`
`Available online at www.sciencedirect.com
`
`.
`Sewis
`ScienceDirect
`
`Advanced Drug Delivery Reviews $9 (2007) 1152~1161
`
`
`
`Advanced
`DRUGDELIVERY
`Reviews
`
`———
`www.clsevier.com/locate/addr
`
`Transdermal skin delivery: Predictions for humans
`from in vivo, ex vivo and animal models *
`
`Biana Godin, Elka Touitou *
`
`Department ofPharmaceutics, School ofPharmacy, Faculty ofMedicine, The Hebrew University afJerusalem, Jerusalem 91120, Israel
`Received 10 May 2007; accepted 20 July 2007
`Available online 16 August 2007
`
`Abstract
`
`The assessment ofpercutaneous permeation of molecules is one of the main steps in the initial design and later in the evaluation of dermal or
`transdermal drug delivery systems. The literature reports numerous ex vivo, in vitro and in vivo models used to determine drug skin permeation
`profiles and kinetic parameters, some studies focusing on the correlation of the data obtained using these models with the dermal/transdermal
`absorption in humans. This paper reviews work from in vifro permeation studies to clinical performance, presenting various experimental models
`used in dermal/transdermal research, including the use of excised human or animal skin, cultured skin equivalents and animals, Studies focusing
`on transdermal absorption of a series of drug molecules and various delivery systems as well as mathematical models for skin absorption are
`reviewed.
`© 2007 Published by Elsevier B.V.
`
`Keywords; Transdermal absorption; Jn vitro-in vivo correlation; Animal skin; Studies in humans; Skin equivalents; Percutancous permeation
`
`Contents
`
`Introduction...2... 1152
`1.
`Issues related to in vitro and in vive skin permeation studies. 6...0 1153
`2,
`Skin structure: human vs. animal models...ce ee ee ae 1153
`3.
`Jn vitro permeation across human skin vs. animal models .. 0... oe ee es 1154
`4,
`The use of tissue culture-derived skin equivalents in transdermal research. ©. 1... ee ees 1155
`5.
`Jn vitro skin permeation studies focusing on delivery systems ..- 2 2...ee 1155
`6.
`7. Animal models for evaluation of skin absorption in humans: molecules... 1...ee 1156
`8. Animal models for evaluation of skin absorption in humans: delivery systems... . 2... 2-1-2 22-2 eee eee eee 1157
`9:
`(Mathematical models of akin absorption on. cae ee ee eee ee ee ewe eee eee 1158
`
`AD.SsSmee encase nog s @aleee EE ea. a en Boat a Cbce Go. SRO page a ge OR Gis MURS ep Bara let ate la scar etal S 1159
`
`Packricvgelaedsrr6 goa: dees gee e288 8 de ED Be ST cep Ee gE CaO oe
`Ee EE ploetghh od EQUE ES RoE eB, ATIER Ose goed Re 1160
`etererctt an ar are nie aaa ts eee: hw a Wha eee TNE se EEE Bice IE ae Ried ngrelal eend a eee 1160
`
`
`
`* This review is part of the Advanced Drug Delivery Reviews theme issue on
`“Erediction ofTherapeutic andDrugDelivery Outcomes UsingAnimalModels”.
`=
`of
`Faculty of Medicine, The Hebrew University of Jerusalem, POB 12065,
`Jomisalem 91120, Isracl. Tel.: +972 2 6758660;fax: +9722 6757611.
`E-mail address: \ouitou@jec.huji-ac.il (E. Touitou).
`0169-409X/S - see front matter © 2007 Published by Elsevier B.V.
`doi:10,1016.addr,2007,07,004
`
`1. Introduction
`
`The assessment ofpercutaneous absorption ofmolecules is a
`V®FY important step inthe evaluation ofany dermal ortransdermal
`drug delivery system. A key goal in the design and optimization of
`dermal or transdermal dosage forms lies in understanding the
`
`0001
`
`Noven Pharmaceuticals, Inc.
`EX2017
`Mylan Tech., Inc. v. Noven Pharma., Inc.
`IPR2018-00173
`
`

`

`5. Godin, E. Touitou / Advanced Drug Delivery Reviews 59 (2007) 1152-1161
`
`1153
`
`factors that determine a good in vive performance. Certainly, the
`most reliable skin absorption data are collected in human studies;
`however, such studies are generally not feasible during theinitial
`development of a novel pharmaceutical dosage form or consid-
`eration ofa new drug candidate. Thus, one ofthe main challenges
`ofbiopharmaceutical research is finding a correlation between ex
`vive, animal and human studies for prediction of percutaneous
`absorption in humans. It is practically impossible to assess the
`skin permeability of materials using in vivo experiments alone.
`Consequently, numerous ex vive and in vitro models are fre-
`quently employed to assess drug skin permeation profiles and
`kinetic parameters. Hence, a method that can consistently cor-
`relate ex vivo and in vivo data to shorten and economize the
`process ofdrug development and minimize the number ofhuman
`studies is critically needed.
`This article begins with a short overview of various aspects
`as well as pros and cons of in vitro and in vive animal models
`for skin permeation. Further, studies evaluating percutaneous
`absorption ofvarious drugs with or without permeation enhance-
`ment techniques are covered. And finally, the use of data from
`experiments in skin cultures and mathematical/pharmacokinetic
`models for predicting transdermal absorption are critically
`discussed.
`
`2. Issues related to in vitro and in vivo skin permeation
`studies
`
`Despite ethical concerns, the use ofanimals or isolated animal
`skin models to assess percutaneous absorption of molecules is
`frequently reported. These models, generally more available than
`human skin, are ofprime importance in basic research to improve
`our understanding of the processes, pathways and driving forces
`ofvarious agents across the skinbarrier. However, due tothe large
`number of animal species described in theliterature, it is quite
`difficult to compare thedata in the field ofdermal and transdermal
`drugdelivery. Variations inmethodology used with a specific skin
`model, such as type of diffusion cells, skin temperature, receiver
`media, application dose and diffusion area, can all significantly
`affect data [1]. Yet, it is important to emphasize that in vitro and
`animal models provide important tools for screening a series of
`drug formulations, evaluation of skin permeation enhancing
`properties and mechanism of action of the carrier systems and
`estimation ofrank ofskin transport for a series ofdrug molecules.
`
`3. Skin structure: human ys. animal models
`
`Skin is the largest body organ, weighing approximately 5 kg
`with a surface area of about two square meters in adult humans
`[2—4]. This multilayered organ has an essential function of
`protecting the body from the surrounding environment, thus
`being an efficient permeation obstacle for exogenous molecules.
`Thebarrier properties ofthe skin lie mainly within its uppermost
`strata, the stratum comeum (SC). This highly hydrophobic layer
`is composed ofdifferentiated non-nucleated cells, comeocytes,
`which are filled with keratins and embedded in thelipid domain.
`Sincetherate limiting step for skin absorption ofmostmolecules
`is considered to be this non-viable layer, percutaneous per-
`0002
`
`meation of molecules is believed to be governed by diffusion
`laws [2]. The extent of skin permeation of a compound may
`depend on the route of absorption. There are three pathways
`which can be involved in the transdermal permeation of che-
`micals: (1) through the intercellular lipid domains in SC; (2)
`throughthe skin appendages; and (3) through the keratin bundles
`in SC [2,5].
`Thelack ofcorrelation in transdermal permeation ofmolecules
`across species or from different application sites in the same
`animal model is due mainly to variations in skin (or SC) thickness,
`in the composition ofintercellular SC lipids and in the number of
`skin shafts. Netzlaffet al. [6] have shown that the amount offree
`fatty acids and triglycerides and the density of hair follicles are
`important factors causing differences between the skin barriers
`among species. As the majority ofmolecules applied onto the skin
`permeate along the SC lipid domain, the organization of these
`regions is very important for the barrier function of the skin.
`The SC lipid composition and organization differ from that of
`other biological membranes, with long chain ceramides, free fatty
`acids, cholesterol and cholesteryl esters being the main lipid
`classes [2—4,7,8].
`To evaluate transdermal absorption of a molecule, the most
`relevant membrane is human skin. Skin from various sources,
`including cosmetic surgery and amputations, has been used for
`the in vitro assessment of percutaneous penetration [9,10].
`However,its availability is limited and animal skin is therefore
`frequently used. A wide range of animal models has been sug-
`gested as a suitable replacement forhuman skin and has been used
`to evaluatepercutaneous permeation ofmolecules, These include
`primates, porcine, mouse, rat, guinea pig and snake models.
`Since the use of primates in research is highly restricted, the
`most relevant animal model for human skin is the pig. Porcine
`akin is readily obtainable from abattoirs and its histological and
`biochemical properties have been repeatedly shown to be
`similar to human skin [11—15], Porcine ear skin is particularly
`well-suited for permeation studies and gives comparable results
`to human skin. Studies examining thickness of various skin
`layers have shown that the SC thickness in pigs is 21-26 um
`[10,12] which is comparable to human skin [10,16]. The viable
`epidermis in porcine ear skin is 66—72 «wm thick [10,12], which
`is very similar to the human epidermal thickness of 70 um
`(shoulder) [17]. The follicular structure of pig skin also resem-
`
`Pv that of humans, with hairs and infundibula extending
`; cm* ofporcine ear skin as compared to 14—32 hairs (except
`eeply into the dermis. An average of 20 hairs are present per
`the forehead area) in humans [12]. Moreover, the vascular
`anatomy and collagen fiber arrangement in the dermis, as well
`as the contents of SC glycosphingolipids and ceramides are
`similar in man and in the domestic pig [18].
`Dueto its availability, skin of rodents (mice, rats and guinea
`pigs) is the most commonly used in in vitro and in vivo per-
`cutaneous permeation studies, The advantages of these animals
`are their small size, uncomplicated handling and relatively low
`cost. There ate a number ofhairless species (nude mice, hairless
`rats) in which the absence of hair coat mimics the human skin
`better than hairy skin [19], In these animals there is no need for
`hair removal (clipping or shaving) prior to the experiment, thus
`
`

`

`1154
`
`B, Godin, E. Touitou / Advanced Drug Delivery Reviews 59 (2007) 1152-1161
`
`avoiding the risk ofinjury to cutaneous tissue. Other models have
`a disadvantage of an extremely high density ofhair follicles and
`require hair removal, Since both issues may affect percutaneous
`absorption ofmolecules, hairyrodent skinis usually not used in in
`vitro permeation studies, although i vivo studies are still
`performed on these species. Among rodents, rat skin has more
`structural similarities to human tissue (Table 1).
`Except for rat skin, rodent skin generally shows higher
`permeation rates than human skin [20—21]. Regarding the rat
`skin, permeation kinetic parameters are frequently comparable
`with human skin,
`Snake skin was also proposed as a membrane in skin per-
`Meation experiments. Differential scanning calorimetry (DSC)
`thermograms and infra-red (IR) spectra showed that the SC of
`snake, porcine and human skins have some similarities in
`structure and components [22]. The distinguishing feature of the
`shed snake membraneis its lack offollicles.
`
`1200
`
`8888
`
`3
`
`
`
`Amountofheptanepermeated
`
`{nmol/cm’)
`
`0
`
`5
`
`10
`
`15
`
`20
`
`25
`
`Time (h)
`
`Fig, 1. Jn vitro permeation profiles of heptane across human (squares) and
`porcine (rhombs) skin (reproduced with permission from Ref. [23]).
`
`4. In vitro permeation across human skin vs. animal models
`
`Various studies have been carried out in an attempt to
`correlate in vitro permeation data in animal and human skin.
`Some of them are reviewed here. Most of reports substantiate
`the value of the pig as an animal model for man in skin
`permeation studies, Singh et al. [23] evaluated skin permeability
`coefficients (Kp) and SC reservoir of three hydrocarbons in
`porcine ear compared to human skin. They reported that pig
`skin was slightly more permeable to the substances with the
`ratios Kp porcine skin/Kp human skin of 1.71, 1.28 and 1.16 for
`heptane, hexadecane and xylene, respectively. The permeation
`profiles ofheptane across human and porcine skin are presented
`in Fig. 1. SC binding ofthe hydrocarbons to porcine and human
`skins was also comparable. The skin permeability (Kp) of
`nicorandil was investigated by Sato and co-authors [21] using
`excised skin samples from hairless mouse, hairless rat, guinea-
`pig, dog, pig, and human. Amongthe tested skins, the Kp values
`ofnicorandil in pigs and humans were in good agreement. The
`authors also found that comparable porcine and human skin
`permeation could be attributed to similar surface lipids, barrier
`thickness, and morphological aspects of the excised pig skin
`samples and human tissue. In another series ofexperiments, the
`in vitro permeability ofpig ear skin was compared with human
`(abdominal) skin and rat (dorsal) skin using both hydrophilic
`(water, mannitol, paraquat) and lipophilic (aldrin, carbaryl,
`fluazifop-butyl) penetrants [13]. Pig skin was found to have a
`closer permeability character than rat skin to human skin,
`particularly for lipophilic penetrants. The authors suggested that
`electrical conductivity measurements across pig skin mem-
`branes could be a valuable tool for evaluating the integrity of
`
`Table 1
`Thickness of skin strata in rat, mice and humans [10]
`
`Rat
`Mouse
`Human
`
`SC, om
`18
`9
`17
`
`Epidermis, xm
`32
`29
`AT
`
`Whole skin, mm
`2.09
`0.70
`2.97
`
`membranes. Sekkat et al. [24] reported that differentially tape-
`stripped, porcine skin could serve as an in vitre model for the
`evaluation of transdermal drug delivery to premature neonates.
`In this study the passive permeation of caffeine, phenobarbital,
`and lidocaine and the iontophoretic delivery of lidocaine across
`tape-stripped porcine skin barriers were tested. The barrier
`function of the tissue was monitored by measuring the trans-
`epidermal water loss (TEWL). For all tested drugs, the per-
`meation behavior correlated well with the skin barrier function
`[24]. The results were sustained by a study on diamorphine in
`vive absorption in premature neonates [25]. lontophoretic lido-
`caine. delivery was precisely controlled,
`independent of the
`barrier capability. Lin et al [22] compared in vitro penetration of
`theophylline, sodium diclofenac and benzoic acid through
`artificial cellulose membrane, animal skin (frog, snake with or
`without scales, nude mice, Sprague—Dawley rat and porcine)
`and human skin. The fastest permeation of substances was
`observed through cellulose membrane and frog skin and the
`slowest through human skin, with benzoic acid beingthe fastest
`penetrant
`through all skin types.
`In the case of sodium
`diclofenac the transdermal permeation flux in porcine SC was
`33 times higher than in intact skin, but in snake and human skin,
`the rate through SC was only 2.2 and 1.6 times higher than
`through intact ones.
`A focus of several reports was to compare transdermal
`permeation kinetics between rodent- and human skin. In a study
`by Royet al. [26] permeability coefficients ofmorphine, fentanyl,
`and sufentanil across full-thickness hairless mouse skin were in an
`order ofmagnitude higher than those found for human epidennis.
`There was no correlation between the enhancement in percuta-
`neous transport caused by SC removal in hairless mice and human
`epidermis. Another study examined permeation characteristics of
`human skin from various sites compared to animal skins, and
`found that shed snake and hairless rat skin showed similar
`permeability to human breast and thigh skin, while Wistarrat and
`nude mouse performed similarly to human cheek, neck, and
`inguinal skin [27]. Ravenzwaay et al [28] evaluated transport of
`compounds with various lipophilicities across rat and human
`skins in vitro and im vive in rats. In all cases the in vitro dermal
`penetration through rat skin was higher than in vive and rat skin
`was approximately 11-fold more permeable than human skin,
`These authors suggested the use of the following equation
`0003
`
`

`

`5. Godin, E. Touitou / Advanced Drug Delivery Reviews 59 (2007) 1152-116]
`
`1155
`
`(Eq. (1)) to estimate transdermal transport through human skin,
`based on the combined use of in vivo and in vitro data:
`
`% Percutaneous absorptionsnan
`= % Percutaneous absorption,; * (human/Jra)
`
`(1)
`
`where / is the percutaneous permeation flux.
`Ina separate study evaluating in vitro percutaneous absorption
`of four antihypertensive drugs in mice and human cadaver skin,
`Ghosh et al. reported that the permeation rate in mice skin was
`much higher than that inhuman skin [29]. Van de Sandt et al. [30]
`reported a multi-center skin permeation trial, comparing the in
`vitro absorption of benzoic acid, caffeine, and testosterone com-
`pounds through human skin (nine laboratories) and rat skin (one
`laboratory) in ten European laboratories. All laboratories ranked
`the absorption ofbenzoic acid through human skin as the highest
`of the three molecules (overall mean flux of 16.54+11.87 pg/
`cm*xh), whilethe absorption ofcaffeine and testosterone through
`human skin was comparable (2.24+1.43 and 1,63+1.94 ug/
`cm? xh, respectively). In this study, no differences were observed
`between the mean absorption through human skin and the one rat
`study for benzoic acid and testosterone, however for caffeine, the
`flux value and the total quantitypermeated across the rat skin were
`higher than the correspondent values in human skin.
`
`5. The use of tissue culture-derived skin equivalents in
`transdermal research
`
`A number oftissue culture derived skin equivalents such as
`living skin equivalent models (LSEs) and human reconstructed
`epidermis (HRE) have been used to measure percutaneous
`absorption. These models generally are comprised of human
`cells grown as tissue culture and matrix equivalents normally
`present in skin, and are utilized as alternatives to animal skins.
`LSEs resemble human skin, having a dermis, epidermis and
`partially-differentiated stratum corneum,but are deficient in skin
`appendages including pilosebaceous units, hair follicles and
`sweat glands [31]. These tissues provide much lower barrier
`properties than the whole skin due to their structure and lipid
`composition. For this reason, the kinetic parameters of skin
`permeation obtained when using LSEs usually highly overes-
`timate flux across human skin. For example, in a study by
`Schmooket al., the permeation characteristics ofhuman,porcine
`and rat skins with the Graftskin® LSE and the Skinethic ®
`HRE models were compared using four low molecular weight
`dermatological drugs with various hydrophilicities [32]. The
`permeation of more hydrophobic compounds (clotrimazole and
`terbinafine) through the skin equivalents resulted in an 800-900
`fold higher flux than through split-thickness human skin. On the
`other hand, transdermal flux of a less hydrophobic compound,
`salicylic acid, was in the same order of magnitude as fluxes
`obtained with human skin, In this study porcine skin performed
`as the most appropriate model for human skin and they
`concluded that reconstituted skin models are not suitable for in
`vitro penetration studies [32]. A similar conclusion was drawn
`from results ofanother study in which Royet al. [33] evaluated
`the in vitro permeabilities of alkyl p-aminobenzoates through
`0004
`
`LSE and human cadaver skin. In the case of cadaver skin, the
`permeability coefficient increased as the carbon chain length
`increased, However, this relationship was not observed in the
`permeability coefficients of these esters across LSE. Moreover,
`LSE showed very low resistance to flux compared to cadaver
`skin as the permeability coefficients ofthese esters through LSE
`were an order ofmagnitude higher than through cadaver skin.
`On the other hand, numerous reports support the use of skin
`equivalents for evaluation ofskin irritation [31,34]. Ina study by
`Monteiro-Riviere and colleagues [35], EpiDerm LSE © was
`found to be morphologically and biochemically comparable to
`nonnal human epidennis, providing a model in toxicological and
`skin metabolism studies. Ponec and Kempenaar [36] reportedthat
`architecture, homeostasis and lipid composition ofreconstructed
`human skin models (EpiDerm ®, SkinEthic ®, Episkin © and RE-
`DED ®) were comparable to native human tissue. Itis noteworthy
`that Colipa, the European Trade Assocation for cosmetic and
`toiletry industry, recommends the use of in vitro reconstructed
`skin equivalents as the preferred testing model for skin irritation
`studies [34]. However, the overall use of skin cultures is likely
`to be limited due to questionable performance as a barrier in
`skin permeation studies, as well as due to their cost and data
`reproducibility.
`
`6. In vitro skin permeation studies focusing on delivery
`systems
`
`Correlation of permeation between animal and human skin
`studies from drug delivery systems and pharmaceutical dosage
`forms has attracted significant attention from the pharmaceu-
`tical industry, academia, and regulatory sectors. Design and
`optimization ofcarriers for active agents is a time- and resource-
`consuming process that is an integral part ofthe development of
`any drug delivery system. Jn vitro tests reflecting bioavailability
`data are required to prove that a new delivery cartier is bio-
`equivalent with or superior to the standard. Mechanistic studies
`with sophisticated carriers are performed in animal and human
`skin to try to predict the future performance ofthe drug delivery
`systems in humans ftom in vitro data.
`Among the drug delivery systems tested were carriers based
`on chemical skin permeation enhancers, specially designed
`vesicles, physical and microinvasive techniques. Touitou et al.
`[37] tested transport oftetrahydrocannabinol from an enhancing
`carrier containing 10% w/w oleic acid/propylene glycol/
`polyethylene glycol 4000/ethanol mixture. In this study drug
`permeation across Sabra-strain rat skin was found to be about
`12.8-fold higher than across human skin. Differing lag times,
`11.5 vs 8.5 h for the rat and human skin, respectively, may point
`toward different diffusion pathways for this drug across the skin
`ofthese two species. Priborsky and Muhlbachova [38] assessed
`the effect of chemical permeation enhancers on the in-vitro
`transport across human skin as compared to animal models. Rat
`skin was ~3.3-4 times more permeable than human tissue.
`Using rat skin, the least potent enhancer was dimethylsulph-
`oxide and the maximum permeation enhancement was observed
`with sodium laurylsulphate, In contrastall the tested enhancers
`performed comparably to human skin. In this study, human and
`
`

`

`1156
`
`B. Godin, E. Touitou / Advanced Drug Delivery Reviews 59 (2007) 1152-1161
`
`Interestingly, the amountoftimolol transported during iontopho-
`guinea-pig skins were not significantly different in the per-
`resis (2 h) was significantly different among the various skin
`meation of N-methyl-2-pyrrolidone. In another study, transder-
`species, but the final quantity of timolol crossing the skin during
`mal delivery of 6-beta-naltrexol,
`the active metabolite of
`
`naltrexone, across human skin and guinea pig skin in vitro 24 h @4biontophoresis and 22 h post-iontophoretic passive
`and in hairless guinea pigs im vivo was assessed from a
`diffusion) was comparablein the different species. According to
`propylene glycol/ buffer mixture [39]. Jn vitro flux ofnaltrexone
`this data, iontophoresis may diminish interspecies variations in in
`was about 2.3 and 5.6 times higher than 6-beta-naltrexol across
`vitro skin permeation studies. Microinvasive techniques (micro-
`guinea pig and human skin, respectively, and 6-beta-naltrexol
`needles, RF skin ablation, etc.) represent another means of skin
`lag times were longer in both skin types (Fig. 2). Jn vivo studies
`permeation enhancement. Recently Wang et al.
`[42] imaged
`in guinea pigs showed that the steady-state plasma level of
`infusion of dye molecules, insulin, polymer microparticles, and
`cells into the skin by brightfield and fluorescence microscopy
`naltrexone was twofold greater than 6-beta-naltrexol, which
`correlated well with in vitro data in guinea pig skin. Rigg and
`followingthe insertion ofhollow glass microneedles into hairless
`Barry [40] investigated the skin permeability of two species of
`rat skin jm vive and human cadaver skin in vitro, Studying the
`snake (Elaphe obsoleta, Python molurus) compared to in vitro
`flow mechanism the authors reported that using both models,
`experimental results for human skin and for hairless mouse.
`partial retraction of the needle by withdrawing 100-300 ys or
`vibrating the microneedle array dramatically increased infusion
`The effect of typical enhancers on the permeabilities of the
`flow rate.
`membranes to a model penetrant 5-fluorouracil (5-FU) was
`evaluated. The studied enhancers were 3% Azone in Tween 20/
`saline, propylene glycol (PG), 2% Azone in PG, and 5% oleic
`acid in PG. The data from snake membranes showed minor
`effects of the enhancers, while for hairless mouse skin, the
`enhancer effects were significant. None ofthe membranes was a
`completely reliable model for human percutaneous absorption
`in assessing the effect of skin permeation enhancers, The
`authors concluded that human skin should be used in skin
`permeation studies and not hairless mouse or snake skin;
`otherwise, misleading results may be obtained.
`Kanikkannan and colleagues [41] evaluated the effect of
`species variation (rat, rabbit, mouse, guinea pig and human) on
`the transdermal iontophoretic permeation of timolol maleate.
`
`7. Animal models for evaluation of skin absorption in
`humans: molecules
`
`In studies conducted in the 1970s and 1980s, transdermal
`absorption of various radio-labeled molecules in human volun-
`teers and animals was assessed [43-45],In these studies, the same
`concentration ofsubstance (4 j1g/cm*) was applied on the forearm
`ofsubjects in an attempt to standardize the application conditions,
`and percutaneous absorption was quantified by following the
`excretion of the tracer for 5 days. Bartek et al. [45] undertook a
`comparative study of percutaneous absorption of haloprogin,
`acetylcystein, cortisone, caffeine and testosterone in vivo in
`various animal species (rats, rabbits, miniature swine) and
`humans. The highest extent of percutaneous absorption was
`observed with haloprogin, with complete absorption in rats and
`rabbits but not in humans andpigs,In rats and rabbitsthe absorbed
`fraction of applied dose followed the order; acetylcys-
`tein<cortisone<caffeine=testosterone<haloprogin. Jn vivo data
`from man and pigs indicated that the order of absorption was:
`acetylcystein <cortisone<haloprogin<testosterone<caffeine.
`The authors concluded that the transdermal absorption in rats and
`rabbits was not predictive for human data, while results obtained
`in porcine model and humans were comparable.
`Using the same technique, Wester and Maibach [46,47] com-
`pared the percutaneous absorption of various molecules between
`thesus monkey and humans. They found that the m vivo
`percutaneous absorption of hydrocortisone,
`testosterone and
`benzoic acid was similar forrhesus monkey and man, For example,
`when hydrocortisone, testosterone andbenzoic acid were applied at
`adose of4 ug/cm2,the absorbed dose was 2.9, 18.4 and 59.2% vs.
`1.9, 13.2 and 42.6% inmonkey vs. humans, respectively. Bronaugh
`and Maibach [48] measured the percutaneous absorption extent of
`five nitroaromatic compounds (p-nitroaniline, 4-amino-2-nitrophe-
`nol, 2,4-dinitrochlorobenzene, 2-nitro-p-phenylenediamine, nitro-
`benzene) in humans and monkeys using both in vitro and in vivo
`techniques. It was found that except for the highly volatile
`nitrobenzene, o0 significant differences were observed in the four
`groups of data. Andersen et ai, used the same methodology
`to evaluate percutaneous absorption of '*C ring-labelled
`0005
`
`0
`
`10
`
`430
`2
`Time (h)
`
`40
`
`~@#850
`
`
`
`0
`
`10
`
`30
`20
`Time (h)
`
`40
`
`50
`
`500
`
`A i
`
`5&
`
`a3
`
`400
`3 300
`E © 200
`100
`0
`
`o 2
`
`3s
`
`5
`
`Ba
`
`2g
`
`500
`40
`&
`E = 300
`3 2
`E £ 200
`2
`100
`0
`
`a E
`
`g
`
`Fig. 2. Cumulative amount of naltrexone (squares, n=7) and 6-beta-naltrexol
`(rhombs, n=8) permeated through the human skin (A) and guines pig skin (B)
`(reproduced with permission from Ref. [39)).
`
`

`

`B. Godin, E. Touitou / Advanced Drug Delivery Reviews 59 (2007) 1152-1161
`
`1157
`
`
`
`(ug/cm2)
`
`Cumulativepenetration
`
`
`
`2
`
`—?— Rat epidermis
`
`—— Human epidermis
`
`¥—
`
`Humanfull skin
`
`—*— Rat full skin
`
`—?*— Perfused pig ear
`
`© Rat full skin
`
`4 Fat epidermis
`
`© Human full skin
`
`4 Human epidermis
`Perfused pig ear
`
`* V
`
`2
`3
`
`x
`§
`Be
`
`hydrocortisone, testosterone andbenzoicacid i vivo in guinea pigs
`and compare the obtained results to previous human data [49]. The
`absorption of hydrocortisone and benzoic acid was similar to the
`published human absorption data, but testosterone was absorbed to
`a greater extent in guinea pigs than in man. Interestingly, in this
`work a thioglycollate based depilatory cream significantly
`increased the extent of transdermal permeation of testosterone.
`Although the above studies [43-49] used radiolabeled molecules
`(whose weakness is the accurate detection of the original
`compound), the clear advantage of these early works was their
`ability to compare skin absorption of a large series of molecules
`using the same experimental protocol.
`Later reports used more advanced analytical methods for
`evaluation and comparison of percutaneous absorption in
`animals and humans. Wester et al. [50] employed inductively
`coupled plasma-mass spectrometry for quantitation of biolog-
`ical samples of boric acid, borax and disodium octaborate
`tetrahydrate after their application on the skin. They compared
`the usefulness of finite and infinite dose permeation method-
`ologies across human skin to absorption data in humans. The
`results from the finite dose model were much closer to the in
`vivo absorption data, while the infinite dose methodology
`differed by 10-fold from the in viva results. Cnubben and
`colleagues [51] measured the percutaneous absorption ofortho-
`phenylphenol, a fungicide, in rats, humans and a perfused pig
`ear model. The drug was applied in a hydroethanolic vehicle
`and samples from in vivo studies were evaluated using capillary
`gas chromatography with MSdetector. Jn vivo results indicated
`that in human volunteers, approximately 27% of the applied
`dose was excreted with urine within 48 h versus 40%excreted
`in rats. Among the im vitro parameters tested, the fraction of
`applied dose most accurately predicted human in viva
`percutaneous absorption of the drug (Fig. 3). With respect to
`the other parameters studied, considerable differences were
`observed between the various in vitro models,
`Skin permeation studies using inadequate protocols will
`generate inaccurate data. Currently used sunfilters are lipophilic
`substances with relatively low molecular weight, thus posses-
`sing a good potential to be systemically absorbed across the
`skin. In fact, for a long period oftime scientists have been aware
`of the issues of potential toxicity caused by the percutaneous
`absorption of chemical sunscreens. Recently these concerns
`have been confirmed in numerous reports [52-54]. However,
`the experimental conditions, such as a hydrophilic receiver fluid
`that is used in many in vitro skin permeation experiments with
`sunscreens, generally do not permit a good clearance of these
`molecules from the skin. For example, one study compared the
`skin penetration of benzophenone-3 (BPH), ethylhexyl meth-
`oxycinnamate, butyl methoxydibenzoyl methane, ethylhexyl
`salicylate and homosalate, from two vehicles, an oil-in-water
`(O/W) emulsion gel and petrolatum jelly, both in vitro and in
`vivo. The receptor fluid used in in vitro experiments was saline
`containing 1.5% BSA and at these conditions none ofthe filter
`agents permeated through the skin and negligible amounts were
`detected in various skin layers after 6 bh ofproduct application.
`Also, the effect of the vehicle was minimal in the in vitro
`permeation experimental setup. On the other hand, in humans
`0006
`
`4-4
`a
`.
`4
`a
`=
`7
`2
`v
`v
`°
`a
`
`3
`-
`s
`¥
`
`.
`°
`
`oO
`
`6
`°
`
`-
`
`=
`
`eo
`@#o
`Aa
`“2° Gali) SAH SAGIHSAUER)
`Kp
`PA
`
`Fig. 3. Transdermal absorption of [“C]ortho-phenylphenol, in vitro-in vivo
`correlation:
`(A) Cumulative amount of [‘C]ortho-phenylphenol (a=6)
`permeated in vitro through human viable skin, rat viable skin, human epidermal
`membranes, rat epidermal membranes and perfused pig ears; (B) Factor of
`difference (FOD) between in vitro and in vivo skin absorption of ['*C]ortho-
`phenylphenol based on the systemically available (SA) amount at 4, 8, 24, and
`48 b after a 4-h exposure period of 120 pig/cm’, the permeability coefficient
`(Kp), and the potentially absorbed dose (PA) in humans and rats (reproduced
`with permission fr