LAUSR

In the effort to define a set of rules useful in tuning the properties for a successful BBB permeation, we statistically analyzed a set of 338 compounds and correlated their experimental in vivo logBB with a series of computed descriptors. Contingency tables were constructed, observed and expected distributions calculated, and chi-square distributions evaluated. This allowed to point out a significant dependence of certain physicochemical properties in influencing the BBB permeation. Over 15 computed descriptors, nine resulted to be particularly important showing highly significant chi-square distribution: polar surface area (66.79; p=1.08 x 10-13), nitrogen and oxygen count (51.17; p=2.06 x 10-10), logP (47.38; p=1.27 x 10-9), nitrogen count (38.29; p=9.77 x 10-8), logD (36.80; p=36.80), oxygen count (35.83; p=3.13 x 10-7), ionization state (33.02, p=3.19 x 10-7), hydrogen bond acceptors (30.80; p=3.36 x 10-6), and hydrogen bond donors (29.29; p=6.81 x 10-6). Other parameters describing the mass and size of the molecules (molecular weight: 11.18; p=2.46 x 10-2) resulted not significant since the population within the observed and expected distribution was similar. Depending on the combination of the significative descriptors, we set a three cases probabilistic scenario (BBB+, BBB-, BBB+/BBB-) that would prospectively be used to tune properties for BBB permeation.