Prediction of the Extent of Blood-Brain Barrier Transport Using Machine Learning and Integration into the LeiCNS-PK3.0 Model.

Abstract

Introduction: The unbound brain-to-plasma partition coefficient (K_p,uu,BBB) is an essential parameter for predicting central nervous system (CNS) drug disposition using physiologically-based pharmacokinetic (PBPK) modeling. K_p,uu,BBB values for specific compounds are however often unavailable, and are moreover time consuming to obtain experimentally. The aim of this study was to develop a quantitative structure-property relationship (QSPR) model to predict the K_p,uu,BBB and to demonstrate how QSPR-model predictions can be integrated into a physiologically-based pharmacokinetic model for the CNS.

Methods: Rat K_p,uu,BBB values were obtained for 98 compounds from literature or in house historical data. For all compounds, 2D and 3D physico-chemical and structural properties were derived using the Molecular Operating Environment (MOE) software. Multiple machine learning (ML) regression models were compared for prediction of the K_p,uu,BBB, including random forest, support vector machines, K-nearest neighbors, and (sparse-) partial least squares. Finally, we demonstrate how the developed QSPR model predictions can be integrated into a CNS PBPK modeling workflow.

Results: Among all ML algorithms, a random forest showed the best predictive performance for K_p,uu,BBB on test data with R² value of 0.61 and 61% of all predictions were within twofold error. The obtained K_p,uu,BBB were successfully integrated into the LeiCNS-PK3.0 CNS PBPK model.

Conclusions: The developed random forest QSPR model for K_p,uu,BBB prediction was found to have adequate performance, and can support drug discovery and development of novel investigational drugs targeting the CNS in conjunction with CNS PBPK modeling.