Determination of drug absorption parameters in Caco-2 cell monolayers with a mathematical model encompassing passive diffusion, carrier-mediated efflux, non-specific binding and phase II metabolism

Intestinal absorption is required for a sufficiently high bioavailability of drugs administered by the peroral route. Several molecular mechanisms are involved in intestinal absorption and can profoundly influence its magnitude including permeation of the mucosa by passive diffusion, transport across the intestinal wall by carrier mediated processes, chemical and enzymatic alteration of the molecule in the intestinal lumen and/or in the enterocyte, dissolution behaviour of the drug and interaction with food ingredients or coadministered drugs at the dissolution and the transport level. These mechanisms typically act simultaneously and each depends on drug molecule and epithelium related chemical and biological factors whose effect is not definitely established. This makes intestinal absorption a rather complex process, which, despite recent advances, is fundamentally still poorly understood. Therefore, experimental verification of drug absorption remains a must in current industrial drug development practice. Prediction of in vivo absorption based on in vitro methodology may help reduce the volume of necessary clinical investigations. Cell culture techniques predominantly employing the Caco-2 cell line have been established in the last decade as a screening and study tool of intestinal absorption. This technique, although widely used in industrial and academic settings, still poses a number of challenges. These include artefacts like adsorption to container surfaces and cellular accumulation, which might lead to an erroneous estimation of the permeability, poor recovery, and faulty mass balance. The estimation of independent parameters for parallel processes (e.g. passive permeability and active efflux) is still not the common procedure in the analysis of permeation data and the usually employed apparent permeability coefficient Papp and efflux ratio (ER) are afflicted with limitations. The objective of the present work was to establish a methodology for determining the contribution of passive diffusion, carrier-mediated transport, enzymatic degradation and non-specific binding/adsorption to drug absorption measured in the Caco-2 cell system and investigating drug-drug interactions for absorption in this system. Transport experiments across the cell monolayer were conducted in bi-directional modus using model drug compounds (Amentoflavone, Verapamil, Digoxin and Quinidine) that are known to be subject to more than one of the above molecular mechanisms. In order to delineate the contribution of these mechanisms, a model for analysing the experimental data was introduced. This model encompassed quantitative expressions based on biophysical or physicochemical principles of the effect of all these mechanisms on transport and described the variation of drug concentration in the different compartments of the cell system as a result of the simultaneous action of these mechanisms. The
resulting system of differential equations was fitted to the experimental data using
regression analysis following numerical integration and relevant parameters reflecting the
quantitative effect of each mechanism involved in absorption were deduced. These
parameters were the passive permeability coefficient, first and zero order carrier
mediated transport rate, first order metabolic rate constant and binding constant. This
was done for the above drugs used individually and for selected binary mixtures.
Amentoflavone exhibited strongly asymmetric transport, which was almost not detectable
in the apical to basal direction and pronounced in the basal to apical direction. This
suggested that Amentoflavone is subject to apical efflux in the Caco-2 cells. This was
partly reversed by GF120918 and Vinblastine which are known inhibitors of Pglycoprotein
(P-gp) and P-gp and MRP2, respectively, indicating that Amentoflavone was
substrate of at least one of these efflux transporters. The active apical efflux was almost
abolished at low temperature. Amentoflavone also underwent phase II metabolism in the
Caco-2 cells. At least two glucoronides and one sulfate were detected by HPLC-MS,
which were hydrolysable by specific enzymes. These metabolites exhibited also apical
efflux. Finally, Amentoflavone was strongly adsorbed to the surface of the Transwell
plates used in cell culture and transport studies. The adsorption and desorption rate
constant and the total number of surface binding sites was determined in blank
experiments using the plates without cells. For this, a model describing the time
dependent concentration decrease of the drug due to adsorption was employed. These
surface adsorption parameters were subsequently used in the model describing the
absorption of Amentoflavone in Caco-2 cells. This procedure allowed the determination
of absorption parameters that were not biased by non-specific binding effects.
Amentoflavone was found to have a rather low passive cell permeability, which combined
with its substantial apical efflux resulted in its marginal absorption in the apical to basal
transport direction. The metabolic rate constant was smaller than the efflux rate constant
but still sufficiently large to produce relevant amounts of metabolite in the course of the
cell absorption experiment.
Verapamil, Digoxin and Quinidine were studied in a wide concentration range individually
and in binary mixtures to determine the significance of the interplay of passive diffusion
and apical efflux on absorption and how this is affected by concomitant administration of
a second drug. Passive permeability coefficients were independent of concentration and
varied among the three drugs over at least a ten-fold range. The rate of apical mass
efflux varied between the three drugs, increased with concentration and seemed to level
off for at least one of the three drugs in the studied concentration range. The dominance
of one mechanism over the other depended on concentration in a different pattern for the
three compounds. Thus, the outcome of the combination of passive permeation and
apical efflux for apical to basal absorption can only be predicted if the passive
permeability coefficient and the concentration dependence of the carrier mediated efflux
rate are known. All three drugs were substrates of apical efflux carriers. In binary
mixtures, they commonly reduced the efflux rate of the concomitant compound. This
reduction was mutual yet its extent varied between the compounds. Hence, these apical
efflux carrier substrates may also act as inhibitors and vice versa exhibiting at least a
partially overlapping specificity.
In conclusion, the introduced model approach and data analysis provide a quantitative
insight into the process of drug absorption which can be used for a better understanding
and potentially as a means supporting the prediction of in vivo absorption based on cell
culture data and the delineated effect of the involved mechanisms.