An equation for the prediction of human skin permeability of neutral molecules, ions and ionic species
Experimental values of permeability coefficients, as log Kp, of chemical compounds across human skin were collected by carefully screening the literature, and adjusted to 37 °C for the effect of temperature. The values of log Kp for partially ionized acids and bases were separated into those for their neutral and ionic species, forming a total data set of 247 compounds and species (including 35 ionic species). The obtained log Kp values have been regressed against Abraham solute descriptors to yield a correlation equation with R2 = 0.866 and SD = 0.432 log units. The equation can provide valid predictions for log Kp of neutral molecules, ions and ionic species, with predictive R2 = 0.858 and predictive SD = 0.445 log units calculated by the leave-one-out statistics. The predicted log Kp values for Na+ and Et4N+ are in good agreement with the observed values. We calculated the values of log Kp of ketoprofen as a function of the pH of the donor solution, and found that log Kp markedly varies only when ketoprofen is largely ionized. This explains why models that neglect ionization of permeants still yield reasonable statistical results. The effect of skin thickness on log Kp was investigated by inclusion of two indicator variables, one for intermediate thickness skin and one for full thickness skin, into the above equation. The newly obtained equations were found to be statistically very close to the above equation. Therefore, the thickness of human skin used makes little difference to the experimental values of log Kp.