Exposure–Response Analyses of Polysomnography and Subjective Sleep Efficacy End Points From the Phase 3 Trials of Lemborexant, a Dual Orexin Receptor Antagonist for the Treatment of Insomnia

This report describes polysomnography and sleep diary exposure-response analyses from Study E2006-G000-304 (Study 304), a 1-month trial of 5- or 10-mg lemborexant, zolpidem, or placebo; and Study E2006-G000-303 (Study 303), a 6-month trial of 5- or 10-mg lemborexant or placebo. Studies 304 and 303 included 1006 (86%) and 956 (68%) (female) participants, respectively; >40% were ≥65 years, with individual lemborexant exposures derived from a previously described pharmacokinetic model. Linear mixed-effects analyses of polysomnography: latency to persistent sleep (LPS), sleep efficiency (SE), and wake after sleep onset (WASO) quantified the change from baseline given lemborexant exposure, time, and covariates, guided by consensus recommendations regarding clinical significance. A small impact of sex, body weight, and race was predicted for LPS and SE, irrespective of treatment. Effect of age on LPS was small; baseline SE was estimated to be 8% higher for a 50-year-old versus an 80-year-old, decreasing to 6% by 1 month. Baseline WASO was 13 minutes longer for Black versus White subjects, corresponding to a 5-minute lower change from baseline at the end of the study. For subjective end points, the statistically significant covariate effects for age, sex, and race were not deemed therapeutically relevant, likely reflecting physiologic sleep pattern changes across age and study subgroups. Both polysomnography and subjective analyses indicated clinically meaningful differences from baseline for both lemborexant treatments, with effects being greater for 10-mg versus 5-mg lemborexant, while indicating that covariate-specific lemborexant dose adjustments are not warranted.