Evaluation of various PAMPA models to identify the most discriminating method for the prediction of BBB permeability

Abstract

The Parallel Artificial Membrane Permeability Assay (PAMPA) has been successfully introduced into the pharmaceutical industry to allow useful predictions of passive oral absorption. Over the last 5 years, researchers have modified the PAMPA such that it can also evaluate passive blood-brain barrier (BBB) permeability. This paper compares the permeability of 19 structurally diverse, commercially available drugs assessed in four different PAMPA models: (1) a PAMPA-BLM (black lipid membrane) model, (2) a PAMPA-DS (Double Sink) model, (3) a PAMPA-BBB model and (4) a PAMPA-BBB-UWL (unstirred water layer) model in order to find the most discriminating method for the prediction of BBB permeability. Both the PAMPA-BBB model and the PAMPA-BLM model accurately identified compounds which pass the BBB (BBB+) and those which poorly penetrate the BBB (BBB-). For these models, BBB+ and BBB- classification ranges, in terms of permeability values, could be defined, offering the opportunity to validate the paradigm with in vivo data. The PAMPA models were subsequently applied to a set of 14 structurally diverse internal J&J candidates with known log (brain/blood concentration) (LogBB) values. Based on these LogBB values, BBB classifications were established (BBB+: LogBB0 >or=; BBB-: LogBB<0). PAMPA-BLM resulted in three false positive identifications, while PAMPA-BBB misclassified only one compound. Additionally, a Caco-2 assay was performed to determine the efflux ratio of all compounds in the test set. The false positive that occurred in both models was shown to be related to an increased efflux ratio. Both the PAMPA-BLM and the PAMPA-BBB models can be used to predict BBB permeability of compounds in combination with an assay that provides p-gp efflux data, such as the Caco-2 assay.