Blood-brain barrier permeability revisited: Predicting intrinsic passive BBB permeability using the...

Passive permeability is a critical parameter for modeling drug uptake across the blood-brain barrier (BBB). In this work, we tested the hypothesis that intrinsic passive BBB permeability (P_0,BBB) is equivalent to intrinsic membrane permeabilities measured in Caco-2 or MDCK assays, and that it can also be predicted using the solubility-diffusion model (SDM). To evaluate this, we re-analyzed brain perfusion data from the literature, excluding measurements limited by processes other than passive membrane permeation, such as flow limitation, paracellular transport, or active transport. The resulting membrane-limited P₀ dataset (N = 84), spanning approximately six orders of magnitude, was well predicted by Caco-2/MDCK permeabilities of experimental origin (RMSE = 0.86; N = 34). SDM predictions based on hexadecane/water partition coefficients estimated by COSMOtherm and various LSER methods showed satisfactory performance (RMSE = 1.73-2.29; N = 84), with notable improvement observed for small molecules (MW < 500 g/mol; RMSE = 1.32-1.93; N = 70). No evidence for a molecular size cutoff was found. The approach was also successfully applied to zwitterionic compounds. The findings demonstrate that P₀ values in the BBB are directly comparable to Caco-2/MDCK assays, and that the SDM is a valuable tool for predicting BBB permeability, with the potential to complement or partially replace more resource-intensive in vitro assays.