Max Tsai (1), Paul Goldsmith (2), Jinhui Xie (1), Tom A. Macek (1)
(1) Takeda Development Center Americas, Inc., Deerfield IL, USA; (2) Takeda Development Center Europe, Inc., London, UK
Objectives: TAK-063 is a potent and selective inhibitor of the phosphodiesterase 10A enzyme expressed primarily in the striatal medium spiny neurons of the basal ganglia and may have therapeutic benefit in treating schizophrenia. Here we describe the modelling of TAK-063 PK and its relationship with adverse events (AEs) observed in Phase 1 studies.
Methods: Data were pooled from two (single and multiple dose) Phase 1, randomised, double-blind, placebo-controlled studies in healthy (Japanese and non-Japanese) subjects and those with schizophrenia (multiple dose only). Following single or daily dosing of TAK-063 or placebo for 7 days, safety and tolerability assessments and PK samples were collected.
1- or 2-compartment model with 1st order absorption and linear or saturable elimination were evaluated as candidate models. Intersubject and residual variability were estimated using various error models. Baseline covariates such demographics, creatinine clearance and disease status (healthy vs schizophrenia subjects) were evaluated using stepwise forward addition and backward elimination.
Incidence of most frequently reported AEs (somnolence and extrapyramidal symptoms [EPS]) was based on a logistic regression model: f[P(AEi=1)]=log[p/(1-p)] = β + fexp where AEi =1 if subjecti has AE at some time during the study and 0 otherwise; β = logit for subjects not on drug (placebo); and fexp = function describing the exposure-response relationship expressed as linear or nonlinear forms.
Modelling was conducted using NONMEM or R.
Results: TAK-063 PK was best characterised using a transit compartment model for absorption and a 1-compartment model for disposition. Oral bioavailability was dose-dependent and decreased with increasing dose. A proportional increase in central volume of distribution with BMI was noted; no other covariates were statistically significant. Model demonstrated a reasonable goodness-of-fit.
The frequency of somnolence and EPS appeared to increase with increasing dose and/or exposure. Linear models demonstrated adequate goodness-of-fit with no substantial model improvement using Emax function. Disease status as a covariate was significant for EPS but not for somnolence.
Conclusions: The incidence of EPS and somnolence increased with increasing exposure to TAK-063, though a non-zero event rate was noted in placebo subjects. The reason for EPS difference at same doses between healthy subjects and those with schizophrenia requires further investigation.
Reference: PAGE 25 (2016) Abstr 5751 [www.page-meeting.org/?abstract=5751]
Poster: Drug/Disease modeling - CNS