QSARs for skin permeation of chemicals

Workers are occupationally exposed mostly by inhalation and by skin contact. Dermal exposure gets considerably more attention in health risk assessments. The exposure via inhalation is reduced due to improved occupational hygiene measures and the contribution of dermal exposure to the total daily exposure is assumed to be increased comparatively.

External dermal exposure or better dermal deposition has been studied extensively and has finally resulted in regulatory default deposition values as mentioned in the European Technical Guidance Documents (EU 2003). One should realise, that dermal deposition is generally studied for chemicals with a relatively slow permeation rate in order to enable analysis of dermal deposition. However, studies into dermal absorption and especially into the dermal absorption rate in the occupational environment are less numerous. The European Technical Guidance Documents (EU 2003) recommends a conservative estimate of 100% dermal absorption for a substance with a molecular weight of smaller than 500. However, the kinetic aspects in this guidance are completely ignored.

This webpage provides a refinement of the dermal absorption model of Wilschut et al. (1995). The refinement includes a revision of the mathematical description of the model and the estimation of the lag time from published validated skin permeation coefficients and from stratum corneum/water partition coefficients of generally apolar substances. The permeation coefficient and the stratum corneum/water partition coefficient together are indispensable for the estimation of the lag time.

A QSAR for estimation of the skinpermeation coefficient through human skin in vitro may be derived from published skin permeation coefficients. The European regulatory risk assessors are convinced, that there are no reliable databases for skin permeation in vitro and in vivo.

Nevertheless, a database of validated skinpermeation coefficients has been published by Vecchia and Bunge (2002a,b). This database was used to derive 3 QSARs for estimating a skin permeation coefficient on the basis of the following starting points and are mathematically formulated as below:

  1. The skin barrier is formed by the stratum corneum, to be considered as an homogeneous membrane.

  2. The skin barrier consists of the stratum corneum but the resistance is caused by the intercellular route through the skin lipids and the transcellular route by diffusion through the corneocytes, indicated with stratum corneum with 2 pathways.

  3. The skin barrier consists of the stratum corneum with the intercellular and trancellular route, but there is an additional resistance due to the viable epidermis.

The difference between the 3 models for the permeation coefficent and the maximum absorption from a saturated aqueous solution is shown for a substance with a molecular weight of 200 dependent on the octanol/water partition coefficient. The maximum solubility in water was estimated from the QSAR for water-solubility of EPI-Suite v3.20 (EPA 2007, equation 19 of WsKowWin.exe). The permeation coefficient and the maximum aqueous absorption have been plotted against the octanol/water partition coefficient both on a logarithmic scale.

Between a log(Kow) of -2 and +4 there is not much difference between the three QSARs. The highest aqueous permeation coefficient is mostly predicted by the QSAR, assuming that the stratum corneum has 2 pathways and that epidermal resistance has hardly any influence on the permeation coefficient. On the basis of experimental data this QSAR is considered to be most close to the measured skin permeation coefficients.

The regression coefficients of the QSAR assuming a stratum corneum with 2 pathways (item 2 above) were derived by means of non-linear regression analysis. The variance of the logarithmic transformed data was explained by the regression equation for 68 percent. Take notice that the effect of the molecular weight is fully different between the 2 pathways. The results are presented in the two tables below.

Residual variance =    0.4686
Degrees of freedom =    177
Fraction of regression explained =    0.682
b 1 = -2.590E+00 (SD for b 1 = 1.337E-01)
b 2 = 7.318E-01 (SD for b 2 = 5.396E-02)
b 3 = -6.832E-03 (SD for b 3 = 5.470E-04)
b 4 = 4.300E-02 (SD for b 4 = 6.158E-02)
b 5 = 1.361E+00 (SD for b 5 = 3.034E-01)

The graphs below provide additional information on the goodness of fit:

QSAR for stratum corneum/water partition coefficient

The aqeous skin permeation coefficient, multiplied with the maximum solubility in water of the chemical of interest, results into the maximum absorption rate in mg/cm2/hour. However, this is only valid in case of steady state of stratum corneum partition, that means that the concentration profile in the stratum corneum does no longer change in contact with the aqueous solution of the chemical of interest. It takes some time for achieving the steady state of maximum permeation and this time may be considered to be the lag time. An estimate of the lag time is only possible, if information on the stratum corneum partition coefficient is available.

Vecchia and Bunge (2002) have also published a database on validated wet Stratum Corneum/Water partition coefficients. With the help of this database a regression line for the wet stratum corneum/water partition coefficient has been developed according to the equation below.

The regression coefficients of the QSAR for the stratum corneum/water partition coefficient have been derived by means of non-linear regression analysis. The variance of the logarithmic transformed data was explained by the regression for nearly 76%. The results are in the two tables below.

Residual variance =  0.1326
Degrees of freedom =  95
Fraction of regression explained =  0.7578
a 1 = 0.7201 (SD for a 1 = 0.1157)
a 2 = 0.4298 (SD for a 2 = 0.02493)

The plot of the observed and estimated stratum corneum/water partition coefficient provides additional information on the accuracy of this regression line.

Still a problem remains for highly water soluble substances with very small octanol/water partition coefficients for instance sucrose (logKow = -3.7), raffinose (logKow about -6) and stachyose (logKow about -9). Because the substance with the lowest octanol/water partition coefficient in this database is water (logKow = -1.38), it is very difficult to extrapolate lower than logKow = -1.4. For stachyose the regression line predicts, that the wet SC/water partition coefficient on a volume basis should be about 0.0001. This seems to be not very probable, because stachyose is a very water soluble compound, the wet stratum corneum contains about 0.8 ml water per ml Stratum Corneum and the partially hydrated stratum corneum about 0.35 ml per ml Stratum Corneum (Nitsche et al. 2006). So a partition on a volume basis of around 0.5 or higher is well possible for stachyose. To extend the use of the regression line to logKow lower than -1.4 the predicted wet SC/water partition coefficient is increased with 0.5 for all cases.

The importance of the stratum corneum/water partition coefficient is obvious from the following equations. The diffusivity and lag time can only be assessed if both the aqueous permeation coefficient and the stratum corneum/water partition coefficient are measured or estimated from a reliable QSAR.

By using the equations above it is possible to get understanding of the kinetics of skin permeation from aqueous solutions.


Transformation from aqueous permeation coefficient to vapour in air permeation coefficient

The permeation coefficient for aqueous solutions (measured or predicted) can be easily converted to a permeation coefficient for gases by multiplication with the partition coefficient of the gas or vapour between water and air or by multiplying with the reciprocal measured dimensionless Henry coefficient. Multipication of the Kpsk-water with the Kwa produces the permeation coefficient of the skin (Kpsk-air, in cm/hour) for gaseous compounds.

For gaseous compounds permeating very fast through the skin, the diffusion from the air to the skin surface may become the rate limiting factor. In order to take this effect into account it is assumed that a layer of stagnant air is present between skin surface and workroom air. The thickness of this air layer for air diffusion modelling purposes is estimated to be 3 cm for light work clothing (Lotens and Wammes in 1993). The air diffusion constant (Dair, in cm2/hour) and the permeation coefficient of the air boundary layer on the skin (Kpair, cm/hour) are to be estimated according to the simple physical diffusion law of Fick. The overall Kpsk-air-air (in cm/hour) is controlled by the diffusion through the air boundary layer on the skin (Kpair) and diffusion through the skin (Kpsk-air).

The skin permeation rate of gases is estimated by multiplication of the Kpsk-air-air with the atmospheric concentration in mg/cm3. The atmospheric concentration is limited to its maximum vapour pressure at ambient temperature.

The advantage of this simple approach is to get some feeling for the skin permeation of vapours, when the permeation coefficient from aqueous solutions is known. Applying this model will predict a high permeation rate for gaseous aniline, nitrobenzene, glycolethers etc. and a low permeation rate for gaseous hydrocarbons like benzene and toluene. This is known from industrial practice, so this model seems to be consistent with real life observations. Not including the air boundary layer in the equation for the permeation coefficient of gases will result in unrealistic high estimates of the skin permeation rate in case of fast skin permeating substances. Model prediction of the permeability of vapours of apolar substances are mostly more accurate then that of the neat liquid. Direct skin contact with low levels of vapour does not result in defatting or irritation of the skin like may occur with neat liquid. Defatting and irritation of the skin are not considered in predicting skin permeation coefficients from only physico-chemical parameters.

Exposure to liquids

Distinction is needed between the following exposure scenarios:

Skin contact with the neat liquid may cause structural changes in the stratum corneum due to defatting or irritant action. Defatting has an impact on the lipid structure of the stratum corneum, irritants will change the protein conformation. Both effects result in loss of the barrier function. Therefore permeation coefficients from aqueous solution do often not apply to dermal contact with the neat liquid in case of irritant and defatting liquids.

Another complicating issue is the occurrence of skin metabolism. If the permeation occurs slower than enzymatic decomposition, skin metabolism may convert the substance in faster permeating metabolites. Prediction based on only physico-chemical properties is no longer feasible.

If the liquid full strength does not change the structure of the stratum corneum, the permeation might be considered as permeation from an aqueous solution saturated with the liquid. The transfer from the stratum corneum to the blood is assumed to occur via a very thin water layer and to be controlled by aqueous diffusion. The water solubility can not be exceeded in the aqueous diffusion process.

In case of contact with aerosols like may occur during painting, the liquid will deposit on the skin. After deposition on the skin the following events may occur dependent on the physico-chemical properties and the dropsize:

The permeation through the stratum corneum becomes higher if the evaporation requires more time. Evaporation and permeation is dependent on physico-chemical properties like vapour pressure, water solubility, octanol/water partition coefficient and molecular weight. If the aerosol deposition is so high that evaporation to dryness of the skin does not occur, the exposure becomes more and more similar to dermal contact with the neat liquid.

In order to get some feeling for the fate of liquid aerosol on the skin, one should estimate the rate of permeation into the stratum corneum and the rate of evaporation at the same time. Material disappears by evaporation and permeation. If the liquid layer has evaporated, the substance continues to evaporate from the stratum corneum and to permeate through the stratum corneum into the blood at the same time. Riviere et al. (1999) studied the dermal absorption of topically dosed jet fuels to isolated perfused porcine skin flaps. They observed, that the naphtalene in a drop of 25 microlitre of jetfuel evaporated for 95% and of dodecane for about 89% in a period of 5 hours. The percentage of naphtalene in the perfusate was 1.4% of the applied dose . In case of dodecane this figure was 0.4%. The remaining resided in the epidermis. This process can only be modelled on the basis of diffusion layers. However, to show the appropriate model would be too detailed for this paper.

Exposure to solids

Dermal exposure to solids may occur via dust of the substance in air. The following processes are relevant:

It has been observed that the deposition rate of solids on clothes and the bare human skin is on the average 10 times higher than the normal dust sedimentation rate on horizontal surfaces. This was observed under standardised conditions and are assumed to be caused by air turbulence due to heat radiation of the body (Byrne et al. 1995). Solid material might also be lost from the skin of the hands due to several manual handlings, but quantitative data are lacking. The transfer of solids from contaminated surfaces of the workfloor to the skin has been observed to be between 1 and 10 % of the surface contamination (Paull 1997, Brouwer et al. 1999). These observations might be used as a rule of thumb to get some feeling for the deposition or transfer of solids to the skin of head and forearms.

How does the transfer of deposited solid material into the stratum corneum occur? Have the solid particles some vapour pressure or are they dissolved in the sweat from the skin. Might it be assumed, that the material is present as saturated solution and will permeate the skin at a rate similar as permeation from an aqueous saturated solution? An alternative assumption is to consider the maximum water permeation rate through the skin. The dermal absorption of a substance can never be higher than the amount dissolved in the maximum amount of water, that can permeate through the skin. These are more or less the considerations to make a worst case estimate of the extent of dermal absorption of a solid substance through the skin.

In the scope of contract research the permeation of phenylglycinamide, applied at a dose of 1000 microgram per cm2 as solid powder on human skin in vitro, was measured. The permeation was no more than 2 % of the applied dose over a period of 24 hours, that is 20 microgram per cm2 per 24 hours (Leeman and van de Sandt 1998).

The maximum permeation rate of phenylglycinamide from an aqueous saturated solution on the basis of the regression model, presented in this paper, was estimated to be 187 microgram per cm2 per 24 hours versus 20 microgram observed. The observed lag time was 4.15 hours and the estimated lag time 2.0 hours according to detailed model developments (see attached download file with underlying data).

As an alternative assumption it was estimated, that over 24 hour no more than 17 microliters water may pass the stratum corneum per 1 cm2. The maximum solubility of phenylglycinamide is 51.3 microgram per microliter. This means that the permeation will never exceed 872 microgram per cm2 per 24 hours.

In practice no more than 20 microgram per cm2 was permeating the skin in 24 hours. This means, that this recent Excel application(2007) for estimating skin permeation overestimated the measured permeation of phenylglycinamide with a factor of 9.

Conclusions on skin permeation modelling

1. Published permeation coefficients of organic substances from aqueous solutions through human skin in vitro appeared to support a theoretical model for simulation of permeation of organic substances through the skin.
2. Modelling of skin permeation requires not only substance properties like the octanol/water partition coefficient and the molecular weight or the molar volume, but should also include diffusion kinetics.
3. Diffusion kinetics may provide additional understanding for the rate of permeation of gases, of liquids and of solid substances dissolved in water.
4. The model applies to non-ionised substances, which do not irritate, do not remove lipids from the skin and permeate faster than the substance is metabolised in the epidermis.
5. Model predictions are accurate within one order of magnitude. This is accurate enough to get some feeling for the contribution of dermal absorption in comparison with absorption by inhalation or ingestion.


Brouwer DH, Kroese R and Van Hemmen JJ, 1999. Transfer of contaminants from surface to hands: experimental assessment of linearity of the exposure process, adherence to the skin and area exposed during fixed pressure and repeated contact with surfaces contaminated with a powder. Applied Occuaptional and Environmental Hygiene 14, 231-239.

Byrne MA, Bell KF, Goddard AJH, Lange C and Roed J, 1995. Aerosol deposition velocity measurements to human body surfaces in Proceedings of the ninth annual conference of the Aerosol Society, Aerosols, their generation, behaviour and application.

EU 2003. European Technical Guidance Documents

Lotens WA and Wammes LJA, 1993. Vapour transfer in two-layer clothing due to diffusion and ventilation. Ergonomics 36(10), 1223-1240.

Leeman WR and Sandt JJM van de, 1998. In vitro percutaneous absorption study with phenylglycinamide through viable human skin membranes. TNO-Nutrition and Food Research Institute, TNO report V98.875, October 1998.

Mackay D, 1982. Correllation of bioconcentration factors. Environmental Science & Technology 16, 274-278.

Nitsche JM, Wang T-F and Kasting GB, 2006. A two-phase analysis of solute partitioning into the stratum corneum. Journal of Pharmaceutical Sciences 95(3), 649-666.

Paull JM, 1997. A proposed risk-based model for the evaluation of surface contamination and the assessment of potential dermal exposure. Thesis submitted to: the School of Hygiene and Public Health of the Johns Hopkins University, Baltimore, Maryland.

Riviere JE, Brooks JD, Monteiro-Riviere NA, Budsabe K and Smith CE, 1999. Dermal absorption and distribution of topically dosed jet fuels Jet-A, JP-8, and JP-8(100). Toxicology and Applied Pharmacology 160, 60-75, 1999.

Vecchia BE and Bunge AL, 2002a. Skin absorption databases and predictive equations. Chapter 3 in Transdermal Drug Delivery, edited by Guy RH and Hadgraft J, Publisher Marcel Dek-ker.

Vecchia BE and Bunge AL, 2002b. Partitioning of chemicals into skin: Results and Predictions. Chapter 4 in Transdermal Drug Delivery, edited by Guy RH and Hadgraft J, Publisher Marcel Dekker.

Wilschut A, Berge WF ten, Robinson PJ and McKone TE, 1995. Estimating skin permeation. The validation of five mathematical skin permeation models. Chemosphere 30(7), 1275-1296.

Underlying databases and Windows application for understanding skin permeation of chemicals

The underlying databases, the program for estimation of the regression coefficients of the 3 QSARs, describing permeation from aqueous solution, and a MS-Windows application, using the QSAR considering permeation through the transcellular and intercellular pathways through the stratum corneum, can be downloaded from this website. Simple manuals have been included. You are free to add or replace records in the database SkinPermBasicData.xls to study the impact of changing records on the regression coefficients of the QSARs for the aqueous permeation coefficients.

Download databases, manuals and application!


Comments on this new approach are highly appreciated. Please send you suggestions for improvement to Wil ten Berge


Return to Homepage