J. Plaann. Pharmacol. 1997, 49:866-872
`Received March 3, 1997
`Accepted June 2, 1997
`
`(~) 1997 J. Pharm. Pharmacol.
`
`In-vitro Permeability of the Human Nail and of a Keratin
`Membrane from Bovine Hooves: Prediction of the Penetration
`Rate of Antimycotics through the Nail Plate and their Efficacy
`
`DIRK !vIERTIN AND BERNHARD C. LIPPOLD
`
`Depattment of Pharmaceutical Technology, Heinrich-Heine-University, Universitdtsstr. l, D-40225 Diissehlorf
`Germany
`
`Abstract
`In contrast to the partition coefficient octanoI/water the molecular size of penetrating drugs has a noticeable
`influence on the permeability of the human nail plate and a keratin membrane from bovine hooves. The
`relationship between permeability and molecular weight is founded on well-established theories. The
`correlation between the pemaeability of the nail plate and that of the hoof membrane allows a prediction of
`the nail permeability after determination of the drug penetration through the hoof membrane.
`The maximum flux of ten antimycotics (amorolfine, bifonazole, ciclopirox, clotrimazole, econazole,
`griseo-*t-ulvin, ketoiconazole, naftifine, nystatin and tolnaftate) through the nail plate ~ on the
`basis of their penetration rates through the hoof membrane and their water solubilities. An effcacy coefficient
`against onychomycoses was calculated frmn the maximum flux and the minimum inhibitory concentration.
`Accordingly, amorolfine, ciclopirox, econazole and naftifine are expected to be especially effective against
`dermatophytes, whereas in the case of an infection with yeasts only, amorolline and ciclopirox are promising.
`
`The influence of molecular size on permeability was investi-
`gated in the present work, after previous studies had shown
`that drug penetration through the nail plate and a keratin
`membrane from bovine hooves is independent of the lipophi-
`licity of the diffusing substance (Mertin & Lippold 1997).
`Siucc the petmtration of non-electrolytes through biological
`membranes is similar to that through polymers, the diffusion
`mechanism has also been transferred (L]eb & Stein 1969).
`Ahhough there is no consistent theory about the diffusion
`in polymers, it is assumed that the thermal movement of
`the polymer chains creates holes which are occupied by the
`diffusing molecules (Kuminis & Kwei 1968; Lieb & Stein
`1969t. The penetration rate is limited by the formation fre-
`quency and the size distribution of these free volumes. On the
`other hand, these factors are influenced by the temperature, the
`nature of the polymer and the interactions of the polymer
`chains with each other and with the diffusing molecules.
`Transferred to biological structures, free volumes can be
`formed by separating lipid bilayers or proteins (Lieb & Stein
`1969).
`Cohen & Turnbull (1959) deduced an exponential
`relationship between the molecular volume VM of the diffusing
`particle and its diffusion coefficient D from statistical
`analysis of the lluctuations of the free volume in super cooled
`liquids:
`
`determine, it is combined with the partition coefficient bar-
`rier/vehicle PCB/v to the permeability coefficient P:
`
`P = Do ¯ e-I~v’~ ’ PC~/v
`
`{2)
`
`Taking the logarithm leads to:
`
`log P = log DO - 2.3/~03 VM q- log PCB!v
`
`(3)
`
`Since both nail plate and hoof membrane are hydrophilic gel
`membranes whose PCB/v is approximately unity (Mertin &
`Lippold 1997), equation 4 follows by combining the constant
`parameters:
`
`log P = k- fl’ ’ VM
`
`(4)
`
`resp. log P = k - t" ¯ MW
`
`where fl" is similar to fl’ and contains a factor which reflects
`the conversion of the molecular volume into the molecular
`weight (MW).
`On the basis of the diffusion of non-electrolytes in polymers,
`Lieb & Stein deduced an empirical equation which can also be
`transferred to biological membranes:
`
`D = DO . MW-~
`
`D = DO ¯ e-/~vM
`
`(I)
`
`resp. log D = log DO - z. log MW
`
`(6)
`
`(7)
`
`{8)
`
`where Do is tire diffusion coefficient of a hypothetical molecule
`with the mole volume of 0 and [3 is a reciprocal value for the
`average free volume (Potts & Guy 1993). As tire diffusion
`coefficient through a biological membrane is difficult to
`
`Correspondence: B. C. Lippold, Department of Phamaaceutical
`Technology, Heinrich-Heine-University, Universiditsstr. 1, D-40225
`Diisseldorf, Germany.
`
`If the PCB/v becomes unity, it follows:
`
`log P = k- z. log MW
`
`The parameter z is called mass selectivity coefficient which
`quantifies the sensitivity of the diffusion coefficient to altera-
`tions of the molecular weight of the diffusing compound. It
`ranges from 1.1 to 3.8 in plastics, from 2.9 to 6.0 in celt
`membranes and from 0.3 to 0.5 in liquids (Lieb & Stein 1969).
`
`ARGENTUM EX1023
`
`Page 1
`
`

`

`PENETRATION OF ANTIMCYOTICS THROUGH NAIL PLATE AND THEIR EFFICACY
`
`867
`
`The higher the value, the higher is the sensitivity to alterations
`of the molecular weight. The power function of Lieb & Stein
`(1969) often provides a satisfactory fit to the experimental
`data, but it is disadvantageous that the exponent z has no
`physical meaning (Potts & Guy 1993).
`However, the single consideration of the molecular volume
`or weight may lead to the wrong prediction of the diffusion
`coefficient. Investigations of the penetration of linear and
`branched paraffins through different polymers show that
`branching reduces the diffusion to a greater extent than an
`increase of the molecular volume (Flynn et al 1974). The size
`as well as the shape of the molecules is important. Due to
`taking the logarithm of the molecular weight, the equation of
`Lieb & Stein (Eqn 6) seems to be less sensitive to neglecting
`the molecular shape than the Cohen-Turnbull correlation
`(Eqn 1) (Flynn et al 1974).
`In this study, the relationship between the permeability of
`the nail plate or the hoof membrane, respectively, and the
`molecular weight of the penetrating substance has been
`investigated to enable the prediction of the nail penetration of
`potential antimycotics.
`Among the nail infections onychomycoses, i.e. infections by
`fungi, are predominant. As antimycotics, which seem to be sui-
`table for topical application, are expected to have low fluxes due
`to their slight water solubility, only their penetration through the
`hoof membrane was studied. The prospective maximum flux
`(JmaO of the antimycotics amorolfine, bifonazole, ciclopirox,
`clotrimazole, econazole, griseofulvin, ketoconazole, naftifine,
`nystatin and tolnaftate through the nail plate was calculated from
`their penetration through the hoof membrane and their water
`solubility. The efficacy of a topically applied antimycotic is not
`only influenced by the maximum flux but also by the antifungal
`potency, which is quantified by the minimum inhibitory con-
`centration (MIC). An efficacy coefficient E is calculated from
`Jmo~ and MIC, which predicts the topical effectiveness of an
`antimycotic against onychomycoses.
`
`Materials and Methods
`
`Chemicals
`Phosphate buffered saline pH 7.4 (Ph. Eur.) and, in the case of
`the antimycotics, a mixture of phosphate buffer pH 7.4 of a
`higher buffer capacity with ethanol (resulting ethanol con-
`centration 42% v/v) were used as media. Since the ethanol
`restrains the dissociation of phosphate, the pH value of the
`mixture is 8.1. The selection of the model compounds was
`reduced to water-soluble substances with the exception of the
`antimycotics.
`Paracetamol was obtained from Boehringer lngelheim
`(lngelheim, Germany), phenacetin and bifonazole from Bayer
`(Leverkusen, Germany), diprophylline from Knoll (Ludwig
`shafen, Germany), chloramphenicol and clotrimazole from
`Caesar & Lorentz (Hilden, Germany), iopamidol from Byk
`Gulden (Konstanz), methyl, ethyl, butyl and hexyl nicotinate
`were obtained from Aldrich-Chemie (Steinheim, Germany),
`octyl nicotinate from the Department of Pharmaceutical
`Chemistry of the University of Diisseldorf, Germany, amor-
`olfine from Hoffmann-La Roche (Basel, Switzerland), ciclo-
`pirox olamine and griseofulvin from Cassella-Riedel
`(Frankfurt, Germany), econazole nitrate from Cilag (Schaff-
`hausen, Switzerland), ketoconazole from Janssen (Beerse,
`
`Belgium), naftifine hydrochloride from Sandoz (Nurcmberg,
`Germany) and tolnaftate from Essex (Munich, Germany).
`HPLC-pure acetonitrile (Acetonitfil Chromasolv) and metha-
`nol (Methanol Chromasolv) were from Riedel-de Hahn
`(Seelze, Germany).
`
`Penetration studies
`The diffusion cells, the preparation of the nails and of the hoof
`membranes, the penetration studies, the analyses, the deter-
`mination of the solubilities and the calculation of the perme-
`ability coefficient P and of the maximum flux Jm,,x have
`already been described in an earlier publication (Mertin &
`Lippold 1997). The antimycotics as well as paracetamol,
`phenacetin and chloramphenicol were presented as saturated
`solutions in their maximum thermodynamic activity. The set-
`ting of the saturation concentrations was guaranteed by sus-
`pending and stirring a surplus of the drug at 32"C for 48 h. Due
`to their very high water solubility, diprophylline and iopamidol
`were able to be used as non-saturated solutions (hoof mem-
`brane: C=1000 mg L-I; nail plate: C=20000 mgL ~).
`With the antimycotics, the donor compartment consisted of the
`drug suspension in ethanol 42% (v/v), pH 8-1. The penetrating
`amount per time and area therefore represented the maximum
`flux. Due to its high solubility in the medium, ciclopirox was
`an exception: it could be dissolved completely in a con-
`centration of 1000 mg L-J. Since the antimycotic with the
`least molecular size had a mole mass of 207, homologous
`nicotinic acid esters served to cover the low molecular weight
`area which ranged from 140 to 230 in a donor concentration of
`1000 mg L 1. Ethanol 42% (v/v), pH 8.1 also served as the
`acceptor medium.
`
`Determination of the dissociation constants
`For the determination of the acid constants of the antimycotics,
`the potentiometric method of Albert & Serjeant (1984) was
`performed. Solutions (0.02-0.10mol) of the antimycotics
`were used due to their slight solubility. The pH values were
`recorded with two decimal places after each addition of the
`titrant at 32:t_1°C and the pK~ value was determined according
`to the Henderson-Hasselbalch equation. Since the titrations
`were carried out in ethanol 42% (v/v), the pH-meter (Digilal-
`pH-Meter 644, Knick, Berlin) with glass electrode (U
`402/165, lngold, Frankfurt) was calibrated with ethanol 42%
`(v/v) containing 0.001 mol benzoate, salicylate and ammo-
`nium buffer solutions. The corresponding pK~ values in etha-
`nol 42% (v/v) are 5.24 (benzoic acid), 3.62 (salicylic acid)
`(Grunwald & Bcrkowitz 1951) and 8.78 (ammonium chloride)
`(Gutbezahl & Grunwald 1953).
`
`Results and Discussion
`
`Permeability and molecular weight
`Table 1 shows the molecular weights (MW) and the perme-
`ability coefficients of the drugs, calculated from the con-
`centration increase in the aeceptor through the nail plate (PN)
`and the hoof membrane (PRO. The corresponding data of the
`homologous nicotinic acid esters (Mertin & Lippold 1997)
`were included in the analysis. Since the antimycotics were only
`investigated in ethanol 42% (v/v), pH 8.1, these results were
`analysed separately from those of the other substances. Figs 1
`and 2 show the correlation between the permeability coeffi-
`
`r
`
`L
`
`Page 2
`
`

`

`868
`
`DIRK MERTIN AND BERNHARD C. LIPPOLD
`
`Table 1. Molecular weights (MW) of antimycotics and other drugs and their permeability coefficients
`through nail plate PN and hoof membrane PH-
`
`MW
`
`PN
`(10-s cm2 S-1)
`
`PH, PH.EtOH resp.
`10_8
`em2 s-1
`
`Medium: aqueous phosphate buffer pH 7.4
`Paracetamol
`151.2
`Phenaeetin
`179-2
`Diprophylline
`254.3
`Chloramphenicol
`323.1
`Iopamidol
`777.1
`
`1.784-0-32
`1.405-0-47
`0.142-t-0-055
`0.1825-0.047
`0-010+0.002
`
`Medium: ethanol-containing phosphate buffer pH 8-1
`Amorolfine
`317-5
`Bifonazole
`310.4
`Cielopirox
`207-3
`Clotrimazole
`344-8
`Econazole
`381-7
`Griseofulvin
`352.8
`Ketoconazole
`531.4
`Naftifine
`287.4
`Nystafin
`926.1
`Tolnaftate
`307.4
`
`n.d.
`n.d.
`n.d.
`n.d.
`n.d.
`n.d.
`n.d.
`n.d.
`n.d.
`n.d.
`
`Results are presented as means 5-s.d., n=4. n.d., not determined.
`
`20-974-5.15
`20-784-5.05
`6.144-2.01
`9.014-2.61
`1.444-0.34
`
`2-03+0.25
`3.054-037
`2-134-0.65
`2-30_I_0.65
`3.374-1.20
`1.004-0-26
`0-844-0.18
`4-08+0-98
`0.104-0-02
`3.444-1.03
`
`-6
`
`-7
`
`0_ 8
`
`d -9’
`
`-10’
`
`-11
`oo
`
`.... . ,
`200 300 400 500
`600
`700
`800
`MW
`
`_71
`
`o_ -8
`
`I
`
`-11
`2.1 " 212 " 213 " 2~4 " 2’.5 " 2~6 " 2’-7" 2’.8 " 219
`Log MW
`
`FIG. 1. Relationship between the logarithm of the permeability
`coefficient P for the nail plate (A, ¯) or the hoof membrane (©, @)
`and the molecular weight (n = 3-8, means, s.d. see Fig. 3). Medium:
`phosphate buffer pH 7.4. P is expressed in cm2 s-] ©, A nicotinic
`acid esters; 1, ¯ remaining substances. Plot according to Cohen &
`Tumbull (1959) (Eqn 5): log PN = --7.296 ~0.003708 MW, r = 0.933
`(nail plate) or log Prl=--6.284 ~).002071 MW, r=0.920 (hoof
`membrane).
`
`FIG. 2. Relationship between the logarithm of the permeability
`coefficient P for the nail plate (~, ¯) or hoof membrane (©, 0) and
`the logarithm of the molecular weight MW (n = 3-8, means, s.d. see
`Fig. 3). Medium: phosphate buffer pH 7.4. P is expressed in cm2 s-l.
`©, ZX nicotinic acid esters; @, ¯ remaining substances. Plot according
`to Lieb & Stein (1969) (Eqn 8): log PN= --0.427 --3.341 log MW,
`r=0.981 (nail plate) or log PH= --2.635 -1.782 log MW, r=0.924
`(hoof membrane).
`
`cient and the molecular weight according to the theory of
`Cohen & Turnbull (log P vs MW) and Lieb & Stein (log P vs
`log MW), respectively, in the aqueous milieu pH 7.4, where
`the investigated substances were nearly undissociated (Table 1,
`upper part).
`There was a linear relationship with a negative slope
`between the permeability coefficient and the molecular weight
`for both the nail plate (generally lower P-values) and the hoof
`membrane. Although giving of the correlation coefficient r is
`only permitted for regressions of the second kind (x and y as
`random variables) (Documenta Geigy 1975), it was never-
`theless considered, as it simplifies a judgement of the rela-
`tionship. The correlation coefficients show that the plot
`according to Lieb & Stein (Fig. 2) was either equal or slightly
`superior to the Cohen-Tumbull plot (Fig. 1). The slopes of the
`nail-plate data and the hoof-membrane data differed in both
`
`correlations by a factor of 1.8 to 1.9. This meant that the
`permeability of the nail plate was about twice as sensitive to a
`change of the molecular size as that of the hoof membrane.
`Both results (the lower permeability, but higher slope in the
`case of the nail plate) could be explained by the denser net-
`work of the nail keratin matrix. This demanded that the
`molecules had to diffuse a longer way due to the greater pore
`tortuosity and the penetration rate was therefore reduced in
`general. On the other hand, the penetration rate in the pores
`was reduced by the increased friction between the diffusing
`molecules and the gel network, which meant that the radius of
`the solvated molecule (rs) became closer to the pore radius of
`the network (rp) (Flynn et al 1974). The close-meshed keratin
`network of the nail plate contains few pores in the order of
`magnitude of the larger diffusing molecules, which are hin-
`dered to a stronger extent than smaller ones. The higher dif-
`
`Page 3
`
`

`

`r
`
`10-7
`
`~- 10-8
`t
`
`%
`
`z lO-S
`a_
`
`PENETRATION OF ANTIMCYOTICS THROUGH NAIL PLATE AND THEIR EFFICACY
`
`869
`
`-7-0
`
`GD
`
`J
`-8.0
`
`._1
`
`9-0
`
`griseofulvin
`
`16-6
`
`"200" " "400’ " "660" " "860" " "1000
`MW
`
`10-1°
`~,/1~
`
`10-8
`
`16-7
`PH (cm2s-1)
`
`FIG. 3. Relationship between the permeability coefficient of the nail
`plate PN and the permeability coefficient of the hoof membrane Pn at
`32°C (n--3-8, means -4- s.d.). © nicotinic acid esters; ¯ remaining
`substances. Log PN=3.723 + 1.751 log PH, r=0.971.
`
`fusional resistance of the nail keratin cannot only be explained
`by the lower swelling in the aqueous milieu compared with
`that of the hoof membrane. With approximately 27% it was
`only slightly below the corresponding value of the hoof
`membrane (36%) (Mertin 1995). A distinct difference
`regarding the structure of the two barriers must be the reason.
`The permeabilities of both the nail plate and the hoof
`membrane derive from the molecular size of the drugs and can
`be therefore calculated. It is not necessary to have information
`about partition parameters as in the case of the stratum cor-
`neum; correlations using the molar volume instead of the
`molecular weight showed that this parameter has no advantage
`(Mertin 1995).
`
`Prediction of the nail permeability
`Since nail plates are only available to a certain extent for the
`preclinical development of topical drugs, it is of interest to
`calculate the expected permeability coefficient of the nail from
`a determined value using the hoof membrane model. Although
`the permeability coefficients of the nail plate and the hoof
`membrane differ from each other, it has been shown that the
`bovine hoof membrane may serve as an appropriate model for
`the nail, because both are hydrophilie gel membranes (Mertin
`& Lippold 1997). As the logarithm of the permeability coef-
`ficient represents the correlating parameter, the nail plate
`permeability of a drug can be derived directly from a plot of
`log P of the nail plate (log Pr4) vs log P of the hoof membrane
`(log Pn) after experimental determination of PH (Fig. 3). The
`drug permeability of the nail plate evaluated by this procedure
`should better correspond to the real value than the direct cal-
`culation using molecular weight, according to equations 5 and
`8, or the parameters in Figs 1 and 2, respectively, since the
`experimental determination of the penetration through the hoof
`membrane considers the characteristics of a substance (e.g.
`interactions with keratin) to a larger extent.
`The hoof membrane is therefore a suitable in-vitro model
`regarding the prediction of the permeability of the nail plate.
`The result is the following equation (Fig 3):
`
`log PN ---- 3.723 + 1.751 log PH
`
`(10)
`
`Penetration of the antimycotics and its prediction
`The permeability coefficients of the antimycotics through the
`hoof membrane in the ethanol-containing medium range from
`
`FIG. 4. Relationship between the logarithm of the permeability
`coefficient of the antimycotics through the hoof membrane PH and
`the molecular weight MW (n =4 means). Medium: ethanol 42% (v/v)
`pH 8.1. Pn is expressed in cm2s . © nicotinic acid esters;
`¯ antimycotics. Plot according to Cohen & Tumbull (1959) (Eqn 5):
`10g PH,EtOH = -- 6.795 -0.002427 MW, r = 0.931.
`
`7.0,
`
`o_m -8-0.
`o~
`
`....i
`
`-9.0.
`
`econazole
`
`¯ :’ ttt. "
`
`ClCoplrox ll~,~.,~.
`
`nystatin
`
`2.0
`
`¯ "2:2" " "2’.4’ " "2[6. " "218. ’ ’ 3"0
`Log MW
`
`IRG. 5. Relationship between the logarithm of the permeability
`coefficient of the antimyeotics through the hoof membrane PH and
`the logarithm of the molecular weight MW (n = 4 means). Medium:
`ethanol 42% (v/v) pH 8-1. PH is expressed in cm2 s 1. © nicotinic
`acid esters; ¯ antimycotics. Plot according to Lieb & Stein (1969)
`(Eqn 8): log Pr].ntoH= --2.224 2.181 log MW, r=0-915.
`
`0.10 x 10-s to 4-08 x 10-8 cm2 s- ] (Table 1, lower part and
`Figs 4 and 5, respectively)¯ The distinct decrease in the
`penetration rate to one-fourth or one-fifth compared with the
`pure aqueous milieu was a result of the de-swelling effect of
`the ethanol. The swelling of the keratin membrane decreased
`from 36% (m/m) to 27% (m/m) due to replacing water by
`ethanol 42% (v/v) (Mertin 1995). The correlation according to
`Cohen & Tumbull (Fig. 4) as well as according to Lieb & Stein
`(Fig. 5) again showed that there was a clear relationship
`between the permeability coefficient and the molecular weight.
`The correlation coefficients were similar to those determined
`in the pure aqueous milieu, but the Cohen-Turnbull plot
`seemed to have a slight superiority. For both correlations the
`regression coefficients of the respective straight lines of the
`data determined in ethanol 42% (v/v) were higher than in the
`case of the aqueous solutions: 0.002427 vs 0.002071 (Cohen &
`Turnbull) and 2.181 vs 1-782 (Lieb & Stein). This was also the
`consequence of the decrease of the membrane swelling in
`ethanol with a higher sensitivity of the permeability towards an
`
`Page 4
`
`

`

`870
`
`DIRK MERTIN AND BERNHARD C. L1PPOLD
`
`Table 2. Physicochemical and antimicrobial properties of the antimycotics.
`
`pKa.EtOH
`
`flEtOH
`
`C~,~
`
`C~7.4
`
`MICD
`
`MICy
`
`Amoroltine
`Bifonazole
`Ciclopirox
`Clotrimazole
`Econazole
`Griseofulvin
`Ketoconazole
`Naftifine
`Nystatin
`
`0.0316 (water)
`0-01) 10
`0.517
`0.0004
`0.0019
`
`6.6 (water)*
`5.11 -t-0.04
`8.074-0.05
`4.74-t-0.04
`5.38-t-0.04
`no acidic or basic groups
`5.20-t-0.10 0.0013
`0.0477
`6.80-t-0.03
`pK,~: about 4.0
`pKa2:7.73-I-0.03
`
`1.00
`0.299 (zwitter ion)
`0.701 (negative)
`
`9995
`0.35
`8590
`3.0
`1020
`10.4
`10.6
`8650
`
`8.8
`0-13
`1020
`2.7
`11.5
`10.1
`7.9
`2.9
`
`18.6
`
`18.5
`
`0.01
`0-1
`2.0
`2-3
`0-35
`3-1
`2-23
`0.55
`
`4-5
`
`0.55
`0.89
`2.0
`35
`100.0
`
`25
`50
`
`3
`
`Tolnaflate
`
`no acidic or basic groups
`
`0.07
`
`0-11
`
`0-55
`
`pKa.EtOH: dissociation constant in ethanol 42% (v/v) (n = 6-9, means 4- max. deviation), fief OH: degree of dissociation in ethanol 42% (v/v) at
`pH 8-1. C~w: water solubility at 32°C (means, n=2) expressed in mg L-1. C~74:solubility in phosphate buffer pH 7.4 at 32°C (means, n =2)
`expressed in mg L- i. MICD, MICy: MIC against dermatophytes or yeasts, respectively, calculated as the geometrical mean of the limits of the
`highest range given in the literature (Plempel & Stetter 1987; Wilson & Ryley 1990; McEvoy & Litvak 1993) expressed in mg L- !. *Hofmann-La
`Roche AG (1992). ED, EY: efficacy coefficients against dermatophytes and yeasts, respectively, taking into account the calculated maximum fluxes
`from water, expressed in cm s i.
`
`alteration of the molecule size due to the denser structure of the
`keratin filaments.
`Antimycotics, which differ to a larger extent from the
`regression line, are labelled in the diagrams. Ciclopirox,
`deviating in both plots, was dissociated at pH 8-1 to about 50%
`(Table 2) and was inhibited as an anion in its penetration
`through the negatively charged keratin membrane due 1o the
`Donnan equilibrium (Mertin & Lippold 1997). A similar
`argument can be applied to nystatin, which was present as an
`anion to 70%. Although griseofulvin had a high affinity
`towards keratin (ICI-Pharma 1981), its rather low permeability
`coefficient was probably not due to the sorption phenomenon.
`It rather represented, as did the deviation of econazole, a
`normal experimental error.
`Since the Cohen-Turnbull correlation led to a better adap-
`tation of the permeability coefficients and was theoretically
`better sustained than the Lieb-Stein plot, it was used in the
`following calculations to predict the penetration of the anti-
`mycotics through the nail plate. A direct calculation of the nail
`plate permeability according to Fig. 3 was not possible due to
`the different substances and media used. Combining the
`regression equations concerning the permeability of the nail
`plate in water (Eqn 11 ) and also the hoof membrane in ethanol
`42% (Eqn 12) resulted in equation 13 after transformation:
`
`log PN = -7.296 0-003708 MW
`
`log PH,EtOH = --6.795 -- 0.002427 MW
`
`log PN = 1.528 x log PH,EtOH -I- 3.085
`
`(11)
`
`(12)
`
`(13)
`
`According to equation 13, the permeability coefficients of the
`antimycotics through the nail plate in an aqueous medium
`could be derived from the experimental data in ethanol 42%
`(v/v). Taking the water solubility of the drug Csw (Table 2)
`into account, the maximum flux through the nail plate was
`calculated according to equation 14:
`
`PN
`J .... = ~. Csw
`
`(14)
`
`The values were standardized to a barrier thickness of
`hB = 1000 pm (Jmax(1000 ~m)). As information about the pH
`value in the nail or its buffer capacity was not available, the
`water solubility C~, instead of the solubility in phosphate
`buffer pH 7.4 was used.
`While the expected permeability coefficients of the various
`antimycotics through the nail only differed by a factor of
`100, the maximum fluxes ranged from 10-8 to
`10-3 mg cm-~ s-I (Table 3) due to the influence of the
`
`Table 3. Permeability coefficients PN and maximum flux Jmax(1000/~m) of the antimycotics through the nail
`plate and their predicted efficacy against dermatophytes ED and yeasts Ey, calculated from the experimental
`data (PH.EtOH) according to equations 13, 14 and 15.
`
`PN
`(cm2 s-l)
`
`Jmax( 1000 ,um)
`(mg cm-2 S-l)
`
`Amorolfine
`Bifonazole
`Ciclopirox
`Clotrimazole
`Econazole
`Griseofulvin
`Ketoconazole
`Naftifine
`Nystatin
`Tolnaftate
`
`2.15 x 10-9
`3-98x 10 9
`2.30 x 10 9
`2.59 x 10.9
`4.66 x 10 9
`7-27x 10 10
`5-52 x 10-m
`6.23 X 10-9
`2-16 x 10 11
`4.g0x 10-9
`
`2.15 x 10-4
`1.39x 10 8
`1-98 x 10 4
`7.77 x 10.8
`4.74 x 10 5
`7.56× 10 8
`5.85 × 10 s
`5-38 X 10-4
`4-02 × 10.9
`3.36x 10 9
`
`E~
`(cm S-1)
`
`2.15 x 10-z
`1-39x 10-7
`9-87 x 10.5
`3-38 x 10-~
`1-35 × 10-4
`2-44x 10-8
`2-62 x 10-~
`9-78 × 10-4
`8.93 x 10-10
`6.11 × 10 9
`
`Ey
`(cm s 1)
`
`3-91 x 10-4
`1-56x 10-~
`9-87 x 10.5
`2-22 x 10-9
`4-74 x I0-7
`-
`2-34 x 10.9
`1"08 X 10 5
`1.34 x 10.9
`_
`
`Page 5
`
`

`

`PENETRATION OF ANTIMCYOTICS THROUGH NAIL PLATE AND THEIR EFFICACY
`
`871
`
`solubility (Table 2). The maximum flux of the investigated
`compounds was influenced to a larger extent by their saturation
`concentrations than by their permeability coefficients. The
`high solubility of the antimycotic salts (amorolfine hydro-
`chloride, ciclopirox olamine, econazole nitrate and naftifine
`hydrochloride) in water, in contrast to buffer pH 7.4 (Table 2),
`resulted as expected in a high maximum flux. So the pene-
`tration inhibition, which the protonated molecules were sub-
`jected to in a keratin membrane, can be over-compensated by
`the solubility improvement (Merfin & Lippold 1997). The
`slightly-water-soluble bascs amorolfinc, econazole and nafti-
`fine profited from this, while ciclopirox already had a high
`basal solubility in water (about 500 mg L t). The calcnlaled
`maximum fluxes of about 10-7 mg cm-z s- ~ (clotrimazole,
`griseofulvin, ketoconazole) were of a medium size, whereas
`only small fluxes could be expected for bifonazole and tol-
`naftate due to their slight solubility, and for nystatin due to its
`high molecular weight. Assuming a buffered milieu in the nail
`(pH 7.4), the maximmn fluxes of amorolfine, econazole and
`naftifine strongly dccrcased, as they now were only slightly-
`water-soluble bases (Mertin 1995).
`
`Prediction of the topical efficacy of the antimycotics
`Not only the flux of thc drug through the nail plate and
`therefore the obtained concentration therein is of importance
`for the clinical success of a topical therapy against onycho-
`mycoses, but also the sensitivity of the fungi towards the
`antimycotic. Numerous investigations show that the antifungal
`in-vitro activity, whose measure is the minimum inhibitory
`concentration (MIC), does not always coincide with the inhi-
`bition in-vivo (Galgiani 1987; Wilson & Ryley 1990; Rex et al
`1993) as the determination of the MIC reacts in a very sensi-
`tive way to the experimental conditions.
`Due to the lack of other parameters, characterizing the
`activity of the antimycotics, the MIC should still lead to an
`estimation of the therapeutic efficacy. It is proportional to the
`drug concentration available at the site of action, which
`depends on the maximum flux, and to the reciprocal of the
`MIC. An efficacy coefficient E is therefore introduced, which
`should be maximum for high therapeutic effecliveness:
`
`E = Jmax( 1000 ltm)
`M1C
`
`(15)
`
`In the literature (Plempel & Stetter 1987; Wilson & Ryley
`1990; McEvoy & Litvak 1993), the MIC is usually given as a
`range and differs for varying strains. To obtain just one char-
`acteristic value, the geometrical mean was calculated from the
`lowest and highest value of the referenced range. As the MIC
`values for the different dermatophyte types on the one hand
`and the yeast types on the other hand do not differ considerably
`from each other, a unified value for dermatophytes and yeasts
`respectively was taken in order to simplify the calculation. In
`the case of contradictory values in the literature, the greatest,
`that is for the therapeutic success most adverse MIC, was
`chosen (Table 2). Dermatophytes are usually more sensitive
`towards antimycotics than yeasts; griseofulvin and tolnaftate
`do nol posses any activity against yeasts. Moulds were not
`taken into consideration, because they are only of secondary
`importance in the case of onychomycoses.
`The efficacy coefficient E (Table 2), which may estimate the
`therapeutic potency of an antimycotic against onychomyeoses
`
`caused by dermatophytes or yeasts, respectively, was calcu-
`lated on the basis of the maximum flux from water (Table 3). A
`high maximum flux through the nail plate, expected for sub-
`stances which possess a high water solubility (amorolfine
`hydrochloride, ciclopirox olamine, econazole uitrate and naf-
`tifine hydrochloride) resulted also in an elevated E value
`against dermatophytes (10-4-10-2 cm. s ~). The high in-
`vitro activity of bifonazole and tolnaftate (Table 2) compen-
`sated for their low flux only to a certain extent. With the
`exception of ciclopirox and nystatin the efficacy coefficients
`towards yeasts decreased due to the lower activity of the
`antimycotics against these micro-organisms. However, eco-
`nazole nitrate, naftifine hydrochloride and especially amor-
`olfine hydrochloride and ciclopirox olamine could be judged
`favourably relative to the other drugs. If the maximum flux was
`delermined by the drug solubility at pH 7.4, only amorolfine
`and ciclopirox showed a high activity against both classes of
`fungi, the former due to its low MIC and the latter due to the
`high solubility of the undissociated form. On the contrary,
`cconazole and naflifine are only indicated for infections by
`dermatophytes. The remaining antimycotics did not seem to be
`suitable for the topical treatment of onychomycoses compared
`with those previously mentioned. Thus, the investigations
`support the fact that among the studied drugs, only amorolfine
`and ciclopirox arc present in the market as topical preparations
`against onychomycoses. However, econazole and naftifinc
`hydrochloride are also potential candidates, if one succeeds in
`excluding the probable buffer effect of the nail interior and is
`therefore able to capitalize on their high solubility in water.
`Although the results should be interpreted with caution they
`suggest that by detemfining the penetration rate of an anti-
`mycotic through hoof membrane in conjunction with calcula-
`tion of the efficacy coefficient E, prediction of the
`effectivcness of an antifungal drug in the topical therapy of
`onychomycoses appears promising.
`
`References
`
`Albert, A., Serjeant, E. P. (1984) The Determination of Ionization
`Constants. Chapman and Hall, New York, pp 14-38
`Cohen, M. H., Turnbull, D. (1959) Molecular transport in liquids and
`glasscs. J. Chem. Phys. 31:1164-1169
`Documenta Geigy (1975) Wissenschaflliche Tabellen. G. Thieme
`Verlag, Stuttgart
`Flynn, G. L., Yalkowsky, S. H., Roseman, T. J. (1974) Mass transport
`phenomena and models: theoretical concepts. J. Pharln. Sci. 63:
`479 510
`GalgianL J. N. (!9871 AntifungaI susceptibility tests. Antimicrob.
`Agents Chcmother. 31 : 1867-1870
`Grunwald, E., Berkowitz, B. J. (t9511 The lneasurement and correla-
`tion of acid dissociation constants for carboxylic acids in the system
`ethanol-water. Activity coefficients and empirical activity functions.
`J. Am. Chem. Soc. 73:4839~,944
`Gutbezahl, B., Grunwald, E (1953) The effect of solvent on equili-
`brium and rate constants. II. The measurement and correlation of
`acid dissociation constants of anilinium and ammonium salts in the
`system ethanol-water. J. Am. Chem Soc. 75:559-574
`Hoflmann-La Roche AG (1992) Loceryl - Standard information fiir
`Krankenhausapotheker., Grenzach Wyhlen
`ICI-Pharma (1981) Fulcin S 500, Fulcin S., Plankstadt
`Kuminis, C. A., Kwei, T. K. (1968) Free volume and other theories. In:
`Crank, J., Park, GS. (eds) Diffusion in Polymers. Academic Press,
`London, pp 107-139
`Lieb, W. R., Stein, W. D. (1969) Biological membranes behave as non-
`porous polymeric sheets with respect to the diffusion on non-
`elcctrolytes. Nature 224:240-243
`
`Page 6
`
`

`

`872
`
`DIRK MERTIN AN