Percutaneous Absorption of Vanilloids: In Vivo and in Vitro Studies

+

+

Percutaneous Absorption of Vanilloids: In Vivo and in Vitro Studies GERALD B. KASTINGX, WILLIAM R. FRANCIS, LISA A. BOWMAN,

AND

GENE O. KINNETT

Received November 16, 1995, from the The Procter & Gamble Company, Miami Valley Laboratories, Cincinnati, OH 45253-8707. Final revised manuscript received April 10, 1996. Accepted for publication April 16, 1996X. Abstract 0 The percutaneous absorption of three highly lipophilic analogs of capsaicinsvanillylnonanamide (VN), olvanil, and NE-21610swas measured in vivo in the CD:VAF rat, and in vitro through excised CD: VAF and SkH:Fz rat skin and human cadaver skin. Absorption and skin metabolism were monitored by radiolabel techniques. The rank order of penetration in all species was VN > olvanil > NE-21610, in accordance with that expected from their physical properties. Rat skin was more permeable than human skin by factors ranging from 4 to 8 for VN, 10 to 20 for olvanil, and ≈10 to 100 for NE-21610. All three compounds were extensively metabolized during passage through fresh SkH:Fz rat skin, with the primary route of degradation for at least two of the compounds involving hydrolysis of the amide bond (the metabolites of NE-21610 were not identified). For the in vitro studies a range of receptor solutions was employed to determine a set of conditions that best mimicked in vivo absorption. The results with phosphate-buffered saline containing a preservative and 1−6% polyoxyethylene-20 oleyl ether (Oleth-20) were in good agreement with in vivo results for all three compounds for periods up to 24 h post-dose; after this time, in vivo absorption rates declined but in vitro rates remained relatively constant. Buffered saline or saline containing 0.5% bovine serum albumin led to marked underestimates of in vivo penetration for olvanil and NE-21610, whereas a 1:1 ethanol: water solution led to gross overestimates of the in vivo absorption rates for all three compounds.

Figure 1sVanilloid structures.

Our program involved the development of nonirritating vanilloids for topical use. The compounds under investigation were extremely lipophilic and, furthermore, were subject to metabolism in the skin, as will be shown. Both of these features can lead to misleading results in in vitro penetration testing if not properly taken into account.12-14 This report describes the in vivo percutaneous absorption and elimination of three vanilloidssVN, olvanil, and NE-21610sin rats and the procedures used to validate in vitro skin penetration studies with these compounds. We report also human cadaver skin penetration rates determined by the validated in vitro method.

Experimental Section

Introduction Vanilloids are a class of compounds structurally and pharmacologically related to capsaicin, the primary pungent principle in hot peppers. Systemic doses of capsaicin are known to produce antinociception in rodents,1,2 an effect that is believed to be mediated through a direct interaction with C-polymodal nociceptors.3 Topical capsaicin produces a variety of antinociceptive and anti-inflammatory effects in skin,4 although its primary classification is that of a counterirritant.5 A review by Carter4 summarizes understanding through 1989 of the cutaneous pharmacology and clinical efficacy of capsaicin in the treatment of psoriasis, postherpetic neuralgia, pruritis, and several other cutaneous disorders. Subsequently, capsaicin has also been found to reduce the severity of both primary and recurrent genital herpes in guinea pigs following either systemic6 or topical7 dosing. In all of these applications, the intrinsic irritancy of the compound remains the limiting factor for more widespread clinical use. Vanillylnonanamide (VN), a close structural analog of capsaicin, has similar physical and pharmacological properties to the natural product.8 It has frequently been used as a surrogate for capsaicin because ease of synthesis and higher available purity. A large number of analogs have also been prepared and tested, including olvanil (NE-19550) and NE21610. The latter compounds are less acutely irritating than capsaicin, yet retain many of its desirable pharmacological properties.9-11 The structures of these compounds are shown in Figure 1. X

Abstract published in Advance ACS Abstracts, November 1, 1996.

142 / Journal of Pharmaceutical Sciences Vol. 86, No. 1, January 1997

ChemicalssCarboxyl-[14C]vanillylnonanamide ([14C]VN, 98%, 180 µCi/mg) was synthesized at Procter & Gamble, Miami Valley Laboratories (Cincinnati, OH). Benzyl-[14C]olvanil (97%, 134 µCi/mg) was purchased from Amersham. Benzyl-[14C]NE-21610 (97%, 109 µCi/ mg) was prepared from [14C]olvanil at Miami Valley Laboratories and converted to the acetate salt. Unlabeled vanilloids were synthesized at Miami Valley Laboratories. All other chemicals were reagent grade. Human SkinsHuman cadaver skin, dermatomed to a thickness of 0.25 mm and stored frozen, was obtained from the Ohio Valley Skin and Tissue Center. Tissue preparation and handling were as previously described.15 Rat SkinsMale CD:VAF rats (haired rats), 200-225 g, were obtained from Charles River Laboratories, Portage, MI. Male SkH: Fz rats (fuzzy rats) were obtained from Harlan/Sprague Dawley Inc., Indianapolis, IN. Haired rats were carefully shaved 24 h prior to each study. For the in vitro studies, dorsal skin was collected immediately following sacrifice, trimmed to a thickness of 0.25 mm with a Padgett electrodermatome, and stored in chilled saline prior to mounting. In contrast to Bronaugh and Stewart13 who had difficulty in obtaining intact membranes of <0.35 mm thickness from haired female Osborne-Mendel rats, we were able to obtain consistent penetration results with either haired or fuzzy rat skin trimmed to 0.25 mm. This difference may be due either to rat strain, age, sex, etc., or to dermatoming technique. We found that shaving carefully 24 h prior to sacrifice, spreading the excised pelt on a moist sponge, and application of uniform tension to the tissue by a second operator during the cutting process were important for obtaining high quality membranes. Physical PropertiessSolubilities were determined by equilibrating a small excess of the radiolabeled solute with 2 g of solvent with stirring for at least 48 h. Octanol solubility, Soct, was determined at room temperature (20 ( 2 °C), and receptor solution solubilities were

S0022-3549(95)00484-9 CCC: $14.00

© 1997, American Chemical Society and American Pharmaceutical Association

+

+

Table 1sVanilloid Physical Properties Property

VN

Olvanil

NE-21610

MW 293 418 461 mp, °C 57−58 43−44 106 log Koct 3.74 8.02 7.16 Soct, mg/ga 354 540 16 pKa (estimated value) 9.5−10.5 9.5−10.5 9.2 Receptor phase solubility @ 37 °C, µg/mL (radiolabel method) PBSb 16 e0.5 e2.1 PBS + 0.5% BSAc 145 36 37 PBS + 6% Oleth-20d 7710 3920 2090 EtOH:H2O (1:1 v/v) 40,800 1030 NDe a Solubility in n-octanol at 20 ± 2°C. b Dulbecco’s phosphate-buffered saline, pH 7.4, preserved with 0.02% sodium azide. c Bovine serum albumin (Sigma). d Polyoxyethylene-20 oleyl ether (Volpo 20, Croda). e Not determined.

Table 2sIn Vitro Skin Penetration Results

Jss,a µg/cm2h Receptor Solution PBS PBS + 0.5% BSA PBS + 6% Oleth-20 EtOH:H2O (1:1 v/v) PBS PBS + 0.5% BSA PBS + 1% Oleth-20 PBS + 6% Oleth-20 EtOH:H2O (1:1 v/v) PBS PBS + 0.5% BSA PBS + 6% Oleth-20 EtOH:H2O (1:1 v/v)

VN

olvanil

SkH:Fz rat 1.7 ± 0.1 0.07 ± 0.01 2.2 ± 0.2 0.35 ± 0.09 2.3 ± 0.3 1.3 ± 0.1 16 ± 2 8±1 CD:VAF rat NDb ND 2.2 ± 0.2 0.35 ± 0.03 3.3 ± 0.5 0.64 ± 0.03 4.0 ± 0.6 0.73 ± 0.14 ND ND Human 0.59 ± 0.14 0.005 ± 0.002 0.56 ± 0.13 0.022 ± 0.005 0.52 ± 0.14 0.059 ± 0.012 2.8 ± 0.8 0.7 ± 0.1

NE-21610 0.008 ± 0.001 0.023 ± 0.007 0.06 ± 0.01 1.1 ± 0.1 ND 0.12 ± 0.01 0.40 ± 0.07 1.1 ± 0.6 ND 0.0011 ± 0.0004 0.008 ± 0.002 0.007 ± 0.002 0.05 ± 0.02

a Steady state flux (7−48 h post-dose) from a near-saturated solution (mean ± SE, n ) 5−13). b Not determined.

measured at 37 °C. Samples were withdrawn after 24 and 48 h, filtered through a 0.45-µm syringe filter, and assayed for the solute by scintillation counting. The octanol/buffer partition coefficient, Koct, of VN was measured at room temperature by a shake flask method with centrifugation and HPLC analysis of both phases.16 For olvanil, Koct was determined by retention time on a calibrated reversed-phase HPLC column.16 The Koct of NE-21610 was estimated from that of olvanil by the fragment constant method.17 Ionization constants were estimated from those of similar compounds listed in the NIST database.18 The pKa values for VN and olvanil are associated with ionization of the phenolic hydroxyl group; that for NE-21610 refers to deprotonation of the protonated amine. Adsorption StudysSkin penetration receptor solutions spiked with 100 ng/mL of [14C]VN, [14C]olvanil, or [14C]NE-21610 were placed in glass scintillation vials and stored under a variety of conditions to check for loss due to glass adsorption. In Vitro Skin PenetrationsThe method was similar to that described previously.15 The skin, either human or rat, was divided into small sections and mounted in modified Franz diffusion cells (0.79 cm2) with low tops open to the atmosphere. The receptor solutions (see Tables 1 and 2) consisted either of a preserved phosphate buffer to which albumin or surfactant was added or an ethanol:water mixture. The receptor wells were magnetically stirred and maintained at 37 °C in a thermostatted block. After an overnight equilibration period, a dose of 10 µL of a near-saturated solution of radiolabeled compound [15% (w/w) for VN and NE-21610, 5% for olvanil] with specific activity of 0.4-1.2 µCi/mg, in propylene glycol was applied to the epidermal surface of each skin sample. The receptor solutions were exchanged at 2, 4, 7, 24, 48, and (in some studies) 72 h post-dose and analyzed by liquid scintillation counting. Penetration data were averaged following a log transformation to remove skewness; thus, results reported are geometric means.19

In Vivo Dermal AbsorptionsGlass dose cells (7 cm2) were attached to the shaved dorsal surface of CD:VAF rats with a cyanoacrylate adhesive. The dose solution (same as in vitro, but with specific activities of 4-10 µCi/mg to ensure adequate sensitivity), 100 µL weighed to 0.1 mg, was applied to the rat’s back; the cell was then covered with a wire screen and wired shut. The rats were fitted with tail cups (to collect feces) and placed in stainless steel metabolism cages. Urine and feces were collected after 6, 24, 48, and 72 h. After 72 h, the rats were sacrificed and selected samples (dose cell rinse, dose site, adjacent tissue, carcass, cage wash) were obtained. Samples were analyzed for radioactivity by scintillation counting after the appropriate preparation procedure, dissolution or combustion, was performed on the solid samples. Absorption was estimated from the cumulative excretion of radiolabel in urine and feces,12 plus a timeadjusted fraction of the carcass and cage wash radioactivity determined at sacrifice. The latter corrections were small, amounting to no more than 15% of the total absorbed radioactivity. CD:VAF rats were selected for the study after an unsuccessful attempt with SkH:Fz rats. We could not keep the glass dose cells glued to the backs of the latter, which resulted in compromised absorption data. The problem may have been related to the size and weight of the dose cells, as Bronaugh and Stewart13 reported no problems with fuzzy rats using a smaller nylon ring as the dose cell.13 Dermal Metabolism StudysSkin from SkH:Fz rats was prepared and mounted in glass diffusion cells as already described for the in vitro penetration studies. Dose solutions consisting of 0.5% (w/v) [14C]VN (26 µCi/mg), 2.5% [14C]olvanil (10.5 µCi/mg), and 7.5% [14C]NE21610 (3.3 µCi/mg) in 1:1 (v/v) propylene glycol:ethanol were prepared and applied to the epidermal surface of each tissue sample at 20 µL per cell and 16 cells per formulation. The receptor solution was Dulbecco’s phosphate-buffered saline (PBS) containing 0.02% sodium azide and 1% bovine serum albumin. After 24 h, the receptor solutions were collected, and a portion of each was assayed for radioactivity by scintillation counting. After ascertaining that there were no leaky cells, the receptor solutions for each compound were combined and extracted three times with diethyl ether. The ether fractions were combined, assayed for radioactivity, and evaporated to dryness under a stream of dry nitrogen. The residue was dissolved in a small amount of ethanol and profiled for purity by radio-TLC. The aqueous fraction was mixed with an equal volume of acetonitrile (to precipitate the protein), centrifuged, and analyzed for metabolites by radio-HPLC.20 Control incubations and analyses were conducted without skin to check extraction efficiency and to ensure that any degradation products found were a product of skin metabolism.

Results Physical PropertiessThe physical properties of the test compounds are shown in Table 1. VN, a moderately lipophilic compound, had a solubility in pH 7.4 PBS of 16 µg/mL. This value was in reasonable agreement with a previous HPLC determination of the solubility of this compound in distilled water (32 µg/mL at 25 °C).16 Olvanil and NE-21610, which are extremely lipophilic because of their long hydrocarbon chains, had much lower solubilities in PBS. The measured values of 0.5 and 2.1 µg/mL, respectively, should be regarded as upper limits because of the possible influence of radiolabeled impurities on these very low level measurements. With a pKa value of ∼10 for the phenolic hydroxyl group, VN and olvanil are not appreciably ionized at pH 7.4; thus, the contribution of the salt forms of these molecules to solubility is negligible at this pH. NE-21610, however, should be >98% ionized at pH 7.4 because of its protonatable amine group, yet its PBS solubility was not much greater than that for olvanil. Additional studies (data not shown) showed that the solubility of NE-21610 did increase in acidic solutions; this fact plus a calculation based on Yalkowski’s equation16 suggest that the PBS solubility of NE-21610 was limited by the free base. Addition of 0.5% BSA to the buffer increased the solubilities of all compounds to >30 µg/mL, whereas addition of 6% Oleth-20 increased solubilities to >2 mg/mL. Ethanol: water (1:1, v/v) was a very good solvent for VN, but was not as effective for olvanil as the surfactant solution.

Journal of Pharmaceutical Sciences / 143 Vol. 86, No. 1, January 1997

+

+

Adsorption StudysLow concentration solutions of [14C]NE-21610 in phosphate buffer lost radioactivity due to glass adsorption (data not shown). This problem was eliminated by addition of BSA or Oleth-20 to the buffer, or by the use of ethanol:water receptor solutions. [14C]VN and [14C]olvanil were not susceptible to this effect. In Vitro Skin PenetrationsThe results of in vitro penetration studies in human skin and the two rat skins are summarized in Table 2. As delivery rates were nearly constant between 7 and 48 h, results are expressed as the mean steady state flux, Jss, over that time period. For the moderately lipophilic compound, VN (log Koct ) 3.74), Jss was only weakly dependent of the choice of receptor phase for the PBS-based receptor solutions. For olvanil and NE-21610, however, Jss values increased in the order: PBS , PBS + 0.5% BSA < PBS + 1% Oleth-20 < PBS + 6% Oleth-20. The ethanol:water receptor solution yielded much higher flux values than any of the PBS-based systems for all three compounds. In Vivo Dermal AbsorptionsThe results of the in vivo absorption study are shown in Figure 2 and Tables 3 and 4. The observed order of absorption was VN > olvanil > NE21610, as was found for the in vitro penetration studies. The distribution of radioactivity between urine and feces was compound-dependent; very little radioactivity was retained in the carcass. Results of the dermal metabolism study (see next section) and related systemic metabolism studies20,21 suggest that the absorbed materials were extensively metabolized, with vanillamine and/or vanillamine conjugates (glucuronide or sulfate) being the primary 14C-labeled metabolites for all three compounds. Rapid conversion into polar metabolites would explain the lack of accumulation of radioactivity in the tissues for these compounds, because the lipophilic parent compounds might otherwise have been expected to partition to a greater extent into fatty tissues. Dermal Metabolism StudysRadiolabel found in the receptor phases following the 24-h in vitro penetration study with radiolabeled vanilloids and freshly excised SkH:Fz rat skin was only partially extractable into ether. The etherextractable fraction was 63% of the total radiolabel for [14C]VN, 27% for [14C]olvanil, and 33% for [14C]NE-21610. The corresponding values for control samples incubated for 24 h without skin were 98, 99, and 96%, respectively. Radioactivity in ether extracts coeluted with parent compound in each case. The radioactivity in aqueous fractions remaining after ether extraction coeluted with vanillamine (a hydrolysis product) for VN and olvanil; for NE-21610 radioactivity was below radHPLC detection limits.

Discussion For times up to 24 h post-dose, in vivo skin penetration rates in the rat for these highly lipophilic compounds were best modeled by in vitro experiments in which the receptor solution consisted of a preserved phosphate buffer containing 6% Oleth-20. Using this buffer, the agreement at 24 h between in vitro penetration of carbon-14 into the receptor solution and in vivo absorption was -18% for VN, +7% for olvanil, and -38% for NE-21610 (Table 4). After 24 h, in vivo absorption rates declined whereas the in vitro rates remained constant; thus, none of the buffers tested accurately mimicked in vivo absorption over the full 72-h duration of the test. By 72 h, the best agreement in two of the three cases (-9% error for VN and -15% for NE-21610) was provided by the 1% Oleth-20 solution. For olvanil, the 0.5% BSA buffer gave the best agreement at 72 h (-6% error) but the 1% Oleth-20 solution overpredicted absorption by 86%. It is evident from the results in Table 2 that much larger errors would be incurred for in vitro studies employing either 144 / Journal of Pharmaceutical Sciences Vol. 86, No. 1, January 1997

Figure 2sComparison of in vivo (n ) 4−5) and in vitro (n ) 5−13) dermal absorption rates of radioactivity associated with [14C]labeled vanilloids in the CD: VAF rat (mean ± SE). Key: (b) in vivo; (O) PBS + 0.5% BSA; (0) PBS + 1% Oleth-20; (9) PBS + 6% Oleth-20.

unmodified phosphate buffer or ethanol:water solutions in the receptor compartment. Assuming that 6% Oleth-20 gave about the correct absorption rates, we found that the use of unmodified phosphate buffer underpredicted the absorption of olvanil and NE-21610 by an order of magnitude in either human skin or fuzzy rat skin. This effect was not seen with the more water soluble permeant, VN. The use of ethanol: water (1:1) in the receptor compartment overpredicted the absorption of all three compounds by a factor ranging from 5 to 18. The latter technique, therefore, should be avoided. In general, our findings with vanilloids confirm the results of Bronaugh and Stewart.12,13 These workers studied the percutaneous absorption of four lipophilic compounds in either haired (Osborne-Mendel) or fuzzy (SkH:Fz) rats. They found the best in vitro/in vivo agreement over a 5-7 day period with a 0.5% Oleth-20 receptor solution and 200-µm-thick skin

+

+

Table 3sDisposition of Radioactivity in CD:VAF Rats 72 h after a Topical Dose of 14C-labeled Vanilloids Percent of Dose Recovereda VN

olvanil

NE-21610

Urine Feces Carcass Cage wash Total absorbed doseb

Sample Type

7.5 ± 1.3 3.4 ± 1.3 0.9 ± 0.6 0.6 ± 0.6 12.7 ± 4.0

3.0 ± 0.6 0.3 ± 0.2 0.13 ± 0.03 0.12 ± 0.03 3.6 ± 0.4

0.5 ± 0.4 1.2 ± 0.8 0.2 ± 0.2 0.1 ± 0.1 2.0 ± 1.2

Dose site wash Dose site skin Cell wash Adjacent tissue Total unabsorbed dosec

51.5 ± 8.6 12.2 ± 3.9 4.1 ± 4.5 0.3 ± 0.4 68.1 ± 6.3

56.5 ± 3.9 25.5 ± 3.3 1.8 ± 0.9 0.1 ± 0.1 83.9 ± 3.2

48.0 ± 6.4 24.1 ± 4.6 9.4 ± 2.5 0.02 ± 0.02 81.5 ± 4.2

Total recovery of 14C

81.2 ± 3.6

87.6 ± 3.5

83.5 ± 3.8

Mean ± SD, n ) 4−5. b Sum of urine, feces, carcass, and cage wash. c Sum of dose site wash, dose site skin, cell wash, and adjacent tissue. a

Table 4sComparison of In Vivo and In Vitro Penetration Results in CD:VAF Rats Percent of Dose Penetrateda Elapsed Time In vivob PBS + 0.5% BSA PBS + 1% Oleth-20 PBS + 6% Oleth-20 In vivoc PBS + 0.5% BSA PBS + 1% Oleth-20 PBS + 6% Oleth-20

VN 24 h post-dose 4.5 ± 0.7 2.3 ± 0.4 3.2 ± 0.5 3.7 ± 0.8 72 h post-dose 12.7 ± 2.0 6.8 ± 0.6 11.6 ± 2.0 14.4 ± 2.4

olvanil

NE-21610

1.4 ± 0.1 0.9 ± 0.1 1.5 ± 0.1 1.5 ± 0.2

0.85 ± 0.26 0.14 ± 0.01 0.38 ± 0.08 0.53 ± 0.14

3.6 ± 0.2 3.4 ± 0.3 6.7 ± 0.4 7.9 ± 1.5

2.01 ± 0.61 0.45 ± 0.05 1.70 ± 0.31 2.63 ± 0.46

a Geometric mean ± SE, n ) 4−5 (in vivo) or 5−13 (in vitro). b Includes radiolabel in urine and feces + 1/3 (carcass + cage wash). c Includes radiolabel in urine, feces, carcass, and cage wash.

samples from the fuzzy rat. Haired rat skin samples could not reliably be cut this thin. However, good agreement was also found for haired rats using 6% Oleth-20 and 300-µm-thick skin samples. They preferred the combination of lower surfactant concentration and thinner skin samples because this combination has less impact on the penetration of the hydrophilic control compound [3H]cortisone. We found that reliable data and good in vitro/in vivo agreement could be obtained with 250-µm-thick haired rat skin samples and 1-6% Oleth-20 solutions. Furthermore, the 250 µm thickness was attractive as it is the standard setting employed by many human skin and tissue banks. A further point of agreement with Bronaugh and Stewart12,13 concerns the aqueous solubility value below which modification of the in vitro receptor solutions becomes necessary. Bronaugh and Stewart had estimated this value to be ≈10 µg/mL.12 We found that receptor solution modification was important for NE-21610 (PBS solubility, 2.1 µg/mL), but not for VN (PBS solubility, 16 µg/mL). Thus our results are consistent with the 10 µg/mL limit. Note that it is apparently advantageous to increase the receptor solution solubility to well beyond this value for poorly soluble compounds to obtain the best agreement with in vivo absorption rates (cf. Table 1 and Figure 2; see also refs 12 and 13). In terms of absorption rate profile, however, we observed in vitro/in vivo differences not described by Bronaugh and Stewart.12,13 Our study examined the time course of absorption over a 3-day period following topical application of what amounted to an infinite dose of each compound. By sequen-

tially monitoring cumulative excretion in vivo and cumulative penetration into the receptor solution in vitro, we found that in vivo absorption rates declined gradually over this period, whereas in vitro absorption rates maintained a steady state. Depletion of the donor solutions would be expected to be similar in vivo and in vitro, because the applied dose per unit area was nearly identical. Hence, it seems unlikely that any choice of in vitro conditions would completely remove this difference. However, the absorption profiles obtained with 1-6% Oleth-20 solutions in the receptor compartment clearly provided useful estimates of vivo absorption over this time period. We would propose that skin turnover in vivo may account for the gradual decline in in vivo absorption rates relative to in vitro. It is possible that a more physiologic concentration of albumin (e.g., 3-10% BSA) would provide in vitro results closer to in vivo than the 0.5% BSA solution we studied. However, there are practical limitations to this approach (the solutions foam copiously, leading to bubble formation) and previous experience has not been encouraging. Bronaugh and Stewart12,13 found that neither 3% BSA nor rabbit serum were nearly as effective as 0.5-6% Oleth-20 for increasing in vitro penetration rates of highly lipophilic permeants. In all species, the rank order of vanilloid penetration was VN > olvanil > NE-21610. This order is in agreement with the predictions of a two-layer penetration model in which molecular weight, octanol solubility, and water solubility are the governing variables (calculations not shown).22 Rat skin was from 4 to >100 times more permeable than human skin; this ratio was compound and species-dependent. The two rat skins had comparable permeability to VN and olvanil, but the haired rat was ≈10 times more permeable to NE-21610 than the fuzzy rat. We speculate that increased follicular delivery of the protonated form of NE-21610 in the haired rat may be responsible for the increased permeability because VN and olvanil are not appreciably ionized at neutral pH. All three compounds were extensively metabolized on passage through freshly excised rat skin. Hydrolytic cleavage of the amide bond was a major route of skin metabolism under these conditions. This result is similar to that seen with vanillamine amides in other rat tissues.20,21 Although it may be argued that skin viability should be maintained for in vitro studies with such compounds (e.g., by using the methods of Collier et al.23), our work shows that good in vitro/in vivo agreement can be obtained without taking such precautions. Bronaugh and Stewart13 came to a similar conclusion in their work with benzo(a)pyrene, another highly lipophilic compound that is metabolized in skin.13 Taken together, these results suggest that the human in vitro skin penetration rates for vanilloids measured in this study are representative of human in vivo absorption rates, despite the probable low metabolic activity of the cadaver skin samples.

Conclusions The use of split-thickness skin (0.25 mm) in vitro and a receptor solution comprising a preserved phosphate buffer containing 1-6% Oleth-20 gave good agreement with in vivo dermal absorption rates in rats for three moderate to highly lipophilic vanilloids. This result confirms the findings of Bronaugh and Stewart12,13 regarding appropriate in vitro penetration methodology for highly lipophilic compounds. Human skin penetration rates for vanilloids were not accurately predicted by either the haired rat or the fuzzy rat, nor was there a constant permeability ratio for rat skin versus human skin. Thus, excised human skin would be preferred versus rodent skin for predicting the human in vivo absorption

Journal of Pharmaceutical Sciences / 145 Vol. 86, No. 1, January 1997

+

+

of similar compounds. Finally, all three compounds tested were absorbed relatively slowly through rat skin and were efficiently excreted once absorbed. There was evidence for hydrolytic metabolism of these compounds in rat skin.

Acknowledgments The authors thank K. R. Wehmeyer for the HPLC analysis of the dermal metabolism samples.

References and Notes 1. LaHann, T. R.; Farmer, R. W. Proc. West. Pharmacol. Soc. 1983, 26, 145-149. 2. Buck, S. H.; Burks, T. F. Trends Pharmacol. Sci. 1983, 4, 8487. 3. Lynn, B. Pain 1990, 41, 61-69. 4. Carter, R. B. Drug Devel. Res. 1991, 22, 109-123. 5. Federal Register, OTC Compilation Part 1. 1979, 44(234):Tab 9, 144-145. 6. Stanberry, L. R. J. Infect. Dis. 1990, 162, 29-34. 7. Stanberry, L. R.; O’Neill, T. P.; Holroyde, M. J.; Berman, E. F. Program and Abstracts of the 30th Interscience Conference on Antimicrobial Agents and Chemotherapy, Atlanta, GA, October, 1990; Abstract No. 720. 8. LaHann, T. R.; DeKrey, L. J.; Tarr, B. D. Proc. West. Pharmacol. Soc. 1989, 32, 201-204. 9. Brand, L.; Berman, E.; Schwen, R.; Loomans, M.; Janusz, J.; Bohne, R.; Maddin, C.; Gardner, J.; LaHann, T.; Farmer, R.; Jones, L.; Chiabrando, C.; Fanelli, R. Drugs Exp. Clin. Res. 1987, 13, 259-265. 10. Dray, A.; Bettaney, J.; Rueff, A.; Walpole, C.; and Wrigglesworth, R. Eur. J. Pharmacol. 1990, 181, 289-294.

146 / Journal of Pharmaceutical Sciences Vol. 86, No. 1, January 1997

11. Janusz, J. M.; Buckwalter, B. L.; Young, P. A.; LaHann, T. R.; Farmer, R. W.; Kasting, G. B.; Loomans, M. E.; Kerckaert, G. A.; Maddin, C. S.; Berman, E. F.; Bohne, R. L.; Cupps, T. L.; Milstein, J. R. J. Med. Chem. 1993, 36, 2595-2604. 12. Bronaugh, R. L.; Stewart, R. F. J. Pharm. Sci. 1984, 73, 12551258. 13. Bronaugh, R. L.; Stewart, R. F. J. Pharm. Sci. 1986, 75, 487491. 14. Bronaugh, R. L.; Collier, S. W. Cosmetics & Toiletries 1990, 105, 86-93. 15. Kasting, G. B.; Bowman, L. A. Pharm. Res. 1990, 7, 136-145. 16. Smith, R. L.; Kasting, G. B. Program and Abstracts of JUC Pharm. Sci. (Am. Pharm. Assoc. and Pharm. Soc. Japan), Honolulu, HI, December, 1987. 17. Rekker, R. F. The Hydrophobic Fragmental Constant; Elsevier: New York, 1977. 18. Martell, A. E.; Smith, R. M. Critically selected stability constants of metal complexes database, Vers. 2.0, NIST Standard Reference Data, Gaithersburg, MD, September, 1995. 19. Kasting, G. B.; Filloon, T. G.; Francis, W. R.; Meredith, M. P. Pharm. Res. 1994, 11, 1747-1754. 20. Wehmeyer, K. R.; Kasting, G. B.; Powell, J. H.; Kuhlenbeck, D. L.; Underwood, R. A.; Bowman, L. A. J. Pharm. Biomed. Anal. 1990, 8, 177-183. 21. Wehmeyer, K. R.; Kasting, G. B., unpublished results. 22. Kasting, G. B.; Smith, R. L.; Anderson, B. D. Prodrugs for dermal delivery: Solubility, molecular size and functional group effects. In Prodrugs: Topical and Ocular Drug Delivery; Sloan, K. B., Ed.; Marcel Dekker: New York, 1992; pp 117-161. 23. Collier, S. W.; Sheikh, N. M.; Sakr, A; Lichtin, J. L.; Stewart, R. F.; Bronaugh, R. L. Toxicologist 1988, 8, 125.

JS950484A