ALPHA-TOCOPHEROL ACETATE PERMEATION - C 8000 � f 6000 c 1 1 4000 .. . :i 'i E � 2000 :I u 0 ♦ Ethanol Solution o 1%Klucel • 3% Klucel f 2 3 4 5 Time (hr) 99 Figure 1. Average (n = 3) cumulative amount of ATA ± standard deviation found in the receiver com­ partment of modified Franz cells for the regenerated cellulose membrane experiments. ANOV A performed on the cumulative AT A amount found in the receiver at each time point showed that the isopropyl myristate formulation produced a significantly larger permeation of ATA through the skin than the other formulations (p 0.001). There was no difference in the cumulative amount detected among the other formulations. There­ fore, it can be concluded that 1 % and 3% Klucel® or ethanol solution and light mineral oil solution without Klucel® did not make any difference in the permeation of ATA through the skin. Small amounts of AT were detected only in some studies and did not increase with time, suggesting that the AT detected was not a metabolite of AT A but may be attributed to the AT already present in the human cadaver skin. DISCUSSION In this study, the permeation through human cadaver skin of ATA was determined using a new mathematical approach. The method used a modified Franz cell apparatus and 95% degassed ethanol in the receiver compartment. Previous ATA permeability studies (14) used Dulbecco's modified phosphate-buffered saline to maintain skin viability and 3% bovine serum albumin to improve ATA solubility, as reported in the literature (19). However, the study reported that the ATA collected in the receiving medium of the Franz cell was always negligible (below detection limits). Another study (21) reported the use of phosphate buffer and Tween 80 as a receiving medium, and again it failed to show ATA in the receiving compartment. The goal of the present study was to find an experimental setup that allowed a fast and relatively accurate method to evaluate the permeability properties of different formulations of AT A across the skin. The mathematical method used (16) requires sink conditions to work. Therefore, the solubility of ATA in the receiver media is a key part of the experiment, even if it compromises the viability of the skin. Solubilization of ATA in Dulbecco's buffer with an increasing percent of ethanol provided unsatisfactory results. Finally, it was decided to use 95% ethanol because ATA is freely soluble in this media. However, the use of a lipophilic receptor fluid has the potentiality to extract lipids from the skin barrier and

100 JOURNAL OF COSMETIC SCIENCE to artificially increase skin permeability (22). Because this artifact would be common to all the formulations tested in this experiment, it would not affect the relative results. Previous studies showed that ATA is slowly metabolized in the viable part of the skin (20). Therefore, if some metabolic capabilities were left in the skin, AT would be detected in the receiver container. For this reason, the samples collected from the receiver container were analyzed with an assay able to quantify both ATA and AT. However, AT was detected only in a few studies and it didn't increase with time. It is then possible that the AT detected was not a metabolite of ATA but came from AT content already existent in the human cadaver skin. ATA was detected in the receiver medium in a time-dependent way, and cumulative­ amount perfused curves could be built (Figure 2). The mathematic model proposed by Bellantone et al. (16) was used in this study because the donor compartment is unstirred. A popular method to estimate skin permeability in vitro is the lag-time method (23), which requires that the drug be held at constant concentration in the donor compart­ ment. However, that method requires a constant concentration at the donor-membrane interface. Because in our experiment the donor is unstirred and possibly substantial depletion of drug occurs due to drugs that cross the skin, the concentration in the donor is not constant and the lag-time analysis does not apply. Figure 3 shows the model fits obtained from the lag-time method (gray line) and equation 1 (black line) for one set of experimental data. It can be seen that equation 1 has a more natural and accurate fit than the lag-time model. In addition, it has been pointed out (16) that the method gives accurate values for the permeability, using experiments that are usually considerably shorter (by half or one-third) than those required by lag time or other steady-state approximation methods. The mathematical model used in this paper to estimate permeability through human cadaver skin (membrane is the rate-limiting step) requires a previous estimate of the diffusion coefficient in the donor compartment D d· For the ethanol solution and the two gel formulations, D d was estimated by permeability studies through a regenerated cellulose membrane. This method was considered appropriate because the similarity between the medium of the donor compartment and the receiver compartment would minimize backflow of fluid to the donor, which could significantly decrease the net release of the drugs into the receiver. A value for D d was obtained for the ethanol donor, and lower limits for D d were obtained for the alcoholic gels. The same approach was attempted for isopropyl myristate and light mineral oil by filling the receiver compart­ ment with isopropyl myristate and light mineral oil, respectively. However, the experi­ ment was technically difficult because isopropyl myristate and light mineral oil, used as receiving medium, could not be injected directly into the HPLC. By UV spectropho­ tometer analysis, the concentration of ATA in the receiver was always below the limit of detection. Hence, the diffusion coefficients of ATA from the isopropyl myristate and light mineral oil solutions were estimated by measurement of the viscosity of the solvents and by calculations. The results of these studies show that, under the experimental conditions used, the formulation of ATA in isopropyl myristate has a permeability though human cadaver skin that is two orders of magnitude larger than those of the other formulations. A possible explanation for this finding is that isopropyl myristate has a blend of polar and non-polar properties, which probably mimic to some extent the complex lipid/polar