Efficacy of Once-Nightly Sodium Oxybate in Patients with Narcolepsy: Post Hoc Analyses of Sensitivity, Effect Size, and Numbers Needed to Treat from the Phase 3 REST-ON Trial.

Abstract

Background: Once-nightly sodium oxybate (ON-SXB; LUMRYZ™; FT218) treatment significantly improved the coprimary endpoints of mean sleep latency on the Maintenance of Wakefulness Test (MWT), Clinical Global Impression of Improvement (CGI-I) rating, and number of weekly cataplexy episodes versus placebo in a randomized, placebo-controlled trial (REST-ON). The objective of these post hoc sensitivity analyses was to evaluate the robustness of treatment with ON-SXB, while accounting for missing participant data. Number needed to treat (NNT) and effect size analyses were conducted to quantify the treatment benefits.

Methods: Participants ≥ 16 years of age with narcolepsy type 1 or 2 were randomized 1:1 to receive ON-SXB (4.5 g [week 1]; 6 g [weeks 2-3]; 7.5 g [weeks 4-8]; or 9 g [weeks 9-13]) or placebo. Sensitivity analyses included completer population, placebo-based multiple imputation (MI) with a missing-not-at-random assumption, analysis of covariance (ANCOVA), and tipping-point-based MI of worsening values until P > 0.05. Mean differences and P-values were calculated for the MWT and number of cataplexy episodes. For CGI-I, odds ratios and P-values were calculated for completers; mean differences (1-7 points; lower values indicate greater improvement) and P-values were calculated using ANCOVA. Effect sizes were calculated using Cohen's d; NNTs were calculated as the inverse of the absolute risk reduction.

Results: In the completer population (ON-SXB, n = 69; placebo, n = 79), all ON-SXB doses demonstrated significant improvements versus placebo for all coprimary endpoints (P < 0.001). All ON-SXB doses demonstrated significant improvements (P < 0.001) versus placebo for all coprimary endpoints when missing values in both treatment arms were imputed from observed values in the placebo arm (i.e., missing data were replaced with placebo data) and when analyzed using ANCOVA. Tipping-point-based analysis on the change from baseline in mean sleep latency on the MWT demonstrated that implausible or nearly implausible baseline MWT assumptions were needed to render the differences between ON-SXB and placebo no longer statistically significant. All doses of ON-SXB had NNTs of three and effect sizes of 0.7-0.9 for MWT response. For the response in terms of number of cataplexy episodes, NNT was six for the 6 g dose and three for the 7.5 g and 9 g doses; the effect sizes were between -0.7 and -0.8. For the Epworth Sleepiness Scale (ESS) response, NNTs ranged from three to six, with a dose-response effect. Effect sizes were between -0.5 and -0.7 for all doses.

Conclusions: These post hoc results demonstrate the robustness of the REST-ON clinical trial efficacy data.

Clinical trial id: NCT02720744.