Prediction of Percutaneous Absorption in Human Using Three ...

362 Vol. 35, No. 3

© 2012 The Pharmaceutical Society of Japan

Biol. Pharm. Bull. 35(3) 362—368 (2012)

Prediction of Percutaneous Absorption in Human Using Three-Dimensional Human Cultured Epidermis LabCyte EPI-MODELTomohiro Hikima,* Noriaki Kaneda, Kyouhei Matsuo, and Kakuji TojoDepartment of Bioscience and Bioinformatics, Kyushu Institute of Technology; Iizuka, Fukuoka 820–8502, Japan.Received September 14, 2011; accepted December 3, 2011; published online December 16, 2011

The objective of this study is to establish a relationship of the skin penetration parameters between the three-dimensional cultured human epidermis LabCyte EPI-MODEL (LabCyte) and hairless mouse (HLM) skin penetration in vitro and to predict the skin penetration and plasma concentration profile in human. The skin penetration experiments through LabCyte and HLM skin were investigated using 19 drugs that have a different molecular weight and lipophilicity. The penetration flux for LabCyte reached 30 times larger at maximum than that for HLM skin. The human data can be estimated from the in silico approach with the diffusion coefficient (D), the partition coefficient (K) and the skin surface concentration (C) of drugs by as-suming the bi-layer skin model for both LabCyte and HLM skin. The human skin penetration of β-estradiol, prednisolone, testosterone and ethynylestradiol was well agreed between the simulated profiles and in vitro experimental data. Plasma concentration profiles of β-estradiol in human were also simulated and well agreed with the clinical data. The present alternative method may decrease human or animal skin experi-ment for in vitro skin penetration.

Key words in silico correlation; reconstructed epidermis; LabCyte EPI-MODEL; hairless mouse

Human skin permeability of drugs in vivo can be estimated from in vitro experiment using human skin. However, some researchers have ethical and cultural difficulty to get human skins. Thus, animal skins have been widely used for pharma-ceutical, toxicological and cosmetic R&D although the skins have a species difference between human. Recently, the 7th Amendment of the EC Cosmetic Directive prohibits the sale of cosmetics and ingredients validated their safety by animal testing with an alternative method.1) Thus, researchers utilize reconstructed human skin instead of animal skin. The ex-periments used reconstructed epidermis is approved by the European Centre for the Validation of Alternative Methods (ECVAM) as the scientifically validated methods for skin cor-rosion and irritation.2,3) Permeability studies with Episkin®, EpiDermTM and SkinEthic® comparison to human and animal skin were investigated and the authors reported that the pen-etration flux through the reconstructed skin was higher than that of skin.4—6) The OECD regulations accepted the results from the reconstructed skin7) but don’t approve it as an alter-native method for in vitro human skin penetration experiment yet.

Skin penetration is widely expressed by the mathematical models on the basis of the process of diffusion and dissolu-tion. Penetration parameters, diffusion coefficient D [cm2/s] and partition coefficient K [—], through stratum corneum (SC) and viable skin (VS) of hairless mouse skin are found to well agree with that of human skin.8) Not only the skin penetra-tion profile but the blood concentration profile in human is well simulated using the penetration parameters determined from hairless mouse skin, the thickness of human SC and VS, the skin enzymatic parameters (first-order kinetic constant k [1/s], Michaelis–Menten constant Km [μg/mL] and maximum rate Vmax [μg/mL/s]) and the pharmacokinetic parameters (elimination rate constant from plasma Ke [1/s] and volume of distribution Vd [mL]) in human. Thus, in vitro/in vivo/in silico correlation may become an alternative method for in vitro penetration experiment of human and animal skin.

This study aims for the establishment of an alternative

method with the three-dimensional cultured human epidermis (CHE), LabCyte EPI-MODEL (Japan Tissue Engineering Co., Ltd.), for human skin penetration study in vitro. First of all, the skin penetration experiments across LabCyte and hairless mouse (HLM) skin with several model drugs were carried out. We then calculated the penetration parameters in SC and VS based on the bi-layer skin model and simulated the drug blood concentration in human using the commercially available PC software SKIN-CAD®.

MATERIALS AND METHODS

Materials β-Estradiol (EST), 17α-ethynylestradiol (ETE), lomefloxacin hydrochloride (LFX), progesterone (PS), timo-lol maleate salt (TM) and verapamil hydrochloride (VRP) were purchased from Sigma Chemical Co. (St. Louis, MO, U.S.A.). Menthyl salicylate (MS) and prednisolone (PN) were purchased from Fluka Chemie GmbH (Buche, Switzerland) and Nacalai Tesque Inc. (Kyoto, Japan), respectively. Caffeine (CF), hydrocortisone (HC), indomethacin (IM), ketoprofen (KP), ketotifen fumarate (KF), 2-naphtol (NP), nicotine (NC), norethisterone (NET), propiverine (PP), propiverine hydro-chloride (PPH), tulobuterol (TB), testosterone (TES), vitamin B12 (VB) and other reagents were purchased from Wako Pure Chemical Industries Ltd. (Osaka, Japan). Molecular weight and Log Ko/w of model drugs were summarized in Table 1.

Animal Skin Preparation Hairless mice (HLM, Kud:Hr− strain, female, 7 weeks old) were purchased from Kyudo Co., Ltd. (Saga, Japan). The relevance of HLM skin as an experi-mental model of transdermal absorption through human skin was discussed.9) Intact skin (IS) and stripped skin (viable skin, VS), SC of which removed by tape-stripping, were used for the skin penetration experiments. VS histologically consists of two layers, viable epidermis and dermis, and is considered as a hydrophilic monolayer. All animal studies conformed to the “Principles of Laboratory Animal Care,” NIH publication #85-23, revised 1996. The mice were sacrificed by an experienced individual.

Regular Article

* To whom correspondence should be addressed. e-mail: [email protected]

March 2012 363

Three-Dimensional Cultured Human Epidermis (CHE) Preparation LabCyte used as CHE in this study was sup-plied and was commercially available from Japan Tissue Engineering Co. (J-TEC, Aichi, Japan). LabCyte was sent from J-TEC under cooling and used within two days from shipment. LabCyte was cultured from human epidermis cell and stratified on polyethylene terephthalate (PET) membrane (thickness: 13.6±0.9 μm, pore size: 0.4 μm) as a supporting layer. LabCyte morphologically resembles human skin10) and consists of SC and viable epidermis (Epi) which is made up from granulocyte, stratum spinosum, and basal layer. The SC was differentiated from the viable epidermis (Epi model) and formed after two weeks cultivation; it was named SCEpi model. The Epi and SCEpi model of LabCyte corresponded to the VS and IS of HLM, respectively. The thickness of Epi (60.6±15.1 μm) was measured from the photo images of LabCyte slices dyed by H&E stain and, on the other hand, the SC thickness assumed 10 μm. Katoh et al. reported that LabCyte can be applied to in vitro skin irritation testing ac-cording to ECVAM validated protocol.10)

In Vitro Skin Penetration Experiments To fit LabCyte (10 mm in diameter) between side-by-side in vitro diffusion cells (KH cell, volume; 5 mL, Vidrex Ltd., Fukuoka, Japan) of 9 mm in diameter, a silicone elastomer adaptor (external diameter; 22 mm, internal diameter; 8 mm, and thickness; 1 mm) was used. LabCyte sandwiched between the adapters was mounted between the KH cells and then the effective diffusion area for LabCyte was 0.50 cm2. Because LabCyte was a fragile membrane against the pressure, we used it for the penetration experiments with PET supporting membrane. On the other hand, the IS and VS of HLM were mounted without adaptor and the diffusion area for HLM was 0.64 cm2. Donor and receptor compartments were filled with solution of a known concentration of model drugs (Table 1) and without

drugs, respectively. Solvent (40% PEG400 solution and phos-phate buffer (PB), pH 7.4) was changed according to Ko/w of model drug and the donor concentration was selected to keep a sink condition of receptor solution (Table 1). Ionic strength of PB was prepared at 0.32 (120 mM) to keep the pH value of solution in all experimental condition. The temperature in the KH cell was maintained at 37°C. Two hundred microliters sample was withdrawn from the receptor and the drug con-centration was assayed by HPLC. Hydrodynamic and thermal characteristics of in vitro skin permeation system was previ-ously defined and calibrated11) and we obtained the intrinsic rate of skin penetration from this cell system as well as Franz diffusion cell.

Analytical Methods All model drugs were assayed by HPLC assembled LC-10Avp system (Shimadzu Co., Ltd., Kyoto, Japan) and the assay conditions were established ac-cording to refs. 12 and 13.

RESULTS AND DISCUSSION

Effect of Cultured Period on the Penetration across LabCyte Three and four weeks cultured CHE, custom prod-ucts offered from J-TEC for this study, were investigated for in vitro penetration experiments of EST, TM and HC as model drugs having a different lipophilicity. The steady-state pene-tration flux, linear portion, and lag-time, x-axis of flux, on the cumulative amount penetrated versus time plot was measured (Fig. 1). Both of flux and lag-time became a plateau after two weeks cultivation period. LabCyte is made up of three layers, SC, Epi (VS) and PET membrane. The SC was formed after two weeks and thickens with cultivation time. Thus, we con-cluded that the SC as a penetration barrier reaches full growth after two weeks. We also carried out the penetration experi-ments through PET membrane and revealed no penetration

Table 1. Physicochemical Properties and Donor Compositions of Model Drugs Used in This Study

Drugs M.W. Log Ko/w Solvent Donor concentration (μg/mL)

NP 144.2 2.7a) 40% PEG400 20000NC 162.2 1.2a) 120 mM PB 10000CF 194.2 −0.07a) 120 mM PB 20000KP 254 0.14b) 40% PEG400 1342±43

KP (PB) 254 0.14b) 120 mM PB 1000EST 272 3.20±0.09 40% PEG400 247±22MS 276 5.87±0.36 40% PEG400 211±23TES 288 3.31a) 40% PEG400 592±49ETE 296 4.06±0.10 40% PEG400 2941±60NET 298 2.96±0.09 40% PEG400 156±4PS 314 3.84±0.05 40% PEG400 159±14IM 358 1.32c) 40% PEG400 424±9

IM (PB) 358 1.32c) 120 mM PB 1700PN 360 1.49±0.04 40% PEG400 2130±62HC 362 1.53a) 120 mM PB 379±2PP 363 2.94d) 40% PEG400 238±17

LFX 388 −0.3e) 120 mM PB 1303±71PPH 406 0.3 d) 40% PEG400 20000KF 426 1.97±0.02 120 mM PB 500TM 433 −1.00c) 120 mM PB 10000VB 1356 −4.33c) 120 mM PB 1000

Values are obtained from refs. 27,a) 28,b) 29,c) 30,d) and 31.e) NP; 2-naphthol, NC; nicotine, CF; caffeine, KP; ketoprofen, EST; β-estradiol, MS; menthyl salicylate, TES; testosterone, ETE; 17α-ethynylestradiol, NET; norethisterone, PS; progesterone, IM; indomethacin, PN; prednisolone, HC; hydrocortisone, PP; propiverine, LFX; lomefloxacin, PPH; propiverine hydrochloride, KF; ketotifen fumarate, TM; timolol maleate salt, VB; vitamin V12.

364 Vol. 35, No. 3

barrier (data not shown).The CHE for research on the market has no appendage, es-

pecially hair follicles. This is the reason that the development of hair follicles has technical difficulties because the reconsti-tution of hair follicle involves several factors14) and has no ne-cessity for the test for skin corrosion and irritation. However, researchers reported that the skin penetration was influenced by the existence of hair follicles.15—17) The steady-state pen-etration rate of model drugs through appendage-free skin of hairless rat was two to four times larger than that of normal skin.15) Aim of our research isn’t to indicate the importance of appendage but is to show the correlation between the penetra-tion parameters through LabCyte and HLM skin. Therefore, we disregard the lack of appendage on LabCyte in this study.

Comparison with the Penetration Flux and Lag-Time across LabCyte and HLM Skin The penetration flux and lag-time for LabCyte and HLM skin were summarized in Tables 2 and 3 and also plotted in Figs. 2 and 3, respectively. The lag-time of MS (molecular weight; 276 and Log Ko/w; 5.87) was obviously one order of magnitude larger than that of other chemicals. MS was reserved in skin for a long time and the MS concentration in SC was 760 times higher than that in donor solution.18) This indicates that there is the interaction (binding) between MS and skin tissue. Thus, we excluded MS results from the following discussion. Solvent influences the penetration flux and lag-time. We investigated the comparison between the permeability for LabCyte and HLM skin under the same in vitro experimental condition. Moreover, we con-firmed that there is no significant difference between the pa-rameters from 40% PEG400 and from PB (p<0.01, two-sided and independent/paired sample t-test). We considered the pen-etration parameters irrespective of solvent in this study. The relationships between LabCyte and HLM parameters were expressed by each equation;

Fig. 1. Variation of Steady-State Penetration Flux (a) and Lag-Time (b) by Incubation Period

Symbol represents β-estradiol (EST, ○), hydrocortisone (HC, □) and timolol (TM, △). Each data point is mean±S.D. of three experiments.

Table 2. Steady-State Penetration Flux and Lag-Time across LabCyte

DrugsFlux (μg/cm2/h) Lag-time (h)

SCEpi Epi SCEpi Epi

NP 62.5±6.55 83.5±14.1 0.45±0.03 0.20±0.04NC 351±33.7 564±42.5 0.24±0.01 0.06±0.01CF 193±24.5 364±33.0 0.28±0.16 0.08±0.02KP 3.74±1.56 8.33±1.56 1.29±0.66 0.53±0.32

KP (PB) 7.72±1.72 24.9±2.51 3.01±1.57 0.95±0.18EST 0.61±0.08 1.67±0.27 1.63±0.29 0.53±0.06MS 0.03±0.004 0.08±0.01 21.4±0.21 8.95±0.25TES 1.88±0.24 6.09±0.56 1.49±0.77 0.37±0.09ETE 4.78±0.40 16.6±0.78 2.77±0.42 0.53±0.10NET 0.49±0.07 1.19±0.10 2.07±0.18 0.74±0.04PS 0.96±0.15 3.05±0.43 2.33±0.81 0.85±0.16IM 0.74±0.02 1.84±0.18 1.63±0.44 0.72±0.12

IM (PB) 22.5±11.6 37.8±7.28 1.80±0.34 0.61±0.01PN 0.36±0.01 2.05±0.17 3.37±1.53 1.12±0.14HC 1.22±0.09 3.56±0.25 2.77±0.24 0.89±0.07PP 8.22±1.52 9.60±0.51 2.19±0.01 1.09±0.34

LFX 1.07±0.08 4.66±0.13 2.45±0.86 0.85±0.08PPH 6.36±0.43 41.7±2.24 5.37±0.81 1.89±0.04KF 14.9±1.00 43.9±2.99 1.82±0.73 0.50±0.12TM 45.3±4.63 127±9.97 2.10±0.15 0.35±0.09VB 0.41±0.12 0.70±0.29 6.20±0.36 1.21±0.13

Abbreviations are indicated in a footnote of Table 1. Data represent mean±S.D. of three experiments.

March 2012 365

Fig. 2. Correlation between Penetration Flux across HLM Skin and LabCyte; the Steady-State Flux across Intact Skin/SCEpi Model (a) and Stripped Skin/Epi Model (b)

Open and closed symbols represent 40% PEG400 solution and PB, respectively. Each data point is mean±S.D. of three experiments.

Fig. 3. Correlation between Lag-Time across HLM Skin and LabCyte; the Lag-Time across Intact Skin/SCEpi Model (a) and Stripped Skin/Epi Model (b)

Open and closed symbols represent 40% PEG400 solution and PB, respectively. Each data point is mean±S.D. of three experiments.

Table 3. Steady-State Penetration Flux and Lag-Time across HLM Skin

DrugsFlux (μg/cm2/h) Lag-time (h)

Intact skin Stripped skin Intact skin Stripped skin

NP 23.8±3.08 160±29.1 0.76±0.15 0.17±0.06NC 111±5.99 3090±348 0.72±0.15 0.02±0.01CF 6.01±1.26 3670±369 0.86±0.23 0.05±0.03KP 1.23±0.30 12.0±2.00 3.23±0.43 0.90±0.18

KP (PB) 0.46±0.10 100±12.9 6.11±1.61 0.46±0.27EST 0.21±0.02 1.38±0.31 4.29±0.19 1.17±0.54MS 0.03±0.004 0.03±0.01 25.3±4.01 15.7±6.80TES 1.51±0.36 3.95±0.97 5.52±0.49 1.49±0.31ETE 1.21±0.12 10.5±7.63 10.5±0.17 0.74±0.26NET 0.15±0.01 0.34±0.08 4.09±1.49 1.31±0.35PS 1.25±0.10 3.35±0.88 5.53±1.54 2.34±0.58IM 0.56±0.10 1.55±0.23 4.83±0.52 2.03±0.17

IM (PB) 1.19±0.35 260±22.0 4.43±0.68 0.25±0.03PN 0.29±0.04 5.67±1.39 6.92±0.73 0.85±0.07HC 0.07±0.02 18.3±5.18 5.25±0.74 0.40±0.05PP 4.50±1.22 7.16±3.88 6.87±0.94 2.73±0.06

LFX 0.21±0.09 72.0±25.6 4.96±0.95 0.28±0.10PPH 0.38±0.13 94.9±40.7 8.55±0.69 2.62±0.75KF 13.1±3.21 182±57.0 2.22±0.82 0.15±0.13TM 1.56±1.36 468±63.7 9.30±1.06 0.42±0.22VB 0.19±0.06 19.0±3.70 19.6±0.37 1.27±0.63

Abbreviations are indicated in a footnote of Table 1. Data represent mean±S.D. of three experiments.

366 Vol. 35, No. 3

1 0.75i SCEpiFlux (3.52 10 ) Flux 0.84r= × × = (1)

1.17v EpiFlux 1.58 Flux 0.93r= × = (2)

0.95i SCEpiLag-time 2.53 Lag-time 0.86r= × = (3)

1.24v EpiLag-time 1.18 Lag-time 0.62r= × = (4)

where subscript i and v represented IS and VS, respectively. The closed plots in Figs. 2 and 3 showed the results on PB as solvent. Correlation between the flux of LabCyte and HLM skin was represented by an expression equation. The flux for SCEpi model was larger than that for intact skin (Eq. 1, Fig. 1a) and, on the other hand, that for Epi model was almost same value with that for stripped skin (Eq. 2, Fig. 1b). Permeability coefficient of drugs for Episkin® was also 10 times larger than that for human epidermis.6) The permeation of hydrophobic compounds through the human reconstructed epidermis model was very higher than through split-thickness human skin.19) The rank order of chemicals penetration through Episkin®, EpiDermTM and SkinEthic® reflected the penetration through human epidermis.20) These results show that the barrier func-tion of SC in CHE doesn’t mature in comparison with hu-man skin. The penetration flux is influenced by the thickness and integrity of SC. Ultra-structure and lipid composition of Episkin®, EpiDermTM and SkinEthic® was investigated.21) Skin

structure of them was similar to human skin and, however, lipid composition was slightly different with human skin. The growth conditions of CHE influenced the SC barrier func-tions.22) Moreover, skin has a species difference and the skin penetration flux also differs among species.23) We have to im-prove the SC barrier function of CHE if we will use the CHE for the skin permeability screening of drugs.

Penetration Parameters We compared the penetration parameters, diffusion coefficient D and skin surface concen-tration C of penetrant, in SC and Epi (VS) calculated using the bi-layer skin model.8) Because the chemical potential in the donor concentration Cd was varied on each drug in this study (Table 1), we used C multiplied partition coefficient K by Cd as the penetration parameter. The relationships between LabCyte and HLM were shown in Figs. 4 and 5, respectively. The D and C in SC and Epi defined by the following correla-tion equation;

1.10SC-HLM SC-LabCyte1.84 0.75D D r= × = (5)

4 1.41VS-HLM Epi-LabCyte(6.54 10 ) 0.96D D r= × × = (6)

1 0.74SC-HLM SC-LabCyte(1.07 10 ) 0.68C C r= × × = (7)

1 0.98VS-HLM Epi-LabCyte(5.89 10 ) 0.93C C r−= × × = (8)

Fig. 4. Correlation between Penetration Parameters, Diffusion Co-efficient DSC (a) and SC Surface Concentration CSC (b), in Stratum Corneum (SC) of HLM Skin and LabCyte

Fig. 5. Correlation between Penetration Parameters, Diffusion Co-efficient DVS (a) and VS Surface Concentration CVS (b), in Viable Skin (VS) of HLM Skin and LabCyte

March 2012 367

Simulation of Skin Penetration and Plasma Concen-tration Profile The skin penetration (Fig. 6) and plasma concentration (Fig. 7) profiles in human were simulated using the commercially available PC software SKIN-CAD®. The experimental data for human skin penetration in vitro were previously reported.12) The skin penetration parameters were

calculated using the preceding correlation equations Eqs. 5 to 8. The plasma concentration was calculated based on the drug dissolved matrix/skin bi-layer/one compartment model. The matrix parameters, thickness, diffusion coefficient, partition coefficient and initial concentration, were referred from Rohr and Seager-Lorenz24) and the pharmacokinetic parameters in human were referred from Baifour and Heel25) and Gilman et al.26) The parameters used for the simulation are given in each figure caption. The simulated profile well agreed with the experimental and clinical data. More studies about other drugs may be required.

CONCLUSION

The penetration parameters, diffusion coefficient and parti-tion coefficient, can be determined from in vitro experiments of LabCyte as well as HLM skin. By using the SKIN-CAD® approach, the body pharmacokinetics for human following transdermal delivery can be well evaluated under clinical con-ditions. We found the usefulness of LabCyte for percutaneous absorption study using the in vitro/in vivo/in silico correlation. This alternative method may replace the in vitro penetration study partly with human or animal skin. Further studies are required to the validity of the prediction approach for the drug penetration through the human skin.

Fig. 6. Prediction of Skin Penetration Profiles of β-Estradiol (EST, a), Prednisolone (PN, b), Testosterone (TES, c) and Ethinylestradiol (ETE, d) across Human Skin in Vitro

Experimental data (plot) referred from ref. 12 were compared with simulated profile (solid line). The penetration parameters in simulation were calculated using Eqs. 5 to 8. DSC; 3.49×10−11 cm2/s for EST, 2.83×10−11 cm2/s for PN, 1.14×10−10 cm2/s for TES and 2.39×10−11 cm2/s for ETE, DVS; 1.22×10−7 cm2/s for EST, 4.23×10−8 cm2/s for PN, 4.52×10−5 cm2/s for TES and 1.43×10−5 cm2/s for ETE, KSC/VS; 5.97 for EST, 1.52 for PN, 6.25 for TES and 10.53 for ETE, Cs; 2.22×103 μg/mL for EST, 1.44×103 μg/mL for PN, 8.82×102 μg/mL for TES and 1.57×103 μg/mL for ETE, L; 1.11×10−1 cm, and h; 2.00×10−3 cm. Each data point is mean±S.D. of three experiments.

Fig. 7. Comparison between Experimental (Plot) and Simulated Profile (Solid Line) of Plasma Concentration of β-Estradiol (EST) in Human

The experimental data and pharmacokinetic parameters in human were referred from refs. 24—26. The penetration parameters of simulated profile were calculated using Eqs. 5 to 8. DSC; 3.49×10−11 cm2/s, DVS; 1.22×10−7 cm2/s, KSC/VS; 5.97, Cs; 2.22×103 μg/mL, L; 2.00×10−2 cm, h; 2.00×10−3 cm, volume of distribution V1; 1.82×105 mL, and elimination rate constant k10; 6.73×10−1 h−1.

368 Vol. 35, No. 3

REFERENCES

1) EC. 7th Amendment to the EC cosmetic Directive (L66/26—35). 27th February 2003, Brussels, Belgium (2003).

2) Zuang V, Balls M, Botham PA, Coquette A, Corsini E, Curren RD, Elliott GR, Fentem JH, Heylings JR, Liebsch M, Medina J, Roguet R, van de Sandt JJ, Wiemann C, Worth AP. Follow-up to the ECVAM prevalidation study on in vitro tests for acute skin irritation. The European Centre for the Validation of Alternative Methods Skin Irritation Task Force Report 2. Altern. Lab. Anim., 30, 109—129 (2002).

3) Fentem JH. Validation of in vitro tests for skin corrosivity. ALTEX, 16, 150—153 (1999).

4) Netzlaff F, Kaca M, Bock U, Haltner-Ukomadu E, Meiers P, Lehr CM, Schaefer UF. Permeability of the reconstructed human epider-mis model Episkin® in comparison to various human skin prepara-tions. Eur. J. Pharm. Biopharm., 66, 127—134 (2007).

5) Asbill C, Kim N, El-Kattan A, Creek K, Wertz P, Michniak B. Evaluation of a human bio-engineered skin equivalent for drug per-meation studies. Pharm. Res., 17, 1092—1097 (2000).

6) Lotte C, Patouillet C, Zanini M, Messager A, Roguet R. Permeation and skin absorption: reproducibility of various industrial recon-structed human skin models. Skin Pharmacol. Appl. Skin Physiol., 15 (Suppl. 1), 18—30 (2002).

7) OECD. Skin absorption: in vitro method. Test Guideline 428 (2004). 8) Tojo K. Mathematical Models of Transdermal and Topical Drug

Delivery, 2nd ed., Biocom Systems Inc., Fukuoka (2005). 9) Simon GA, Maibach HI. Relevance of hairless mouse as an experi-

mental model of percutaneous penetration in man. Skin Pharmacol. Appl. Skin Physiol., 11, 80—86 (1998).

10) Katoh M, Hamajima F, Ogasawara T, Hata K. Assessment of hu-man epidermal model LabCyte EPI-MODEL for in vitro skin ir-ritation testing according to European Centre for the Validation of Alternative Methods (ECVAM)-validated protocol. J. Toxicol. Sci., 34, 327—334 (2009).

11) Tojo K. Design and calibration of in vitro permeation apparatus. Transdermal controlled systemic medications. (Chien YW ed.) Mercel Dekker, Inc., New York, pp. 127—158 (1987).

12) Hikima T, Maibach HI. Skin penetration flux and lag-time of ste-roids across hydrated and dehydrated human skin in vitro. Biol. Pharm. Bull., 29, 2270—2273 (2006).

13) Kaneda N. The use of three-dimentional human reconstructed epidermis for skin permeation study in vitro. M.S. Thesis, Kyushu Institute of Technology (2010).

14) Weinberg WC, Goodman LV, George C, Morgan DL, Ledbetter S, Yuspa SH, Lichti U. Reconstitution of hair follicle development in vivo: determination of follicle formation, hair growth, and hair qual-ity by dermal cells. J. Invest. Dermatol., 100, 229—236 (1993).

15) Illel B, Schaefer H, Wepierre J, Doucet O. Follicles play an impor-tant role in percutaneous absorption. J. Pharm. Sci., 80, 424—427 (1991).

16) Essa EA, Bonner MC, Barry BW. Human skin sandwich for as-sessing shunt route penetration during passive and iontophoretic

drug and liposome delivery. J. Pharm. Pharmacol., 54, 1481—1490 (2002).

17) Frum Y, Eccleston GM, Meidan VM. Factors influencing hydrocor-tisone permeation into human hair follicles: use of the skin sand-wich system. Int. J. Pharm., 358, 144—150 (2008).

18) Hikima T, Ashizuka Y, Natakeyama N, Tojo K. In vitro examina-tion of drug reservoir effect in epidermis. Yakuzaigaku, 60 (Suppl.), 259 (2005).

19) Schmook FP, Meingassner JG, Billich A. Comparison of human skin or epidermis models with human and animal skin in in-vitro percutaneous absorption. Int. J. Pharm., 215, 51—56 (2001).

20) Schäfer-Korting M, Bock U, Diembeck W, Düsing HJ, Gamer A, Haltner-Ukomadu E, Hoffmann C, Kaca M, Kamp H, Kersen S, Kietzmann M, Korting HC, Krächter HU, Lehr CM, Liebsch M, Mehling A, Müller-Goymann C, Netzlaff F, Niedorf F, Rübbelke MK, Schäfer U, Schmidt E, Schreiber S, Spielmann H, Vuia A, Weimer M. The use of reconstructed human epidermis for skin ab-sorption testing: Results of the validation study. Altern. Lab. Anim., 36, 161—187 (2008).

21) Ponec M, Boelsma E, Weerheim A, Mulder A, Bouwstra J, Mommaas M. Lipid and ultrastructural characterization of recon-structed skin models. Int. J. Pharm., 203, 211—225 (2000).

22) Batheja P, Song Y, Wertz P, Michniak-Kohn B. Effects of growth conditions on the barrier properties of a human skin equivalent. Pharm. Res., 26, 1689—1700 (2009).

23) Panchagnula R, Stemmer K, Ritschel WA. Animal models for trans-dermal drug delivery. Methods Find. Exp. Clin. Pharmacol., 19, 335—341 (1997).

24) Rohr UD, Saeger-Lorenz K. 17β-Estradiol matrixpatch removal and reapplication in postmenopausal women: theoretical predictions with an oscillating diffusion coefficient model. J. Pharm. Sci., 91, 822—844 (2002).

25) Balfour JA, Heel RC. Transdermal estradiol. A review of its phar-macodynamic and pharmacokinetic properties, and therapeutic efficacy in the treatment of menopausal complaints. Drugs, 40, 561—582 (1990).

26) Gilman AG, Goodman LS, Rall TW, Murad F. The Pharmacological Basis of Therapeutics. Micmillan, New York (1985).

27) Hansch C, Leo A. Substituent Constants for Correlation Analysis in Chemistry and Biology. John Wiley & Sons, New York (1979).

28) Tashiro Y, Shichibe S, Kato Y, Hayakawa E, Itoh K. Effect of lipo-philicity on in vivo iontophoretic delivery. I. NSAIDs. Biol. Pharm. Bull., 24, 278—283 (2001).

29) Hikima T, Ohsumi S, Shirouzu K, Tojo K. Mechanisms of synergis-tic skin penetration by sonophoresis and iontophoresis. Biol. Pharm. Bull., 32, 905—909 (2009).

30) Ogiso T, Iwaki M, Hirota T, Tanino T, Muraoka O. Comparison of the in vitro skin penetration of propiverine with that of terodiline. Biol. Pharm. Bull., 18, 968—975 (1995).

31) Rodgers T, Rowland M. Physiologically based pharmacokinetic modelling 2: predicting the tissue distribution of acids, very weak bases, neutrals and zwitterions. J. Pharm. Sci., 95, 1238—1257 (2006).