Modelling of the penetration of chemicals through skin - Final report

The aim of several studies carried out between 2010 and 2015 was to identify the chemical properties of importance for skin penetration and to build a model based on these properties that can be used as a tool to make assessments of the need for decontamination. The statistical tool Multivariate Data Analysis (MVDA) was used both for the selection of probe chemicals as well as for the evaluation of measured penetration data. Primary modelling showed that the size of a molecule, which is also related to the substance's volatility, dominates the skin penetration so strongly that additional important properties are difficult to identify. To try to elucidate any additional properties of importance for skin penetration a representative group of chemicals of similar molecular size was assembled. The inclusion of secondary properties should result in models with higher prediction abilities. The widely varying properties of the substances chosen resulted in extensive experimental measurements and methodological development, e.g. investigation of the effects of various receptor media (the liquid that collects chemicals that penetrate the skin) and the development of a new method for continuous measurement of skin penetration. The modelling carried out shows: ? None of the nine parameters chosen to characterize the breakthrough curve gave a good enough model. Modelling using the whole breakthrough curves gives a slightly better model. ? The best model in this study is built on twelve properties, five are literature values and seven computed values. Five are related to solubility, three to polarity, one to heat capacity, and three are molecular size parameters. ? The model is good at predicting the breakthrough curves for the chemicals on which the model is built, but cannot predict breakthrough curves for chemicals based on the evaluated parameters. Summing up; the extensive experimental measurements and the modelling carried out have not given the expected result, i.e. including secondary properties in the modelling does not give a better model than one based only on the primary property, molecular size.