Predicting blood-brain barrier penetration from molecular weight and number of polar atoms

A simple three-descriptor model to predict blood-brain barrier is derived from a training set of 78 compounds: log BB = -9.880 x 10(-6)M(W)(2) + 7.339 x 10(-3) M-W-0.2268n(pol)-0.1143 (n = 78, r(2) = 0.74), where log 1313 is the logarithm of the ratio of the steady-state concentration of the compound in the brain to concentration in the blood, M-W is the molecular weight, n(pol) is the number of polar atoms (oxygen, nitrogen, and attached hydrogen), n is the number of compounds, and r is the correlation coefficient. The model is validated through use of leave-one-out procedure and an external test set (25 compounds). The model is suitable for the rapid prediction of the blood-brain barrier penetration of drug candidates because of its predictive ability and simplicity. (C) 2008 Elsevier B.V. All rights reserved.