A Time-to-event Modeling Approach Supporting the Safety Profile of New Deutetrabenazine (AUSTEDO) Titration Schemes in Patients with Tardive Dyskinesia and Huntington’s Disease
Benjamin Guiastrennec (1), Dongwoo Kang (2), Rajendra Singh (2)
(1) IntiQuan GmbH, Basel, Switzerland; (2) TEVA Pharmaceuticals, West Chester, PA, USA.
Objectives: Deutetrabenazine is a vesicular monoamine transporter 2 (VMAT2) inhibitor used for the treatment of chorea associated with Huntington’s disease (HD) and tardive dyskinesia (TD) in adults. The starting daily dose of 12 mg is followed by an up-titration of 6 mg weekly based on clinical efficacy and tolerability, up to a maximum recommended daily dose of 48 mg. Depending on the optimal dose required for individual patients, it may take up to 6 weeks to reach clinical efficacy. Thus, longer duration of the titration phase to reach efficacious dose can lead to poor compliance and drop-out of patients from the therapy before clinical efficacy could be reached. In order to support a new rapid titration scheme and to provide additional information to healthcare providers (HCPs) on safety of deutetrabenazine, an exposure-response (ER) analysis was conducted. In this context, a time-to-event (TTE) modeling and simulation approach was used to investigate the profiles of selected safety endpoints with weekly dose increments of 12 mg instead of 6 mg of deutetrabenazine.
Methods: The TTE model development utilized a pooled dataset from 3 Phase 2/3 placebo-controlled, double blind studies in patients with TD and HD in which deutetrabenazine was up-titrated based on efficacy and tolerability according to a pre-defined scheme. Then a maintenance phase for a total study duration of 12 weeks was followed. Daily PK exposures parameters (i.e., average concentration (Cavg), maximal concentration (Cmax), and minimal concentration (Cmin)) were predicted for the total concentration of the active deutetrabenazine metabolites (i.e., (α+β)-HTBZ) for each patient using previously developed and qualified population PK models. The analysis focused on the time-to-first-event, in which somnolence (i.e., most frequent adverse event in the studied population) and central nervous system (CNS)-related adverse events (i.e., somnolence, insomnia, akathisia, suicidal ideation, and/or parkinsonism) were explored. For these 2 endpoints, the model development was conducted in 2 sequential steps. First, baseline hazard models (i.e., exponential, Weibull, Gompertz, and log-normal) were evaluated in R (v4.1.3) with the flexsurv (v2.1) and survival (v3.2.13) packages. Second, time varying PK exposure parameters and covariates were evaluated as predictors on the selected hazard model via a stepwise covariate modeling (SCM) approach (P≤0.05) run in NONMEM (v7.5.1). TTE simulations were conducted under the 6 mg and 12 mg dose weekly up-titration assumptions and compared for models which included an effect of PK exposures.
Results: In total, 463 and 464 TD/HD patients were included in the safety analysis for somnolence and CNS-related AEs, respectively. In the analysis population, incidences were 5.39% and 9.94% for somnolence and CNS-related AEs, respectively. In both cases, baseline hazards were best described by log-normal models, implying that the hazard increased and then decreased over time (non-monotonically). The SCM did not select any predictor for somnolence but included an effect of daily Cmax with a positive slope for CNS-related AEs – translating into a marginal increased risk at higher concentrations. Kaplan-Meier (KM) simulations were performed for the selected CNS-related AEs model on a set of 1,000 virtual patients assigned to placebo, 6 mg weekly titration and 12 mg weekly titration. The KM simulations revealed that despite an effect of PK exposure parameters on the baseline hazard, the net effect between the 2 titration schemes was negligible.
Conclusions: This model-based safety analysis considered 2 safety endpoints. For somnolence, no statistically significant predictors could be identified based on the available data. However, due to the small number of observed events, the results should be interpreted with caution. Although, for CNS-related AEs, a significant effect of Cmax was identified on the baseline hazard, the KM simulations revealed that this effect was negligible on probability of AEs. This analysis provides additional information on the safety profile of deutetrabenazine under different titration regimens to the HCPs, which may help on selection of rapid titration in the clinical practice.