Application of modelling of cell monolayer permeation data to generate input parameters compatible with in vitro-in vivo translation of blood-brain-barrier disposition of drugs

Adequate translation of parameters from in vitro experiments to in vivo requires a proper characterisation and mechanistic representation of the systems. The aim of this study was to apply a mechanistic model to better characterise the data from experiments involving in vitro cell lines transfected with ABCB1 genes. In vitro bi-directional transport data (compound recoveries, apparent permeabilities (Papp) and efflux ratios (ER)) were generated from mock and transfected cell lines of Madin-Darby canine kidney type 1 (MDCK I) and type II (MDCK II), as well as a pig kidney-derived cell lines (LLC-PK₁). These were transfected/transduced with either rat (LLC-PK₁/Mdr1a and MDCK I/Mdr1a) or human (LLC-PK₁/MDR1 and MDCK II/MDR1) P-glycoprotein (P-gp) cDNAs. The compounds tested included donepezil, loperamide, phenytoin, quinidine, risperidone, and verapamil. Also, a mechanistic model was used to analyse the data. ER was <2 for mock cell lines and >2 for transfected cell lines, particularly for compounds that are P-gp substrates. The mechanistic model provided a better fit to the observed data, enabling more accurate estimation of both passive and active transport parameters. Nonspecific binding estimates were slightly higher on the apical side (mean: 0.76; range: 0.49-1.00) compared to the basolateral side (mean: 0.71; range: 0.43-0.90). Passive and active clearance values were approximately two-fold higher with the mechanistic model than with the conventional model (passive: 0.17-5.65; active: 1.31-4.88). Overall, the mechanistic model provides parameter estimates that are consistent across species and cell lines, making them more suitable for translation to in vivo systems.