Abstracts

Rationale: TAK-935 is a potent and selective cholesterol 24S-hydroxylase (CH24H) inhibitor that converts brain cholesterol to 24-hydroxycholesterol (24HC), a co-agonist at the NMDA receptor. 24-HC can drive glutamatergic over-activation by modulation of astrocytes and N-methyl-D-aspartate (NMDA) channel activity, implying a potential role in CNS diseases such as epilepsy. TAK-935 is being developed for rare pediatric epileptic diseases. Based on adult clinical data, a population Pharmacokinetic/Enzyme-Occupancy/Pharmacodynamics (PK/EO/PD) model was developed using an integrated understanding of the relationships between dose/exposure and brain enzyme occupancy or changes in plasma 24HC levels, as PD measure. Utilizing allometric scaling principles this model was used to guide dose selection in adults (≥18 years), adolescents (≥12 and < 18 years) and children (≥2 and < 12 years).  Methods: Based on 4 Phase 1 studies in healthy adults, a population PK model was developed to describe the complex disposition profile of TAK-935 using nonlinear mixed effect modeling. This model consists of a two compartmental model with delayed absorption and dose as a covariate on clearance, inter-compartmental clearance and peripheral volume of distribution to account for PK nonlinearity. Model predicted TAK-935 concentrations were then used to develop an Emax concentration-EO model, whereby predicted TAK-935 plasma concentrations were linked via an effect site compartment to brain enzyme occupancy data obtained from a PET imaging study. In addition, plasma concentrations of 24HC were linked via an indirect inhibitory response PK/PD model.  Results: Population modeling analyses included 1727 PK, 20 EO and 2270 PD observations from 104, 11 and 99 subjects, respectively. The PK/EO/PD model adequately described the relationships between observed TAK-935 plasma exposures, brain enzyme occupancy and changes in plasma 24HC in adults. Model validation used standard goodness-of-fit diagnostics, parameter precision and visual predictive check. The PK/EO/PD model was then used to simulate individual TAK-935 exposures, enzyme occupancy and 24HC plasma concentration at steady state for different dosing regimens in pediatric subjects. To adjust for varying body weight across the age range of interest, clearance and central volume of distribution were allometrically scaled. Model-based simulations suggested that doses ≥400 mg/day can provide sufficient target enzyme occupancy across the 30-80 kg weight range (Figure 1). In addition, twice daily  was superior to once daily dosing with respect to target engagement above the pre-specified threshold of 65% for efficacy based on preclinical information, and to lower peak to trough fluctuations during the dosing interval (Figure 2).  Conclusions: This model-based approach allowed integration of all pertinent PK and PD data using quantitative methods to guide selection of dose and regimen for optimal response in pediatric subjects, while ensuring an acceptable safety and tolerability profile. Funding: Takeda Pharmaceuticals