Ann. occup. Hyg., Vol. 47, No. 2, pp. 168-172, 2003
© 2003 British Occupational Hygiene Society
Published by Oxford University Press
Letters to the Editor |
Reply
Occupational Medicine Division, Department of Clinical Medicine and Immunological Sciences, University of Siena, Italy
Received 28 August 2002; in final form 7 October 2002
The letter of Professors Kissel and Bunge ‘Comment on Sartorelli et al. (1998): invalid calculation of permeability coefficients’ and the reply of Doctors Semple, Cherrie, Brouwer and Dick allow me to discuss some key issues of dermal exposure assessment.
I premise that finite dose experiments are not the golden standard to calculate Kp. Nevertheless it is commonly considered reasonable to calculate Kp when steady-state (or pseudo steady-state) conditions are observed during a finite dose experiment. This can be considered a good estimate of the true Kp that can only be obtained in infinite dose experiments. Steady-state is considered achieved when the linear regression between cumulative dose in the receptor and time shows a very good correlation. These conditions were reached in our experiments as recognized by Professors Kissel and Bunge. Finite dose experiments are frequently adopted in occupational and environmental medicine because they are more realistic.
The first criticism was about the applied dose and the vehicle used. A volume of 20–30 µl applied chemicals is very often used in in vitro experiments with common diffusion cells. Acetone has been used in a number of percutaneous penetration studies in literature, while in other papers volatile vehicles (usually solvents such as xylene) have been used. This is due to the lipophilicity of many occupational toxicants.
Based on theoretical evaluations only, Kissel and Bunge consider a volume of 30 µl too small to completely cover 1.77 cm2 of skin. Researchers carrying out an in vitro experiment always observe if the skin surface is covered by the vehicle; consequently the applied volume is chosen and not just based on theory. Bronaugh and Stewart (1984) reported: ‘The 1.13 cm2 area of skin was completely covered by the vehicle (20 µl of acetone)’, and Franz (1975) in a well-known paper stated: ‘The volume added was 10 µl/cm2, which is sufficient to spread across the entire exposed surface (1 cm2).’ In the EU project ‘Evaluation of in vitro predictions of dermal absorption of toxic chemicals’ (EDETOX) that involves 12 institutes and organizations from seven different European countries, 25 µl has been considered the standard volume of applied chemicals. So Kissel and Bunge’s assertion is belied by a number of experimental observations over the last 30 years. If they would like to, readers can do a 10 min experiment themselves.
But the key issue of the letter is the calculation of Kp. Professors Kissel and Bunge state that to calculate Kp by dividing absorption flux by the concentration applied to the skin (measured as nmol/cm2) is wrong because this represents a constant with units h–1 and not cm/h.
If I knew that readers were interested in these aspects I would explain that to calculate Kp using applied concentration expressed as nmol/cm3 is not correct in our experimental conditions (i.e. chemical applied in a small volume of volatile solvent that evaporates leaving a film of chemical on the surface of the skin), which were chosen to simulate occupational dermal exposure. Kissel and Bunge also consider the original concentration in acetone inappropriate. Because an estimate of the applied concentration is needed to obtain comparable data, in our study we used concentration expressed in nmol/cm2 as an estimate of applied concentration in Fick’s law. So our results were expressed as Kp estimates, while the true Kp is only that obtained in infinite dose experiments that unfortunately do not replicate occupational exposure conditions.
I acknowledge the incorrect use of terminology: the term Kp in the specific case was not correct and I thank Professors Kissel and Bunge for their specification. This is partly due to the existing confusion in defining the permeability/absorption constant we measured. Kissel and Bunge reported the definition of ‘mass absorption rate constant’, Kt to indicate this parameter, but Kt has been used by other authors to mean something different (Guy and Hadgraft, 1989). VanRooij et al. (1993) defined this ‘absorption rate constant’ while others use ‘transfer factor’, TF, or ‘transfer constant’, TC. For clarity in the text I will use Ka to indicate our absorption rate constant.
If terminology is certainly important and Kp and Ka values are different, it is impossible to ascribe a different meaning to these two constants in our experimental conditions. In fact, when a given volume of solution is applied (i.e. finite dose), Ka is not only proportional to Kp, but there is also a simple correction factor ( f ):
f can be calculated as follows:
If D = applied dose (i.e. total amount of applied chemical in nmol), taking the following into account:
and considering that the applied volume is 30 µl = 0.03 cm3, and the exposure area is 1.77 cm2, this gives f as follows:
Multiplying our Ka by 0.01694915, Kp (calculated using concentration expressed as nmol/cm3) is obtained. So the statistical analysis is not affected by changing the type of concentration, and moreover I consider this change neither necessary, nor correct to calculate dermal uptake in working situations. The recalculation of permeability constants could be useful for comparing our results with literature Kps, taking into consideration that Kp is not an accurate estimate of skin permeability when the applied volume is negligible as in the working conditions reproduced in our experiment.
The focal point is occupational dermal exposure. In the workplace skin exposure normally occurs in one of three pathways: immersion, deposition or surface contact.
Immersion of hands in a liquid phase of the chemical is usually done when wearing gloves, so it does not represent a very important pathway. On the other hand, splashing and chemical residue transfer from a contaminated surface are very common in skin exposure cases. These two pathways are exactly reproduced by using our experimental conditions.
This explains our choice and makes it evident that the criticisms of Professors Kissel and Bunge are biased as they are based on a major mistake. Dermal exposure is commonly defined in the same way as respiratory exposure, with the amount of material reaching the skin and available for absorption referred to as skin loading (i.e. nmol/cm2) rather than concentration (Fenske, 1993). In fact, concentration of contaminants after deposition on the skin is not known and it is affected by evaporation, but it is possible to measure contamination as skin loading by using sampling techniques, such as patches, washing, wiping, etc.
As reported in our paper, the mass uptake through the skin can be estimated as follows (Cherrie and Robertson, 1995):
Msk = Kp Esk
where Esk = dermal exposure measured as [area]2 x [time].
Using Kp calculated from concentration expressed as nmol/cm3, cm/h x nmol/cm2 = nmol/cm/h is obtained but does not make sense. On the other hand, Ka suits a predictive model of percutaneous penetration very well, obtaining nmol/cm2/h exactly, which is the definition of the skin loading rate that has the same units of flux at the steady state. Skin loading and absorption rates are very often used in in vivo experiments and also in these cases Ka represents the only way to compare skin permeability measurements. Constants expressed as h–1 are also included in some PBPK models to describe the passage through the different skin layers.
Kissel and Bunge’s methodology is perfectly correct in pharmacology (where concentration of drugs is well known and the dose can be considered as infinite) and maybe in a large part of environmental studies, but is not entirely adaptable to occupational dermal exposure assessment. Calculating dermal uptake from skin loading is the only possible approach in occupational hygiene.
Professors Kissel and Bunge also claim that the general applicability of our algorithm is very unlikely because of variation in percutaneous penetration rates with skin loading changes; in other words, Ka would not be predictive as Kp because of existing differences in thermodynamic activity of different compounds.
Thermodynamic activity depends on solubility of the chemical in the vehicle and in the skin. So it depends on lipophilicity of the compound that is expressed as Kow (octanol–water partition coefficient) and [water] (water solubility). Actually, our model takes into consideration these parameters and the multiple regression analysis between K, ln Kow and [water] was highly significant, confirming that Ka depends on lipophilicity of the compound. It is implicit that QSAR predictions should be limited to the same classes of compounds.
Kissel and Bunge affirm that Ka decreases increasing skin loading while Kp is independent of concentration as long as the skin is not damaged.
In 1998 Dr Zendzian from USEPA stated, in a private mail exchange, that my model was inapplicable because of changes in Kp at different applied doses of chemicals. He declared that he had plenty of data to prove this in both in vivo and in vitro. Driven by these comments, I recently observed that mercury Kp (calculated using the applied volume as quotient) decreases, increasing the applied concentration even if the barrier was not damaged (paper submitted).
This phenomenon represents an overall limit of percutaneous penetration that does not depend on what type of permeability/absorption constant is considered. Absorption rates in general seem to be affected by increasing the applied dose, probably because the solubility into the stratum corneum is a rate-limiting process. Metabolism, formation of stratum corneum reservoir, and loss of sink conditions can also play a role in this. Biological processes always have some limits. It is evident that for mercury (and perhaps for other compounds) the variability of Kp and other penetration/absorption constants such as Ka limits the use of QSAR and other predictive models that are restricted to a range of applied concentrations. Anyhow, this range is reasonably broad.
On the other hand, Kp cannot be considered as a constant for the chemical. Kp for the same chemical in different vehicles is different because of changes in thermodynamic activity (depending on solubility of the chemical in the vehicle and in the skin). It derives that predictive models based on Kp can only predict percutaneous penetration from a specific vehicle.
The letter of Professors Kissel and Bunge is puzzling. It is not clear if the most important problem is considered terminology, the use of Ka in predictive models or our experimental conditions. Two other papers are cited (Sartorelli et al., 1997, 1999) whose aim was not to produce percutaneous penetration data for skin exposure assessment but to compare penetration of chemicals as neat compounds and in commercial products. Kissel and Bunge claim that experimental conditions reproducing occupational skin exposure are not valid. At the same time they apply theoretical principles with a limited practical significance for dermal exposure assessment in occupational medicine. They cite studies that found that absorption rates of solvent-deposited solids decrease linearly (vary inversely) with skin loading above a critical value representing complete coverage. This does not influence percutaneous penetration in the working situations and in our experimental conditions, where the applied volume can be considered negligible. On the other hand, Professor Bunge herself has proved that penetration rates of a pure solid chemical are almost the same as from an aqueous solution (Bunge and Ley, 2002).
Semple et al. (2003) acknowledged that the use of our regression equation led to an error in their model predictions because they inappropriately calculated flux by using constants with units of h–1 (instead of cm/h). In their model, to correct this error, fluxes can be estimated by multiplying our absorption constants Kas by the factor.
In the absoption part of Semple et al.’s model there is another problem. They stated ‘Although solvent is evaporating with time, the thickness of the paint layer will also be reducing (changing from wet film thickness to dry film thickness) and so the model assumes that the solvent concentration on the skin or clothing remains approximately constant and is the same as that of the initial paint mix.’ This assumption does not take into account that thermodynamic activity and Kp change when the solvent is a dry film on the skin surface. In these conditions dermal uptake depends on skin loading and the only way to calculate fluxes is by using Ka. An alternative is represented by the use of absorption percentages instead of absorption fluxes.
Moreover, we did not suggest using our model for risk assessment because it can only predict percutaneous penetration with regard to those particular experimental conditions. In particular we never compared Cercopitecus skin penetration to human skin penetration. The aim of the study was to confirm that models based on experimental data give better statistical results that those based only on theoretical assumptions. At the end of the article we stated ‘Our results suggest that skin notation can be defined by Kp and lag time data which is partly experimental and partly calculated using models, such as the present, based on the results of in vitro experiments. To do this, it will be necessary to obtain internationally accepted penetration rates....This will require the development of a standard in vitro test from which to derive predictive models.’ In this respect there are no differences between the opinions expressed in Semple et al.’s letter and those presented in our paper.
I apologize for the misleading use of terminology, and that readers of our article misunderstood the proposed model as a practical tool rather than a proposal of a different approach in the specific field.
In their reply Semple et al. reanalysed our data by comparing fluxes with other published information. In general, this approach is hampered by the fact that flux directly depends on the applied dose. In particular, Dutkiewicz and Tyras (1968) demonstrated that toluene absorption was proportional to its concentration, but for xylene they only reported a range of absorption rates in volunteers without the respective applied concentrations. Thus, those results are not comparable with any other percutaneous penetration data in literature. On the other hand, the aim of the excellent paper by Kezic et al. (2001) was to obtain percutaneous penetration rates under non-steady-state conditions. These data are not comparable to steady-state absorption rates.
The other manuscript cited by Semple et al. (Wilkinson and Williams (2001) contains a number of percutaneous penetration data regarding solvents, obtained in a very well-conducted in vitro study. In particular, for neat xylene they reported the following results shown in Table 1. From these, we can derive the following values:
|
m-xylene Ka = 1.80 x 10–3
o-xylene Ka = 1.35 x 10–3
These results can be compared to the following Ka obtained from our model:
m-xylene Ka = 6.83 x 10–3
o-xylene Ka = 7.07 x 10–3
By multiplying our Ka by f we obtain a permeability constant comparable to the Wilkinson and Williams’s Kp:
m-xylene Kp = 11.58 x 10–5
o-xylene Kp = 11.98 x 10–5
Semple et al.’s evaluations are inaccurate: there is not an order of magnitude between our permeability/ absorption constants and the experimental ones. Differences can be easily explained by the different in vitro model (dermotomed human skin, flow-through cells, receiving phase added with PEG 20 instead of full-thickness monkey skin, static cells, receiving phase added with bovine albumin, respectively) and the huge difference in the applied dose. Moreover, QSARs are limited, as explained. Our model was based on percutaneous penetration data of polycyclic aromatic hydrocarbons (PAHs), organophosphorus insecticides and phenoxyacetates and it is very difficult to use it to predict absoption of very different compounds such as solvents.
VanRooij et al. (1993) measured, in human volunteers, the surface disappearance of PAHs contained in coal tar by detecting the luminescence signal. They found for PAHs a skin absorption rate constant (h–1) of 0.066 (mean of different anatomical sites of human skin). In Table 2, considering the relative contribution to the response signal, absorption rate constants of some PAHs are reported compared to our in vitro Ka.
|
In a human volunteer study (Griffin et al., 1999), 1046 nmol/cm2 of chlorpyrifos was administered to the forearm skin, and an absorption rate of 1.30 nmol/ cm2/h was measured. In this case Ka = 1.24 x 10–3, while in our in vitro study for chlorpyrifos we obtained Ka = 1.28 x 10–3.
These differences are certainly acceptable for in vivo/in vitro comparisons and can be explained by the different permeability of human and monkey skin.
Even if our in vitro methods were not standardized and an animal model was used, our results and predictions seem reliable in a wide range of situations.
Finally, models for dermal risk assessment should take into consideration experimental results and issues related to dermal exposure, not only mathematical calculations, otherwise they could be misleading. Professors Kissel and Bunge consider that only thermodynamic activity of compounds regulates percutaneous penetration. However, this is not a simple process but the effect of biological, chemical and, to a lesser extent, physical factors. This is demonstrated by the large intra- and inter-individual variability of percutaneous absorption and the number of uncertainties that prevent the use of dermal occupational exposure limits in quantitative risk assessment (Sartorelli, 2002). On the other hand, mathematical parameters represent a way to compare and to predict (with some limits) percutaneous penetration data and they should be considered as a very useful tool.
In my opinion, the use of predictive models in dermal exposure assessment should be based on solid foundations in occupational hygiene. Attention should be focused on exposure pathways in the workplace. Dermal exposure models should incorporate details of the mode of exposure and must be validated using well-characterized exposure scenarios (Van Hemmen, 1997). This is the way followed in the EU research project ‘Risk assessment for occupational dermal exposure to chemicals’ (RISKOFDERM) that includes a qualitative survey of processes, tasks, populations and determinants of dermal exposure throughout Europe as starting point (Sartorelli, 2002). In this sense the paper of Semple et al. is certainly adequate.
REFERENCES
Bronaugh RL, Steward RF. (1984) Methods for in vitro percutaneous absorption studies III: hydrophobic compounds. J Pharm Sci; 73: 1255–8.[Web of Science][Medline]
Bunge A, Ley E. (2002) Dermal exposure to powdered solids and acqueous solutions: are they different? Poster presented at the International Conference on Occupational and Environmental Exposure of Skin to Chemicals: Science and Policy, Washington, DC, 8–11 September 2002.
Cherrie JW, Robertson A. (1995) Biologically relevant assessment of dermal exposure. Ann Occup Hyg; 39: 387–92.
Dutkiewicz T, Tyras H. (1968) Skin absorption of toluene, styrene and xylene by man. Br J Ind Med; 25: 243.[Web of Science][Medline]
Fenske RA. (1993) Dermal exposure assessment techniques. Ann Occup Hyg; 37: 687–706.
Franz TJ. (1975) Percutaneous absorption. On the relevance of in vitro data. J Invest Dermatol; 64: 190–5.[CrossRef][Web of Science][Medline]
Griffin P, Mason H, Heywood K, Cocker J. (1999) Oral and dermal absorption of chlorpyrifos: a human volunteer study. Occup Environ Med; 56: 10–13.
Guy RH, Hadgraft J. (1989) Structure–activity correlations in percutaneous absorption. In Bronaugh RL, Maibach HI, editors. Percutaneous absorption. New York: Marcel Dekker. pp. 95–109.
Kezic S, Monster A, van de Gevel I, Kruse J, Opdam J, Verbek M. (2001) Dermal absorption of neat liquid solvents on brief exposures in volunteers. Am Ind Hyg Assoc J; 62: 12–8.
Sartorelli P. (2002) Dermal exposure assessment in occupational medicine. Occup Med; 52: 151–6.[Abstract]
Sartorelli P, Aprea C, Bussani R, Novelli MT, Orsi D, Sciarra G. (1997) In vitro percutaneous penetration of methyl-parathion from a commercial formulation through the human skin. Occup Environ Med; 54: 524–5.
Sartorelli P, Cenni A, Matteucci G, Montomoli L, Novelli MT, Palmi S. (1999) Dermal exposure assessment of polycyclic aromatic hydrocarbons: in vitro percutaneous penetration from lubricating oil. Int Arch Occup Environ Health; 72: 528–32.[CrossRef][Web of Science][Medline]
Semple S, Brouwer DH, Dick F, Cherrie JW. (2001) A dermal model for spray painters. Part II: Estimating the deposition and uptake of solvents. Ann Occup Hyg; 45: 25–33.
Semple S, Cherrie JW, Brouwer DH, Dick F. (2003) Reply. Ann Occup Hyg; 47: 167–168.
Van Hemmen JJ. (1997) Dermal exposure and risk assessment. Ann Occup Hyg; 41: 729–31.
VanRooij JGM, DeRoos JHC, Bodelier Bade MM, Jongeneelen FJ. (1993) Absorption of polycyclic aromatic hydrocarbons through human skin: differences between anatomical sites and individuals. J Toxicol Environ Health: 38: 355–68.
Wilkinson S, Williams F. (2001) In vitro dermal absorption of liquids. Contract Research Report 350/2001. UK: HSE Books.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||