235

This paper reviews various techniques used to assess permeation of drugs into and through skin
under in vitro conditions. There have been a wide variety of diffusion cells designed for in vitro mea-
surement of skin permeation. These cells have generally been designed in one of two ways: side-by-side
(bichambers) and vertical in viva mimic diffusion cells. A primary goal of in vitro permeation studies
is the prediction of skin permeation in viva. In this regard, side-by-side diffusional cells arc useful in
delineating mechanisms of permeation under controlled conditions, but are of more limited usefulness
in predicting skin permeation in viva. Vertical cells are more versatile because a wide variety of exper-
imental conditions can he used to gain information useful in the evaluation of formulations ultimately
destined for clinical use. A number of diffusion cell designs are reviewed and some of their advantages
and disadvantages are discussed. In addition to diffusion cells used to assess passive diffusion, ionto-
phorctic- and phonophoretic-enhanced skin permeation in vitro is also considered. Until the use of
differentiated keratinocytes in culture becomes inexpensive and the relationships between skin per-
meation in vitro and in viva are established, skin permeability will be measured using excised animal
or human skin in diffusion cells.

Key words: Percutaneous absorption; Diffusion cells; In vitro techniques; Human skin, Rodent skin

tion parameters important in understanding

permeation of drugs across the skin. Ideally, drug
The delivery of drugs into and through the skin permeation through skin under in vitro condi-
(skin permeation) has been an imponant area tions can be used to predict percutaneous ab-
of research for many years. With the relatively sorption in humans. If such predictions are valid,
recent introduction of transdermal drug delivery the amount of human experimentation could be
systems, interest has paralleled real and poten- abridged. This review will focus on the tech-
tial orofits from marketinn of such deliver SYS- niques used currently to evaluate the topical and
tern;. In vitro skin penne~tion experimenis are transdennal dosage forms under in vitro
performed at the beginning of most studies in- conditions.
volving percutaneous absorption. These experi- A joint FDA/AAPS workshop reported on
ments permit the idcntitication of the formula- principles and practices of in vitro percutaneous
penetration studies with respect to bioavailabil-
ity and bioequivalence [ I 1.A number of factors
important in in vitro percutaneous penetration
studies are outlined in this report [ I]. Some of
236

these factors are expanded upon in this review dermis) gave consistently higher permeability
on in vitro skin permeation techniques. when frozen at - I7 “C and subseauentlv thawed
than the same skin samplesexa&ed2prior to
Percutaneous absorption storage. The studies mentioned above used a sin-
gle model permeant and skir.fromoolyone site.
It has also been recommended that glycerol be
used in the storage of frozen skin to inhibit the
formation of ice crystals which can disrupt cell
Ideally, fresh human skin should be used in envelopes [6]. There is considerable variation in
percutaneous absorption studies when research permeability observed within and between spec-
will ultimately lead to clinical applications. While imens [ 7 and references therein]. Interestingly,
it has been difficult to obtain human skin on reg- it was noted that variability in viva was lower on
ular basis, there are LOWcommercial organ sup- average than that observed in in vitm experi-
pliers who provide skin (dermatomed) as they ments [7].
do many other body organs. rnformation re- For most compounds, the stratum corneum is
quired (age, body site, sex, race, cause of death, the primary diffusional barrier toward drugs,
frozen or stored at 4-5’ C) for proper evaluation particularly hydrophilic compounds [ 81.There
of the results obtained using skin is given with may be instances when partitioning of hydro-
each sample. It is possible that skin obtained from phobic drugs from the stratum corneum into the
commercial organ suppliers may be treated with viable tissues may lead to low transdermal fluxes.
a variety of solvents, particularly if the skin is This situation has been addressed theoretically
prepared for cryogenic freezing. Such treatments bv Guv and Haderaft 191. Most drugs are nor-
I , _ __

can lead to extraction of lipids and proteina- mally cleared by the microvasculature, which lie
ceous materials as well as alter phase or confor- close to the stratum corneum at a depth of 150-
mational structures in the skin. Skin thickness 200 pm from the skin surface [IO]. The micro-
can vary considerably and the investigator should circulation begins close to the dermal/epidermal
measure the thickness of the skin prior to per- junction (the average thickness ofthe epidermis
formingpermeation experiments. Removalofthe is approximately 40 to 50 /Irn [ 11 I). Under in
remaining dermis leaving the viable epidermis is vitro conditions, the microcirculation system is
also used by investigators. Generally, frozen skin absent: therefore, an important aspect of proper
obtained from organ donors has limited meta- design of in vitro permeability experiments is the
bolic activity following thawing. ability to ensure skin conditions similar to those
There are a host of physiologic factors that that exist in viva.
must be considered when evaluating the perme-
ability of human skin under in vitro conditions.
Models for human skin
These factors have been reviewed recently by
Siddiqui [ 2 1. One issue is the storage of the skin
prior to use in the in vitro permeability experi- Animal skin
ment. It is common to freeze human skin for
storage until needed for experimental purposes. Researchers have for many years used skin ex-
In one case, it was found that there was no sig- cised from rodents and other animals. Skin from
niticant difference between the permeability of animals is much easier to obtain, the age and sex
human skin stored frozen for over 1 year and the of the animals can be controlled, and large num-
same skin stored without freezing(fresh) 131. bers of samples can easily be obtained. The pri-
Franzhas alsoconcludedthat freezingfor up to mary problem with using rodentskinas a model
3 monthsdoesnot damagethe barrierproperties for human skin is that it can overestimate per-
of excised skin 141. In contrast, Swarbrick et al. meation relative to that in human skin [ 12-171.
[ 51 found that skin (stratum corneum plus epi- This problem is partly associated with the effects
 237

prolonged(generally24 h
of hydration wherein Analysis of skin permeability
or longer) exposure of rodent skin (hairless
mouseskin is the most notorious) to aqueous There arc a numberof factorsthat can affect
donorandreceptorphasesbringsabouta marked drug release, uptake into the skin, diffusion
diminutionin the barrierpropertiesof the skin through the skin, and transport of the drug into
[IS]. The primary difference between human the systemic vasculature. The design of a topi-
skin and rodent skin is the lipid composition and cally applied dosage form requires ioformation
organization in the stratum corneum. Some spe- regarding these factors if the dosage is to func-
cies of rodentskinmaybe useful in studying per- tion effectively. Therefore, design ofin vitro per-
meation of compounds ifthe total exposure time meation experiments must permit accurate col-
is I2 h or less. It has been suggested that hairless lection of this information and, if the primary
mouse skin can, when using limited amounts of goal of these experiments is to predict the perme-
acetone, be used to provide relevant guidelines ability of the drug in humans, to duplicate the
for risk assessment calculations and bioavaila- conditions that occur in viva.
bility determinations 1 18) although this conclu- Percutaneous absorption experiments gener-
sion is limited to using acetone in small volumes ally require a specific apparatus such as a side-
to deposit a penetrant onto skin. Interestingly, by-side or flow-through apparatus (see below).
many patents are issued based on data collected In these cases, the appearance of drug in the re-
using rodent skin; it is possible that the utility of ceptor solution or a decline fmm the donor phase,
these patents may be limited in clinical practice. is monitored as a function of time. Data CO&
lected are expressed as permeation profiles. There
Other models are a number of relationships used to describe
the permeation of drugs through skin. While the
A reliable model for human skin has been a basis for these relationships can be very com-
highly desirable goal for a number of years. The plex, it has been foundthat the amountof drug
goal remains elusive: nonetheless, significant ad- permeating through skin over time can be de-
vances are being made in the area of tissue cul- scribed in a reasonably straightforward manner.
ture. For example, human keratinocyte cultures The amount of drug appearing in the receptor
grown at the air-liquid interface have been found fluid when using diffusion cells with an infinite
to develop substantial barrier properties to water dose can be described with the following rela-
diffusion [ 191, Reconstructed human epidermis tionship [23 1:
has been used to examine the nitroglycerin and
sucrose permeability 1201. Houk and Guy have M,=C,K,d
reviewed the literature on various membrane
models such as eggshell membranes, composites,
laminates, zeolites, silastic and organic liquid
membranes [21]. Reconstituted stratum cor- where M, is the amount of drug passing through
neum tilms have also been examined as a poten- a unit area of the skin in time, I, from a drug sat-
tial model for skin transport [221. However, urated delivery system; C, is the concentration
preparation of a reconstituted membrane using of the diffusing material just within the mem-
a surf&ant to disaggregate stratum corneum into brane surface (usually taken as the concentra-
cells requires intact human skin, which could be tion in the vehicle). K. is the skin/vehicle oar-
used directly in permeabilityexperiments.At iition coefficient, 0, is%Ibsion cokflicient, &d
present,the mostcommonmeansto evaluatethe d is the diffusionalpath length.BecauseD, rep
in vitro permeabihfyof skin involvesthe use of resentsthe net propenicsof the variousstrataof
excised human skinobtainedduringplasticsur- the skin, it is an apparentdiirusioncoelficient.
gery or cadavers and from common laboratory Using Eq. I, it is possible to schematically rep-
animals such as rodents. resent the appearance of drug in the receptor fluid
238

of in vitro permeation cells (Fig. I ). The utility as a single term called the permeability coeffi-
of Eq. 1 is limited in practice. The permeation of cient, P, such that:
drug through skin gradually increases until a
steady-state condition is reached where I&co so
that&. I can be rewritten in a more usable form:
Using a #mposite term such as P, it is easier
to describe the relative effects of changes in a
formulation as it is difficult to identify specific
which describes the amount of drug penetrating changes in diffusivity, partitioning, and path
per unit time under steady-state conditions (the length. II should be noted that skin is a hetero-
linear portion of the curve in Fig. 1). The steady- geneous membrane and hence a variety of sim-
state flux is obtained from equation 3 by differ- plifying assumptions have been made in the der-
entiating M, with respect to t: ivation of the above equations. The validity of
these assumptions is the subject of research in the
area of skin permeation. Osborne [24] has de-
scribed a number of ~mpu~~ional methods
us&l in predicting skin permeability.
The lag time, &, can be used to estimate DSby Ideally, the in vitro permeation experiments
extrapolating the linear portion of the cumnla- should provide accurate data to permit determi-
ti,ve penetration curve to the axis where drug re- nation of parameters such as lag time, permea-
lease=O, such that: bility coefficient, apparent diffusion coefli-
cicnts, and the amount of drug permeating
through the skin over time at steady state. The
most common technique used to gather this type
of data involves diffusion c&s. The following
ivhere BS is an apparent value because of as-
section describes various diffusion cells used and
sumptions concerning the length of the permea-
some advantages and disadvantages of each.
tion pathway through the skin.
It is commcm to express l&D,/6 of equation 4
Diffusion cells for measuring skin
permeation in vitro

Key design and use parameters

A wide variety of ~ffusionai systems have been

developed for use with rate-limiting membranes
(e.g., excised skin from a laboratory animal or
from a human). These diffusion cells generally
have common elements: two chambers, one con-
taining the active agent (donor vehicle) and the
other containing a stirred receptor solution, sep-
arated by a piece of excised skin or other mem-
brane. The c&s are generally wanged in side-
by-side or vertical con~gumtions. In the case of
side-by-side chambers, both chambers should be
mixed homogeneously. Mixing is most com-
monly accomplished by magnetic stir bars; how-
ever, adequate mixing in diffusion cells can be a
 239

problem in several types of diffusion cell design. mg/l or less may demonstrate limited partition-
Controlling the temperature has been accom- ing in vitro in aqueous receptor solutions.
plished using water-jackets or simply submerg- A technique used to obviate the problem of
ing the entire cell assembly into a water bath. poor in vitro/in viva correlation when using full
Temperature gradients can be introduced if the thickness skin in vitro to evaluate hydrophobic
membrane flange is not properly heated; such compounds is the addition of sohtbilizing agents
temperature gradients could be exacerbated by to the receptor solution, Examples of useful sol-
systems with incomplete fluid agitation. How- ubilizing agents include PEG-20 oleyl ether, oc-
ever, small temperature variations will probably toxynol-9 (T&on X-100) [27], bovine serum
not significantly affect relative rates of albumin (3% in buffer) [32], Poloxamer 188
penetration. [33],PEG400 [34],andethanoI [35].Anim-
A key concern in the use of diffusion cells is portant requirement ofsolubilizingagents is that
whether there is sufficient agitation to prevent they must not alter the inherent permeability
local concentrations of drug and to minimize . .
orooerties of the skin. The addition of PEG 400
static diffusion boundary layers. The goal, as far to the receptor solution reportedly leads to a sig-
as possible, is to maintain sink conditions during nificant alteration of the inherent barrier prop-
the experiment. It has been shown that as agita- erties of human skin [ 361. Also, methanolic and
tion in the receptor phase increases, the thick- ethanohc receptor solvents can damage full-
ness of the unstirred boundary layer decreases: thickness rat skin in vitro as assessed by variable
the permeation rate therefore increases [ 25,263. cortisone fluxes [27]. It has been suggested that
Sufftcient agitation of the receptor fluid can isopropyl myristate is a potential receptor-phase
minimize the unstirred boundary layers, thereby solttbilizing agent [ 371. While certain properties
minimizing diffusional resistance. Under in viva (bipolarity, inertness) make isopropyl myristate
attractive as a solubilizing agent, it is immiscible
conditions, it is generally assumed that diffu-
with water, can potentially extract lipophilic
sional boundat,y layers in dermal capillaries are
components of skin, and has been found to in-
insignificant. The calculation of diffusion coef-
crease the flux of theophyliine in propylene gly-
ficients will include a contribution from the
co1 following pretreatment of hait!% mouse skin
boundary layer barrier and hence result in an ap-
parent diffusion coefficient. [38,39]
Hydration is another variable to consider in
The problem of boundary layer effects can be
long-term permeation experiments under in vi-
exaggerated when relatively hydrophobic drugs
tro conditions. While some studies have found
and chemicals are studied in vitro. When full-
the effects of hydration in long-tetm permeation
thickness rodent skins are used in in vitro per- experiments to be negligible [ 38,401, other stud-
meation experiments, the results indicate that ies have demonstrated that certain rodent skins,
permeation is underestimiated compared with in particular hairless mouse skin, are susceptible
that observed in viva [27-291. It has been hy- to hydration leading to changes in permeation
pothesized that compounds of low water solubil- rates over time [ 11,18,41].
ity have low partition coefficients from the lipoi-
da1 domains of the stratum comeum and the Diff&ion cells for measuring in vitro permeation
aqueous environment of the viable epidermis or
the aqueous receptor fluid [27]. Low partition A myriad of cell designs have been used over
coefftcients lower the overall permeation rate ac- the past 30 years. However, most designs fall into
cordingly. In viva, the drug is absorbed by the one of two general categories: side-by-side diffi-
blood in capillaries that lie at a depth of about sion cells and in viva mimic diffusion cells. The
I50-200pm from the surface ofthe skin [30,31]. examples provided herein are far from exhaus-
Bronaugh and Stewart [27] have suggested that tive: they were chosen to illustrate some of the
compounds having an aqueous solubility of IO basic design considerations and some potential
problems encountered during the routine use of
diffusion cells for in vitro skin oermeation

Side-by-side diffusion cells usually comprise

two chambers wherein one chamber contains the Sampli
permeant in solution and the other contains the
receptor solution. These two chambers are sepa-
rated by a membrane (skin in percutaneous per-
meation experiments). The contents of one or
both chambers can be agitated to ensure ade-
quate dispersion of the drug molecules and to
minimize the static diffusion boundary layers.
Configurations of side-by-side diffusion cells in-
clude T-shapes [42] and identical Lshapes [43-
461. Most cells are composed of glass as illus-
trated by the modified conical flask design of
Wurster et al. [47] (see Fig. 2). This cell has
several drawbacks, most notably the inability to
agitate the solutions internally although the au-
thors propose gently shaking the entire appara-
tus in the plane ofthe membrane. The apparatus
Locking N”,
is suspended in a water bath for temperature
control. While this design was shown to be ade- Fig. 3. Side-by-side permeability cell showing general fea-
quate for rapidly diffusing compound (sarin), its IUTCS. Mixingsaccomplirhedrhrough rotationofTeCmsrir-
broader applicability is unknown. rers mounted pcrpcndicular to the membrane. A synchron-
ous motor drives the stirrers in each chamber at the same
A considerably more complex design is that of
raw. RedrawnfromFlynnand Smith 1481.
Flynn and Smith [48] (see Fig. 3). The cham-
bers, unlike those used by Wurster et al. [47], are manufactured from brass. These chambers
are held together with a nut-and-thread assem-
bly. The membrane is held clamped between O-
rinas seated in the flanze surfaces. Relativelv
l&e Teflon stirrers are mounted vertical to the
membrane; these stirrers are in turn mounted
onto shafts which protrude from each chamber.
These protruding stirrer shafts are attached to
gears interlinked with a synchronous motor. Re-
moval of the chamber fluid is accomplished
through sampling ports in both chambers. A per-
forated screen can be used to SUPPOSE the mem-
brane if required. The entire apparatus can be
immersed in a water bath for control of temper-
ature. As designed, the device requires relatively
large amounts of membrane, which limits its
usefulness to skin available in large amounts.
 241

While many of the features of this side-by-side called Valia-Chien cell, hasbeen studied exten-
are well-suited for assessing skin permeation un- sively by Chien and coworkers [44-46). These
der in vitro conditions, the complexity of the de- cells are composed of two identical horizontal
sign (motors, shafts, cogs) make the widespread chambers, 0.9 cm in diameter and 3.8 cm in
use of this system limited despite its advantages leasth: the stoppered sampling ports are vertical
over many other in vitro diffusion cells. (see Fig 5). The active surface arca, suf~ciently
Another de&n for s&&v-side diffnsjon ceils small for use with biomembmaes such as human
is shown in Fig.4 149] Twdglass chambers with skin, is 0.64 cm’. As can be seen in Fig. 5. the
a relatively smalt diffusional area are tempera- cylindrical portions of the ceils are enclosed in a
ture-controlled by immersion in a water bath. water jacket although the membrane connecting
Both chambers arc stirred with Teflon-coated flange is open to ambient temperatures. Fluid
magnetic bars; a magnetic stirrer is positioned mixing is reduced in the constricted region of the
below the side-by-side cell system to produce flange; however, mixing and temperature equi-
synchronous stirring of the bars in tsth cells. AS librium are reached relativelv ranidlv in this cell.
with the example shown in Fig. 3, the membrane As noted by Smith and Naigb [37],ihedesignof
can be supported with a stainless steel mesh. This cehs as shown in fig 5 has led to the apparent
mesh can &so be used to prevent distention of requirement of a computational correction to ac-
the membrane ifthe receptor chamber is used to count for the nonideal design. Again, such cor-
measure vapor diffusion through skin [ 501. Re- rections lead to, at best, only minimal changes in
ceptor and donor sampling is accomplished, the interpretation of data obtained with Valia-
through ports located on each chamber; each port Chien diffusion cells.
can be sealed with Pamfilm if needed. Depend- A combination ofa magnetically-stirred donor
ing on the height of the fluid in the sampling chamber and a flow-through receptor c!ramber
ports, mixing may be inadequate if a substantial has aiso been developed [ 511 (seeFig. 6). A
portion of the receptor (or donor) solution is unique feature of this design is the location of
within the sampling port. the receptor fluid inlet positioned at the center
Another side-by-side diffusicn cell, the so-
 Diffuusiicncells designed to mimic in viva
conditions

Systems that parallel conditions found in viva

are normally vertical with the bottom chamber
desianed to hold a receotor &id. The bottom
chamber is agitated or recycled in an attempt to
maintain siok conditions tbrouphout the experi-

.l_i ment. An advantage of the vertical cell design is

the ability t I vary the nature of the donor vehi-
cle. A film ol material can be applied by solvent
evaporation; ointments, pastes, synthetic mem-
branes in series with skin, and entire transder-
ma1 devices can also be studied. Atmospheric
J’ conditions (e.g., humidity) can be controlled in
these cells as well. Sequential treatments, such as
pretreatment with an enhancer followed by de-
position ofa drug, are easily accomplished using
vertical cells. It is also possible to conduct infi-
nite and finite dose experiments. The actual ex-
perimenlal design will vary depending on the type
of formulation under investigation. For in-
stance, testing of drug permeation from ti topical
Fig. b. Flon-through diffusion cell. Key: A, application tube:
vehicle to deliver corticosteroids is accom-
B, removal abe; C, donor companmcnt stopper; D. neu-
prenerubbe)washer, E, donor companmenl; F, skin spcci- plished through semiinfinite dose [ 521 or Iinite
men clampe. between Perrpex blocks; G. strrer; H, clamp- dose techniques [23,53].
mg screw: 1. c:utlet tube; .I. acceptor compartment; K, inlet There have been a numbe; of vertical cells de-
tube. I -drawn from Aslley and Levine 151 1.
signed and tested over the past 25-30 years. One
of the earlier cells is that of Coldman et al. [ 541
of the memb ane while the effluent is collected
from the peri+?hery. This diffusiou cell was I+
portedly rapat:e of efficiently removing the pen-
elrant from the membrane/receptor solution in-
terface thus maintaining maximum sink
conditions.
Respite the dissimilarity between side-by-side
designs and assessing skin permeation under in
viva conditions, a large amount of the skin per-
meation data has been collected skin using the
side-by-side chambers. Side-by-side designs ex-
pose the membrane (skin) to solvent on both
sides throughout the experiment leading to po-
tential salvation effects. Nonetheless, the data
collected in side-by-side chambers is useful ifthe
limitations of the design are acknowledged. Fig. 1. Glassdiffusionccllcons~ingof a lower~hamberwitb
a side arm for sampling cfrcceplor phase. Key: A, skin sper-
Measurement of permeation rates under condi- imcn: B. Tetlon piecer holding skin; C, clamp; D, mcepmr
tions similar to those encountered m viva re- chamber: E, polyethylene sail: F, Teflon coated magnetw bar.
quires a different cell design as explained below. Redrawn from Coldman et al. [%I].
 243

shown in Fig. 7. This static cell is composed pri-

marily of glass with a side arm for sampling. A
Tenon-coated stirring bar is attached to a poly-
ethylene sail to provide mixing of the receptor
solution. The skin is held in place by a Teflon
disk on a flat ground glass surface at the top of
the receptor chamber. The exposed surface area
of this cell was 0.30 cm* and the receptor cham-
ber volume was IO ml, a portion of which is lo-
cated in the side-arm (mixing may not be ade-
quate in the side-arm). Fluid mixing and mass
transfer characteristics of this diffusion cell have
not been fvlly iovesti~ted.
Barry and coworkers [ 111 have, in addition to
a side-by-side design, examined a vertical cell
(see Fig. 8). This cell is merely one half of the
bide-by-side chambers (see Fig. 2) turned vetti-
tally with a donor chamber placed on top of the
half-cell. As expected, cell components are those upper part of chamber near the area of contact
of the cell shown in Fig. 2 (glass chambers, Te- with the membrane. In the original design, the
tlon magnetic stir bar, glass side arm sealable with cell was static and therefore had a single sam-
Pamfilm, stainless steel membrane support). pling port (unstoppered). The central part ofthc
Similar to the diffusion cell of Coldman et al. receptor chamber is enclosed in a water-jacket for
[ 541, a relatively large portion of the receptor temperature comcol. Portions of the receptor
solution is located in the poorly stirred side-arm. chamber and the entire donor mmpartment are
The Franz diffusion cell is one of the most open to ambient conditions. As with most verti-
widely used systems for in vitro skin permeation cal systems, the receptor chamber is agitated with
studies. First disclosed in 1978 [ 531 and subse- a Teflon-coated magnetic stir bar.
quently marketed, this cell has a small donor A number of modifications have been intro-
compartment and a dumbbell-shaped receptor duced into the original design by Franz. O-Ring
chamber (see Fig. 9). The bottom portion ofthe flanges have been added; a second side-arm has
dumbbell-shaped chamber communicates with a been added to permit flow-through operation, the
narrower cylindrical tube which widens in the donor compartment can be sealed, and it can be
made in a variety of active surface area diame-
ters. While the Franz cell is widely used, it has
several potential drawbacks, most notably rela-
tively poor mixing hyd~d~amics~ Poor mixing
results from the fact that agitation in the lower
bulb must be transmitted through the narrow
cylinder. There is considerable resistance to lam-
inar fluid flow throueh the constricted oortion of
the receptor chamber leading to a staiic bound-
ary layer at the interface between the membrane
and the receptor solution.
The poor mixing properties in the receptor
chamber of Franz cells have been studied [ 55].
It wasfound that the time to complete mixing, as
measured by homogeneous dye dispersion, was
 sign: the skin is sandwiched between two areas
of ground glass. The authors report that no leak-
age of material was observed under any experi-
mental conditions used [ 551.Using the time for
homogeneous dispersion, both cell designs gave
nearly instantaneous mixing ( < 30 s).
Flow-through systems offer an alternative to
sampling ports; by replacing the entire contents
of the receptor chamber on a continuous basis,
sink conditions are more easily maintained. As a
result, flow-through cell design coupled with a
vertical chamber represents conditions similar to
those encountered in viva [ 561. One of earliest
flow-through cells was that of Marzulli [ 571 (see
Fig. 11). The cylindrical design may not have
optimum hydrodynamics: the inlet to the recep-
tor solution is not directed toward or at the
membrane. Removal of the permeant may not
he complete in the Marzulli design. An obvious
improvement in the design would be introduc-
tion of a stirrer [ 581.At the same time, stagnant
diffusion layers are probably only of major con-
cern when using relatively lipophilic permeants.
A relatively complicated permeation cell for
assessing penetration-evaporation was designed
with a flow-through system and a magnetic stir
bar [59] (see Fig. 12). The receptor chamber
(thermostatically jacketed) is stirred with a

FIN. IO. Flaw-throughdiffusioncells.The uppercell hasa re-

ccptorYU,U,,,~of53 ml; lower cell hasa rcceplorvolumeof
3.0 In,. Redrawn from Gummer et a,. ,551.

inadequate in the side-arm and the upper por-

tion of the dumbbell-shaped receptor cell in that
homogeneity was not reached until 30 mm had
passed in some cases [ 551.Based on the data Annular Thrust--.
collected with the Franz cell, two flow-through Block
in vitro penetration cells were designed to ob-
viate the problem of poor mixing (see Fig. IO).
Two types of cells were prepared to accommo-
date two different surface areas. The central de-
sign feature of these cells is the receptor cham-
ber. Itsdiameter is wider than that of the Franz
cell to achieve rapid and evep stirring. As can be
seen in Fig. IO, these cells feature a flow-through
receptor chamber. O-Rings are absent in this de-
 245

receptor chamber.T&n used in magnetic slir-

rem could also absorb hyd~phohic sot&es as
well.
Often, biolo&$cmembranes are fragile and
thereforedifficult to handle.In particular,meas-
uring the permeability of sheets of stratum cor-
neum can be a formidable task. Maintaining a
specific water content in the Stratum corneum
can also be a demanding prospect. While sup-
port structures have been used to maintain the
integrity of the stratum comeurn (see for io-
stance Figs. 4 and S), a new technique has been
developed recently to control the integrity and
water content of stratum corneum Sheets 157)
(see Fig. 13). It is possible to control the humid-
ity such that the Stratum comeum inside the
sandwich adopts a water content which is in
thermodynamic equilibrium with salt solutions
outside the membrane. A minor drawback of this
system is the need to determine not only the
permeability of stratum comeum, hut also that
of the silicone sandwich.

The diffusion celIs described in the previous

sections were designed to meaSure the passive
/- Silicon Adhesive 6 Silicone Membrane diffusion of drugs. Iontophoresis and phonopho-
resis are techniques used to increase the trans-
cutaneous flux of drugs. Iontophoresis is defined
as the increase in permeation rate of a molecule
induced by an applied current through the skin
[61]. While the concept is relatively old, it has
received considerable attention recently because
it appears possible to use iontophoresis to de-
liver certain macromolecules, viz., peptides and
magnetic stir bar. The donor chamber was de- proteins, through skin [K&63]. ~o~o~hn~sis is
signed to control evaporation from the surface of defined as the increase in fate of a solute thrmtgh
the skin by forced, warm air ventilation. skin observed under the influence of art ultra-
A problem with many Row-through systems is sonic perturbation [ 64,651.
the entrapment of air bubbles under the surface Diffusion c‘?llsused in iontophoresis enperi-
of the skin. These bubbles can reduce the diffu- ments are simtlar to those used to aSSeSSpassive
sional area substantially and should therefore be diffusion except electrodes have been added, one
minimized by degassing of solvents or the use of in the donor chamber and one in the receutor
bubble traps. Another problem associated with compartment [66]. While this typeofappar&
the use of flow through Systems is absorption of works well experimentally, electrodes can not be
drugs, in particular hydrophobic drugs, into placed on opposite sides of the skin in viva.
plastic tubing used to coIlect the eftIuent from the GlikfeId et al. 1671designed an iontophoretic cell
246

suboptimal under certain conditions coupled

with recent advances in cell design should ensure
that data collected during in vitro permeation
experiments is valid and therefore useful in mak-
ing predictions about drug delivery in viva from
topically applied formulations.

Acknowledgement

The excellent technical drawings of Kraig

Murray are gratefully acknowledged.

Fig. I4. Inmo~horelic diffusion cell constructed of glass. Ef-

fecuve surface area is 0.8 cm’ with a donor phase volume of
about 0.5 cm’. Rcccptor phase volume is appr@ximately 7 I J.P. Skelly,V.P.Shah, H.1. Maibach,R.H. Guy, R.C.
cm1 A and B arc the elecaodc chambers. Redrawn from Wcster,G. Plynn.and A. Yacobi,FDAand AAPSre-
Glikfeld cl al. (671. port of the workshop on principles and practices of in
vitro oercutancoua Denetration studies: relevance of
for in vitro studies to account for the in viva sit- bio&ilability and ~bioequivalence, Pharm. Rex, 4
(1987) 265-267.
uation (see Fig. 14). In this design, both elec- 2 0. Siddiqui, Pbyncochemical, physiological, and math-
trodes are applied to the same side of the mem- ematical considerations in optimizing percutaneous ab-
brane (skin). Such a design should permit better sorption of drugs, Crirical Rev. Ther. Drug Carrier Syc
approximation of drug delivery using iontopho- terns, 6 (1989) I-38.
resis under in viva conditions. The cell used is 3 S.M. Harrison. B.W. Barry and P.H. Dugard, Effects of
freezing on human skin permeability, J. Pbam. P&r-
based on that of Gummer et al. [ 551 and hence maw,., 36 (1984) 261-262.
should exhibit properties similar to those claimed 4 T.J. Franz, Pwwtaneous absorption. On the relevance
for the diffusion cell shown in Fig. 10. of in vitro data, J. Invest. Dermatol., 64 (1975) l90-
Phonophoresis is rarely studied under in vitro 195.
conditions. Most experiments involve the use of 5 J. Swarbrick, G. Lee and I. Brom, Drug permeation
through human skin: 1. Et&cl of slorage conditions of
existing commercially available ultrasonic skin, J. Invest Dermntol., 78 (1982) 63-66.
equipment. These systems are too large to use 6 G.B. Kasting, R.L. Smith and E.R. Cc=aper,in: B. Shrcot
unde: in vitro conditions, particularly with hu- and H. Schaefer (Eds. ), Pharmacologyandthe Skin, Vol.
man skin. Design of an in vitro diffusion cell ca- I, Skin Pharmacokinetics. Urger, Bawl, 1¶67, pp. 138-
pable of varying experimental parameters im- 153.
portant in phonaphoretic drug delivery would Variations
D. Southwell. B.W. Barry and R. Woodford,
inpermeabilityofhumanskinwithinandbeiweenspec-
probably help accelerate researc III this area. imens, Int. J. Pharm., I8( 1984) 299-309.
LH. Blank and R.J. Scbeuplein, Transpon into and
within theskin, Br. J. Dermatol., 81 SuppI. 4 (1969) 4-
10.
R.H. Guy and I. HadgraIl, Physicochemical aspects of
A variety of diffusion cell designs have been percutanwus absorption and its enhancement. Pharm.
used to assess skin permeation in vitro. Due to Rcs., 5 (1988) 753-758.
problems of inadequate receptor chamber agita- G.M. MacKee, MB. Sulzherger, F. Herrmann and R.L
Baa, Histologic studies on pemutawxs penetration
tion and temperature control, and sampling of
with special reference to vehicles, J. Invest. Dermatol.,
drugs from poorly stirred side arms, compari- 62 (1445) 80-U.
sons of published data can prove problematic. J.T. Whitton and J.D. Everall, The thickness ofthe epi-
Acknowledging that some cell designs may be dermis, Br.3. Dermatol.,Y9 ( 1973) 467-416.
12 R.C. wesccrand P.K. Noonan. Ilekvonca of animal
wJd& far perc”Ia”eourabsoqmor..I”*. J. Phaml.. 7
(1980) 99-l IO.
13 R.L. Bronaugh,R.F. Slcwan and E.R. Congdon,Melh-
adsfor in wro percutaneowabsorp:ionstudiesII. An- 30 H. Schaefer,A. Zcschand C. Sliiagen, Skin Pemxabil-
tmsl modelsfor human skin, Toxic& Appl. Pharma- ity, SpringerVerlag.Berlin, 1982,p. S49.
cot..62 ( 19821481-488. 31 B.W. Barry, D~rrn~lol~jc~ Formulationr i%rcuta*
14 R.C Scoti, M.‘W~ik~~=~d P.H. Dueard, A eompariron neO”sAwxplm”, Marcel DekLcr,New York, L983,p.
of the in vitro permeability propertiesof human and 9.
sameIsb~~%cq animal skins.Inl. J. Cos.SIX, 8 ( 1986)
32 D.W.C. Bmwn and A.G. Ulsamer. Percutaneous penrr
It?+- I Y4.
Iration of hexachloraohene as relatedto rec~~bx sobllu-
15 J.R. Bond and B.W. Barry. Hairless mwse rkin is t&n-
ited as a model for assessing the effectsof penerrarion
33
enhan~s in human skin, I. Invest. l&mat&, 90
(1988) X10-813.
J.R. Bond and B.W. Bany, Vamagingeffecl ofacetone
on the permeability barrier ofhairlessmouseskrncom- 34
pared~viviththatofhumanskin, Ini, 1. Warm.,41 (19R81
. ._.
91-w
P. Catp and D.R. Friend, fransdermal debveryof h-
vonorges~rcl. Ylfl. Ef&ct ofenbaneersonw skin, hair- 35 R.C. Scolh D.E. Leahy,J.R. Evansand E. Bole, In vitro
leamouseskin,hairlcssguincapigskin,and humanskin, percutaneous model for qunnlitative structure-abscrp-
lnt. I. Pharm., 5S (1990) 93-102. lion studies,J. Pharm. Pharmacol.,38 (SuppI.) ( 1986)
18 KS. Hi”<, CD. Hodson,CR. Lorenceand R.H. Guy, 66P.
In vitm perccukeneous penelration: evaluation of the 36 P.P. Sarpotdar, 3.L. Garkiland R.P. Gmnini, Effemof
ulilily of hsirlesr mouseskin, J. Invest. Dcnnatol., 93 polyclhylcnc &cd 400 on the penew&m of dwgr
(1989) 87-91. lhroughhumanskin in vitro, 1. Pharm. Sci., 75 ( 1986)
I¶ V.H.W. Mak, M.B. Cumpslone, A.H. Kennedy, C.S. 26-28.
Harmon, R.H. Guy and R.O. Pals, Barrier function of 37 E.W. Smith andJ.M.Haigh, In vitro systemsforassess-
humankeraiwcyte culluresgrown et thhcair-liqrid in- ment of drug release from topical formulations and
te&ce.J. Invert. ilermatol..96 (1991 t 323-327. *r*nsmembranepermeation.in: R.L. Bronaughand H.1.
20 M. Ponec.P.JJ. Wauh~~Fe~ris~A. Burpsr.J. Kemae- Maibach (Ed%.). Perculaneour Abswption: Me&J-
near and &.t. B&de, Nitmgiycerin and &rose per&
nisms- Me(hodol~gy- Drug Delivery, 2nd Ed., Marcel
ability as quabty marken for remnsmrwxed humanepi-
dcrmis.Skin Phsrmacol.,3 (1990) 126-135. Dekker. New York, 1989,pp. 465-508.
36 K.B. Sloan,K.G. Sivcr and S.A.M. Koch, The cNectOf
21 J. Houk and R.H. Guy, Membranemodelsfor skin pen-
elmiion studies,Chem. REV.,88 (1988) 455-471. vehicleon thedilfusionof salicylicacid thmugbhairless
22 w.R. Kock, 8. Berner, J.L. Bums and D.L. Bissett, mouseskin. I. Pkarm. Sci., 75 (1986) 744-749.
Prcparalion and charactcriaalinn of a nconrtituted 39 R.P. Wanair, KG. Siver and K.B. Sloan,Ihe solubitity
stratum comeam iilm as a model membranefor skin pammeterof vehiclesils a predictorof relative vehicle
mu,sp+,~ Arch. Dermatol. Rex, 280 ( 1988) 252-256. effect on the difTwion of 6.me~ploputinc, Inc J.
23 M.I. Foremanand 1. Kelly, The difiwivn ofnandroione Pharm.,36(1987)2ll-zU.
throughhydratedbum;lncadaverskin. Br. i. Dermaxd., 40 D. SOU,bWC,, B”dB.W. Barry, Pcne,rationcnh*“cemcnt
95(19?6)265-270. in human skin: efTeet of%pyrrolidone, dimethyl form-
24 D.W. Osborne,Co~~pulalionaimetltodsfor predicling amide sod increasedhydration of finite dose peniwa-
skin ~rrnc~b~j~y, Pharm. Manuf. April ( 1986) 41-48. tion of aspirin and caffeine,lnt. J. Pharm., 22 (1984)
25 R.G. Stehleand W.L Higuchi, In vilm model for tranb 291-298.
port of solutesin Ibreepbase systemII: Experimental 41 J.R. Bond and B.W. Barry, Hairless mouseskin is lim-
considerations.3.Pharm. Sci., 61 (1972) 1931-1935.
iwd as a model for asrcnangcffeclsof penetrationen-
26 E.G. Lowing and D.B. Black. Ditliwion layer eficcts
hancen in humanskin.1. Invest.Dermatol.. 90 ( 1988)
on penncation of pbmylbutazwe ihrougb polydime-
8
ibytsiloxane,J. Pharm. Sci., 63 (1974) 399-1402. SIO-SI3.
27 R.L. Bronaughand R.F. Stewa~ Methods for in vitro 42 M. Washitake.Y. Takasbima,S. Tanaka,T. Anmo and
pcreutaneollsabsorplion studieslli. Hydmphobiccom- 1. Tanaka, Dmg pameation thmugb egg shell mem-
oounds.J. Pharm.Sci.. 73 11984) 12%1258. braou, Cbem.Phatm. Bull., 28 (1980) 2855-2861.
28 k.L. B&m@ and R.F. &art; Methods for in vitro 43 A. Dyer,G.G. Hayes,J.G. Wilsonand R. Catwall, Dif-
percufaneons absorptionstudier IV. Preparalion of rhe fusionthroughskin and model syrlemz,Int. 1. Cormet.
barrier1a~w.J. Pharm Set.,75 (1986) 487-491. Sci.. I (1979) 9i-100.
248

44 K. Tojo, J.A. Masi and Y.W. Chien, Hydrodynamic 56 R.C. Wcstorand H.I. Maibach. Animal models iix per-
characterislicsofa” in vitrodrug pcrmealion cell. Ind. coianeous absorption, in: H.1. Maibach and N. I..owe
Eng. Chem. Fondam., 24 ( 1985) 368-373. [Eds.), Models m Derma,“l”gy. Vol. 2, S. Karger, Bw,,
45 K. Toio. M.M. Ghannsm. Y. Sun and Y.W. Chic”. In 1985, pp. 159-169.
wtr” apparatus for cootroiled release studies and inlrin- 57 F.N. Marzulli, Uarrxrs Lo ski” penerrau”“. 1. Invest.
sic rate ofpermeation, J. Controlled Rel., I( 1
1985 197- Dermatol., 39 (1962) 387-393.
203. 58 I.H. Blank, Pcnctral~on of low-molecular-weight alco-
46 Y.W. Chien and K.H. V&a, Development ofa dynamic hols intothe skin, 1. Invest. Dermatol.. 43 (1964) 415-
ski” permeation system for long-term permead”” 61”*- 120.
ies. Drug Dev. Ind. Pharm., 10 ( 1984) 575-599. 59 C.S. Hawkitls and W.G. Reifenrath, Influence of skin
47 DE. Worsler, J.A. Osrrenga and LE. Mstheson. Jr., ~““nx, pcnaration cell fluid. and panition coefftcient
Sari” transport across excised human skin L: Rrmeabil- on in vitro skin permeaatio”. J. Pharm. Sci., 75 f 1986)
sty sod ~b~“~,i”” cbaracreristics, I. Phann. Sci., 68 378-38 I.
(1979) ~406-1#9. 60 H.LG.M. Tiemcsscn, HE. B”dd+ and H.E. Jungingw,
48 G.L. Flynn and E.W. Smith, ~Semb~nediffosion !: De- Silicone membrane sandwich method to measure drug
sign and teslinE of new multifeatured diffusion cell, J. tmospon through isolated human sIralum corneom
Pharm.Sci.,60(1971) 1713-1717. having I fixed wiitcr conrem, lot. J. Pharm.. 56 ( 1969)
49 D. Sollfhwelland B.W. Barry, Penelralionenhancers for 87-94.
human skin: mode of acdon of &pyrrolid”ne and di- 61 RR. Burnecte, lontophorcsir, in: J. Hadgraft and R.N.
melhylformamide on panition and diffusion of model Guy IEds.). Transdcrmal Drug Dchvery, Dcvclap-
compounds water. n-alcohols, and caffeine, J. Invest. menial Issuesand Rwarch Initiatives. Marcel Dekker.
Dermatol., 80 (1983) 507-514. New York, 1989, pp. 247-291.
R.W. Barry. SM. Harrison and P.H. Dugard, Corrcla- 62 0. Siddiqul, Y. Sing. J.-C. Liu and Y.W. Chien, hcili-
~2”” of thermodynamic activity and vnpour diffwon tatcd aansdwmal transpon of insulin. J. Pharm. Sci.,
rhmogh human skin for the model compound, benzy! 76(1987)341-345.
alcohol, I. Pharm. Pharmacol.. 37 (1985) 84-90. 63 B.R. Meyer.W. Kries,J, Eschbach.Y. GX%nm,S. Rosen
J.P. Astlcv and M. LEvinc. ENeci ofdimethvl sulfoxrde and S. Lb&is, Surnssful iransdwmai adminis~taiion of
~~~p~“~i~ doses of 8 polypcptidc L” normal homan
uotunreers, Clin. P~rma~ol. Then., 44 (1988) 607-
M. Aimworth. Methods far measuring percutaneous 612.
absorption, J. Sot. Cosmet. Chcm., I1(
1960) 69-78. 64 D.M. Skauen and G.M. Zcntner, Phonophoresis. Inl.1.
T.J. Franz, The none dose technique as a vahd m vilr” Pharn,., 20,1984) 235-245.
model for the study of percutaneous absorption, Cum 65 P. Tyle and P. Agawala, Drug de!ivery by iontopbo-
Probl. Dermalol., 7 (1978) S&h& resis, Pharm. Res., 6 (1984) 355-381.
M.F. Coldman. B.J. Paulsen andT. Higuchi, Enhance- 66 M. Clcmessey, 0. Couarraze and C. Hcrrcnkncchl, lon-
men, of “ercu~~neous absomtion hv the “se of ““la- ,“Dh”reS,S, an allcrna,l”e 1” pass,w lra”rderma! dew-
tilemonvbladle systems as vehicles, j, Pharm. Sci., 58 cry: analysis of physicochfmical mechanisms, STP
(1969) 1098-1102. Pharms Sci., I( 1991) 24-37.
C.L. Gummer, R.S. Him and H.I. Maibach. The skin 67 P. Ghkfeld, C. Cullander, R.S. Hinzand R.H. Guy, New
penelntion ccl: a d&s” update, inl. 1. Phann.. 40 syswns for in vim” sloda of ionlophoresis, Pharm.
(1987) 101-104. Res., 5 ( 1988) 443-447.