Evaluation of bottom-up modeling of the blood-brain barrier to improve brain penetration prediction via physiologically based pharmacokinetic modeling.

Predicting the brain penetration of drugs has been notoriously difficult; however, recently, permeability-limited brain models have been constructed. Lead optimization for central nervous system compounds often focuses on compounds that have low transporter efflux, where passive permeability could be a main driver in determining cerebrospinal fluid (CSF)/brain concentrations. The main objective of this study was to evaluate the translatability of passive permeability data generated from different in vitro systems and its impact on the prediction of human CSF/brain concentrations using physiologically-based pharmacokinetic (PBPK) modeling. In vitro data were generated using gMDCK and parallel artificial membrane permeability assay-blood-brain barrier for comparison and predictions using a quantitative structure-activity relationship model were also evaluated. PBPK modeling was then performed for seven compounds with moderate-high permeability and a range of efflux in vitro, and the CSF/brain mass concentrations and Kpuu were reasonably predicted. This work provides the first step of a promising approach using bottom-up PBPK modeling for CSF/brain penetration prediction to support lead optimization and clinical candidate selection.