The Organisation for Economic Co-operation and Development (OECD) guidelines specify the use of split skin (e.g., dermatomed to about 0.5 mm) or the full thickness skin (not more than 1 mm thick) in percutaneous absorption studies [7]. However, the inter- and intra-variability of full skin thickness from human donors and skin animal species might compromise the comparison of skin absorption parameters (e.g., percentage of dose absorbed, apparent flux), whereas the use of dermatomed skin requires specialized staff [15, 16]. Therefore, taking into account its relative constant thickness, reconstructed epidermis was proposed for the determination of percutaneous uptake [13]. However, the permeation of caffeine and testosterone (i.e., reference compounds recommended by OECD in percutaneous penetration studies) through different reconstructed epidermis models was found higher than that determined through human epidermal sheet and full thickness porcine skin questioning the validity of such models in percutaneous absorption studies [13].

The present study examined the relevance of heat separated epidermis (HSE) in percutaneous absorption studies by using complementary procedures including caffeine permeation and transepidermal water loss (TEWL) measurements, as recommended by the OECD in test guidelines, in which TEWL allocates the evaluation of stratum corneum thickness [8], skin barrier function in chemical testing [7, 9] as an alternative to tritiated water permeation [5] and skin electrical resistance assay [7]. The permeation of caffeine as exogenous compound was analysed concomitantly to the water transport (i.e., TEWL) taken as an endogenous standard of dermatomed skin and HSE permeability [3].

Fresh dermatomed skin specimens (∼500 μm in thickness) were prepared from human donors (three males, 68 ± 18 years) collected by the Northern California Transplant Bank (Oakland, CA, USA), then stored in media at 4°C and finally used within 12–48 h. Epidermis was separated from dermatomed skin specimens (2 cm × 6 cm) by heating to 55°C for 30 s in phosphate buffer saline solution. HSE was carefully removed from underlying tissue by using forceps, then transferred to petri dishes filled with phosphate buffer saline solution until use within 1 h. Both dermatomed skin and HSE (2 cm × 2 cm) were maintained in static vertical glass diffusion cell (surface area: 0.78 cm2) kept at 37°C insuring a skin surface temperature of 32°C. The surface of the skin was opened to ambient air (24 ± 2°C under relative humidity of 30 ± 10%). The receptor compartment was filled with 3.5 ml of phosphate buffer saline solution continuously stirred with a magnetic stirrer. Therefore, dermatomed skin and HSE specimens were allowed to equilibrate for 2.5 h, then caffeine (Sigma, St Louis, MO, USA) aqueous solution (4 mg/ml) was deposited onto the surface (caffeine dose: 100 μg; caffeine dose per area unit: 128 μg/cm2). The receptor fluids were withdrawn from the receptor compartments at regular intervals of time (0–21 h). The volume of fluid withdrawn was replaced by 800 μl of fresh receptor fluid. The top of the donor compartment was uncapped in order to avoid occlusion. Concomitantly, TEWL was measured using an unventilated evaporimeter (Vapometer Delfin SWL3 type, Delfin Technologies Ltd., Finland) for 21 h [11, 14]. Each permeation sample, filtered through 0.45 μm filter was analysed by high performance liquid chromatography (HPLC, Hewlett Packard serie 1100) system. Detection was performed at λ 272 nm. Caffeine assays were performed by using Eclipse® XDB C18 5 μm (4.6 × 150 mm) column (Agilent, Santa Clara, CA, USA). The mobile phase was a mixture (30/70, v/v) of methanol and 25 mM phosphate buffer solution (pH 5.2). Sample volume injected was 10 μl. The retention time of caffeine was 3.86 ± 0.03 min. For concentrations between 0.5 and 500 μg/ml, the chromatograms were linear with a factor correlation of 0.999 (P < 0.001).

Taking into account the maximum of caffeine solubility in water (i.e., ∼22 mg/ml), the ex vivo study was performed under sink conditions in the receptor fluid reservoir (i.e., maximal theoretical concentration of caffeine in receptor solution, defined as the initial dose per receptor volume, ∼30 μg/ml). Therefore, rates of percutaneous absorption were calculated as followed [1]:

$$ J_{t} \, = \,\frac{1} {A} \cdot \frac{{{\text{d}}Q}} {{{\text{d}}t}}\, = \,\frac{1} {A} \cdot \frac{{[Q_{t} {\text{ }} - {\text{ }}Q_{0} ]}} {{(t{\text{ }} - {\text{ }}0)}} $$
(1)

where Q t and Q 0 (Q 0 = 0) were respectively the cumulative amount of caffeine assayed in receptor fluid at time t and t = 0, A was the skin surface area, and J t was the apparent flux of caffeine through skin specimens at time t. In order to compare the extent of permeation through different skin specimens (i.e., dermatomed skin versus HSE), the rates of percutaneous absorption were linearised as followed:

$$ {\text{Log }}{\left( {J_{t}} - J_{21\,{\rm h}} \right)}\, = \,{\text{Log }}{\left( {\Updelta J} \right)}\, = \, - k \cdot t $$
(2)

where J 21 h was the flux calculated at 21 h and t < 21 h.

Similarly, the variations of TEWL through skin specimens as a function of time were plotted as followed:

$$ {\text{Log }}{\left( {{\text{TEWL}}_{t} - {\text{TEWL}}_{{21{\text{ h}}}} } \right)}\, = \,{\text{Log }}{\left( {\Updelta {\text{TEWL}}} \right)}\, = \, - k' \cdot t $$
(3)

where TEWL21 h was TEWL measured at 21 h t < 21 h.

The relationships between the logarithm of caffeine and TEWL flux variations versus time exposure were tested using regression analysis. The chosen level of significance was P < 0.05.

The profiles of percutaneous absorption of caffeine through dermatomed skin and HSE are in Fig. 1. The convex profiles of the caffeine permeation, exhibiting a linear increase then a plateau phase, suggested finite dose (10 and ∼23% of the initial dose absorbed through dermatomed skin and HSE, respectively up to 21 h) and receptor sink conditions [10]. However, taking into account that caffeine exposure to the skin was experienced with a small donor volume (25 μl), the finite dose conditions should be re-considered as finite vehicle undergoing a dramatic decrease of caffeine fluxes whatever the nature of skin specimens. This interpretation was reinforced by the visual inspection of skin surface after 21 h showing detectable caffeine residues. Although no statistical differences were shown, the extent of caffeine permeation through skin specimens was ranged according to the thickness of the membrane (dermatomed skin: 11.5 ± 11.5 μg/cm2; HSE: 29.4 ± 36.2 μg/cm2 after 21-h skin exposure) confirming the relationship between the pathlength of diffusion and the total amount permeated. In consequence, the use of HSE instead of dermatomed skin in percutaneous absorption studies enhanced quantitatively the extent of permeation data in relation to the tissue thickness.

Fig. 1
figure 1

Cumulative amounts of caffeine absorbed (Q) through dermatomed skin (filled circle) and heat separated epidermis (HSE) specimens (open circle) as a function of time. Each data is the mean ± standard error of the mean of 9 experimental determinations. The initial dose applied was 128 μg cm−2. In respective to membrane thickness, higher permeation of caffeine through HSE was shown as compared to dermatomed skin

Full size image

Thus, the question arises whether the qualitative characteristic of permeation differed between the dermatomed skin and HSE. Figures 2 and 3 reported the logarithmic variation of caffeine and water fluxes through dermatomed skin and HSE as a function of time, respectively. Clearly, the normalized flux-time courses of exogenous (caffeine) and endogenous (water) compounds were comparable through dermatomed skin and HSE suggesting similar permeability properties in both skin models.

Fig. 2
figure 2

Logarithmic variation of caffeine flux (Log (ΔJ)) through dermatomed skin (filled circle) and HSE specimens (open circle) as a function of time (Eq. 2, see text for details). Caffeine fluxes at 21 h (J 21 h) through dermatomed skin and HSE specimens were respectively 0.58 ± 0.57 μg cm−2 h−1 and 1.42 ± 1.73 μg cm−2 h−1. Each data is the mean ± standard deviation of nine experimental determinations. Logarithmic variations of caffeine fluxes were comparable in both HSE and dermatomed skin suggesting a similar behaviour of both skin models in finite donor volume conditions

Full size image
Fig. 3
figure 3

Logarithmic variation of transepidermal water loss (Log (ΔTEWL)) through dermatomed skin (filled circle) and HSE specimens (open circle) as a function of time (Eq. 3, see text for details). TEWL at 21 h (TEWL21 h) through dermatomed skin and HSE specimens were respectively 5.5 ± 6.7 g m−2 h−1 and 4.0 ± 3.0 g m−2 h−1. Each data is the mean ± standard deviation of nine experimental determinations. Logarithmic variations of TEWL were comparable in both HSE and dermatomed skin suggesting a similar behaviour of both skin models barrier function in finite donor volume conditions

Full size image

From conventional understanding, the main barrier to water permeability is primarily located in the stratum corneum which should be thus identical in dermatomed skin and HSE. However, lower TEWL was found through dermatomed skin as compared to HSE. Such discrepancies might be explained by the water-binding proteins (e.g., collagen) and free amino-acids in the dermis which restricted subsequent water transfer to superficial epidermis. Therefore, higher water transport through HSE was assumed by enhanced availability of unbound water for epidermal matrix.

Interestingly, the water fluxes through both membranes presented a dramatic decrease during skin permeation studies suggesting that the skin barrier function of both skin models varied likewise over time. The reduction of water fluxes during the skin exposure might be explained by the progressive dryness of the skin in spite of receptor fluid beneath the innermost parts of the membranes. During HSE preparation, the extraction of free amino-acids and/or the rearrangement of stratum corneum lipids, leading to a decrease of water holding capacity, might explain the characteristic brittleness of HSE observed at the end of the exposure [2]. Therefore, the decrease of caffeine fluxes (Fig. 2) might be explained by the finite volume of vehicle used and the concomitant change of skin barrier properties.

The suitability of TEWL measurements was investigated in vitro as a barrier integrity test for intact and punctured HSE [6]. The relevance of TEWL was questioned since no difference was detected through intact and punctured HSE as revealed by using an inert membrane (i.e., Teflon®), and because the transdermal flux of flufenamic acid was clearly affected by the mechanical damage. However, TEWL was only measured immediately before skin permeation study. As a result, punctual TEWL measurement (i.e., before topical treatment) and course-time permeation data seemed difficult to compare. For these reasons, in our study, both TEWL and permeation data were compared parallely all over the experiment to detect potential skin properties modifications. As observed, TEWL values decreased dramatically along with the time suggesting that both water and model drug fluxes should be considered concomitantly for more appropriate data analysis in transdermal studies. Therefore, as depicted in Fig. 4, our study supported previous works emphasising the usefulness of TEWL to correlate and predict the percutaneous absorption of various drugs [4, 12].

Fig. 4
figure 4

Ex vivo relationship between TEWL and apparent caffeine flux through dermatomed skin (filled circle) and HSE specimens (open circle). Each data is the mean of seven experimental determinations (standard deviations are omitted for clarity reason). Caffeine permeation through HSE and dermatomed skin was strongly correlated to TEWL used as an endogeneous standard for the assessment of skin barrier properties

Full size image

In conclusion, the present study emphasized the feasibility of HSE use in percutaneous absorption studies. Permeation data of caffeine ponderated by TEWL measurements used as endogenous standard of permeability, showed no significant difference between dermatomed skin and HSE. Consequently, the data analysis in percutaneous absorption studies must considerer the apparent flux of exogenous compound comparatively to water flux measured as endogenous standard of skin integrity.