J. Pharm. Pharmacol. 1983, 35:28-33
`Received June 18, 1982
`
`0022-3573183t’010028-06 $02.50/0
`© 1983 J. Pharm. Pharmacol.
`
`Physicochemical characterization of the human nail:
`permeation pattern for water and the homologous
`alcohols and differences with respect to the stratum
`corneum*
`
`KENNETH A. WALTERSt, GORDON L. FLYNN* AND JOHN R. MARVEL~
`
`College of Pharmacy, University of Michigan, Ann Arbor, Michigan 48109, U.S.A., and 2 Dermatological Division, Ortho
`Pharmaceutical Corporation, Raritan, New Jersey U.S.A.
`
`In order to develop a basic concept of the permeability of the human nail plate and thus
`create a better understanding of the toxic potentials and therapeutic possibilities of
`substances applied to the nail, avulsed cadaver nails have been placed in specially
`constructed diffusion chambers and their permeation by water and the n-alkanols through
`dodecanol, all in high aqueous dilution, has been investigated. The permeability coefficient
`of water is 16.5 × 103 cm h-l and that for methanol is 5.6 x 103cmh i. Ethanol’s
`permeability coefficient measured 5.8 × 10-3 cm h 1. Permeability coefficients decreased
`systematically thereafter to a low value of 0.27 x 10-3 em h-1 at n-octanol. The middle
`chain length alkanols, n-pentanol through n-octanol, have similar permeability coefficients
`but n-decanol and n-dodecanol show higher rates of permeation. The data suggest that, as a
`membrane, tile hydrated human nail plate behaves like a hydrogel of high ionic strength to
`the polar and semipolar alcohols. Declining permeability rates appear linked to decreased
`partitioning into the complex matrix of the plate as the compounds become hydrophobic.
`The results for n-decanol and n-dodecanol introduce the possibility that a parallel lipid
`pathway exists which favours the permeation of these exceedingly hydrophobic species.
`
`Apparently, no cvidcnce exists concerning funda-
`mental permeation mechanisms and possible influ-
`ences of chemical structure on transport across the
`nail plate. To an extent its permeability properties
`have been inferred without foundation from the
`behaviour of other horny tissues. In order to make a
`priori judgements concerning toxic risk and thera-
`peutic benefit of substances brought in contact with
`the nail, some baseline information on this tissue is
`nccded.
`We have shown it possible to determine nail plate
`permeability coefficients using standard diffusion
`cell techniques (Walters et a11981). Results obtained
`for water agreed well with literature data on water
`transpiration through the nail plate (Burch & Winsor
`1946; Spruit 1971; Baden et al 1973). In pursuant
`studies the techniques have been extended to the
`permeation of some n-alkanols. These are useful
`prototype compounds with systematically varying
`oi[/water (o!w) distribution coefficients and diffusion
`
`* Correspondence. This work supported through the
`generosity of Ortho Pharmaceutical Corporation, Raritan,
`N.J., U.S.A.
`
`+ Present address, Fisons Limited. Pharmaceutical Divi-
`sion. Research & Development Laboratories, Bakewell
`Road, Loughborough. Leicestershire LEII 0QY. U.K.
`
`coefficients. Such structural influences on physico-
`chemical properties, when considered together with
`relative permeabilities, have helped decipher the
`barrier mechanisms of several membranes (Blank
`1964; Scheuplein 1965; Hwang et al 1976; Ho et al
`1977: Behl et al 1980; Durrheim et al 1980; Flynn et
`al 1981). Previous studies of the alkanol’s permea-
`tion of skin are especially notable as these provide
`evidence that the stratum corneum acts to some
`extent as a hydrophobic continuum (barrier) (Blank
`1964; Scheuplein 1965: Behl et al 1980; Durrheim et
`al 1980; Flynn et al 1981). Similar studies on the
`human nail plate presented here are comparably
`revealing as, unlike the stratum corneum, the nail
`becomes less permeable to the n-alkanols as their
`hydrophobicity is increased. At extreme hydro-
`phobicity there is increased permeability. The
`mechanistic significance of these general observa-
`tions is considered.
`
`MATERIALS AND METHODS
`
`Materials
`Tritiated water and radiolabelled alcohols were
`obtained from New England Nuclear ([311]water,
`[3H]methanol, [14Clethanol, [14C]butanol), Cali-
`fornia Bionuclear ([14Clpropanol. [laC]pentanol,
`
`ACRUX DDS PTY LTD. et al.
`EXHIBIT 1040
`IPR Petition for
`U.S. Patent No. 7,214,506
`
`1 of 6
`
`

`
`NAIL PERMEATION BY’ HOMOLOGOUS ALCOHOI.S
`
`29
`
`[laC]heptanol, [laC]dodecanol) and ICN ([14C]hex- compound, tl is the nail plate thickness and q. is the
`anol, [14C]octanol, [LaC]decanol). All radiolabelled
`diffusional lag time obtained by linear regression of
`compounds were diluted with saline (0.9% NaC1
`the steady state slope of uptake versus time plots.
`Irrigation Solution, Abbott Labs) before use. The The nail plates used in these studies were measured
`alkanols were diluted to trace concentrations. 1@a with a micrometer and averaged 0.54 mm in thick-
`molar or less.
`ness.
`
`Permeation procedures
`Details of the diffusion cell and permeation pro-
`cedures have been given previously (Waiters et al
`19811. Briefly, trimmed human nail plate sections*
`were placed between two halves of a diffusion cell. A
`known amount of a radiolabelled permeant was
`placed in the donor chamber and samples were taken
`at predetermined intervals from the receptor cham-
`ber. Isotope activity was monitored using a Beckman
`LS 9000 liquid scintillation counter.
`The permeation behaviours of [3H]water and
`[3H]methanol and [14C]alkanols in dilute solution
`were followed as a function of time at 37 °C. In all
`cases two permeants were applied with different
`radiolabels. Generally methanol was run as a tri-
`tiated compound along with a ~4Cqabelled co-
`permeant. Methanol thus served as a reference and it
`is important to note that the increased values for the
`permeability coefficient of decanol and dodecanol
`were obtained concurrently with normal methanol
`data.
`Permeability coefficients (P) were calculated
`from:
`
`p V(dC/dt)
`A. ~ C
`
`(i)
`
`where V is the volume of the receiver half ceil, dC/dt
`is the rate of change in concentration in the
`pseudo-steady state portion of the receiver concen-
`tration versus time plot, A is the diffusional area and
`AC is the concentration differential of permeant
`across the membrane. V(dC/dt) gives the diffusional
`flux in mass per unit time. The diffusion cells with
`nail plate membranes in place were scrupulously
`checked for intercompartmental leakage using sol-
`uble but impenetrable polyethyleneglycol markers
`and no leaks were evident.
`Diffusivities of the permeants in the nail plate
`tissue were calculated from the non-stationary state
`periods using:
`
`h2
`Deft = ~I,
`
`(2)
`
`Where D~ff is the effective diffusivity for a given
`
`* Fresh cadaver nails generously supplied by Dr T. M.
`Oelrich, University of Michigan, School of Medicine.
`
`RESULTS AND DISCUSSION
`Permeability coefficients of water and the saline
`diluted n-alkanols are given along with diffusion lag
`times in Table 1. Fig. 1 shows the relationship
`between the logarithms of the permeability coeffi-
`cients and the alkyl chain lengths of the alcohols. An
`unusual pattern is observed with minimum per-
`meability coefficient values at intermediate alkyl
`chain length.
`
`Table 1. Nail plate permeability data for water and
`n-alkanols.
`
`Permcability~
`Lag. time
`coefficient
`(h)
`Permeant
`(cm h 1 x 10~)
`(s)
`Water
`16.5
`_+ 5.9 (6)
`900 + 100
`Methanol
`5.6
`_+ 1.2 (26) 1790 _+ 200
`Ethanol
`5.8 + 3.1 (8)
`27311 + 200
`n-Propanol
`0.83 + 0.15(4)
`4020 + 350
`n-Butanol
`0.61 + 0.27 (4)
`3470 + 350
`n-Pentanol
`0.35 + 0.07 (6)
`27110 _+ 250
`n-Hcxanol
`0.36 _+ 0.23 (51
`3540 = 3(1(I
`n-Heptanol
`0-42 _+ 0.12 {41
`2520 _+ 300
`n-Octanol
`0-27 + 0-03 (4)
`2120 +_ 150
`n-Decanol
`2-5 4- 1.7 (101 2090 + 150
`n-Dodecanol 4.1 4- 2-7 (8)
`23(10 + 150
`
`Effective
`diffusionb
`constant
`(D,.,) cm2 s
`x 10v
`5.4
`2.7
`1.3
`1.2
`1.4
`1-8
`1.4
`1.9
`2.2
`2.1
`2-1
`
`a. Data includc standard deviation and ( ) number of
`experiments.
`
`hz
`b. From t~ = 6D- (Mean value for h = (1-54 mm)
`
`Fig. 2 shows the effective diffusivities of the
`permeants in the nail plate tissue as a function of
`alkyl chain length,
`
`Theoretical considerations
`The nail plate’s barrier properties are governed by its
`anatomical construction and its physicochemical
`properties and a proposed model mnst be support-
`able in terms of both. The model developed here,
`although speculative, fulfills these requirements.
`The plate consists of a laminate of sheets of
`keratinized cells (Caputo & Dadati 1968; Forslind
`1970; Forslind & Thyresson 1975). Like the stratum
`
`2 of 6
`
`

`
`NAIL PERMEATION BY
`
`HOMOLOGO US ALCOI tOLS
`
`31
`
`lipid domain. The quantities hc and hM are the
`summed thicknesses of the cytoplasmic laminae and
`membrane lamellae passed through transcellularly.
`Added together these yield the total nail plate
`thickness. Common values of diffusivity (DM) and
`partition coefficient (KM) are given for the lipid
`elements of the two distinct routes but the 5"hM<<hl.
`even in the absence of a significant tortuosity factor.
`Finally, for all but the most polar permeants,
`DyaKMYh¢>>DcKcYhv as D~.<DM (likely) and
`Zh~a< <’Phi,. Therefore:
`
`Equation 5 gives the flux per unit area for typical
`permeants in terms of physically meaningful mass
`transfer parameters. The bracketed quality is a
`complex mass transfer coefficient or ’permeability
`coefficient’. Using the symbol P for the permeability
`coefficient, the statement J/A = PAC. applies
`generally for such mass transfer systems. The opera-
`tive parallel pathways are indicated by the two
`separate coUections of terms comprising P.
`The permeability coefficient profile for a homolo-
`gous series of permeants will depend upon how the
`diffusivities and partition coefficients in eqnation 5
`are affected by variation of length of the alkyl chain.
`It is impossible to predict how Dc and K~, might be
`affected as very little is known about solubility and
`molecular mobility in the nail’s dense, scmicrystal-
`line protein phases. Based on general considerations
`(Flynn et al 1974) and on partitioning behaviour of
`long chain fatty acids between the naiI plate and
`water (Baden 1970), KM may be assumed to follow
`the general o!w homologue partitioning pattern:
`
`log Kxl,o - log KM,o + =n
`
`(6)
`
`where K~.,, is the partition coefficient of the homo-
`logue of chain length n between a water immiscible
`phase and water and log K~.o is the Y-intercept of a
`plot KM.,, versus n. The term. =. is the slope of the
`log K versus n plot. The nature of the intercellular
`material is such that ~>0.3 and therefore KM.n can
`be expected to grow in an exponential fashion as the
`alkvl chain is extended.
`
`Relative permeabifin’ o1" rite n-alkanols through nail
`plate and stratum corneum
`The alkvl chain lenoth dependencv of permeabilitv
`
`of the n-alkanols (including water at n = 0) is shown
`in Fig. 1. From water to n-octanol permeability
`coefficients decrease systematically and by an overall
`factor of about 60. This observation rules out the
`possibility that a lipid pathway is involved for these
`permeants. Over the same alkyl chain length span
`diffusivities are also decreasing, but only several-
`fold. The lag time based diffusivities are on the order
`10-7 cm2 s 1 a magnitude which is physically
`plausible.
`A nominal molecular size sensitivity for diffusivity
`is evident considering the narrow spread in apparent
`D~ (Deffccm.e) values. Thus, there must be another
`cause for the decline in the permeability coefficients
`with increasing chain length. According to equation
`5 partitioning provides the only alternative basis for
`the declining trend: it would be necessary for the
`partition coefficients between the protein domain of
`the nail (keratin) and the external water to decrease
`about 25-fold. To our knowledge there is no
`precedent for such behaviour in a mass transfer
`framework under circumstances where the external
`media are aqueous. There are, however, some
`observations supporting the concept that the keratin
`matrix has a decreasing ability to dissolve the
`alkanols as the homologous series is ascended.
`Tillman & Higuchi (196i) note that the solvating and
`softening abilities of solvents for callus strips are in
`the o~der water > methanol >> ethanol. A great
`deal of work has been done on the sorption of
`solvents into hair and wool fibres and, according to
`Harrison & Speakman (!958), the fine structure of
`wool seems to be inaccessible to molecules larger
`than n-propanol. On the basis of the present work, it
`would appear that exclusion associated with increas-
`ing hydrophobicity is more a thermodynamic than a
`kinetic (molecular size) phenomenon. Hair (wool),
`callus and nail seem to have more in common
`chemically and physically with each other than they
`do with the stratum corneum (Baden 1970; Baden ct
`al 1973) and thus inferences drawn from the cited
`works have good probability of being applicable to
`the nail. Furthermore, at 10 7 cm2 s i the effective
`diffusivities are too large for an approximately
`25-fold factor to be incorporated and hidden in some
`complex way. It therefore appears that, to a rough
`first approximation, the nail plate acts as a concen-
`trated hydrogel to the alkanol permeants through
`n-octanol. The behaviour suggests there is a positive
`free energy change accompanying the transfer of a
`methylene group from the external, water medium
`to the intracellular protein phase.
`Skin permeation of the alkanols throuc, h n-octanol
`
`3 of 6
`
`

`
`¢-
`
`E
`u
`v
`
`%
`
`y__
`
`25
`1:3
`
`E
`
`13_
`
`J
`
`2O
`
`10
`
`1.0
`
`0.1
`
`,,,~ --Water
`
`K. A. V~ AI,TERS ET AI_
`
`]
`
`T )
`I/
`
`J
`
`. /
`
`The above picture allows consideration of diffu-
`sion across the nail plate at its first level of
`organizational complexity. For a given permeant,
`the principle mass current may either pass directly
`through the cell units stacked upon one another and
`separated by the intercellular substance, or may flow
`mainly around the cell contents by way of the
`interconnecting, extra-cellular lipid network. The
`first of these possibilities involves alternating pas-
`sages through two distinctly different domains.
`Based on the lipid content of the nail being totally
`extracellular and 1% of the total volume and on
`individual cell dimensions of 30 um diameter (hexa-
`gonal) and 1 !tm thickness, the latter route would
`offer a fractional area for diffusion of 5 x l0-a. At
`this percentage composition and with these cell
`dimensions, the calculated width of the region
`between cells is approximately 100 A, in reasonable
`agreement with ultra-microscopic estimates (Zaias &
`Alvarez 1968; Hashimoto 1971b). Allowance has to
`be made for diffusional path of the extracellular
`II 21 31 4 5 61 I i 71 8 91 i ll0 111 112
`route to be tortuous, having a path length up to, but
`not exceeding, 15 times (1/2 cell diameter), the nail’s
`width.
`The following equation can be formulated to
`describe flux across the nail plate as described using
`the general principles outlined by Flynn et al (1974):
`
`Alcohol alkyt chain length
`
`FIG. 1. Relationship between the logarithm of the per-
`meability coefficient and alkyl chain length of the alcohol
`
`corneum, the cytoplasmic keratin mass is partially
`crystalline and partially amorphous. In section, thin
`lipid seams are seen to separate the cell layers. This
`lipid is from the original cell membranes and is
`apparently supplemented by intercellular deposition
`of so-called membrane coating granules during tire
`plate’s formation (Hashimoto et al 1966; I Iashimoto
`1971a,b).
`
`2 5
`dl
`
`u 4
`C~
`
`~×3
`~57
`
`wE
`r;-i u 1
`
`i ~ I I I
`t /
`I I I [ I
`1 2 3 4 5 67 8 9 10 11 12
`length
`Alcohol atkyl chain
`
`FIr3.2. Effecive diffusivities of the permcants as a function
`of alkyl chain length of the alcohol,
`
`~- LERe + 5"RM +
`
`AC
`
`(3)
`
`Here J is the total flux (mass/time), A, is the area of
`application, and f~ and ft. are the fractional areas
`available for the transcellular and lipid routes
`respectively. The terms Rc and R~4 are the summed
`resistances of the two types of lamina encountered
`transcellularly while RI. is the resistance of the
`extracellular lipid route. The term AC is the driving
`force for the mass transfer process measured as the
`concentration difference across the nail plate.
`Equation 3 can be made more explicit by including
`the estimated fractional areas and by defining the
`resistances in terms of effective thicknesses (h),
`diffusivities within phases (D), and partition coeffi-
`cients (K):
`
`J
`y 0-9995DcDMKcKM
`A ~ DMKMZhc +DcKcZhM +
`L
`
`(4)
`
`5 X 10-4DMKM -]A
`5
`hl.
`The subscript c is used to indicate the intracellular
`protein domain and the subscript M the extracellular
`
`C
`
`4 of 6
`
`

`
`32
`
`K. A. WALTERS EI’ AL
`
`is strikingly different (Blank 1964; Scheuplein 1965;
`Behl et al 1980; Durrheim et al 1980; Flynn et al
`1981). As the alkyl chain length is increased,
`permeability coefficients also increase, with the
`increase being essentially exponential past ethanol.
`Such behaviour signifies that the stratum corneum
`functions for the most part as a hydrophobic mem-
`brane to these permeants, with increased perme-
`ability coefficients the result of increased partition-
`ing into some critical hydrophobic phase within the
`horny structure. The stratum corneum’s far greater
`lipid content undoubtedly plays a role here, setting
`its membrane behaviour apart from that of the nail
`and perhaps other cornified tissues.
`The striking increase in nail plate permeability
`coefficients from n-octanol to n-dodecanol signals a
`change in diffusional mechanism. Now permeability
`is increasing with increased hydrophobicity of the
`permeants, an unmistakable indication of the
`emergence of a functionally lipid pathway. Equa-
`tions 4 and 5 were formulated with such a route in
`mind, namely a route through the intercellular
`seams, with the collection of terms,
`5 x 10-aDMKM/hL, representing the route’s diffusive
`contribution. Regardless of whether the placemcnt
`of the lipid pathway is anatomically proper, the
`route’s essential trait is an exponentially increasing
`distributioning with increased alkyl chain length, as
`described in equation 6. Even a fractionally minor
`lipid route will assume rate-controlling proportions
`at an appropriately long alkyl chain length providing
`no other competitive pathway has a partitioning
`sensitivity. The transition to lipid pathway control
`occurs with but a slight increase in effective diffusiv-
`ity based on the effective diffusivities of n-decanol
`and n-dodecanol. It would appear that the route is
`not tortuous as these effective diffusivities do not
`allow for great non-linearity in path. The appearance
`of this route so late on the alkyl chain length profile is
`consistent with its limited fractional area, which we
`estimate to be about 1/100 of that of the same
`pathway in stratum corneum. With a methylene unit
`partitioning factor (=) of t>0.3 as suggested by
`literature data (Baden 1970), the alkvl fragment
`necessary to place the permeation process into
`extracellular control would have to be five to six
`carbons longer than it would for the stratum cor-
`neum, and this is very close to what is experimentally
`observed. Assuming a fractional area of 5 × 10 a
`and otherwise using the permeability data in Table 1,
`the partition coefficient between the extracellular
`lipid substance and water necessary to account for
`dodecanol’s high permeability coefficient can be
`
`estimated, ie. P-~ fLDMKM/hM, and KM is about
`6000. This value is certainly not too large considering
`the extended length of the alkyl chain.
`
`It is thus evident that the nail plate as a membrane
`behaves in a manner widely different from the
`epidermis (stratum corneum). There are other dif-
`ferences. The absolute rate of water transpiration
`through the nail is faster than through the intact skin
`(-~10 times) and, if the rate is thickness-normalized,
`the ratio is approximately 1000 in favour of the nail.
`
`Over a wide polarity range nail plate permeability is
`inversely related to polarity while the reverse is true
`for the stratum corneum. The declining nail plate
`permeability appears related to decreasing affinity of
`the keratin matrix for the higher alkanols. Attempts
`to confirm this supposition by way of equilibrium
`partitioning were not entirely successful.
`At extrcme hydrophobicities (~>Cs) a new path-
`way for diffusion through the nail becomes evident.
`And from n-octanol to n-dodecanol equilibrium
`partition coefficients increased exponentially, a
`trend generally supportive of the lipid character of
`the route. A partition coefficient for n-dodecanol of
`131 _+ 24 was obtained. If it is assumed that this Ci2
`homologue is concentrated exclusively in the lipid
`domain, which overall is estimated to occupy
`approximately 0.01 volume fraction, then dodecan-
`ol’s intrinsic partition coefficient would be 13 000.
`
`This value is only a little over twice the value
`computed from the permeability coefficient using a
`fractional area for diffusion of 5 × 10-4. Given all
`uncertainties, this is in reasonable if not fortuitous
`accord.
`
`It seems likely to us that the alkanol permeability
`pattern of the nail plate reflects general nail behav-
`iour and thus suggests how other low molecular
`weight organics might permeate. If this supposition
`is true, then very polar compounds might be surpris-
`ingly easily delivered through the nail plate to
`underlying tissues. The fact that urea can be used to
`chemically loosen and separate the nail plate from its
`bed is a supporting observation (Farber & South
`1978). The low incidence of problems associated
`with the use of powerful hydrophobic organic
`solvents in nail laquer seems equally reinforcing.
`Certainly toxic and irritant properties of substances
`measured via patch tests on skin cannot be extrapol-
`ated to the nail. Moreover, physicochemical criteria
`governing the selection of therapeutic candidates to
`treat nail disorders would seem to be very different
`from the established criteria used for drug selection
`for the skin.
`
`5 of 6
`
`

`
`NAIL PERMEATION BY HOMOLOGOUS ALCOHOLS
`
`33
`
`REFERENCES
`
`Baden, H. P. (1970) J. Invest. Dcrm. 55:115-122
`Baden, H. P., Goldsmilh, L. A., Fleming. B. 11973)
`Biochim. Biophys. Acta. 322:269-278
`Behl, C. R., Flynn. G. L., Kurihara. T., Smith, W..
`Higuchi, W. I., rio, N. F. H., Pierson, C. L. 11980) J.
`Invest. Derm. 75:346-352
`Blank, I. H. (1964) Ibid. 43:415-4211
`Burch, G. E., Winsor. T (1946) Arch. Derm. Syph. 51:
`39-4l
`Caputo, R., Dadati, E. (1968) Arch. Klin. Exp, Derm. 231:
`344-354
`Durrheim, H., Flynn, G. L., Higuchi, W. I., Behl, C. R.
`(1980) J. Pharm. Sci.. 69:781-786
`Farber, E. M., South, D. A. (1978) Cutis 22. 689-692
`Flynn, G. L., Durrheim, II.. Itiguchi, W. I. (1981) J.
`Pharm. Sei. 70:52 56
`Flynn, G, L., Yalkowsky. S. H.. Roseman, T. J. 11974)
`Ibid. 63:479-5111
`Forslind, B. (1970) Acta. Derm. Venereol 50:161-168
`Forslind, B,, Thyresson, N. 11975) Arch. Derm. Forsch.
`254:199-204
`
`Harrison, D., Spcakman. J. B. (1958) Textile Rcs. J. 28:
`1005 1007
`Hashimoto. K. (1971a) Ultrastructure Res. 36:391-410
`Hashimoto, K. (1971b) Arch. Derm. Forsch. 240:1-22
`Hashimoto, K., Bernard. G. G.. Nelson, R., Lever, W. F.
`(1966) J. Invest. Derm. 47:2115-217
`Ho, N. F. H., Park, J. Y., Morozowich, W., ttiguchi, W. 1.
`11977) In: Roche, E. B. (ed.) "Design of Biopharmaceut-
`ical Properties through Prodrugs & Analogs American
`Pharmaceutical Association, Washinglon, pp 131",-227
`Hwang, S., Owada, E.. Yotsunganagi, T., Suhardja, L.,
`Ho, N. F. H.. Flvnn, G. L.. Hignchi, W. I. (1976) J.
`Pharm. Sci. 65:1i74-1578
`Scheuplein, R, J. 11965) J. Invest. Derm. 45:334-346
`Spruit, D. 11971) Ibid. 56:359-361
`Tillman, W. J.. Higuchi, T. (1961) J. Invest. Derm. 37:
`87-92
`Waiters, K. A., Flynn, G. L, Marvel, J. R. 11981) J.
`Invest. Derm. 76:76-79
`Zaias, N., Alvarez, J. 11968)J. Invest. Derm. 51:12/>136
`
`6 of 6