topical terbinafine, in vivo , in Man
Ingo Alberti1,2, Yogeshvar N. Kalia1,2, Aarti Naik1,2, Richard H. Guy1,2,3
1Centre Interuniversitaire de Recherche et d'Enseignement "Pharmapeptides", Campus universitaire, F-74166 Archamps (France); 2Laboratoire de Pharmacie Galénique, Section de Pharmacie, Université de Genève, CH-1211 Genève 4 (Switzerland); 3To whom correspondence should be ad-dressed
Published as short communication in Pharmaceutical Research 2001, Vol. 18, No. 10, 1472-1475.
Abstract
Purpose: The objective of this study was to assess and, ultimately, to predict the in vivo cutaneous bioavailability of a model drug, terbinafine (TBF). Specifically, the goal was to demonstrate that the SC uptake of TBF following a short skin contact could be used to predict the so-called dermatopharmacokinetic profile of the drug over time.
Methods: The volar forearms of human volunteers were treated for 0.5, 2, and 4 hours with a ethanol-isopropyl myristate solution of TBF. Post-application, stratum corneum (SC) at the treated sites was partially removed by sequen-tial adhesive tape-stripping, and assayed for TBF using HPLC. The concentration-SC depth profile was fitted to a model based upon Fick's second law of diffusion and the characteristic diffusion parameter (D/L2, where D is the drug's effec-tive diffusivity across the SC of putaeffec-tive thickness L) and the drug's SC/vehicle partition coefficient (K) were deduced.
In addition, the concentration profile was integrated over the entire SC to obtain the total TBF quantity in the barrier at each time.
Results: While K was completely unaffected by time, the diffusion parameter was significantly higher at 30 minutes than at 2 and 4 hours, when it was constant. This may be due to a "burst" effect at short times, due to the uptake and permeation of ethanol. However, even when D/L2 following a 30 minute exposure is used to predict the total TBF in the SC at 2 and/or 4 hours, the agreement between theory and experiment was remarkably good.
Conclusions: A single, short-time experiment may be all that is needed to construct at least part of the stratum corneum concentration versus time dermatopharmacokinetic profile suggested recently as an objective and quantitative measure of cutaneous bioavailability and bioequivalence.
Keywords: Cutaneous bioavailability; Dermatopharmacokinetics; Stratum corneum; Skin absorption; Tape stripping.
Introduction
he effectiveness of a topical derma-tological formulation should be evaluated in terms of the in vivo cutaneous bioavailability of the drug. Usually, however, bioavailability is de-fined as the total amount of drug reaching the general circulation from an administered dos-age form [1], and while this definition works
adequately for transdermal delivery, intended for systemic pharmacological effect, its rele-vance to dermatological drugs applied and targeted to local skin sites is less clear.
To be specific, even if there is an evident relationship between the amount of drug ap-plied to the skin and the resulting (albeit, typically very low) blood levels [2], the sig-nificance of the observation to the cutaneous
T
availability and efficacy of the active agent is rarely, if ever, established. Thus, the objective evaluation of topical drug bioavailability (and, hence, of the bioequivalence or lack thereof between formulations) must involve the target tissue itself, that is, one or more components of the skin at or in the region of the application site.
Whereas the drawing of a blood sample may now be considered a routine procedure, a skin biopsy (which is the most direct, if not the most subtle, method by which to sample the tissue of interest) is not. Clearly, for the routine determination of a "dermato-pharmacokinetic" drug profile as a function of time, repeated biopsies are difficult to per-form. Practical, and generally applicable, al-ternative strategies have not been abundantly proposed [3] and only the vasoconstriction assay [4] for corticosteroids has achieved any level of acceptance (despite the fact, it should be said, that this is not a true measure of drug availability, being based on the observa-tion/quantification of a local pharmacody-namic effect which is presumed to be propor-tional to the drug's presence).
Most recently, the U.S. Food and Drug Administration (FDA) has proposed a new approach, the so-called dermatopharmacoki-netic (DPK) methodology [3,5]. The idea is to evaluate topically applied drug levels in the stratum corneum (SC) in vivo as a function of time post-application and post-removal of the formulation, so as to generate a SC
concentra-tion versus time profile from which such
"classical" measures as maximum drug level, time to reach this maximum, and the area un-der the curve can be obtained (i.e., in an analogous fashion to blood level measure-ments for an orally administered drug, for example). The DPK method assumes that: (a) in normal circumstances, the SC is the rate-determining barrier to percutaneous absorp-tion, (b) the SC concentration of drug is di-rectly related to that which diffuses into the underlying viable epidermis, and (c) SC drug levels are more useful and relevant for assess-ing local, dermatological efficacy than plasma concentrations.
While methodological and validation is-sues for the DPK approach remain to be an-swered, it is clear that there is an attractive logic behind this idea. However, examination of the draft guidance [5], which has been pub-lished, reveals that this methodology will be quite labour-intensive even for relatively sim-ple evaluations of bioequivalence. Our objec-tive, therefore, is to begin an examination of whether experiments (following, in general, a DPK methodology) of relatively short dura-tion can be analyzed to produce physico-chemical parameters that describe drug trans-port in the SC and which can therefore be used to predict drug uptake as a function of time.
Specifically, using the antimycotic drug, terbinafine (molecular weight = 291 Da;
log[octanol/water partition coefficient] = 3.3)
[6,7], we have measured, in vivo, in man, the SC concentration versus depth profile follow-ing a 30-minute exposure to a simple formula-tion. The data were analyzed with a non-steady state diffusion equation to yield the SC/vehicle partition coefficient (K) of the drug and its characteristic diffusion parameter (D/L2, where D is the drug's diffusivity across the SC of thickness L). With those values, the same diffusion equation was used to predict the integrated quantity of drug in the SC fol-lowing treatment periods of 2 and 4 hours, and the predictions were then compared to experiment. In addition, the longer time data were evaluated independently to determine whether K and D/L2 were sensitive to the du-ration of drug treatment.
Materials and methods
Chemicals
Terbinafine (TBF) as the free base was supplied by Novartis Pharma (Basel, Switzer-land). It was dissolved at 500 mg/mL in a 50:50 v/v vehicle consisting of isopropyl myristate (≥95%; IPM: Siegfried, Sofingen, Switzerland) and analytical grade absolute ethanol (≥99.8%; ETOH: Fluka, Buchs, Swit-zerland). The high TBF concentration was chosen so as to minimize drug depletion from the formulation within the period of the ex-periment, and to maximize TBF presence in the SC for its analytical detection.
All solvents used were of HPLC grade: ul-tra-pure deionized water, acetonitrile (Sigma-Aldrich, Steinheim, Germany), and tetrahy-drofuran (THF, Sigma-Aldrich, Steinheim, Germany). Triethylamine (TEA) and tetrame-thylammonium hydroxide pentahydrate (TMAH), used as buffers, were also from Sigma-Aldrich.
Human subjects
Four healthy caucasian volunteers (2 fe-male, 2 fe-male, 24-41 years) with no history of dermatological disease participated in this study, which was approved by the ethical commitee of the local university hospital.
Written consent was obtained from all sub-jects. The application sites, covering individ-ual skin areas of 7x1 cm2, were located on the non-hairy, volar aspect of the forearm, 4 cm from the wrist.
Treatment protocol
A single dose of 700 µL of TBF solution (i.e., 100 µL/cm2, or 45.7 mg TBF/cm2) was applied on a cellulose swab (Tela, Basel, Switzerland), then affixed to the skin via an adhesive polyurethane film (Opsite, Smith-Nephew, Hull, UK) under an occlusive poly-ester film (Scotchpak, 3M, St. Louis, MN).
After 0.5, 2 or 4 hours, the patch was re-moved and the excess formulation was cleaned off by gently blotting the skin with three dry cellulose swabs. Solvent, which could further affect the percutaneous
penera-tion of the drug [3], were not used in the sur-face cleaning procedure.
SC sampling protocol
SC is the target tissue of TBF and the tis-sue sampling compartment for DPK approval.
Cutaneous bioavailability was therefore as-sessed by measuring the drug concentration in this membrane layer, by partially removing and subsequently analyzing (see below) mi-croscopic layers (0.5-1 µm) of the SC. This procedure is relatively non-invasive and painless, and was easily performed by simply applying a commercially available contact-sensitive adhesive tape (polypropylene back-ing and acrylic adhesive, Scotch Book Tape, 3M, St. Louis, MN) on the treated site, press-ing it to the skin by rubbpress-ing six times back and forth with a spatula, and then removing it.
In order to remove SC only from the treated area, this site was delimited by other strips of tape, so that only a window of 7x1 cm of skin was exposed. No tape-strips were discarded:
all drug not removed in the surface cleaning process was considered "bioavailable".
Up to 20 SC layers were removed. To check the status of skin barrier function, tran-sepidermal water loss (TEWL) measurements were performed with an Evaporimeter EP1 (Servomed, Stockholm, Sweden) during the stripping procedure. Tape stripping was stopped if TEWL reached 50 g/m2h, even if fewer than 20 strips had been removed.
Each tape was carefully weighed before and after the procedure with a 10-µg precision balance (Mettler AT 261, Greifensee, Swit-zerland). The TBF levels were normalized with respect both to the removed volume and the SC depth, which was obtained from the TEWL measurements, using a method de-scribed elsewhere [8], assuming a SC density of 1 g/cm3 [9].
Extraction and analysis of TBF in the tape strips
Analysis of the removed SC samples was performed by HPLC. The tapes were placed in capped polypropylene tubes (the first 10 individually, the last 10 combined by pairs, thus providing up to 15 samples) containing 7 mL of a 80:20 mixture of acetonitrile and TEA 0.72 mM at pH 2.5. After 16 hours agi-tation, the supernatant was passed through a 0.45 µm filter (Nalgene, Nalge, Rochester, NY) and transferred into 1-mL vials, for HPLC quantification. This treatment did not dissolve the adhesive matrix of the tapes, and the extracted impurities did not interfere with the drug's chromatographic peak. Validation of the extraction (in quintuplicate) was carried out using the same procedure, by spiking tape stripped samples of untreated SC with 100 µL of a 10 mg/mL solution of TBF. Drug recov-ery was 96.6±1.9%.
TBF was determined using an isocratic HPLC method. The system consisted of a
model 600 pump, an autosampler 717 Plus (Waters-Millipore, Milford, MA), a 12-cm Partisphere RP-18 column (Whatman, Clifton, NJ), and a model 486 UV detector at 280 nm (Waters-Millipore, Milford, MA).
The mobile phase was a mixture of acetoni-trile, THF and TMAH 1.59 mM pH 7.8 (50:14.3:35.7 v/v). At a flow rate of 2 mL/min and at room temperature, the retention time of TBF in a 20 µL sample was about 6 min. Peak recording and data processing were performed with the built-in system manager. TBF amounts were determined using the area-under-the-curve (AUC) method from calibra-tion plots generated with the pure compound.
The detection limit was 0.5 µg/mL.
Experimental strategy and data analysis The SC concentration (Cx) versus depth (x) profile of TBF was first determined fol-lowing a 30-minute treatment of the skin, and the data were fitted to the appropriate solution of Fick's second law of diffusion [10]:
C K C 1 x
which assumes that the applied drug concen-tration (Cv) remains constant for the period of treatment (t) and that the viable epidermis acts as a perfect sink for the drug. The third boundary condition is that the SC contains no drug at t=0. The fitting procedure permits values of K and D/L2 to be deduced. With these parameters, Eq. (1) can be integrated
across the SC thickness (i.e., from x=0 to x=L) to provide an effective AUC for the drug in the SC at any time t:
This overall uptake, of course, is the metric to be evaluated in the proposed DPK method as a function of time. We therefore used Eq.
(2) and the K and D/L2 values from the 30-minute experiment to predict AUC at t=2 and t=4 hours. These predictions were then com-pared to actual experiments performed on the same subjects. In addition, the SC concentra-tion versus depth profiles at 2 and 4 hours were evaluated independently using Eq. (1), each time generating best-fit values of K and D/L2 which could then be compared with the results at t=30 minutes to reveal any time-dependent changes in these key trans-port/uptake parameters.
Results and discussion
The SC concentration of TBF versus depth profiles for four subjects following a 30-minute treatment are shown in Figure 1. Be-cause different subjects had different SC thickness and because the tape-stripping proc-ess will rarely, if ever, permit a reproducible amount of SC to be removed either per strip, or per subject, the concentration profiles are presented as a function of normalized SC
depth to facilitate comparison between the results. To give an idea of the variability ob-tained, the removal of 35% of the SC from the 4 subjects employed required, respectively, 5, 8, 7, and 10 strips for subjects G, F, E and B.
The best fits of Eq. (1) to the data are drawn through the individual points in Fig. 1.
The deduced values (mean±SD) of K and D/L2 from these
measure-ments were 0.77 (±0.34) and 2.28 (±0.87) × 10-5 s-1, respec-tively. These parameters were
then used with Eq.(2) to predict AUC follow-ing 2 and 4 hours of TBF application and, finally, relevant experimental data were ob-tained to compare with the predictions. This comparison is shown in Table I, and indicates that no significant difference (p>0.05) be-tween experiment and prediction, at either t=2 hours or t=4 hours, was observed.
Figure 1: Concentration profile of TBF across human SC in vivo following a 30-minute application in a vehicle con-taining 50:50 v/v ethanol/IPM under occlusion. The lines of best fit of Eq. (1) through the individual experimental data points are shown.
Table I: Comparison between the experimentally determined values of AUC (mean±SD; n=4) following 2 and 4 hours of TBF application, and the predic-tions based upon K and D/L2 results determined from the data obtained after a 30-minute application.
Treatment time (h)
Experimental
101∗AUC (M) Predicted 101∗AUC (M)
0.5 2.60±1.09 -
2 4.28±1.81 4.76±1.95 a
4 4.17±1.03 5.21±2.68 a
a Experimental value is not significantly different (p>0.05) from the corre-sponding predicted result.
1 0.8
0.6 0.4
0.2 0
2
1.5
1
0.5
0
Normalized SC depth
TBF concentration in SC (M) Subject B
Subject E Subject F Subject G
0.5 h
The individual concentration profiles of TBF at 2 and 4 hours are presented in Figures 2A and 2B, respectively. Compared to Figure 1, it can be seen that the drug permeates fur-ther into the SC with increasing application time and that the profiles are tending towards
linearity for the longest contact period. As has been previously reported [11] we also found that with increasing exposure time more SC was removed from all subjects by the same number of tape-strips. At t=0.5 hours, 58±5%
of the SC was taken off in 20 strips; at 2 hours
Figure 2: Concentration profile of TBF across human SC in vivo following either a (A) 2-hour or (B) 4-hour applica-tion in a vehicle containing 50:50 v/v ethanol/IPM under occlusion. The lines of best fit of Eq. (1) through the indi-vidual experimental data points are shown.
1 0.8
0.6 0.4
0.2 0
2
1.5
1
0.5
0
Normalized SC depth
TBF concentration in SC (M) Subject B
Subject E Subject F Subject G
2 h (A)
1 0.8
0.6 0.4
0.2 0
2
1.5
1
0.5
0
Normalized SC depth
TBF concentration in SC (M) Subject B
Subject E Subject F Subject G
4 h (B)
and 4 hours, the amounts removed were 74±9% and 81±9%, respectively. It seems reasonable to suppose that the combined ef-fects of occlusion, and of vehicle penetration into the SC, on the cohesivity of the SC are important factors in this observation.
The continuous lines through the data in Figures 2A and 2B represent the independent best fits of Eq. (1) to each profile. The result-ing values of K and D/L2 obtained from the data at 2 hours and 4 hours, together with those deduced by the same analysis at 0.5 hours, are presented in Table II. These results reveal an interesting effect apparent at short application times. It is seen that the diffusion parameter (D/L2) is significantly higher at 0.5 hours than at longer times (when it becomes constant). A possible explanation for this ob-servation is that ethanol, which makes up an important part of the vehicle used, rapidly and significantly enters the SC post-application of the dosage form and transiently facilitates TBF transport in the outer layers of the bar-rier. After this initial "burst" effect, it appears, the drug diffusion slows to a more moderate pace. Clearly, from a
mecha-nistic point of view, this result warrants further investigation.
On the other hand, the SC/vehicle partition coeffi-cient of TBF is not influenced by the contact time between the skin and the formulation,
the results suggesting a rapid equilibrium of TBF at the interface.
With respect to the study's more practical objective, however, it is nevertheless remark-able that, despite the derivated D/L2 value used from the t=0.5 hour application, the pre-dicted AUC results at longer times are in good agreement with the experimental find-ings (Table I). Although based on a limited set of observations, it is tempting to conclude therefore that a short-time experiment may provide sufficient information with which to construct a major part of a drug's DPK profile.
This deduction is supported, at least in part, by some earlier work in our laboratory tar-geted at the facile prediction and assessment of dermal exposure to toxic chemicals [12].
Of course,these preliminary observations re-quire additional work to be performed, ideally examining other drugs and other vehicles.
Also, while we have so far addressed the "up-take" aspects of the DPK profile, it will be important also to consider the "elimination"
phase of drug from the SC post-removal of the delivery system (as is also predicated in
Table II: Values of K and D/L2 (mean±SD; n=4) deduced from the best fit of Eq. (1) to the SC concentration profiles of TBF shown in Figures 1 and 2.
Treatment time
a ANOVA (p<0.05) indicates that the value at 0.5 h is significantly greater then the values at both 2 and 4 hours; however, the data at 2 and 4 h are not significantly different from one another.
b ANOVA indicates no significant differences between these three values.
the draft FDA guidance document [5]).
In conclusion, we believe that the results of this work, while representing only an initial effort, offer a practical and potentially insight-ful approach to address, to quantify and, ulti-mately, to optimize topical drug bioavailabil-ity. The method may prove sufficiently sensi-tive to detect subtle vehicle-induced effects on skin permeation and pragmatic enough to have utility in efficient cutaneous bioavail-ability/bioequivalence studies.
Acknowledgements
We thank Novartis Pharma AG (Basel, Switzerland), for financial support, for pro-viding the HPLC method, and for the contin-ued input of Dr. Jean-Daniel Bonny. We are also grateful to Galderma (Paris, France), for partially sponsoring this study and for the contributions of Dr. Christopher Hensby.
References
1. Borsadia S, Ghanem AH, Seta Y, Higuchi WI, Flynn GL, Behl CR, et al. Factors to be considered in the evaluation of bioavailability and bioequivalence of topical formulations. Skin Pharmacol 1992; 5:129-145.
2. Guy RH, Hadgraft J, Maibach HI. A pharmacoki-netic model for percutaneous absorption. Int J Pharm 1982; 11:119-129.
3. Shah VP, Flynn GL, Yacobi A, Maibach HI, Bon C, Fleischer NM, et al. Bioequivalence of topical
de-rmatological dosage forms--methods of evaluation of bioequivalence. Pharm Res 1998; 15:167-171.
4. Barry BW. Bioavailability of topical steroids.
4. Barry BW. Bioavailability of topical steroids.