Abstract

Predicting the detailed kinetics of drugs in thebrain from molecular properties and in vitro data alonecould potentially be a cost-effective screening tool in drugdiscovery. Whilst this currently appears to be beyond thestate of the art, we propose that a physiologically-basedmodel of the brain offers the best way forward in thisendeavour, although several problems are outstanding.Such a model will require compound specific parameters,such as the diffusive and active transport components ofthe blood-brain barrier permeability and the degree ofbinding in the brain cells. These parameters may comedirectly from an in silico approach (such as QSAR) oralternatively may be derived from in vitro assays, such asdrug binding in a brain homogenate or drug transportacross an MDR1-MDCK or caco-2 cell monolayer. Thechallenge here is translating the in vitro measurements intothe in vivo parameters for use in the physiologically-basedmodel. Furthermore, having built the model it must bevalidated against actual in vivo drug concentrations in thebrain. For rat, microdialysis data is available, but theproblem of validating a model of the human brain remains.