Biologics and Behcet's: Disease Diversity and Trial Design

In this issue of OII, a team of talented investigators from the US National Institutes of Health reports that a promising biological therapy, daclizumab, failed to improve the outcome of the uveitis associated with Behcet's disease over standard therapy in a well-designed, prospective, randomized trial. We can learn much from this report.

Behcet's disease was once one of the most virulent forms of uveitis imaginable. In 1970, Mamo reported that most Lebanese patients with Behcet's disease had visual acuity of 20/200 or less within one year of the onset of eye symptoms.¹ Subsequently, well-designed, randomized, controlled trials established the efficacy of either azathioprine² or cyclosporine³ for the uveitis associated with Behcet's disease. More recently, both alpha interferon⁴ and anti-tumor necrosis factor alpha therapy with infliximab⁵ have been reported to be dramatically effective in this disease. But daclizumab therapy, which is directed against the IL-2 receptor on activated T lymphocytes, has also been shown in well-designed, prospective trials to be an effective therapy for uveitis.^6–8 So why does it fail in Behcet's disease?

Many uveitis trials are underpowered, meaning that appreciating efficacy could be missed just by virtue of the study size. Most uveitis trials are not randomized and are retrospective. The report by Buggage and colleagues in this issue has none of these failings. It is randomized and prospective. Although a larger trial might have revealed statistical significance in favor of the biologic, this is very unlikely when the trend is that untreated patients fared better than those treated with daclizumab. Traditionally, many journals have been reluctant to publish “negative” results, i.e., trials which fail to reach the primary endpoint. Thankfully, the more recent trend is to recognize that a well-designed negative trial provides valuable information for clinicians.

Immunologic disease has different phases. In animal models, it is always far easier to prevent a disease as opposed to treating effectively an established disease. It should be apparent that what initiates an immune response and what sustains damage within a localized tissue might differ substantially. Similarly, what could maintain remission in relatively inactive disease could differ from what is able to treat active disease. The failing in this trial is unlikely to be due simply to this explanation since, for the majority of patients, the trial was a test of the ability to sustain remission and daclizumab had previously demonstrated efficacy in other forms of uveitis for which remission had been established.

Therapeutic trials can fail to achieve the primary endpoint for additional reasons. The investigators may have picked an inadequate dosage. The route of administration may have been wrong. The randomization may have biased the trial. Indeed, Buggage and colleagues suggest that one of the daclizumab patients had many more attacks before starting daclizumab relative to any of the other patients in the trial. While this explanation is plausible, the authors could have performed a subset analysis excluding this patient. The authors suggest that the success of conventional therapy made it difficult to show that daclizumab was efficacious. Although this is plausible, another study which succeeded in showing efficacy using infliximab for Behcet's was quite similar.⁹ And one could hypothesize that different ethnic groups might be optimally treated with different pharmacologic therapies, a hypothesis that finds support in the emerging field of pharmacogenomics.

Most trials of novel therapies for uveitis equate all forms of uveitis, an expediency that facilitates recruitment but obfuscates the interpretation of the trial results. Imagine if a rheumatology trial took patients with any form of inflammatory arthritis and lumped the response of patients with gout with the response of patients with rheumatoid arthritis. Or if a chemotherapy trial took on all comers with malignancy regardless of histological type. To its credit, the present study by Buggage et al. took cognizance that not all forms of non-infectious uveitis should be equated by studying a specific form of uveitis, i.e., that associated with Behcet's disease. Many uveitis experts treat a disease such as serpiginous choroiditis more aggressively than intermediate uveitis. The pathogenesis of an acute attack of anterior uveitis in association with ankylosing spondylitis must differ from the chronic inflammation that characterizes the chronic iritis that afflicts some children with juvenile idiopathic arthritis. Since daclizumab works for other forms of uveitis, the failure of daclizumab to benefit patients with Behcet's disease might be a major clue to the pathogenesis of this syndrome.

So while the negative study of Buggage et al. has several plausible explanations, the appreciation that the pathogenesis of Behcet's disease might be substantially different from other forms of uveitis emerges as the most likely, at least from the perspective of your editorialist.

If so, there seems a clear moral to this story. The ideal uveitis trial should study a single disease. We desperately need trials to study patients with birdshot choroidopathy, the chronic uveitis of juvenile idiopathic arthritis, intermediate uveitis, and serpiginous choroiditis, among other uveitis entities. Many forms of inflammation do have a final common pathway, but the differences may be at least as important as the shared features. Just as we celebrate diversity among cultures, we need to recognize diagnostic diversity. Our therapeutic armamentarium will be most finely honed if we recognize the differences among inflammatory diseases and design our trials accordingly.

Supported in part by a Senior Scholar Award from Research to Prevent Blindness and the Stan and Madelle Rosenfeld Family Trust.