Nonlinear Quantitative Structure-Property Relationship Modeling of Skin Permeation Coefficient

Abstract

The permeation coefficient characterizes the ability of a chemical to penetrate the dermis, and the current study describes our efforts to develop structure-based models for the permeation coefficient. Specifically, we have integrated non-linear, quantitative structure-property relationship (QSPR) models, genetic algorithms (GAs), and neural networks to develop a reliable model. Case studies were conducted to investigate the effects of structural attributes on permeation using a carefully characterized database.Upon careful evaluation, a permeation coefficient data set consisting of 333 data points for 258 molecules was identified, and these data were added to our extensive thermophysical database. Of these data, permeation values for 160 molecular structures were deemed suitable for our modeling efforts. We employed established descriptors and constructed new descriptors to aid the development of a reliable QSPR model for the permeation coefficient. Overall, our new nonlinear QSPR model had an absolute-average percentage deviation, root-mean-square error, and correlation coefficient of 8.0%, 0.34, and 0.93, respectively. Cause-and-effect analysis of the structural descriptors obtained in this study indicates that that three size/shape and two polarity descriptors accounted for approximately 70% of the permeation information conveyed by the descriptors.

Discovering and Developing Molecules with Optimal Drug-Like Properties

“…The physicochemical properties generally preferred for common routes of delivery have been reviewed (Mathias and Hussain 2010) and key parameters are summarized in Table 2.9. Further details may be found in recent reviews for transdermal (Neely et al 2009;Paudel et al 2010;Watkinson 2013), intranasal (Chapman et al 2013), and sublingual/buccal delivery Goswami et al 2013;Lam et al 2013). Strict limitations around human efficacy dose (Table 2.10) for nonoral routes of administration are due primarily to permeation limitations or dose volume constraints at these sites.…”

Modeling and Prediction of Solvent Effect on Human Skin Permeability using Support Vector Regression and Random Forest

“…Among the most famous classical models is the Potts and Guy's model (11), which predicts the permeability coefficient (k p ) by linear regression using octanol-water partition coefficients (log P) and molecular weights of permeants. Recently, in silico prediction models of skin permeability, based on machine learning algorithms and matrices containing the observed permeabilities and physicochemical features of permeants, have also been reported (12)(13)(14)(15)(16)(17)(18).…”

Design of improved permeation enhancers for transdermal drug delivery

“…Recently,13, 14 we have developed nonlinear, quantitative structure–property relationship (QSPR) models to predict skin permeation and skin sensitization accurately, based solely on the molecular structure of the chemical being considered. Overall, our nonlinear QSPR models are capable of predicting skin permeation data within 8% absolute average deviation, and skin sensitization probability is classified with 98% accuracy.…”