Objectives The occupational hazard of chemical absorption via dermal route was frequently assessed by evaluating the skin permeability (Kp) of chemical. In addition to in vitro testing using human skin, the quantitative structure-activity relationship (QSAR) has been employed as an alternative source to providing Kp. In the early Kp QSARs the octanol-water partition coefficient (log K_OW) and molecular weight (MW) were commonly applied as dominant properties to describe transdermal transport of chemical. This study examined the efficacy of log K_OW-/MW-based QSARs in Kp prediction.

Methods One hundred and fifty-eight chemicals of known Kp determined in vitro using human skin were used to evaluate the goodness of fit (R²) of the model estimates approximating the measured Kp for six log K_OW-/MW-based Kp QSARs reported in Wilschut et al. (1995) and Mitragotri (2002). A new Kp QSAR consisting of additional descriptors for the same compounds was developed by identification of key descriptors from a pool of 3,224 descriptors supported by Dragon^® followed by stepwise regression.

Results For the log K_OW-/MW-based Kp QSARs, the regression of model estimates against experimentally determined Kp yielded R² of 0.314 to 0.744, with the lowest value observed for the model employing log K_OW alone. In the new QSAR, in addition to log K_OW the electrostatic distribution in the molecular space appeared to be a significant factor affecting Kp, while the MW exerted its influence as a sub-domain, thus under constraints, of antineoplastic properties.

Conclusions As the Kp QSAR continues to evolve, attention may be required of on interpreting the limitations of MW as a Kp descriptor. Four of the investigated Kp QSARs show a R² close to 0.7 or higher when predicting Kp, suggesting a consistent performance of these models to serve as a tool of dermal hazard characterisation.