Background
Behçet’s disease (BD) is a multisystem inflammatory disease characterized by oral ulcers, genital ulcers, pustular or nodular skin lesions, uveitis or retinal vasculitis, and pathergy. The cardinal manifestation of BD is oral ulceration. Despite well-proven, effective therapies for mucocutaneous manifestations [
1], many patients continue to exhibit refractory oral and genital ulcerations.
The relative importance of innate and adaptive immunity in the pathogenesis of disease is incompletely understood with most evidence suggesting joint roles. IL-1 is a central cytokine in innate immunity and its blockade is often effective in the treatment of autoinflammatory diseases caused by disruptions of innate pathways. Case reports and series suggest efficacy of IL-1 blockade in mucocutaneous disease [
2‐
5]. Furthermore, initial promising results with gevokizumab - a monoclonal IL-1β blocking antibody - in inflammatory eye disease has generated further interest in IL-1 blockade in this disorder, although notably, mucocutaneous flares did occur in this study [
6].
While there are encouraging data on the use of IL-1 blockade in the treatment of mucocutaneous BD, so far these have all been from retrospective studies or case reports. The objectives of this two-phase, adaptive study design were to evaluate the safety and efficacy of IL-1 blockade with the IL-1 receptor antagonist anakinra, in controlling oral and genital ulcers in an American cohort of patients with refractory mucocutaneous manifestations of BD.
Methods
Patients
Adult patients with BD as defined by the International Study Group Criteria were eligible for participation. The trial was a single-center study conducted at the National Institutes of Health (Bethesda, MD, USA). Patients were recruited by study staff from an observational cohort of patients with BD from February 2012 to February 2014. Patients had active mucocutaneous disease defined by physician documentation of at least one oral or genital ulcer in the month prior to enrollment with direct observation of an ulcer at one month prior to enrollment and also at the time of enrollment. Patients were on a stable or decreasing dose of steroids, nonsteroidal anti-inflammatory medications, colchicine, or disease-modifying antirheumatic drugs for 4 weeks. Exclusion criteria included (1) organ or life-threatening disease including ocular inflammation within 3 months; (2) treatment with tumor necrosis factor inhibitors within 8 weeks; (3) infection including tuberculosis, hepatitis B or C, and human immunodeficiency virus; (4) history of a severe or chronic medical condition including congestive heart failure, malignancy, or uncontrolled asthma; (5) significant cytopenia; and (6) pregnancy or breastfeeding.
Study design
This study was a two-stage adaptive design. Seven patients could be enrolled in the initial phase. If 5 of 7 patients met the primary outcome of complete response between months 3 and 6, the second study phase would be initiated with a total of 20 patients enrolled into the study. In the second phase, patients meeting complete response criteria would be randomized to continuation of study drug vs placebo for 6 months. If five of seven patients did not meet the complete response criteria by 6 months, patients meeting complete or partial response criteria could be continued on open-label therapy for the remainder of the trial. The full study protocol is listed on Clinicaltrials.gov (NCT01441076).
Treatments
All patients were initially treated with anakinra 100 mg daily via subcutaneous injections. In other types of autoinflammatory disease, increasing the dose of anakinra above 100 mg per day is often required to obtain disease control [
7]. Consequently, if oral or genital ulcers persisted after the first month of treatment, anakinra was increased to 200 mg daily. For patients with ulcers at month 6, anakinra was further increased to 300 mg daily. At each study visit, glucocorticoids could be decreased up to 20% of the total dose per judgment of the investigator.
Assessments
Study visits occurred at baseline, day 10, month 1, and then monthly up to 16 months. A safety evaluation was performed on day 7 and at all study visits. Patients were evaluated by a multi-disciplinary team including rheumatology, dermatology, gynecology, ophthalmology, oral medicine, and gastroenterology as appropriate. The following questionnaires and disease assessment indices were collected at each visit: physician and patient visual analog scales (VAS), BD current activity form (BDCAF), BD quality of life (BDQOL), and Behçet’s syndrome activity scale (BSAS) [
8‐
10]. Patients maintained daily diaries recording the number and severity of oral and genital ulcers. Severity of ulcers was scored on a scale from absent (0) to severe (4).
End points
The pre-specified primary outcome was complete response defined as absence of oral and genital ulcers on physical examination on two consecutive monthly visits between months 3 and 6. To meet the complete response criteria, patients could not at any point during the study have evidence of organ-threatening disease or ocular inflammation as defined by the Standardization of Uveitis Nomenclature (SUN) criteria. Partial response was defined as a decrease in the number of genital or oral ulcers at the current and previous study visits compared to the baseline number of ulcers. Treatment failure was defined as lack of complete or partial response at 6 months. Patients not completing the initial phase of the study were considered treatment failures.
The pre-specified secondary outcomes included the number of physician-observed ulcers, number and severity of patient-reported ulcers, patient and physician global scores, and standardized disease activity scores.
Adverse events
Adverse events were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) v4.0. An independent study safety officer met with the principal study investigators every 6 months to review adverse events.
Statistical analysis
Criteria for proceeding beyond the initial study phase were based on the Simon’s optimal two-stage design. The probability of a response rate worthy of further investigation was set as 80% and the probability of a response rate not worthy of further investigation was 50%. Based on a two-sided significance level of 0.05 and power of 80%, up to 7 patients were needed in the first stage and an additional 13 patients in the second stage; if 5 of 7 patients did not respond during the initial stage, the second stage would be terminated. If 14 of 20 patients responded in the total population, the treatment would be considered successful.
Summary statistics were calculated including mean and median values, ranges, standard deviations, and frequency distributions. The primary outcome was evaluated using descriptive statistics. For the secondary outcomes, two-sample comparisons were performed with the Wilcoxon signed rank test for paired continuous measures and Fisher’s exact test for binary measures. Patients’ diary data were analyzed using repeated-measures analysis of variance to account for within-subject correlation. Severity of ulcer pain was assessed by the comparing the percentage of days in which the patient-reported severity score was 1 vs >1.
Discussion
BD is a heterogeneous disorder with clusters of disease and ethnic subtypes. Conflicting data are emerging on the efficacy of IL-1 blockade in this disorder. In this study, we systematically treated patients with oral and genital ulcers with increasing doses of anakinra and reported standardized clinical outcomes.
BD shares clinical features with autoinflammatory diseases, including the lack of autoantibodies, periodic episodes of inflammation, mucosal surface involvement, and neutrophilic infiltration. The prototypical cytokine in autoinflammation is IL-1 and blockade of this cytokine is highly effective for many autoinflammatory diseases. High levels of IL-1β have been observed in serum and synovial fluid from patients with BD [
11,
12], and polymorphisms in IL-1 are reported in some cohorts of patients, although this was not confirmed in a larger gene-wide association study [
13].
Case reports and series suggest a mixed effect of IL-1 blockade on clinical manifestations of BD. The largest series to date is a retrospective multicenter study of 30 patients with varied manifestations of BD treated with either anakinra 100 mg daily or canakinumab 150 mg every 6–8 weeks; 13 of 30 patients achieved complete remission after 12 months [
4]. For mucocutaneous disease alone, an additional series described a rapid improvement in nine patients treated with anakinra 100 mg daily; however, relapses were common ultimately resulting in a poor response in seven patients [
5]. A smaller series of three patients with refractory mucocutaneous disease reported a prompt and sustained response to canakinumab in all patients [
3]. For eye disease, the most promising study was an open-label study of seven patients with refractory disease treated with gevokizumab [
6]. Unfortunately, initial optimism was tempered by a phase III multicenter study, EYEGUARD-B, which failed to reach its primary endpoint.
We prospectively demonstrated a partial response of oral and genital ulcers in BD to anakinra. This was evidenced by objective measurements of ulcers and standardized disease activity outcomes. It is possible that our strict complete response criterion of no ulcerations on two separate visits may have been too stringent, as improvement in refractory disease may be clinically relevant even without a complete response. This is underscored by the recent positive results for apremilast in the treatment of mucocutaneous BD, for which the primary study outcome was reduction in the number of ulcers rather than sustained complete response [
14]. While our criterion was chosen based on prior pivotal trials [
15], the development of valid and widely accepted outcome measures is an active area of research in BD [
16]. Anakinra was well-tolerated and adverse events were few.
Patient-reported outcomes and daily diaries were used to capture different elements of disease activity. IL-1 blockade reduced the number and severity of ulcers for both physician-based and patient-reported outcomes. These findings may be more relevant than a complete response, given that a complete response is seldom observed, even in successful studies. The physician-observed ulcer burden was relatively minimal after baseline. In contrast, in the periods between physician-monitored visits, patients recorded a substantial ulcer burden in daily diaries. Future clinical trials in BD should strongly consider incorporating patient-reported outcomes into the trial design, as perceptions of illness may differ between patients and physicians.
We escalated the dose of anakinra beyond the standard Food and Drug Administration (FDA)-approved dose of 100 mg daily for treatment of rheumatoid arthritis. We feel this is critical as high doses are often necessary in other autoinflammatory diseases, with doses up to 10 mg/kg having an adequate safety profile [
17]. As we observed improvement from anakinra 100 mg to 200 mg daily but no further improvement with further escalation, we determined that the optimal dose is 200 mg daily and our study did not suffer from inadequate dosing. We studied only patients with active mucocutaneous findings and do not advocate anti-IL-1 treatment for more severe manifestations of disease.
Limitations include that our study was small, unblinded, and ultimately single-arm. However, by using a two-phase adaptive design, we were able to reach our conclusions while minimizing the cost and effort that would be required for a larger study. While there was a difference in degree of clinical response to anakinra among patients in this trial, the small sample size precludes identification of potential subsets of patients who may be more or less responsive to anakinra. Similarly, it also precludes a more thorough evaluation of potential confounders including concomitant medications and important lifestyle factors such as smoking.
Given the relapsing-remitting nature of this disease, an element of benefit could be that we minimized the effects of regression to the mean, by requiring that patients have ongoing active disease with physician-observed ulcers persistent for a month prior to enrollment. This is a similar strategy to that used in a recent phase II study of apremilast in mucocutaneous BD [
14].
All patients resided in the USA and the majority was female, which may limit the generalizability of findings to other cohorts. Because of this, we used strict enrollment criteria to improve the homogeneity of our cohort. The release of data on the numbers of oral ulcers in EYEGUARD-B would be helpful to validate these findings in a separate cohort.
Our results confirm those in the literature but with some important differences. Similar to prior studies, most patients benefited from treatment with anakinra, but improvement in disease was neither complete nor sustained. In contrast to prior studies, our study was prospective, it was based on objective measurements of disease activity, and it addressed the issue of insufficient dosing with dose escalation. As some patients did achieve benefit, we concur with case reports that a complete response is possible.
Our results parallel those of apremilast in the treatment of refractory mucocutaneous BD. Apremilast significantly reduces ulcer burden from a mean of 3.0 ulcers at baseline to 0.5 ulcers after 6 months. In the present study, which was not placebo-controlled, we observed a similar magnitude of reduction in oral ulcer burden from 3.5 ulcers at baseline to 1.5 ulcers after 3 months and 0.5 ulcers after 6 months (n = 4 patients) of treatment with anakinra. Differences in trial design may contribute to differences in study populations between these studies. Unlike the apremilast trial, patients taking >10 mg daily prednisone or receiving concomitant immune modulating therapy were not excluded from study participation.
Acknowledgements
We would like to acknowledge the generosity of our patients in participating in this study.