Abstract

Within drug discovery, it is desirable to determine whether a compound will penetrate and distribute within the central nervous system (CNS) with the requisite pharmacokinetic and pharmacodynamic performance required for a CNS target or if it will be excluded from the CNS, wherein potential toxicities would mitigate its applicability. A variety of in vivo and in vitro methods for assessing CNS penetration have therefore been developed and applied to advancing drug candidates with the desired properties. In silico methods to predict CNS penetration from chemical structures have been developed to address virtual screening and prospective design. In silico predictive methods are impacted by the quality, quantity, sources, and generation of the measured data available for model development. Key considerations for predictions of CNS penetration include the comparison of local (in chemistry space) versus global (more structurally diverse) models and where in the drug discovery process such models may be best deployed. Preference should also be given to in vitro and in vivo measurements of greater mechanistic clarity that better support the development of structure-property relationships. Although there are numerous statistical methods that have been brought to bear on the prediction of CNS penetration, a greater concern is that such models are appropriate for the quality of measured data available and are statistically validated. In addition, the assessment of prediction uncertainty and relevance of predictive models to structures of interest are critical. This article will address these key considerations for the development and application of in silico methods in drug discovery.