In silico prediction models for blood-brain barrier permeation

The ability to permeate across the blood brain barrier (BBB) is essential for drugs acting on the central nervous system (CNS). Thus, for speeding up the drug discovery process in the CNS-area, it is of great importance to develop systems that allow rapid and inexpensive screening of the BBB-permeability properties of novel lead compounds or at least small subsets of combinatorial CNS-libraries. In this field, in silico prediction methods gain increasing importance. Starting with simple regression models based on calculation of lipophilicity and polar surface area, the field developed via PLS methods to grid based approaches (e.g. VolSurf). Additionally, the use of artificial neural networks gain increasing importance. However, permeation through the BBB is also influenced by active transport systems. For nutrients and endogenous compounds, such as amino acids, monocarboxylic acids, amines, hexoses, thyroid hormones, purine bases and nucleosides, several transport systems regulating the entry of the respective compound classes into the brain have been identified. The other way round there is striking evidence that expression of active efflux pumps like the multidrug transporter P-glycoprotein (P-gp) on the luminal membrane of the brain capillary endothelial cells accounts for poor BBB permeability of certain drugs. Undoubtedly, P-gp is an important impediment for the entry of hydrophobic drugs into the brain. Thus, proper prediction models should also take into account the active transport phenomena.