Predicting Chemical Absorption From Complex Chemical Mixtures

Jim E. Riviere
Center for Chemical Toxicology Research and Pharmacokinetics
North Carolina State University

Significant progress has been made on predicting dermal absorption/penetration of topically applied chemicals by developing quantitative structure permeability relationship (QSPR) models based on linear free energy relations (LFER). However, all of these efforts have employed chemicals applied to the skin in aqueous or single solvent systems, a dosing scenario that does not mimic occupational, environmental nor pharmaceutical exposure. We have explored two different approaches to account for chemical mixture / formulation component effects on skin permeability, one being a computational approach using QSPR models and the second an experimental approach using an array of membrane coated fibers (MCF). Both models employ penetration data obtained from in vitro porcine skin diffusion cells dosed in mixtures of up to 5 components. The first approach uses hybrid QSPR equations that describe individual chemical penetration based on the molecular descriptors for the chemical modified by a mixture factor (MF) that accounts for the physiochemical properties of the vehicle/mixture components. The MF is calculated based on percentage composition of the vehicle/mixture components and physical chemical properties selected using principal components analysis. The MCF array approach is based on the similarity in the absorption mechanisms of the MCF membranes and the stratum corneum of the skin. A set of probe compounds is used to detect the relative molecular interaction strengths of chemicals [using the same LFER framework above] with the vehicle and three MCFs (polydimethylsiloxane for lipophilic, polyacrylate for polarizable, and CarboWax for polar interactions). This provides a linkage between the skin permeability (log kp) and MCF partition coefficients (log Kf). A predictive model was established via multiple linear regression analysis of the data matrix of experimentally measured log kp value and log Kf values, as well as a separate model for predicting mixture effects on changes in LFER strength coefficients caused by the mixture components for both MCF and skin. These results suggested that both approaches are capable of dissecting out the specific physical chemical interactions present that modulate dermal chemical absorption from a mixture. This approach holds promise to allow permeability assessments of unknown compounds to be made using either computational or high-throughput experimental methods.

For further details see Riviere and Brooks. SAR and QSAR Environ. Res. 18: 31-44, 2007; Xia et al. Toxicol. Appl. Pharmacol. 221: 320-328, 2007; Riviere et al. Toxicol. Sci. 99: 153-161, 2007; Baynes et al. Chem. Res. Toxicol. 21: 591-599, 2008. (Supported by NIH R01 OH 07555).

Back to the program