Available online at www.sciencedirect.com
`
`SCIENCE@DIAECT•
`
`ELSEVIER
`
`European Journal of Pharmaceutical Sciences 21 (2004) 471--477
`
`IUJIOPIAN JOUllNAL OP
`
`PHAR~IACEUTl~AL
`SCIENCES
`
`www .e lsev ier .com/locate/ej ps
`
`In vitro permeation of several drugs through the human nail plat1e .. : ~---,
`l.al)e
`relationship between physicochemical properties and nail
`EXHIBIT NO. 5
`permeability of drugs
`Yoichi Kobayashi a, Tsunehisa Komatsu•, Machiko Sumi a, Sachihiko Numajiri a-..•k•--~lili.J
`Misao Miyamotob, Daisuke Kobayashi a, Kenji Sugibayashia, Yasunori Morimoto•·'
`
`" Facuf(v i?{Pharmaceulical Sciences, Josai University, J-1 Keyakidai, Sakado, Saitama 350-0295. Japan
`b Nissan Chemical Co., ltd, 3-7-1 Ka11da-Nishiki-cho. Chiyoda-ku, Tokyu 101-0054, Japan
`
`Received I 0 April 2003: received in revised fonn ! 0 October 2003; accepted 17 November 2003
`
`Abstract
`
`The objectives of the present study arc to clarify the relationship between the physicochc1nical properties and the nail penncability of drugs
`through human nail plates. Ho1nologous p-hydroxybenzoic acid esters were used to investigate the relationship between the octanol/watcr
`partition coetficient and the penneability coeflicient of several drugs. The nail permeability was fOund to be independent of the lipophilicity
`of a penetrating drug. However, the nail pcnneabi\ity of several 1nodcl drugs was found to 1narkedly decrease as their 1nolecular weights
`increased. The nail permeability of an ionic drug was found to be significantly lower than that of a non-ionic drug, and the nail penneability of
`these drugs markedly decreased as their molecular \veights increased. The permeation ofa model drug, 5-0uorouracil (5-FU), through healthy
`nail plates was also dctcnnincd and coinparcd with that through nail plates with fungal infections. The drug permeation through a nail plate
`pecrcas_t;_Q _with an incr~?SC i.1.1. n~il_ pl_~tc thick1_1i;:S;5._Nail pl~t.es_with fl~pgµl iQfections .e~.J:tjbited approximately the S<!-lne 5-FU pcnneation as
`hcUlthy. iiU.ii" plates. We suggest thai the penneability of a drug is 1nalnly influencl!d by its molecular weight and penneability through nails
`with fungal infection can be esti1nated from data on healthy nail pcnneability.
`© 2004 Elsevier B.V. A!\ rights reserved.
`
`Klywords: Nail; Nail permeation; Fungal nail phttc; Onychomycosis; Antifungal agents
`
`1. Introduction
`
`Onychomycosis has been treated 111ainly with oral anti(cid:173)
`fungal medication (Piepponen et al., 1992; Vi liars and Jones,
`1992). This oral therapy, however, sometin1es has severe
`systemic side effects, which interrupt treatment (Wilson and
`Plunkett, 1962). On the other hand, it is well known that
`topical treatment is not \\l·idely used in onychon1ycosis ther(cid:173)
`apy. The anticipated low levels of nail penetration and per(cid:173)
`meation during topical antifungal drug exposure are very
`significant factors in onychotnycosis therapy. Only a few
`drug penneation studies have been performed on the hun1an
`·nail plate and, as a consequence, the mechanisn1s behind
`healthy and fungal nail permeation have yet to be confirmed.
`Walters et al. ( 1983) have suggested that the nail plate be-
`
`~ Concsponding author. Tel.: +81-49-271-7685;
`fax: +81-49-285-5863.
`E-mail address: morimoto((_&josai.ar:.jp (Y. Morimoto).
`
`0928-0987/$ - sec front matter~) 2004 Elsevier I3.V. All rights reserved.
`doi: 10. ! 016/j.cjps.2003.11.008
`
`haves like a hydrophilic gel membrane and that there is an
`additional lipophilic route, as revealed by an in vitro pen(cid:173)
`etration study of homologous alcohols through the human
`nail plate. Mertin and Lippold ( 1997) also suggested that
`the nail plate behaves like a hydrophilic gel 1ne111brane and
`that the dissociation of a penetrating drug leads to a re(cid:173)
`duction in the penetration rate, as revealed by an in vitro
`penetration study of homologous nicotinic acid esters, ben(cid:173)
`zoic acid and pyridine through the hu1nan nail plate and
`bovine hoof membrane. Since they used healthy nails fron1
`dead men and women for the in vitro pern1eation study, they
`were probably not able to cany out n1any permeation stud(cid:173)
`ies using human nail plates. We have developed a modified
`side-by-side diffusion cell using nail tip pieces fro1n healthy
`volunteers in order to investigate the nail penetration niech(cid:173)
`anisn1 and enhancing systen1. Fron1 the results we obtained
`using our in vitro technique, we suggest that dn1g diffu(cid:173)
`sion in the upper layer of the hun1an nail plate is the niain
`barrier to nail permeation and that the full-thickness nail
`
`CFAD v. Anacor, IPR2015-01776, CFAD EXHIBIT 1076 - Page 1 of 8
`
`

`
`472
`
`Y. Kobaya,;hi el al. I European Journal of Pharmace11/ical Science:; 2 J (2004) 471-477
`
`plate behaves like a hydrophilic gel membrane (Kobayashi
`et al., 1999). In addition, we found that N-acetyl-t-cysteine
`and 2-mercaptoethanol are able to enhance drug permeation
`through the human nail plate (Kobayashi et al., 1998). How(cid:173)
`ever, it is very difficult to evaluate antifungal drug perme(cid:173)
`ation, because of the ve1y low nail pern1eability involved.
`In order to clarify the nail permeation mechanism of drugs
`through human nail plate, we studied the relationship be(cid:173)
`tween the physicochemical properties and the healthy nail
`pern1eability of several drugs. To observe the effect of oc(cid:173)
`tanol/water partition coefficients on nail Permeability coeffi(cid:173)
`cients, homologous p-hydroxybenzoic acid esters were used
`as model drugs. The effects of the molecular weight and drug
`dissociation on nail permeability were also investigated in a
`nail permeation study using several model drugs. We inves(cid:173)
`tigated the relationship between the flux of a model drug,
`5-ftuorouracil (5-FU), and the healthy or fungal nail plate
`thickness.
`
`nail plates were used to make a co1nparison between the
`different drug permeabilities. After healthy and fungal nail
`pieces were hydrated for a day, they were used to evaluate
`drug permeability.
`
`2.3. Determination oj'solubilities and octanol!water
`partition coefficients
`
`The drug suspensions \Vere mixed with a magnetic stin·er
`at 37°C. After 12 and 24h, each suspension was subjected
`to filtration (Ekicrodisc 3; German Sciences Japan, Ltd.,
`Tokyo). The filtrate was immediately diluted with methanol
`or acctonitrile to obtain samples for analysis. No difference
`in drug solubility was observed between 12 and 24h. The
`octanol/water partition coefficient of the drugs (K0 w) was
`defined as the solubility ratio in octanol/water at 37"C. A
`few of the values were taken from the literature (Morimoto
`et al., 1992).
`
`2. Experimental
`
`2. 1. Materials
`
`Antipyrine, 5-FU, p-hydroxybenzoic acid esters (methyl
`ester, MP; ethyl ester, EP; propyl ester, PP; butyl ester,
`BP; amyl ester, AP; hexyl ester, HP) and sodium nicotinate
`were obtained from Tokyo Kasei Kogyo Co. (Tokyo, Japan).
`Aminopyrine, barbital sodium, benzoic acid, ethanol, pro(cid:173)
`caine hydrochloride, pyridine and sodium benzoate were ob(cid:173)
`tained fro.m Wako Pure Chemical Industries (Osaka, Japan).
`Deuterium oxide was obtained fro1n Merck Co. (Darmstadt,
`Gem1any). lsoproterenol hydrochloride, lidocaine, lidocaine
`hydrochloride and mexiletin hydrochloride-were obtained
`from Sigma Chemical Co. (St. Louis, MO, USA). Crocona(cid:173)
`zole hydrochloride was supplied by Shionogi & Co. (Osaka,
`Japan). Isosorbide dinitrate was supplied by Toko Pharma(cid:173)
`ceutical Ind. Co. (Tokyo). All other reagents were obtained
`fron1 commercial sources.
`
`2.2. Preparation of the nail plate
`
`I-lealthy nail tip pieces were obtained from the fingers and
`toes of healthy volunteers (15 1nales and 5 fen1ales; 1nean
`age 25 years, range 20-45) using nail clippers. Nail pieces,
`which had been allowed to grow for at least one month,
`were used in this permeation study. Fungal nail plates
`were supplied from Saiseikai Central Hospital (Tokyo) and
`Saitama Medical Center (Saitama Medical School, Saitama,
`Japan). The thickness of the healthy and fungal nail pieces
`was measured with a rnicron1eter (Mitutoyo Corp., Japan)
`equipped with pointed metal attachn1ents. l-Iealthy nail
`plates, having a thickness of about 400 µm (350--450 µm),
`were used to evaluate the effect of the octanol/water par(cid:173)
`tition coefficient, molecular weight and dissociation of the
`penetrant on the nail permeability. Both fungal and healthy
`
`2.4. Permeation studies
`
`A piece of nail plate was sandwiched between two
`adapters n1ade of polypropylene with an 0-shaped ring
`(effective diffusion area 0.049 cm2) and mounted in a
`side-by-side diffusion cell ( 1.5-2.5 ml) with a water jacket
`connected to a water bath at 37°C (Kobayashi et al., 1998).
`The dorsal nail plate side was filled with a drug suspension
`or solution (almost all dn1gs were applied as suspensions,
`but for drugs with a high solubility, the dorsal nail plate side
`was filled with drug solution. The maximun1 flux was eval(cid:173)
`uated frotn the {lux obtained with the applied concentration
`and drug solubility. In both cases, the drug concentration in
`the dorsal nail plate side was almost constant throughout the
`nail permeation studies) and the ventral nail plate side was
`filled with distilled water. No preservative was added be(cid:173)
`cause the receiver solution was clear even at the end of the
`experin1ent. Drug permeation was n1easured by sampling
`the solution on the ventral nail plate side at pre-determined
`tin1es. The experimental period was 5-17 days, because of
`the low degree of nail permeability of the drngs used.
`
`2. 5. Analytical 1nethods
`
`Deuterium oxide was quantified from the intensity of
`·the 0-D stretching vibrational band at 2512 cm- 1 with
`an infrared spectrophotometer (260-30, Hitachi, Tokyo)
`(Ilatanaka et al., 1993). Ethanol was measured by GC
`as described previously (Kobayashi et al., 1997). Other
`dn1gs were determined by l·IPLC. Sample solutions were
`injected
`into an l-IPLC consisting of a pump system
`(LC-IOA, Shimadzu Seisakusho, Kyoto, Japan), an UV
`detector (SPD-IOA, Shimadzu), a chromatopack (C-R5A,
`Shimadzu), a system controller (SCL-IOA, Shimadzu), an
`auto injector (SIL-JOA, Shi1nadzu), and a reverse-phase
`column (lnertsil ODS 250 mm x 4.6 mm i.d., GL Sci(cid:173)
`ences Inc., Tokyo). The 1nobile phase, which consisted
`
`CFAD v. Anacor, IPR2015-01776, CFAD EXHIBIT 1076 - Page 2 of 8
`
`

`
`Y. Kobayashi et al. I European Journal of Pharmace11tical Sciences 21 (2004) 471-477
`
`473
`
`of methanol/0.1 o/o phosphoric acid or acetonitrile/0.1 %
`phosphoric acid mixtures, with or without ion-pair chro(cid:173)
`matography reagents (sodiutn l-hexanesulfonate or sodiun1
`dodecylsulfate), was pumped at ~ow rate ranging from I to
`1.5 ml/min. No other peak apatt from the drug and inter(cid:173)
`nal standard was observed on the J-IPLC recording. It was
`clear that the pern1eating dn1gs were stable and no n1aterial
`leached from the nail plates during the permeation studies.
`
`3. Results and discussion
`
`3.1. Influence o.f the octano/!1vater partition coefficient on
`the permeability coefficient
`
`The steady-state permeation of drugs
`the
`through
`solution-diffusion membrane is characterized by Fick's law:
`
`l=
`
`DmKmCv
`h
`
`(!)
`
`in which J is the steady-state flux, Km is the membrane/donor
`vehicle partition coefficient of the drug, Dm is the diffusion
`coefficient of the drug in the 1nembrane, Cv is the concen(cid:173)
`tration in the donor solvent and h is the men1brane thick(cid:173)
`ness. The permeability coefficient (P) can be characterized
`as follows:
`
`P= DmKm.
`
`"
`
`(2)
`
`To predict the skin.permeability of drugs, Potts and Guy
`( 1992) carried out their analysis using a mathen1atical 1nodel
`based on pern1eant size (molecular volun1e (MV) or molec(cid:173)
`ular weight (MW)) and octanol/water pa1tition coefficient
`(K0 w ). The functional dependence of Dm on MW is expo(cid:173)
`nential and it can be characterized by:
`
`Dm = D 0 exp(-,8 MW)
`
`(3)
`
`900
`
`Cu1nulative amount
`
`800
`i 700
`-2r, 600
`2
`"' 500 M
`§ 400
`" JOO
`~ ;; 200
`100
`
`~
`
`0
`
`0
`
`where D0 represents the diffusivity of a hypothetical
`molecule having zero molecular weight and fi is a constant.
`The relationship between Km and Kow is better expressed as:
`
`Km = [Kow]/
`
`(4)
`
`where the coefficient,/, accounts for the difference between
`the partitioning domain presented by octanol and that pre(cid:173)
`sented by the membrane lipids. A combination of Eqs. (2),
`(3) and (4) yields Eq. (5):
`log P = Jog(D0 / h) +flog Kow -
`
`fl' MW
`
`(5)
`
`where ,B' = ,B/2.303,
`The effect of the octanol/water partition coefficient on
`the permeability coefficient of each drug was investigated.
`Fig. I shows a typical permeation profile of one of the
`model drugs, p-hydroxybenzoic acid n1ethyl ester, through
`the healthy human nail plate. Steady-state fluxes of homol(cid:173)
`ogous p-hydroxybenzoic acid esters can be observed from
`4.5 to 6.5 days allowing calculation of the nail permeability
`coefficients.
`Table I shows the physicochen1ical paran1eters and nail
`permeability coefficients of homologous p-hydroxybenzoic
`acid esters used as model drugs in this investigation. The
`molecular weights of homologous p-hydroxybenzoic acid
`esters covered a narrow range from 152.12 to 222.28. The
`logarithm values of the octanol/water partition coefficients
`varied widely from 1.53 to 4.25.
`Fig. 2 shows the relationship between the octanol/water
`partition coefficient and
`the penneability coefficient
`of p-hydroxybenzoic acid esters. The pern1eability of
`p-h}idroXybenzoic acid est6rs did not increa-se with an in(cid:173)
`crease in lipophilicity. If a nail plate behaves as a lipid par(cid:173)
`tition membrane, the slope([) should be clearly greater than
`O; but it was nearly zero (f = -0.160). Multiple regres(cid:173)
`sion analysis was carried out to clarify which paran1eters
`(molecular weight or octanol/water partition coefficient)
`contribute to the permeability of p-hydroxybenzoic acid
`
`Flux
`
`180
`
`150
`
`':;:;
`0
`
`120
`
`" E
`u "' 90
`2 ,
`0 c
`~ ::;
`
`60
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`Time (day)
`
`•
`
`30
`
`0
`
`0
`
`2
`
`3
`
`4
`
`6
`
`7
`
`Time (<lay)
`
`Fig. \. Typical permeation profile of p-hy<lroxybenzoic acid methyl ester:; through the healthy human nail plate. Each value represents the mean± S.E.
`(!! =4).
`
`CFAD v. Anacor, IPR2015-01776, CFAD EXHIBIT 1076 - Page 3 of 8
`
`

`
`474
`
`Y. Kobayashi et al. I European Journal of Pharmace11/ical Sciences 21 (2004) 471-477
`
`Table I
`Physicochcmical parameters and nail permeability coefficients (h
`400 µm) of p-hydroxybcnzoic acid esters
`
`Drug
`
`MW
`
`logKowa
`
`pb
`
`in the human nail plate has been found to be 1nuch lower than
`in the stratum corneum of skin (Walters and Flynn, 1983).
`Thus, there does not appear to be an additional Iipophilic
`route in the hun1an nail plate.
`
`152.15
`
`1.53
`
`166.18
`
`2.23
`
`180.20
`
`2.75
`
`194.23
`
`3.13
`
`p-Hydroxybenzoic acid
`methyl ester (MP)
`p-Hydroxybcnzoic acid
`ethyl ester (EP)
`p-Hydroxybenzoic acid
`propyl ester (PP)
`p-Hydroxybenzoic acid
`butyl ester (BP)
`p-Hydroxybcnzoic acid
`amyl ester (AP)
`p-Hydroxybcnzoic acid
`hcxyl ester (HP)
`Each value represents mean ± S.E. (11 = 4).
`" K0w: octanol/watcr partition coefficient.
`b P: permeability cocllicicnt (x 107 cm/s).
`
`208.25
`
`3.65
`
`222.28
`
`4.25
`
`3.68 ± 0.08
`
`2.43 ± 0.48
`
`2.01 ± 0.35
`
`2.38 ± 0.32
`
`2.24 ± 0.39
`
`1.24 ± 0.32
`
`3.2. Influence of the molecular lveight and dissociation on
`the permeability coefficient
`
`Nail permeability was found to be independent of the
`lipophilicity of the drug used. From the findings in this early
`study, the coefficient,}; in Eq. (5) can be assumed to equal
`zero. Thus, substitution of .f = 0 in Eq. (5) gives a sin1ple
`1nodel as follows:
`
`log P =log ( ~0)- fl' MW.
`
`(6)
`
`We investigated the nail permeability of several model
`drugs having various molecular weights. Because the influ(cid:173)
`ence of drug dissociation on the nail permeability is not
`clearly understood, the nail pern1eability of non-ionic and
`ionic drugs was evaluated separately.
`Table 2 su1nn1arizes the pKa, pl-I of the donor solution,
`n1olecular weight and penneability coefficient of the model
`drugs used in this investigation. All the model drugs were
`in the non-ionic form in the donor solution. The 1nolecular
`weight of the model drugs ranged from 20 to 240. The nail
`penneability of drugs having a molecular weight of above
`240 could not be determined, because of the low nail perme(cid:173)
`ability. The nail penneation data of p-hydroxybenzoic acid
`esters (pK, = 8.4, donor pl-I = 3.8-5.8) were added to
`this investigation because they exist in non-ionic forn1 in the
`donor solution.
`-
`According to Eq. (6), Fig. 3 shows the relationship
`between the pern1eability coefficient and the niolecular
`weight of each non-ionic drug. The logarithm of the nail
`penneability coefficient decreased as the molecular weight
`of the penetrating drug increased. A linear relationship
`(r = -0.860, P < 0.01) existed between the permeability
`coefficient and the molecular weight of the model drug,
`the slope (fl') and the intercept (log(D0!h)) being 0.00856
`
`Table 2
`Physicochcmical parameters and nail permeability coefficients (h
`400 µm) of non-ionic model drngs
`
`Drug
`
`pK"
`
`Deuterium oxide
`Ethanol
`Pyridine
`£3cnzoic acid
`5 -Fl uorourncil
`Antipyrine
`Aminopyrine
`Lidocaine
`lsosorbide dinitrate
`
`5.19
`4.19
`8.0, 13.0
`1.50
`5.00
`7.92
`
`rH"
`
`7.81
`7.43
`7.35
`3.21
`4.65
`6.37
`8.44
`10.21
`5.72
`
`MW
`
`20.0
`46.l
`79.1
`122.I
`130.1
`188.2
`232.3
`234.3
`236.1
`
`b
`Pnon
`
`45.52 ± 4.30
`19.81 ± 2.21
`6.36 ± 0.40
`12.84 ± 0.05
`2.08 ± 0.13
`0.53 ± 0.07
`0.09 ± 0.02
`0.39 ± 0.14
`1.5 l ± 0.29
`
`Each value represents mean± S.E. (5-FU, 11 == 30; other drugs, 11 == 3-4).
`" The pH in the donor solution.
`b /',w 11 : perme<1bility coefficient (x 107 cm/s) of non-ionic drugs.
`
`esters. It was suggested that the molecular weight makes
`a greater contribution to the permeability coefficient than
`the octanol/water partition coefficient. The F-values of the
`molecular weight or octanol/water partition coefficient were
`1.9399 or 0.1058, respectively.
`From our findings, it was evident that nail plates behave
`as a hydrophilic gel rne1nbrane rather than a lipophilic parti(cid:173)
`tion membrane. Although this suggestion agreed with the in(cid:173)
`vestigations of Walters et al. (1983) and Mertin and Lippold
`( 1997), an additional lipophilic route suggested by Walters
`et al. ( 1985) could not be found in the human nail plate. It
`has been reported that lipids are present in the dorsal and
`ventral plates, but only at very low levels in the intermediate
`plate which forms the main nail body (Jarrett and Spearn1an,
`1966; Kobayashi et al., 1999). In addition, the lipid content
`
`-4
`
`-5
`
`'.:? -6
`1
`'" ..5 -7
`
`0..
`
`-8
`
`2
`
`3
`Log Kow
`
`4
`
`5
`
`Fig. 2. Relationship between the permeability coefficient (P) and the
`octmtoJ/water partition coefficient (K,)w) of p-hydroxybcnzoic acid esters.
`Each value represents the mcan±S.E. (11 == 4). The abbreviations attached
`lO the closed circle (e) represent the model d1ugs in Table l.
`
`CFAD v. Anacor, IPR2015-01776, CFAD EXHIBIT 1076 - Page 4 of 8
`
`

`
`Y. Kobayashi et al. I European Journal of Phannar..:eutical Sciences 21 (2004) 471-477
`
`475
`
`-4
`
`-5
`
`-6
`
`':;;' " -7
`
`3-
`0.
`"' -8
`
`0
`..l
`
`-9
`
`-10
`
`-II
`()
`
`100
`
`200
`M.W.
`
`300
`
`400
`
`Fig. 3. Relationship between the permeability coefficient (P) and the
`molecular weight (MW) of non-ionic and ionic dn1gs. Each value repre(cid:173)
`sents the mean± S.E. (5-FU, n = 30; other drugs, 11 = 3-6). The dotted
`line represents the 95'Yo confidence interval of the regression line. The
`closed circle C•) represents the non-ionic drugs in Table 2. The open
`squurc (0) represents the ionic drugs in Table 3.
`
`and -5.260, respectively. This finding suggests that nail
`permeability depends on the molecular weight of the pen(cid:173)
`etrating dn1g, i.e. the drug diffusivity in the nail plate. The
`nail permeability of non-ionic drugs (P11011 ) can be pre(cid:173)
`dicted approxin1ately fron1 the linear regression (log ? 11011 =
`. -0 .. 00856 MW -,- 5 .260) and molecular weight of each .
`drug.
`Table 3 summarizes the pKa, pI-1 of the donor solution,
`molecular weight and pern1eability coefficient of the other
`model drugs. The model drugs existed in ionic form in the
`donor solution. The molecular weights of the ionic drugs
`used in this investigation ranged from 120 to 312.
`Fig. 3 shows the relationship between the permeability
`coefficient and the molecular weight of the ionic drug. The
`logarithm of the nail penneability coefficient decreased as
`
`Table 3
`Pbysicochemical parameters and nail permeability
`400 µm) of ionic model drugs
`
`coefficients (h
`
`Drug
`
`pKa
`
`pH a
`
`MWh
`
`l'illnc
`
`Sodium bcnzoate
`4.19
`8.12
`4.85
`7.21
`Sodium nicotinate
`7.91
`10.29
`Barbital sodium
`Mexiletin hydrochloride
`9.0
`4.87
`lsoproterenol hydrochloride
`8.57
`4.06
`Li<locaine hydrochloride
`7.86
`4.33
`Procaine hydrochloride
`8.8
`5.27
`Crocon<izole hydrochloride
`6.0
`1.96
`Each value represents mean ± S.E. (11 = 4-6).
`" The pH in the donor solution.
`b The molecular weight with an ionic t01111 of the drug.
`.c 1\m: pcnneability coctncient ( x 107 cm/s) of ionic drugs.
`
`121.1
`122.1
`183.2
`179.3
`211.2
`235.3
`237.3
`3 11.8
`
`0.910 ± 0.136
`0.606 ± 0.204
`0.135 ± 0.016
`0.202 ± 0.057
`0.084 ± 0.013
`0.031 ± 0.003
`0.110 ± 0.016
`0.017 ± 0.009
`
`the molecular weight of the penetrating ionic drug increased.
`A linear relationship (r = -0.966, P < 0.01) also existed
`between the permeability and the molecular weight of the
`ionic drug; the slope (fJ') and intercept (log(D0!h)) were
`0.01030 and -5.907, respectively. The nail permeability of
`ionic drugs (Pion) can also be predicted from the linear re(cid:173)
`gression (log P; 0 , = -0.0!030MW - 5.907) and molecular
`weight of the ionic form. Multiple regression analysis was
`carried out to clarify which para1neters (molecular weight
`or degree of dissociation) contribute to the permeability of
`model drugs. The degree of dissociation of the model dn1gs
`was calculated fron1 the pKa and pl-I in the donor solution.
`It was found that the molecular weight 111akes a greater con(cid:173)
`tribution to the permeability coefficient of model drugs than
`the degree of dissociation. The F-values of the 1nolecular
`weight or degree of dissociation were 10.9254 or 0.8599,
`respectively.
`The permeability coefficient (12.84 x 10- 7 cm/s) of the
`non-ionic form of benzoic acid, which is an acidic drug, was
`about I 0 times higher than that (0.91 x 10-7 cm/s) of its
`ionic form. The permeability coefficient (39 .31 x 1 o-9 c1n/s)
`of the non-ionic form of lidocaine, which is a basic dn1g,
`was also about IO times higher than that (3.10 x 10-9 cm/s)
`of its ionic fonn. In addition, the regression lines, which
`show the relationship between the permeability coefficient
`and the molecular weight of the non-ionic and ionic drugs,
`were parallel to each other. Furthern1ore, the regression line
`of the non-ionic form was about I 0 times higher in position
`than that of the ionic forn1, as shown in Fig. 3. These results
`make it clear that dissociation leads to a reduction in nail
`permeability,· irrespective of the charge on the dn1g. It is·
`thought that the decrease in the pe1n1eability of ionic drugs
`is caused by a small increase (about 100) in the apparent
`molecular weight due to ion hydration.
`In the investigation of Walters et al. ( 1983), they con(cid:173)
`cluded that there was little or no dependence of htunan nail
`permeability on the dissociation of miconazole. However,
`Mertin and Lippold (I 997) contradicted this conclusion be(cid:173)
`cause they found that the dissociation of benzoic acid and
`pyridine leads to a reduction in their penetration rate through
`bovine hoof membrane. They concluded that the decreased
`permeation caused by dissociation, is due to the Donnan
`effect or electrostatic repulsion between the keratin nlem(cid:173)
`brane and the diffusing niolecule. The conclusions that we
`drew from our investigation using hun1an nail plate agree
`with those of the latter researchers. The forn1er researchers
`used citrate/phosphate buffer solution to evaluate the ef(cid:173)
`fect of rniconazole dissociation on nail penneability. Their
`buffer solution has a high ionic strength and is composed of
`compounds exhibiting different ionic fo1n1s (mono-, di- and
`tri-ionic forms) with changing pl-I, co1npared with that (con(cid:173)
`stant ionic strength, /t = 0.158) used in the investigation
`of the latter group. The diffusion coefficient of the tri-ionic
`forms for citrate or phosphate is lower than that of the di-,
`mono-, and non-ionic forms (Southard et al., 1991 ). It seems
`that the decrease in pern1eability due to dissociation could
`
`CFAD v. Anacor, IPR2015-01776, CFAD EXHIBIT 1076 - Page 5 of 8
`
`

`
`476
`
`Y. Kobayashi et al. I European Journal of Phw·mace111ical Sciences 21 (2004) 471-477
`
`Table 4
`5-FU flux (J) through fungal nail plates from eight patients
`
`Patient number
`
`Age/gender
`
`I
`2
`3
`4
`5
`6
`7
`8
`
`52/fcmulc
`
`38/male
`42/fcmalc
`
`48/male
`30/male
`
`a h: nail thickness (µm).
`b J: flux (µ.g/cm 2/h).
`
`N'
`
`502
`570
`626
`665
`820
`930
`955
`I !58
`
`.fa
`
`6.25
`13.07
`15.4 l
`9.67
`5.93
`4.26
`3.2J
`0.51
`
`plate thickness. A linear relationship (r = 0.796, slope =
`0.642 µ,g/cm/h, intercept= -3.38 µ,g/cm 2/h, P < 0.01) ex(cid:173)
`isted between the 5-FU flux(.!) through the healthy nail plate
`and the reciprocal of the nail plate thickness (l/h) according
`to Eq. (1) (Fig. 4). In thick nail plates (800-1200µ,m), the
`5-FU flux through fungal nail plates was very similar to that
`through healthy nail plates. However, the 5-FU flux through
`thin fungal nail plates (50()-700 µ,m) tended to be a little
`higher than that through healthy nail plates. No significant
`difference (P = 0.05, Fisher's pairing !-test) was observed
`between the permeability-thickness products (P x h) of
`the healthy and fungal nail plates, calculated using Eq. (2).
`The mean thickness of the fungal nail plates used in this
`study was a little thicker than that of the healthy nail plates
`used. A fungal nail plate, particularly the deepest layer
`in the nail plate (ventral nail plate), generally becomes
`thicker than a"healthy nail plate (Sagher, 1948; Jillson and
`Piper, t 957). The 5-FU flux through the fungal nail plate
`can be estimated from a change in nail plate thickness in
`Fig. 5.
`
`JO
`
`D
`
`not be confirmed due to the interaction of various ions in
`the buffer solution. We did not use a buffer solution because
`it would have complicated the assessment of dn1g diffusiv(cid:173)
`ity in the nail plate. The diffusivity of rv-dicarboxylic acids
`decreased about 5% following complete ionization (Albery
`et al., 1967). In a comparison of the diffusivity of zwitteri(cid:173)
`onic glycine and neutral glycolamide, which have the sa~e
`n1olecular forn1ula, the diffusivity of the ionic compound is
`nearly I Oo/o less than that of the neutral con1pound (Flynn
`et al., 1974). We suggest that the decrease in permeability
`is caused by a decrease in diffusivity due to ion hydration
`rather than the Donnan effect or electrostatic repulsion be(cid:173)
`tween nail keratin and the penetrating drug.
`
`3.3. Comparison between healthy and fungal nail plate
`permeation of the drugs used
`
`Nail plates have different thicknesses in the fingers and
`toes of the hun1an body. To evaluate drug permeation through
`healthy and fungal nail plates, 5-FU was selected as a model
`drug. 5-FU was used because nail permeation can easily be
`detennined and because it is comparatively soluble in water
`( 17. l mg/ml). In this permeation study, 5-FU was suspended
`in all the donor vehicles.
`Fig. 4 shows the relationship between the 5-FU flux and
`the healthy nail plate thickness. 5-FU penetrates healthy nail
`plates of thickness ranging from 225 to 1050 µ,m at a flux
`of 1-28 µ,g/cm 2/h. The 5-FU flux increased as the nail plate
`thickness decreased.
`The 5-FU permeation through fungal nail plates from
`eight patients was investigated (Table 4) antl compared
`with that through nail plates from healthy volunteers. The
`5-FU flux through the fungal nail plate ranged from 0.5 to
`15.5 µg/c111 2 /h and tended to increase with a decrease in nail
`
`30
`
`D
`
`D
`B
`'e B D
`D D
`'1:i ~ D
`D
`
`'ti
`
`D
`
`D
`
`• D
`
`D
`IDo D
`D Ro
`D
`of-,~~~~~~~~~~
`0
`200
`400
`600
`800
`I 000 1200
`h (µm)
`
`Fig. 4. Relationship between the 5-FU flux and heallhy nail plate thickness
`(µm) (11 = 78).
`
`of--'-'"T--~~~~~~~~~
`20
`JO
`0
`10
`40
`50
`
`l/h (cm")
`
`Fig. 5. Comparison between the healthy and the fungal nail plutc !luxes of
`5-FU. The closed circle(•) represents fungal nail plate (11 = 8), the open
`square (0) represents healthy nail plate (11 = 78); h: nail thickness (cm).
`
`CFAD v. Anacor, IPR2015-01776, CFAD EXHIBIT 1076 - Page 6 of 8
`
`

`
`Y. Kobayashi et al. I European Journal of Pharmaceutical Sciences 21 (2004) 471-477
`
`477
`
`In this investigation, very heavy fungal nail plates could
`not be used for two reasons. Firstly, the thickness of the nail
`plate cannot be detem1ined because it is ve1y uneven. Sec(cid:173)
`ondly, the uneven nail plate collapses in water. The flux of
`drug through a very heavy fungal nail plate may be higher
`than that through a healthy nail plate because of nail de(cid:173)
`struction by fungi.
`We found that the healthy nail permeability depends on
`the diffusivity of penetrant. Since an increase in nail thick(cid:173)
`ness leads to a decrease in healthy or fungal nail flux, the
`fungal nail pern1eability may also depend on the diffusivity
`of penetrant. As a result, we suggest that the pern1eability
`through healthy and fungal nail plates is not significantly
`different and the fungal nail pern1eability can be estimated
`from healthy nail penneability data.
`
`4. Conclusion
`
`In the present study, we investigated the in vitro nail
`permeation of several model drugs. Nail permeability was
`analyzed using a simple model based on the octanol/water
`partition coefficient and the 1nolecular weight of the drug.
`Although nail permeability was independent of the oc(cid:173)
`tanol/water partition coefficient of the penetrating drug, it
`n1arkedly decreased with increasing 111olecular weight. The
`dissociation of the drug led to a decrease in nail pern1eabil(cid:173)
`ity. It appears that the decrease in the permeability of ionic
`drugs is caused by a small increase (about 100) in the ap(cid:173)
`parent n1olecular weight due to ion hydration. It may be that
`the p·ermeabiiity Of the fungal- nail plate is approXhnately ·
`the same as the pern1eability of the healthy nail plate.
`
`Ackno\vledgements
`
`The authors wish to thank the volunteers at Josai Univer(cid:173)
`sity for supplying the healthy nail samples and the Saitama
`Medical Center (Saitama Medical School) and the Saiseikai
`Central I-Iospital for supplying the fungal nail san1ples.
`
`References
`
`Albeiy, W.J .. Greenwood, A.R., Kibble, R.F, 1967. Diffusion cocflicicnts
`of cal'boxyric acid. Trans. Faraday Soc. 63, 360--368.
`
`Flynn, G.L., Yalkowsky, S.H., Roseman, T.J., 1974. Mass transport phe(cid:173)
`nomena and models: theoretical concepts. J. Phann. Sci. 63, 479-
`510.
`Hatanaka, T., Shimoyarna, M., Sugibayashi, K., Morimoto. Y., 1993.
`Effect of vehicle on the skin penncability of drugs: polyethylene
`glycol-400-water and ethanol-water binaiy solvents. J. Controlled Re(cid:173)
`lease 23, 247-260.
`Jarrett, A., Spearman, R.l.C., 1966. The histochemistry of the human nail.
`Arch. Dermatol. 94, 652-657.
`Jillson, O.F, Piper, E.L., 1957. The role of the saprophytic fungi in the
`production of eczematous dcnnatitis. J. 11west. Dc111mtol. 28, 137-145.
`Kobayashi, Y., Nakamura, H., Sugibayashi, K., Morimoto, Y .. 1997. Esti(cid:173)
`mation of action site of1.-lactic acid·clhano!-isopropyl myristatc mixed
`system for its enhancing effect on the skin pcnneation of kctotifcn.
`Int. J. Pharm. 156, 153-162.
`Kobayashi, Y., Miyamoto, M., Sugibayashi, K., Morimoto, Y., 1998.
`Enhancing effect of N-acety!-t-cysteine or 2-mercaptocth:mol on the
`in vitro peimeation of 5-ftuorouracil or tolnaftalc through the human
`nail plate. Chem. Phann. Bull. 46, l 797- t 802.
`Kobayashi, Y., Miyamoto, M., Sugibayashi, K., Morimoto, Y., 1999. Drug
`penneation through the three layers of the human nail plate. J. Phann.
`Pharmcol. 51, 1-9.
`Mertin, D., Lippold, 8.C., 1997. In vitro permeubilit