论文部分内容阅读
The presentation introduces a new hybrid in vitro and in silico model for the blood brain barrier.The model is based on an assay kit containing reconstituted brain phospholipids as well as on polar surface area estimates.A statistical model is used to predict plasma/brain equilibrium distribution (logBB) and the compounds free fraction in brain.The model has been build on a validation data set of 60 compounds targeting the CNS and other indication areas.We found the model with four input parameters to be highly significant with a p-value of less than 10-13.Leave one out cross validation revealed an average prediction error of 0.2 logBB units,which is comparable to in vivo logBB measurement errors.Predicted free fractions in the brain correlated with in vivo free fraction in rat with r2=0.94.Statistical tests for relative importance of variables revealed that affinity to brain lipids is orthogonal to polar surface area.Based on these two estimates form the TRANSIL Brain Absorption assay kit,it is possible to predict the drug concentration available at the receptor site in the brain which is the ultimate measure of interest.