Rilonacept (IL-1 TRAP) for treatment of colchicine resistant familial mediterranean fever (FMF): a randomized, multicenter double-blinded, alternating treatment trial

Abstract

Background/Purpose: There is no proven treatment alternative for pts with familial Mediterranean fever (FMF) whose disease is resistant to, or do not tolerate colchicine. Since pyrin has an important role in IL-1b regulation we hypothesized that IL-1 inhibition would decrease the number of FMF attacks in these pts. We aim to determine if rilonacept, a fusion protein that binds and neutralizes IL-1, decreases the number of FMF attacks compared to placebo. Methods: FMF pts >=4 years of age with at least 1 FMF attack per month despite adequate doses of, or intolerant to colchicine were recruited at 6 U.S. sites. Pts received two 3-month courses of rilonacept (Arm A) at 2.2 mg/kg (max 160 mg) by weekly SC injection and two 3-month courses of placebo (Arm B). Pts were randomized to 1of 4 treatment sequences (ABAB, BABA, ABBA, BAAB). Escape visits were allowed to switch arms (blinding was maintained) for pts with at least 2 attacks within a course. The primary outcome was the difference in the frequency of FMF attacks between rilonacept and placebo courses. Responders were defined as pts with a >40% difference. Data were analyzed by signed rank tests and Bayesian methods. Result: Fourteen pts were randomized, 8 males and 6 females, mean ((plus or minus)SD) age 24.4(plus or minus)11.8 yrs (range 4.5-47.3; 4 pts <18 yrs), disease duration 17.5(plus or minus)12.6 yrs, with 3.1(plus or minus)2.0 attacks per month at baseline. Eleven pts completed 12 months and 3 discontinued early (1 each due to lack of efficacy, distance from study site and lost to follow-up). Among the 12 pts who completed at least 2 treatment courses, the mean number of attacks per month was 1.0(plus or minus)1.2 on rilonacept vs. 1.8(plus or minus)0.9 on placebo (P=0.027). The risk ratio for attacks (ArmA/ArmB) by Bayesian analysis (consistent for all prior assumptions) was 0.57(plus or minus)0.17 (2.5-97.5 percentile estimates of 0.31-0.98, significant lower rate in Arm A). There were 8 responders; all 4 non-responders were adults. Significant exploratory outcomes include the number of treatment courses without attacks (29% Arm A vs 0% Arm B, P=0.004), courses with >50% decrease in attacks compared to baseline (75% Arm A vs 35% Arm B, P=0.006), the proportion of time pts were treated with Arm A (57.5%) vs Arm B (42.5%, P=0.05) and fibrinogen levels (P=0.04). No significant differences were seen in the proportion of days in attack (Arm A 0.15(plus or minus)0.2 vs Arm B 0.22(plus or minus)0.23, P=0.08), number of escape courses (25% Arm A vs 46% Arm B, P=0.13), patient/parent (P=0.07) and physician global assessment (P=0.14), the modified Armenian Severity Score (P=0.1) and other acute phase reactants. There were significant differences in physical but not psychosocial aspects of quality of life between treatment arms (P=0.01 and 0.64, respectively). The median time to develop 2 attacks was >90 days (not reached by end of course) in Arm A vs 36 days in Arm B (p=0.009). There were 2 respiratory infection SAEs, 1 on rilonacept and 1 on placebo. Injection site reactions were significantly more frequent with rilonacept (P=0.04) but no differences were seen in other adverse events, including infections. Conclusion: Rilonacept significantly reduces the number of FMF attacks and has an acceptable safety profile. IL-1 inhibition is a treatment option for most colchicine resistant FMF pts and especially children.