experienced at least one more event while on etanercept; six subjects reported both a new event and a subsequent flare event while receiving etanercept. Rates and confidence intervals for the uveitis events in the double-blind and open-label extensions are included in table 3.

In double-blind, placebo-controlled trials, the uveitis rate with etanercept was lower than with placebo (8.6 and 19.3 per 100 subject years, respectively; p = 0.03) (table 3). In the double-blind active comparator study, uveitis event rates were similar for etanercept and sulfasalazine (10.7 and 14.7 per 100 subject years, respectively; p = 0.49).

In the combined analysis to determine the long-term uveitis rate, 1074 subjects received etanercept for a total exposure of 1136.9 subject years (table 3). Seventy-six subjects receiving etanercept reported 136 uveitis events resulting in a rate of 12.0 events per 100 subject years.

No uveitis event led to a study discontinuation or was considered to be a serious adverse event by the study investigator. Events were mostly (96%) mild to moderate in severity and resolved with no treatment or topical therapy. One subject, while being treated for uveitis, experienced a severe myocardial infarction and subsequent complications that resulted in death.

Discussion

Uveitis reports from eight clinical trials of etanercept in subjects with AS conducted in the USA, Europe, Latin America and Asia were pooled and assessed. Subjects who reported uveitis at baseline had a longer mean disease duration than those who did not and at least 91% were HLA-B27 positive. The overall uveitis rate for etanercept was lower than that for placebo, but similar to that for sulfasalazine. The long-term pooled uveitis rate for etanercept was similar to the double-blind rates. These data indicate that in subjects with AS, the rate of uveitis was not increased with etanercept use and did not change with extended duration of treatment.

The overall uveitis rates for etanercept and placebo in this analysis are similar to the flare rates reported in a smaller pooled analysis comparing the flare rates of etanercept, infliximab and placebo.14 In that study, the rates for etanercept and infliximab were both significantly lower than the rate for placebo (p = 0.05). In a 12-week study23 evaluating occurrence of flares in subjects with AS treated with adalimumab, the flare rate of 7.4 per 100 subject-years calculated for adalimumab was similar to the etanercept rate (8.6) in the current analysis of double-blind studies. A retrospective study (n = 46) of patients with spondyloarthritis who had at least one uveitis flare found that the uveitis rate with etanercept (58.5 per 100 patient-years) was similar to that seen before treatment (54.6) but higher than that for infliximab (9.0 per 100 patient-years).15 However, limitations of the comparison include the retrospective study design and small sample size.

Sulfasalazine has previously been shown to significantly reduce the rate of uveitis flares in subjects with AS.13 In our analysis, the rate of uveitis events was similar between etanercept and sulfasalazine (10.7 and 14.7, respectively; p = 0.49), suggesting that etanercept is an effective treatment option for patients with AS with a history of uveitis.

Subjects who reported one or more uveitis event during this study had a longer disease duration and were almost always HLA-B27 positive (91–100%) than their respective pooled groups (79–82%). These results concur with that of a Finnish HLA study,6 in which 98% of subjects with AS and uveitis were HLA-B27 positive.

Approximately, 85% of the subjects with AS who experience uveitis are diagnosed with acute anterior uveitis.1 2 It generally affects men and is unilateral, painful and self-limiting. The cases reported here were predominantly seen in men (83%), were mild or moderate in severity and resolved with topical therapy in all but one subject. This finding differs from the study by Coates et al where they report on five patients with AS who develop new-onset uveitis as a complication of TNF-blocker treatment.24 In that series, the uveitis occurred predominantly in female patients, was bilateral, responded poorly to treatment and required discontinuation of the anti-TNF agent. In another analysis of uveitis cases that occurred in any patients treated with a TNF blocker in the USA, which were reported to the World Health Organization, the rate of uveitis was higher with etanercept than with infliximab or adalimumab.25 This study, however, relied on non-randomised voluntary reporting and was linked to limited clinical information. Further, the authors mention that their findings do not support the use of infliximab (or adalimumab) over etanercept.25

A limitation of our study is that a prior history of uveitis was not specified on the case report form. Patients were not specifically asked whether they had a history of uveitis; thus, the reported rates of 13–22% in our pooled analysis may be an underestimate of the true background incidence of uveitis. Also, the exposure to placebo and sulfasalazine were considerably shorter than to etanercept. These limitations may introduce bias in the interpretation of the results, leading to an underestimation of the incidence of flares and an overestimation of the incidence of new events for the subjects receiving placebo or sulfasalazine.

The current analysis indicates that clinicians can safely choose etanercept for the treatment of AS. These results also indicate that patients who experience uveitis during treatment with etanercept can expect the event to be mild to moderate in severity and to resolve with topical therapy. For those rare patients with refractory or bilateral symptoms, further ophthalmological evaluation, intervention and withdrawal of the drug should be considered.