Background/Purpose: Two similarly designed phase 3 randomized clinical trials (PRESURGE-1 and PRESURGE-2) in gout patients initiating urate-lowering therapy (ULT) showed that subcutaneous treatment with the IL-1 antagonist rilonacept 80 or 160 mg weekly resulted in a significant reduction in the mean number of gout flares (GFs) per patient by 71%-80% relative to placebo over 16 weeks. Analyses were performed to estimate the number needed to treat to benefit (NNTB).

Methods: Using pooled data for the placebo and rilonacept treatment groups (n~160 per pooled group), NNTB was estimated using 1) classical calculations based on binary outcomes, 2) graphical analyses based on the distributions within each treatment group of the number of flares experienced by each patient, and 3) the estimated distributions of flares per patient in the placebo and rilonacept group using negative binomial and Poisson distributions, and the distribution of the difference between groups.

Results: Approximately half the placebo patients (48.4%) did not have a GF, but the other half averaged 2.2 GFs/patient. Approximately 30% more patients on rilonacept vs placebo had less than 1 GF, and approximately 25% more patients on rilonacept vs placebo had less than 2 GFs.  NNTBs are ~4 based on either one of these individual binary outcomes. However, considering NNTB based on these binary outcomes individually does not account for the partial overlap of subgroups that benefit, nor for uncaptured flare reduction among patients with multiple GFs. Graphical analysis suggests that almost all of the patients destined to flare would likely have benefited from rilonacept treatment. A patient who would have had one GF would likely have none, and those destined to have multiple GFs would have fewer. With an estimated mean reduction in GFs of 74% with rilonacept, patients destined to flare would average about 1.6 fewer GFs.  Combining these benefits shows that of 2 patients initiating ULT, 1 would not be expected to flare, even without prophylaxis, but the other patient would be expected to flare and would benefit from rilonacept, resulting in an NNTB of ~2. The difference in the negative binomial distributions and the difference in the Poisson distributions (Skellam distribution) independently confirm an NNTB of ~2; conditional on benefit with rilonacept, the mean benefit is about 1.65 fewer GFs. The most common category of adverse event (AE), infections, was balanced among treatment groups; the most frequent treatment-related AE, injection site reaction, was greatest with rilonacept 160 mg. The incidence of SAEs was low and similar across treatment groups.

Conclusion: This analysis shows that approximately 50% of patients with gout who are initiating ULT would benefit from treatment with rilonacept, resulting in an NNTB of ~2.  Conditional on benefit with rilonacept, patients would experience about 1.6 fewer GFs with treatment.