- Email: [email protected]
Analysis of in vitro release through reconstructed human epidermis and synthetic membranes of multi-vitamins from cosmetic formulations
Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 461–467
Contents lists available at ScienceDirect
Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba
Analysis of in vitro release through reconstructed human epidermis and synthetic membranes of multi-vitamins from cosmetic formulations Simone Gabbanini a , Riccardo Matera b , Claudia Beltramini a , Andrea Minghetti a , Luca Valgimigli b,∗ a b
BeC S.r.l., R&D division, Via C. Monteverdi 49, 47100 Forlì, Italy University of Bologna, Faculty of Pharmacy, Dept. Organic Chemistry “A. Mangini”, via San Giacomo 11, 40127 Bologna, Italy
a r t i c l e
i n f o
Article history: Received 22 September 2009 Received in revised form 7 January 2010 Accepted 14 January 2010 Available online 21 January 2010 Keywords: Vitamins RHE Cosmetics Percutaneous absorption HPLC
a b s t r a c t A convenient method for in vitro investigation of the release of lipid- and water-soluble vitamins from cosmetic formulations was developed. The permeation of (d)-␣-tocopherol (vitamin E), retinyl acetate (pro-vitamin A), ascorbic acid (vitamin C) and pyridoxine (vitamin B6) through SkinEthic® reconstructed human epidermis (RHE), and synthetic polyethersulfone and polycarbonate membranes was studied in vitro using a Franz-type diffusion apparatus, coupled either to a spectrophotometer for continuous reading (dynamic setting) or to HPLC-DAD analysis of the receptor medium (static setting). O/W and W/O emulsions were compared with simple aqueous solutions for their kinetic of vitamins release, to evaluate the influence of the cosmetic formulation on the bioavailability of active ingredients. Results indicate that synthetic membranes offer a limited barrier to the diffusion of vitamins, but may provide information on the release ability of the formulation. Penetration was more effective when water was the external phase of the formulation, i.e. W/O emulsions were less effective in the release of vitamins than O/W emulsion or aqueous solutions. RHE (17 days old) offered a significantly higher barrier to penetration of vitamins, as expected for native human epidermis. The relative ranking in coefficient of permeability (Ps (cm/h)) was: ascorbic acid > pyridoxine retinyl acetate > ␣-tocopherol ∼0, the absolute values depending on the formulation. The method herein described showed to be a practical and convenient tool for the quality-control and efficacy evaluation of cosmetic formulations. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Vitamins are common ingredients of cosmetic formulations and a number of claims are based on the implicit assumption that these bio-active molecules are effectively released from the formulation into epidermis and possibly through epidermis into the derma and subcutaneous tissues. However studies assessing the skin absorption of vitamins released from cosmetic formulations are extremely rare in the literature and mostly limited to retinoids (vitamin A) [1] or vitamin E. [2] The lack of specific obligation for bioavailability studies and the prohibition of animal testing for cosmetic formulations, expressed in current EU regulations [3] do not stimulate the objective evaluation of these efficacy-related parameters. The term cosmeceuticals was coined in mid-eighties to indicate a particular category of personal-care products that could be placed at the
Abbreviations: RHE, reconstructed human epidermis; egf, epidermis growth factor; W/O, water in oil; O/W, oil in water; Ps, coefficient of permeability; LOD, limit of detection; LOQ, limit of quantitation; DTT, (dl)-Dithiothreitol; TFA, trifluoroacetic acid. ∗ Corresponding author. Tel.: +39 051 2095683; fax: +39 051 2095688. E-mail address: [email protected] (L. Valgimigli). 0731-7085/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jpba.2010.01.029
border-line between cosmetics and pharmaceuticals [4]: cosmetic products containing active ingredients meant to have beneficial physiologic effects resulting from their pharmacologic properties [5]. Since there is no legal definition of cosmeceuticals, these products are often formally classified as cosmetics or in some cases as drugs, the distinction being sometimes arbitrary and varying from country to country. For instance skin-protectants (such as those based on vitamin E) are classified as drugs in USA and as cosmetics in Europe. This category of products has been exponentially increasing its importance during last two decades and a relevant percentage of products currently found in EU market could be classified as cosmeceuticals. However a scientifically sound evaluation of their efficacy is often lacking, as have been lamented by authoritative investigators [5,6]. In compliance with EU regulations and in line with OECD [7] and COLIPA guidelines [8] we aimed to set-up a convenient method, based on a custom designed Franz-type diffusion apparatus, for in vitro evaluation of the kinetic of release of vitamins form cosmetic/cosmeceutical formulations. In order to avoid the use of animal skin [9–12], which would conflict with the ethical principles of EU regulations, and to evaluate a convenient substitute for human epidermis from cadaver or surgery [11,13,14], we tested SkinEthic® reconstructed human epidermis (RHE). Previous studies
462
S. Gabbanini et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 461–467
have shown that RHE is reasonably similar to native human epidermis both in terms of morphology and lipid composition [15,16]. It was found less selective than native epidermis to the permeation of specific drugs [15], however it has been indicated as a valid substitute for in vitro testing of topical formulations [17,18,19]. Synthetic membranes have also been employed for release kinetics studies [20,21], so we compared RHE to synthetic polyethersulfone and polycarbonate membranes. Two different settings have been initially compared for the analysis of vitamins in the receptor fluid: continuous spectrophotometric reading by circulating the fluid through a flow cell (dynamic setting, see Appendix A), or discontinuous sampling followed by HPLC-DAD analysis (static setting). The former has lately been abandoned due to limited performance. Three different types of cosmetic formulation were considered in this investigation: O/W emulsions (creams), W/O emulsions (ointments) and aqueous solutions (lotions), to evaluate the influence of the formulation in the release of lipid- and water-soluble vitamins. Ascorbic acid (vitamin C), pyridoxine hydrochloride (vitamin B6 ), retinol acetate (pro-vitamin A), and ␣-tocopherol (vitamin E) were chosen as representative exempla of commonly employed waterand lipid-soluble vitamins. 2. Materials and methods 2.1. Materials The standard vitamins used for method validation were: ascorbic acid (99.9%), pyridoxine hydrochloride (99.9%), (dl)␣-tocopherol (99.6%), retinol acetate (97.4%) from Supelco (Bellefonte, PA, USA); trifluoroacetic acid (TFA, ≥99.0%), (dl)Dithiothreitol (DTT; >99.0%), sodium phosphate dibasic anhydrous (≥99%), sodium chloride (≥99.5%), magnesium sulfate hexahydrate (>99.0%), isopropanol (≥99.8%), methanol (≥99.8%) from Sigma, (St. Louis, MO, USA); potassium dihydrogen phosphate (≥99.5%) from Merck (Darmstadt, Germany). Reconstituted Human Epidermis 12 days old 4.0 cm2 , and Growth Medium (1.5 mM calcium chloride, 25 mg/mL gentamycin, 5 mg/mL insulin, 1 ng/mL egf) were purchased from SkinEthic Laboratories (Nice, France). Polyethersulfone membranes (0.2 m, Supor® 200, Pall corp., New York, USA) and polycarbonate membranes, (0.01 m, Whatman, GE Healthcare, Uppsala, Sweden) were from FAVS s.n.c. (Bologna, Italy). 2.2. Apparatus and chromatographic conditions 2.2.1. Franz-type diffusion cells Franz-Type diffusion cells (diffusion surface: 1.54 cm2 ; internal volume: 14.8 ± 0.1 mL), designed in our labs to optimize the use of SkinEthic® RHE disks, were previously described [19] and are shown in Appendix A (Fig. S1). Homogenous concentration of the analytes within the receptor compartment was ensured during preliminary tests by monitoring the diffusion of blue-colored crystal violet solutions. 2.2.2. HPLC-DAD analysis All analyses were carried out using a Thermo Accela Pump equipped with Accela Autosampler and Surveyor Photo-Diode Array detector (Thermo Scientific, San Jose, CA, USA). The LC was performed on a Synergi C18 Hydro-RP 4 m column (150 mm × 4.6 mm I.D.) equipped with a guard column (C18, 4.0 mm × 3.0 mm) from Phenomenex (Torrance, CA) operating at 30 ◦ C. Two solvent systems were employed: A = CH3 OH + TFA (0.25 mL/L) and B = H2 O + TFA (0.25 mL/L). The injection volume was 5 L. Analyses of the lipid-soluble vitamins (method LM) were performed isocratically at 1.0 mL/min for 13 min with the mobile phase A-B (80:20, v/v). Retinol acetate and ␣-tocopherol were detected at
320 and 290 nm, respectively. Analyses of the water-soluble vitamins were performed with method WM, eluting at 1.0 mL/min in the following gradient program for 17 min: t = 0, A–B (0:100, v/v); t = 3, A–B (0:100, v/v); t = 10, A–B (10:90, v/v); t = 15, A–B (10:90, v/v); t = 17, A–B (0:100, v/v). Ascorbic acid and pyridoxine were detected at 245 and 290 nm, respectively. 2.3. Assay procedure 2.3.1. Calibration for HPLC-DAD analysis Standard solutions in five levels from 0.001 to 0.100 mg/mL were analyzed in triplicate. Ascorbic acid and pyridoxine hydrochloride were dissolved in pH 7.4 phosphate buffered saline solution with (dl)-Dithiothreitol (DTT, 0.4 mg/mL), whereas (dl)-␣-tocopherol and retinol acetate solutions were prepared in methanol. The LOD was determined from repeated analyses, as the concentration that yields a signal-to-noise (S/N) ratio of at least 3:1 and the LOQ was obtained as the concentration that yields S/N ratio of 10:1. Linearity and LOD/LOQ data are collected in Table 1. 2.3.2. Validation of the analytical procedure Standard receptor medium solutions containing three-levels concentration of either lipid-soluble vitamins or water-soluble vitamins were subjected to five replicate analyses for three nonconsecutive days (15 replicates for each level). Solutions were stored at +4 ◦ C in the dark to minimize oxidation and thermalphotochemical degradation of vitamins. Accuracy was evaluated from % recovery with respect of theoretical content, while precision data were obtained from intra-day and inter-day % relative SD. To test the robustness standard solutions were repeatedly analyzed (n = 5) intra-day using the method described in Section 2.2.2 (setting 1) and changing the chromatographic conditions (setting 2) as follows: T = 25 ◦ C (−5 ◦ C), flow rate 0.8 mL/min (−20%) and decreasing the content of TFA in the mobile phase to 0.20 mL/L (−20%). To test the selectivity, vitamins content was determined in the donor compositions (W/O emulsions, O/W emulsions and aqueous receptor medium) as detailed in Section 2.3.3. The receptor medium was phosphate saline buffer pH 7.4 (8 g/L NaCl, 0.2 g/L KCl, 0.2 g/L KH2 PO4 and 1.15 g/L NaHPO4 in bi-distilled water), 3% w/w of polysorbate-20 and 0.25% w/w of isopropanol and DTT 0.4 mg/mL [22]. 2.3.3. Preparation and analysis of the cosmetic formulations (donor) Four creams being Oil in Water or Water in Oil emulsions, each containing the water-soluble vitamins, or the lipid-soluble vitamins and DTT 0.4 mg/mL [22], were prepared as described in Appendix A and stored at 4 ◦ C before use. The nominal vitamin C and B6 content were 1.7% (w/w) each, whereas the vitamin A acetate and E were 1.8% and 3.0% (w/w), respectively. The exact amount of vitamin content was assessed by HPLC-DAD analyses. Briefly, water-soluble vitamins containing creams (100 mg) were suspended in CH3 OH (10 mL), then 1 mL of this suspension was diluted up to 10 mL with H2 O + TFA (0.25 mL/L). The suspension was filtered with a syringe through 0.45 m syringe filter and analyzed. The lipid-soluble vitamins containing creams (100 mg) were suspended in CH3 OH (10 mL), diluted up to 10 mL with the same solvent and filtered (0.45 m) prior to analysis. 2.3.4. Study of vitamins’ stability in the donor and receptor media Standard receptor medium solutions, W/O emulsions and O/W emulsions, each containing DTT 0.4 mg/mL and the water-soluble (B6 , C) or lipid-soluble (A, E) vitamins were stored at 4 ◦ C (in the
S. Gabbanini et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 461–467
463
Table 1 Vitamins five levels calibration parameters expressed as [area] vs [concentration in mg/mL], linearity and LOD/LOQ values for HPLC-DAD analysis. Analyte
LOD [g/mL]
LOQ [g/mL]
Slope
Ascorbic acid Pyridoxine Retinol acetate ␣-Tocopherol
0.24 0.03 0.10 1.50
0.81 0.12 0.34 5.10
1.519 ± 0.015 E+07 1.199 ± 0.003 E+07 3.973 ± 0.082 E+07 1.924 ± 0.019 E+06
dark) and 34 ◦ C under normal day-lighting and periodically analyzed for 48 h. 2.3.5. Preparation of reconstructed human epidermis Twelve days old reconstructed human epidermis (RHE) held on polycarbonate disks were revitalized by addition of 4 mL of growth medium for skin for day (1.5 mM calcium chloride, 25 mg/mL gentamycin, 5 mg/mL insulin, 1 ng/mL egf) in sterile petri boxes under laminar flux cabinet and incubated at 37 ◦ C in 5% CO2 atmosphere in a CO2 incubator for five days prior to investigation. 2.3.6. In vitro diffusion experiments Diffusion cells were mounted on a Magnetic 6 Stirrer and connected to a water bath at 34 ◦ C. The receptor medium had the composition described in Section 2.3.2. The receiver compartment, equipped with 10 mm × 2 mm stirring bar, was filled with receptor medium and the skin membranes were cut from plastic support and mounted with their polycarbonate filter with the epidermal side facing upward into the donor compartment. With synthetic membranes (polyethersulfone or polycarbonate) the same procedure was followed. Donor and receptor chambers were watertight closed with a metallic clamp, and a dose of 0.5–1.0 mL of the cosmetic formulation (donor) was applied to the surface of the membrane. At time intervals of 30–60 min within 5–24 h from application, a 100 L aliquot of the receptor medium was withdrawn and immediately replaced with an equal volume of fresh buffer [13]. Samples that could not be analyzed immediately were sealed under nitrogen and stored at +4◦ C. 2.3.7. Data analysis The cumulative amount of substance diffused through a unit of surface of skin (Q; g/cm2 ) was obtained by HPLC analysis of the sampled receptor medium, corrected for the analytes subtracted at every sampling [12], and for the degradation of vitamins (A and C) in the receptor medium.
R2
Intercept −20,930.2 3783.2 6447.6 −1334.5
± ± ± ±
8213.0 1929.8 20,423.0 1578.9
0.9997 0.9999 0.9983 0.9998
were sufficiently selective to allow the analysis of vitamins directly in the cosmetic formulations (O/W and W/O emulsions) without preliminary sample treatment other than dilution (see Section 3.2). 3.2. Validation of analytical methods and analysis of donor compositions We tested accuracy, precision and robustness on standard solutions containing three levels of each target vitamin. The content of each vitamin was analyzed for three non-consecutive days (15 replicate analyses): accuracy and precision were evaluated from mean percent recovery and % RSD as detailed in Table 2. Overall the performance of the analytical method was judged adequate to the subsequent kinetic investigation. Only with the two lower concentrations of vitamin C recoveries were lower than expected and tended to decrease in the analysis performed on subsequent days. Indeed the inter-day RSD was significantly larger than intra-day. We attributed it to air-oxidation of ascorbate in the standard solutions, which could not be completely suppressed by DTT (vide infra). To overcome this source of error, care was taken to analyze receptor samples containing ascorbate immediately upon withdrawal (or within 1 h) during the kinetic investigation. Robustness was tested by performing two parallel sets (n = 5, intra-day) of analyses on three-levels standard vitamins solutions (containing 0.4 mg/mL DTT) using our method (setting 1) or by changing sensitive parameters like temperature, flow rate, and eluent composition by a reasonable amount (setting 2: temperature, −5 ◦ C; flow rate, −20%; and [TFA], −20%; see Section 2.3.2). Significantly different (p < 0.05) recovery values are marked in Table 2 with an asterisk. Only in 4 out of 12 samples % recovery was significantly different and differences were always within ±5%, indicating a good robustness on the method. To test the selectivity, the same analytical methods was used to determine the actual content of vitamins in each donor formu-
3. Results and discussion 3.1. HPLC analysis In order to be suited to the kinetic investigation the analysis protocol had to be sufficiently quick to keep pace with hourly sampling and avoid column changing. We employed a Synergi Hydro-RP C18 column (with polar end-capping), which was suited to both polar and apolar analytes, and preferred separate methods for lipid-soluble (LM) and water-soluble (WM) vitamins, with run times as short as 13 min and 17 min (Fig. 1) for lipid- and water-soluble vitamins respectively. According to the experimental design the diffusion kinetics of lipid- and water-soluble vitamins could be studied separately (with sampling frequency as short as 20–30 min) or, if needed, combined in a sequence WM → LM on the same column, with an overall analysis time of 45 min, including column conditioning. Despite their simplicity, these methods offered excellent resolution (selectivity coefficient ␣ was 2.87 and 2.20 for water- and lipid-soluble vitamins respectively), linearity and retention time repeatability (% RSDs were: 4.62; 2.69; 1.83 and 1.25 for vitamin C, B6, A and E respectively, in 20 replicate analyses), and
Fig. 1. Typical chromatograms obtained by HPLC-DAD analysis of lipid-soluble (A and B) and water-soluble (C and D) vitamins with methods described in Section 2.2.2. The detection wavelengths are A = 320 nm, B = 290 nm, C = 290 nm, D = 245 nm; and number correspond to 1 = retinol acetate, 2 = ␣-tocopherol, 3 = pyridoxine hydrochloride, 4 = ascorbic acid.
464
S. Gabbanini et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 461–467
Table 2 Validation of the analytical methods for the analysis of vitamins in the receptor solution. An asterisk indicates statistically different results (p < 0.05; two-sided) between settings (see text). Analyte
Nominal concentration [mg/mL]
Accuracy: average % recovery (n = 15)
Precision: intra-day (n = 5)/inter-day (n = 15) % RSD
Robustness: average % recovery ± % RSD (n = 5) Setting 1/setting 2
Ascorbic acid Level 1 Level 2 Level 3
0.0020 0.0080 0.0800
85.8 92.7 105.3
2.58/10.51 1.31/7.86 0.18/3.53
93.4 ± 1.7/96.1 ± 1.0* 96.6 ± 1.1/96.3 ± 0.8 104.1 ± 0.3/101.6 ± 0.8*
Pyridoxine Level 1 Level 2 Level 3
0.0021 0.0084 0.0840
102.2 104.0 107.0
1.79/4.56 0.45/0.76 0.15/1.50
101.4 ± 1.2/99.7 ± 0.9 95.2 ± 1.1/96.3 ± 0.8 101.3 ± 0.2/106.1 ± 0.4*
Retinol acetate Level 1 Level 2 Level 3
0.0040 0.0200 0.0400
97.1 101.9 103.1
0.88/5.17 0.25/3.16 0.54/2.48
96.8 ± 0.6/97.0 ± 0.5 100.2 ± 0.2/99.8 ± 0.3 102.5 ± 0.3/103.6 ± 0.8
␣-Tocopherol Level 1 Level 2 Level 3
0.0044 0.0220 0.0440
108.6 101.6 103.8
4.16/6.87 1.75/2.73 0.82/1.52
102.4 ± 1.7/103.1 ± 1.0 101.9 ± 1.2/102.3 ± 0.8 100.2 ± 0.3/103.6 ± 0.8*
Table 3 HPLC-DAD replicate analyses (intra-day, n = 5) of the vitamins incorporated aqueous solutions and O/W or W/O emulsions used as donor in the diffusion experiments. Peak purity test indicates the method selectivity (max score = 999.9). Analysis methods and sample treatment are reported in Sections 2.2.2 and 2.3.3 respectively. Emulsion/vitamins
Analyte
Average peak purity ± SD
Average content [mg/mL] ± SD
W/O/water-soluble
Ascorbic acid Pyridoxine
Average Rt [min] ± SD 2.33 ± 0.01 6.68 ± 0.08
999.0 ± 0.1 998.6 ± 0.8
16.48 ± 0.20 17.73 ± 0.08
O/W/water-soluble
Ascorbic acid Pyridoxine
2.33 ± 0.01 6.60 ± 0.07
999.0 ± 0.2 998.4 ± 0.5
16.16 ± 0.21 18.09 ± 0.10
W/O/lipid-soluble
Retinol acetate ␣-Tocopherol
5.06 ± 0.12 11.11 ± 0.07
987.4 ± 16.3 973.2 ± 32.8
18.07 ± 0.20 33.26 ± 0.11
O/W/lipid-soluble
Retinol acetate ␣-Tocopherol
5.05 ± 0.11 11.09 ± 0.07
977.3 ± 26.1 970.3 ± 26.9
17.62 ± 0.12 32.10 ± 0.13
Aqueous solution
Ascorbic acid Pyridoxine
2.47 ± 0.30 6.46 ± 0.07
993.6 ± 10.3 996.9 ± 3.1
17.04 ± 0.08 16.90 ± 0.11
lation prepared for the subsequent kinetic investigation (O/W and W/O emulsions), after dilution in methanol or methanol/water and filtration (see Section 2.3.3). Overall no significant interference was suffered from the matrix or the many components in the emulsions, and DAD peak purity was >940 (94%) for any vitamin in any composition. This served also to provide analytical data for CD values to be used in the kinetic investigation. Results are collected in Table 3. 3.3. Vitamins’ stability in donor/receptor media Oxidative degradation of ascorbic acid in aqueous solution has been reported to proceed with apparent zero-order kinetics with recovery (linearly) dropping from 100% to 1% in less than 24 h at 25 ◦ C and pH 6.9 [23]. From the reported data it is possible to extract an empirical equation describing % recovery decay of the form: % recovery = 100 − kt, with k = −8.23 (%/h). To maintain ascorbic acid in the reduced form following the report by Shephard et al. [22], we added 0.4 mg/mL DTT to any sample or emulsion, however some progressive loss of ascorbate recovery was observed during method validation (standards solutions were kept at 4 ◦ C for 1–2 days between analyses at 30 ◦ C) despite the presence of DTT. To investigate the stability of all tested vitamins under our experimental settings we analyzed at time intervals donor (O/W and W/O emulsions) and receptor (pH 7.4) standard preparations containing 0.4 mg/mL DTT and either water-soluble or lipid-soluble vitamins stored at 4 ◦ C in the dark or at 34 ◦ C in a thermostatted bath (to reproduce the condition used for diffusion kinetics) for 48 h. While pyridoxine and ␣-tocopherol were perfectly stable, both ascorbate
and retinol acetate were significantly degraded in receptor solution at 34 ◦ C, degradation following apparent zero-order kinetics (Appendix A, Fig. S3), as previously described. The measured rate constants for degradation were k = 0.68 (%/h) and k = 0.86 (%/h) for ascorbic acid and retinol acetate, corresponding to recoveries of 68% and 59% respectively at 48 h. The rate of ascorbic acid degradation was decreased by over 10 times by DTT (despite the higher pH and temperature, by comparison with literature data [23]) however it was not completely suppressed. At 4 ◦ C in receptor solution in the dark only ascorbic acid showed some (slower) degradation with apparent k = 0.17 (%/h) corresponding to 92% recovery at 48 h. The stability of retinol acetate under these conditions suggests that photo-degradation rather that oxidation is responsible for its decay. Only ascorbic acid showed instability in the donor formulations at 34 ◦ C with decay rate constants of 0.21 and 0.32 (%/h) in W/O and O/W emulsions respectively, confirming previous observations on the dependence of ascorbic acid degradation on the medium [24]. All analytes were instead stable in the W/O and O/W emulsions at 4 ◦ C for 48 h. Degradation data of ascorbic acid and retinol acetate in receptor at 34 ◦ C were used to correct the results from diffusion kinetics. 3.4. Diffusion through synthetic membranes We operated under “infinite dose” conditions and avoid significant decrease of analytes’ concentration in the donor during the diffusion experiment. The trans-membrane diffusion profile was determined by HPLC analysis of the receptor medium, sampling
S. Gabbanini et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 461–467
Fig. 2. Typical kinetic plots of the cumulative amount Q of vitamins released from O/W emulsion through polyethersulfone membranes under static setting. Plots are obtained from one single experiment each for water-soluble and lipid-soluble vitamins. Legend: 䊉 = pyridoxine, = ascorbic acid, = ␣-tocopherol, = retinol acetate.
from each diffusion cell at regular intervals. Under the steady-state approximation, the diffusion process occurs in accordance with Fick’s first law (Eq. (1)). dQ = PS (CD − CR ) dt
(1)
Here PS (cm/h) is the skin permeability coefficient and CD and CR (g/cm3 ) are the drug concentrations in the donor and receptor chambers, while Q is the cumulative amount diffused through a unit of skin surface (g/cm2 ). Integration of Eq. (1) gives: Q = PS (CD − CR )t
(2)
Fig. 2 shows that with polyethersulfone membranes the quantity Q varied linearly with time, in agreement with Eq. (2) assuming that CD CR (infinite dose condition). Some deviation from linear behaviour was observed during the diffusion of water-soluble vitamins from O/W emulsion due to the very rapid diffusion, being faster than re-equilibration of vitamins in the viscous donor emulsion. The Flux J (g/cm2 /h) calculated from the slope of Q vs time during the first part of the kinetic study, afforded the skin permeability coefficient (Eq. (3)). PS =
J CD
(3)
465
Results obtained with polyethersulfone membranes are collected in Table 4. In general water-soluble pyridoxine hydrochloride (B6 ) and ascorbate (C) were released much more rapidly and effectively than lipid-soluble retinyl acetate (A) and ␣-tocopherol (E), qualitatively paralleling their solubility in the aqueous receptor fluid, which appeared to be the main driving force of the diffusion process. Interestingly the release of vitamins depended on the type of cosmetic formulation. Formulations having water in the external phase (O/W emulsions or lotions), released vitamins significantly more rapidly than W/O emulsion, although the actual contribution of the formulation depended on the physicochemical properties of the analyte. O/W emulsions were as effective as lotions in the release of water-soluble vitamins and approximately three times as effective as W/O emulsions, regardless of the structure of the vitamin. Conversely the permeability coefficient of retinyl acetate was only marginally increased (+20%) on changing from W/O to O/W emulsions. Somewhat surprising was instead the relative increase (five times) recorded for ␣-tocopherol. These findings can be rationalized considering the overall diffusion process: donor → membrane → aqueous receptor: water-soluble vitamins clearly benefit from an aqueous donor (or external phase) where they are more soluble maximizing the diffusion gradient (CD − CR ). Conversely with the very apolar ␣-tocopherol (log P = 12.2) having oil in the external phase (W/O emulsions) brings to very unfavourable partition (donor/receptor), which is not balanced by the larger concentration gradient. With less apolar retinyl acetate (log P = 7.9) the two factors, diffusion gradient and partition, almost compensate each other, bringing to only minor changes in Ps value. These findings suggest that polyethersulfone membranes, although being more permeable than more sophisticate models of human epidermis (vide infra), would provide a convenient and economical model to study the intrinsic release properties of different formulations. Polycarbonate membranes (0.01 m pore size) under identical experimental settings offered almost no barrier to the diffusion of vitamins. With aqueous donor the diffusion was faster than we could reliably monitor and brought to equilibration of the donor and receptor compartments (CD = CR ) within the experiment time. This result confirms that polycarbonate membranes are not a valuable model to study the release of active ingredients from cosmetic formulations. 3.5. Diffusion through reconstructed human epidermis With RHE membranes experiments were conducted as described for synthetic membranes (see Section 3.4). Due to the slower permeation encountered with RHE, the kinetic study was extended for 10 h with a follow-up the next day up to 24 h.
Table 4 Donor concentration (CD ), Lag-time (Lt ), flux (J) and permeability coefficient (PS ) through polyethersulfone membranes and RHE membranes of vitamins released from cosmetic formulations at 34 ◦ Ca Donor concentrations (CD ) is reported in Table 3. Polyethersulfone membranes b
RHE membranes
J × 10 [mg/cm /h] 3
2
PS × 10 [cm/h]
Lt [h]b
J × 103 [mg/cm2 /h]
PS × 105 [cm/h]
0.35 0.25 0.02 0.037
3.86 ± 0.32 7.54 ± 1.05 4.20 ± 1.03 ∼∞
43.52 ± 3.87 28.42 ± 2.52 0.050 ± 0.009 ∼0
264.1 ± 23.5 160.3 ± 14.2 0.28 ± 0.05 ∼0
0.84 2.55 0.02 0.04
1.05 ± 0.12 2.10 ± 0.24 6.15 ± 1.38 >21
141.42 ± 6.92 85.04 ± 6.64 0.063 ± 0.019 ∼0
875.1 ± 42.8 470.1 ± 36.7 0.36 ± 0.11 ∼0
57.82 ± 2.46 58.94 ± 2.20
0.95 ± 0.14 1.62 ± 0.18
145.11 ± 7.72 51.65 ± 4.34
851.6 ± 45.3 305.6 ± 25.7
3
Formula
Analyte
Lt [h]
W/O emulsion
Ascorbic acid Pyridoxine Retinol acetate ␣-Tocopherol
∼0 ∼0 0.31 ± 0.23 4.12 ± 1.34
309.8 390.0 5.18 1.33
± ± ± ±
5.6 4.2 0.32 1.22
19.31 22.86 0.29 0.040
± ± ± ±
O/W emulsion
Ascorbic acid Pyridoxine Retinol acetate ␣-Tocopherol
∼0 ∼0 0.45 ± 0.15 0.34 ± 0.26
966.1 1061.10 6.09 9.74
± ± ± ±
13.2 44.55 0.41 1.21
61.69 60.81 0.35 0.30
± ± ± ±
Aqueous solution
Ascorbic acid Pyridoxine
∼0 ∼0
a b
Mean ± SD (n = 3). Estimated from extrapolation of the diffusion curve to Q = 0.
984.4 ± 41.8 996.0 ± 37.2
466
S. Gabbanini et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 461–467
Fig. 3. Kinetic plots of the cumulative amount Q of vitamins released from O/W emulsion through RHE membranes under static setting. Plots are averaged from three experiments each for water-soluble (A) and lipid-soluble (B) vitamins. The large error bars account mainly for variation in lag-time. Legend: = ascorbic acid, = pyridoxine, = retinol acetate, 䊉 = ␣-tocopherol.
Typical kinetic plots are depicted in Fig. 3, showing reasonably linear diffusion profiles as previously discussed for synthetic membranes, which suggests that permeation occurs mostly by passive, diffusion-like, transport processes. With RHE, significant deviation from linear behaviour was observed only for ascorbate in the final portion of the diffusion experiment, conceivably due to airoxidation of the vitamin in the receptor solution. Large lag-times were always observed for the permeation through RHE, particularly for lipid-soluble vitamins, possibly due to the high affinity for the Stratum Corneum (SC) and phospholipids’ bi-layers in epidermis, as previously reported both for RHE and native epidermis. [1,2] Indeed ␣-tocopherol gave some detectable diffusion in the receptor only after 22 h, resulting in no measurable human skin permeability under our experimental settings. Conversely, we attributed the shorter lag-times necessary to establish the diffusion of ascorbic acid and pyridoxine to the predominance of the intercellular diffusion route for these small water-soluble molecules. [25] Results are summarized in Table 4. In general RHE was much more selective than synthetic membranes and permeability coefficients were up to two orders of magnitude lower than those recorded with polyethersulfone membranes. Particularly ascorbic acid and pyridoxine had Ps values ∼10 and ∼20 times lower on RHE, while for retinol acetate the release was ∼100 times slower. The high selectivity of RHE suggests it is a much more valuable model to mimic the behaviour in vivo than synthetic membranes. As mentioned, for ␣-tocopherol the permeability was decreased so much to prevent measurement with our approach. To a qualitative level the role of the formulation in determining the release through RHE was similar to that observed with polyethersulfone. This very relevant point has been addressed with contradicting outcomes in the literature: while some investigations using RHE or animal skin failed detecting a clear role of the formulations in the trans-epidermal absorption of tocopherols, [26] others are in line with our findings. [27] 4. Conclusions In the present study a sensitive, accurate and reproducible analytical method is described to investigate the kinetics of diffusion of lipid- and water-soluble vitamins contained in cosmetic formulations, through Reconstructed Human Epidermis or synthetic polyethersulfone membranes in vitro. The described method and instrumental setting are sufficiently simple and cost-effective to
be suited to the quality-efficacy assessment of cosmetic formulations. Due to the limited dispersion of experimental kinetic results (good repeatability), the use of RHE favourably compares with the larger variability normally encountered with native human epidermis, while overcoming the problems of limited availability. Although polyethersulfone membranes are less selective barriers to the diffusion of vitamins than RHE, they can be a convenient model to evaluate the intrinsic release ability of the formulation before testing on RHE. Conversely polycarbonate membranes offer no significant barrier to the release of active molecules from cosmetic formulations. Acknowledgments This work was financially supported by BeC srl (Forlì, Italy) and MIUR (Rome, Italy). The authors are grateful to Prof. Lorenzo Rodriguez (University of Bologna) for helpful discussion. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jpba.2010.01.029. References [1] C. Antille, C. Tran, O. Sorg, J.-H. Saurat, Penetration and metabolism of topical retinoids in ex vivo organ-cultured full-thickness human skin explants, Skin Pharmacol. Physiol. 17 (2004) 124–128. [2] A. Mavron, V. Raufast, D. Redoules, Skin absorption and metabolism of a new vitamin E prodrug, ␦-tocopherol-glucoside: in vitro evaluation in human skin models, J. Control. Rel. 100 (2004) 221–231. [3] EU regulation 76/768/EEC, February 2003. [4] A. Kligman, The future of cosmeceuticals: an interview with Albert Kligman, MD, Dermatol. Surg. 31 (2005) 890–891. [5] D. Kligman, Cosmeceuticals, Dermatol. Clin. 18 (2000) 609–615. [6] C.R. Thornfeldt, Cosmeceuticals: separating fact from vodoo science, Skinmed 4 (2005) 214–220. [7] (a) OECD, Skin Absorption: In Vitro Method, Test Guideline 428, 2004.; (b) OECD, Guidance Notes for the Estimation of Dermal Absorption Values, 2008.; (c) OECD Guidance Document for the Conduct of Skin Absorption Studies Number 28, 2004. [8] (a) Colipa Guidelines for Percutaneous Absorption/Penetration, 1997.; (b) Colipa Guidelines for the Evaluation of the Efficacy of Cosmetic Products, Revised Version, May 2008.; (c) Colipa ID 7: Serious about Cosmetics—Serious about Alternative Methods. Updated Version, May 2007. [9] M.K. Das, A. Bhattacharya, S.K. Ghosal, Effect of different terpene-containing essential oils on percutaneous absorption of trazodone hydrochloride through mouse epidermis, Drug Deliv. 13 (2006) 425–431.
S. Gabbanini et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 461–467 [10] A. Mittal, U.V.S. Sara, A. Ali, M. Aquil, The effect of penetration enhancers on permeation kinetics of nitrendipine in two different skin models, Biol. Pharm. Bull. 31 (2008) 1766–1772. [11] S. Schreiber, A. Mahmoud, A. Vuia, M.K. Rübbelke, E. Schmidt, M. Schaller, H. Kandárová, A. Haberland, U.F. Schäfer, U. Bock, H.C. Korting, M. Liebsch, M. Schäfer-Korting, Reconstructed epidermis versus human and animal skin in skin absorption studies, Toxicol. In Vitro 19 (2005) 813–822. [12] D. Howes, R. Guy, J. Hadgraft, J. Heylings, U. Hoeck, F. Kemper, H. Maibach, J. Marty, H. Merk, J. Parra, D. Rekkas, I. Rondelli, H. Schaefer, U. Tauber, N. Verbiese, Methods for assessing percutaneous absorption, in: ECVAM Workshop Report 13, ATLA 24 (1996) 81–106. [13] J. Reichling, U. Landvatter, H. Wagner, K.-H. Kostka, U.F. Schaefer, In vitro studies on release and human skin permeation of Australian tea tree oil (TTO) from topical formulations, Eur. J. Pharm. Biopharm. 64 (2006) 222–228. [14] K. Cal, M. Sopala, Ex vivo skin absorption of terpenes from Vicks VapoRub® ointment, Med. Sci. Monit. 14 (2008) PI19–PI23. [15] F. Netzlaff, C.-M. Lehr, P.W. Wertz, U.F. Schaefer, The human epidermis models EpiSkin, SkinEthic and EpiDerm: an evaluation of morphology and their suitability for testing phototoxicity, irritancy, corrosivity, and substance transport, Eur. J. Pharm. Biopharm. 60 (2005) 167–178. [16] M.D. Ynsa, E. Gontier, A. Mavon, P. Moretto, M. Rosdy, Comparative study between reconstructed and native human epidermis using nuclear microscopy, Nucl. Instr. Meth. Phys. Res. B 249 (2006) 710–714. [17] M. Schaefer-Korting, U. Bock, W. Diembeck, H.-J. Duesing, A. Gamer, E. HaltnerUkomadu, C. Hoffmann, M. Kaca, H. Kamp, S. Kersen, M. Kietzmann, H.C. Korting, H.-U. Kraechter, C.-M. Lehr, M. Liebsch, A. Mehling, C. MuellerGoymann, F. Netzlaff, F. Niedorf, M.K. Ruebbelke, U. Schaefer, E. Schmidt, S. Schreiber, H. Spielmann, A. Vuia, M. Weimer, The use of reconstructed human epidermis for skin absorption testing: results of the validation study, ATLA 36 (2008) 161–187.
467
[18] M. Schaefer-Korting, A. Mahmoud, S.L. Borgia, B. Brueggener, B. Kleuser, S. Schreiber, W. Mehnert, Reconstructed epidermis and full-thickness skin for absorption testing: influence of the vehicles used on steroid permeation, ATLA 36 (2008) 441–452. [19] S. Gabbanini, E. Lucchi, M. Carli, E. Berlini, A. Minghetti, L. Valgimigli, In vitro evaluation of the permeation through reconstructed human epidermis of essentials oils from cosmetic formulations, J. Pharm. Biomed. Anal. 50 (2009) 370–376. [20] D. Pabla, H. Zia, A comparative permeation/release study of different testosterone gel formulations, Drug Deliv. 14 (2007) 389–396. [21] Y. Frum, G.M. Eccleston, V.M. Meidan, Evidence that drug flux across synthetic membranes is described by normally distributed permeability coefficients, Eur. J. Pharm. Biopharm. 67 (2007) 434–439. [22] A.B. Shephard, S.C. Nichols, A. Braithwaite, Moisture induced solid phase degradation of l-ascorbic acid: Part 1: a kinetic study using tristimulus colorimetry and a quantitative HPLC assay, Talanta 48 (1999) 585–593. [23] M.-L. Chang, C.-M. Chang, Simultaneous HPLC determination of hydrophilic whitening agents in cosmetic products, J. Pharm. Biomed. Anal. 33 (2003) 617–626. [24] R. Austria, A. Semenzato, A. Bettero, Stability of vitamin C derivatives in solution and topical Formulations, J. Pharm. Biomed. Anal. 15 (1997) 795–801. [25] B.W. Barry, Dermatological Formulations Percutaneous Absorption, Marcel Dekker Inc., New York, 1983. [26] M.S. Roberts, K.A. Walters (Eds.), Dermal toxicity assessment, 2nd edition, Informa Healthcare USA Inc., New York, 2008. [27] F. Dreher, F. Fouchard, C. Patouillet, M. Andrian, J.T. Simonnet, F. Benech-Kieffer, Comparison of cutaneous bioavailability of cosmetic preparations containing caffeine or ␣-tocopherol applied on human skin models or human skin ex vivo at finite doses, Skin Pharmacol. Appl. Skin Physiol. 1 (Suppl. 15) (2002) 40–58.
Contents lists available at ScienceDirect
Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba
Analysis of in vitro release through reconstructed human epidermis and synthetic membranes of multi-vitamins from cosmetic formulations Simone Gabbanini a , Riccardo Matera b , Claudia Beltramini a , Andrea Minghetti a , Luca Valgimigli b,∗ a b
BeC S.r.l., R&D division, Via C. Monteverdi 49, 47100 Forlì, Italy University of Bologna, Faculty of Pharmacy, Dept. Organic Chemistry “A. Mangini”, via San Giacomo 11, 40127 Bologna, Italy
a r t i c l e
i n f o
Article history: Received 22 September 2009 Received in revised form 7 January 2010 Accepted 14 January 2010 Available online 21 January 2010 Keywords: Vitamins RHE Cosmetics Percutaneous absorption HPLC
a b s t r a c t A convenient method for in vitro investigation of the release of lipid- and water-soluble vitamins from cosmetic formulations was developed. The permeation of (d)-␣-tocopherol (vitamin E), retinyl acetate (pro-vitamin A), ascorbic acid (vitamin C) and pyridoxine (vitamin B6) through SkinEthic® reconstructed human epidermis (RHE), and synthetic polyethersulfone and polycarbonate membranes was studied in vitro using a Franz-type diffusion apparatus, coupled either to a spectrophotometer for continuous reading (dynamic setting) or to HPLC-DAD analysis of the receptor medium (static setting). O/W and W/O emulsions were compared with simple aqueous solutions for their kinetic of vitamins release, to evaluate the influence of the cosmetic formulation on the bioavailability of active ingredients. Results indicate that synthetic membranes offer a limited barrier to the diffusion of vitamins, but may provide information on the release ability of the formulation. Penetration was more effective when water was the external phase of the formulation, i.e. W/O emulsions were less effective in the release of vitamins than O/W emulsion or aqueous solutions. RHE (17 days old) offered a significantly higher barrier to penetration of vitamins, as expected for native human epidermis. The relative ranking in coefficient of permeability (Ps (cm/h)) was: ascorbic acid > pyridoxine retinyl acetate > ␣-tocopherol ∼0, the absolute values depending on the formulation. The method herein described showed to be a practical and convenient tool for the quality-control and efficacy evaluation of cosmetic formulations. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Vitamins are common ingredients of cosmetic formulations and a number of claims are based on the implicit assumption that these bio-active molecules are effectively released from the formulation into epidermis and possibly through epidermis into the derma and subcutaneous tissues. However studies assessing the skin absorption of vitamins released from cosmetic formulations are extremely rare in the literature and mostly limited to retinoids (vitamin A) [1] or vitamin E. [2] The lack of specific obligation for bioavailability studies and the prohibition of animal testing for cosmetic formulations, expressed in current EU regulations [3] do not stimulate the objective evaluation of these efficacy-related parameters. The term cosmeceuticals was coined in mid-eighties to indicate a particular category of personal-care products that could be placed at the
Abbreviations: RHE, reconstructed human epidermis; egf, epidermis growth factor; W/O, water in oil; O/W, oil in water; Ps, coefficient of permeability; LOD, limit of detection; LOQ, limit of quantitation; DTT, (dl)-Dithiothreitol; TFA, trifluoroacetic acid. ∗ Corresponding author. Tel.: +39 051 2095683; fax: +39 051 2095688. E-mail address: [email protected] (L. Valgimigli). 0731-7085/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jpba.2010.01.029
border-line between cosmetics and pharmaceuticals [4]: cosmetic products containing active ingredients meant to have beneficial physiologic effects resulting from their pharmacologic properties [5]. Since there is no legal definition of cosmeceuticals, these products are often formally classified as cosmetics or in some cases as drugs, the distinction being sometimes arbitrary and varying from country to country. For instance skin-protectants (such as those based on vitamin E) are classified as drugs in USA and as cosmetics in Europe. This category of products has been exponentially increasing its importance during last two decades and a relevant percentage of products currently found in EU market could be classified as cosmeceuticals. However a scientifically sound evaluation of their efficacy is often lacking, as have been lamented by authoritative investigators [5,6]. In compliance with EU regulations and in line with OECD [7] and COLIPA guidelines [8] we aimed to set-up a convenient method, based on a custom designed Franz-type diffusion apparatus, for in vitro evaluation of the kinetic of release of vitamins form cosmetic/cosmeceutical formulations. In order to avoid the use of animal skin [9–12], which would conflict with the ethical principles of EU regulations, and to evaluate a convenient substitute for human epidermis from cadaver or surgery [11,13,14], we tested SkinEthic® reconstructed human epidermis (RHE). Previous studies
462
S. Gabbanini et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 461–467
have shown that RHE is reasonably similar to native human epidermis both in terms of morphology and lipid composition [15,16]. It was found less selective than native epidermis to the permeation of specific drugs [15], however it has been indicated as a valid substitute for in vitro testing of topical formulations [17,18,19]. Synthetic membranes have also been employed for release kinetics studies [20,21], so we compared RHE to synthetic polyethersulfone and polycarbonate membranes. Two different settings have been initially compared for the analysis of vitamins in the receptor fluid: continuous spectrophotometric reading by circulating the fluid through a flow cell (dynamic setting, see Appendix A), or discontinuous sampling followed by HPLC-DAD analysis (static setting). The former has lately been abandoned due to limited performance. Three different types of cosmetic formulation were considered in this investigation: O/W emulsions (creams), W/O emulsions (ointments) and aqueous solutions (lotions), to evaluate the influence of the formulation in the release of lipid- and water-soluble vitamins. Ascorbic acid (vitamin C), pyridoxine hydrochloride (vitamin B6 ), retinol acetate (pro-vitamin A), and ␣-tocopherol (vitamin E) were chosen as representative exempla of commonly employed waterand lipid-soluble vitamins. 2. Materials and methods 2.1. Materials The standard vitamins used for method validation were: ascorbic acid (99.9%), pyridoxine hydrochloride (99.9%), (dl)␣-tocopherol (99.6%), retinol acetate (97.4%) from Supelco (Bellefonte, PA, USA); trifluoroacetic acid (TFA, ≥99.0%), (dl)Dithiothreitol (DTT; >99.0%), sodium phosphate dibasic anhydrous (≥99%), sodium chloride (≥99.5%), magnesium sulfate hexahydrate (>99.0%), isopropanol (≥99.8%), methanol (≥99.8%) from Sigma, (St. Louis, MO, USA); potassium dihydrogen phosphate (≥99.5%) from Merck (Darmstadt, Germany). Reconstituted Human Epidermis 12 days old 4.0 cm2 , and Growth Medium (1.5 mM calcium chloride, 25 mg/mL gentamycin, 5 mg/mL insulin, 1 ng/mL egf) were purchased from SkinEthic Laboratories (Nice, France). Polyethersulfone membranes (0.2 m, Supor® 200, Pall corp., New York, USA) and polycarbonate membranes, (0.01 m, Whatman, GE Healthcare, Uppsala, Sweden) were from FAVS s.n.c. (Bologna, Italy). 2.2. Apparatus and chromatographic conditions 2.2.1. Franz-type diffusion cells Franz-Type diffusion cells (diffusion surface: 1.54 cm2 ; internal volume: 14.8 ± 0.1 mL), designed in our labs to optimize the use of SkinEthic® RHE disks, were previously described [19] and are shown in Appendix A (Fig. S1). Homogenous concentration of the analytes within the receptor compartment was ensured during preliminary tests by monitoring the diffusion of blue-colored crystal violet solutions. 2.2.2. HPLC-DAD analysis All analyses were carried out using a Thermo Accela Pump equipped with Accela Autosampler and Surveyor Photo-Diode Array detector (Thermo Scientific, San Jose, CA, USA). The LC was performed on a Synergi C18 Hydro-RP 4 m column (150 mm × 4.6 mm I.D.) equipped with a guard column (C18, 4.0 mm × 3.0 mm) from Phenomenex (Torrance, CA) operating at 30 ◦ C. Two solvent systems were employed: A = CH3 OH + TFA (0.25 mL/L) and B = H2 O + TFA (0.25 mL/L). The injection volume was 5 L. Analyses of the lipid-soluble vitamins (method LM) were performed isocratically at 1.0 mL/min for 13 min with the mobile phase A-B (80:20, v/v). Retinol acetate and ␣-tocopherol were detected at
320 and 290 nm, respectively. Analyses of the water-soluble vitamins were performed with method WM, eluting at 1.0 mL/min in the following gradient program for 17 min: t = 0, A–B (0:100, v/v); t = 3, A–B (0:100, v/v); t = 10, A–B (10:90, v/v); t = 15, A–B (10:90, v/v); t = 17, A–B (0:100, v/v). Ascorbic acid and pyridoxine were detected at 245 and 290 nm, respectively. 2.3. Assay procedure 2.3.1. Calibration for HPLC-DAD analysis Standard solutions in five levels from 0.001 to 0.100 mg/mL were analyzed in triplicate. Ascorbic acid and pyridoxine hydrochloride were dissolved in pH 7.4 phosphate buffered saline solution with (dl)-Dithiothreitol (DTT, 0.4 mg/mL), whereas (dl)-␣-tocopherol and retinol acetate solutions were prepared in methanol. The LOD was determined from repeated analyses, as the concentration that yields a signal-to-noise (S/N) ratio of at least 3:1 and the LOQ was obtained as the concentration that yields S/N ratio of 10:1. Linearity and LOD/LOQ data are collected in Table 1. 2.3.2. Validation of the analytical procedure Standard receptor medium solutions containing three-levels concentration of either lipid-soluble vitamins or water-soluble vitamins were subjected to five replicate analyses for three nonconsecutive days (15 replicates for each level). Solutions were stored at +4 ◦ C in the dark to minimize oxidation and thermalphotochemical degradation of vitamins. Accuracy was evaluated from % recovery with respect of theoretical content, while precision data were obtained from intra-day and inter-day % relative SD. To test the robustness standard solutions were repeatedly analyzed (n = 5) intra-day using the method described in Section 2.2.2 (setting 1) and changing the chromatographic conditions (setting 2) as follows: T = 25 ◦ C (−5 ◦ C), flow rate 0.8 mL/min (−20%) and decreasing the content of TFA in the mobile phase to 0.20 mL/L (−20%). To test the selectivity, vitamins content was determined in the donor compositions (W/O emulsions, O/W emulsions and aqueous receptor medium) as detailed in Section 2.3.3. The receptor medium was phosphate saline buffer pH 7.4 (8 g/L NaCl, 0.2 g/L KCl, 0.2 g/L KH2 PO4 and 1.15 g/L NaHPO4 in bi-distilled water), 3% w/w of polysorbate-20 and 0.25% w/w of isopropanol and DTT 0.4 mg/mL [22]. 2.3.3. Preparation and analysis of the cosmetic formulations (donor) Four creams being Oil in Water or Water in Oil emulsions, each containing the water-soluble vitamins, or the lipid-soluble vitamins and DTT 0.4 mg/mL [22], were prepared as described in Appendix A and stored at 4 ◦ C before use. The nominal vitamin C and B6 content were 1.7% (w/w) each, whereas the vitamin A acetate and E were 1.8% and 3.0% (w/w), respectively. The exact amount of vitamin content was assessed by HPLC-DAD analyses. Briefly, water-soluble vitamins containing creams (100 mg) were suspended in CH3 OH (10 mL), then 1 mL of this suspension was diluted up to 10 mL with H2 O + TFA (0.25 mL/L). The suspension was filtered with a syringe through 0.45 m syringe filter and analyzed. The lipid-soluble vitamins containing creams (100 mg) were suspended in CH3 OH (10 mL), diluted up to 10 mL with the same solvent and filtered (0.45 m) prior to analysis. 2.3.4. Study of vitamins’ stability in the donor and receptor media Standard receptor medium solutions, W/O emulsions and O/W emulsions, each containing DTT 0.4 mg/mL and the water-soluble (B6 , C) or lipid-soluble (A, E) vitamins were stored at 4 ◦ C (in the
S. Gabbanini et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 461–467
463
Table 1 Vitamins five levels calibration parameters expressed as [area] vs [concentration in mg/mL], linearity and LOD/LOQ values for HPLC-DAD analysis. Analyte
LOD [g/mL]
LOQ [g/mL]
Slope
Ascorbic acid Pyridoxine Retinol acetate ␣-Tocopherol
0.24 0.03 0.10 1.50
0.81 0.12 0.34 5.10
1.519 ± 0.015 E+07 1.199 ± 0.003 E+07 3.973 ± 0.082 E+07 1.924 ± 0.019 E+06
dark) and 34 ◦ C under normal day-lighting and periodically analyzed for 48 h. 2.3.5. Preparation of reconstructed human epidermis Twelve days old reconstructed human epidermis (RHE) held on polycarbonate disks were revitalized by addition of 4 mL of growth medium for skin for day (1.5 mM calcium chloride, 25 mg/mL gentamycin, 5 mg/mL insulin, 1 ng/mL egf) in sterile petri boxes under laminar flux cabinet and incubated at 37 ◦ C in 5% CO2 atmosphere in a CO2 incubator for five days prior to investigation. 2.3.6. In vitro diffusion experiments Diffusion cells were mounted on a Magnetic 6 Stirrer and connected to a water bath at 34 ◦ C. The receptor medium had the composition described in Section 2.3.2. The receiver compartment, equipped with 10 mm × 2 mm stirring bar, was filled with receptor medium and the skin membranes were cut from plastic support and mounted with their polycarbonate filter with the epidermal side facing upward into the donor compartment. With synthetic membranes (polyethersulfone or polycarbonate) the same procedure was followed. Donor and receptor chambers were watertight closed with a metallic clamp, and a dose of 0.5–1.0 mL of the cosmetic formulation (donor) was applied to the surface of the membrane. At time intervals of 30–60 min within 5–24 h from application, a 100 L aliquot of the receptor medium was withdrawn and immediately replaced with an equal volume of fresh buffer [13]. Samples that could not be analyzed immediately were sealed under nitrogen and stored at +4◦ C. 2.3.7. Data analysis The cumulative amount of substance diffused through a unit of surface of skin (Q; g/cm2 ) was obtained by HPLC analysis of the sampled receptor medium, corrected for the analytes subtracted at every sampling [12], and for the degradation of vitamins (A and C) in the receptor medium.
R2
Intercept −20,930.2 3783.2 6447.6 −1334.5
± ± ± ±
8213.0 1929.8 20,423.0 1578.9
0.9997 0.9999 0.9983 0.9998
were sufficiently selective to allow the analysis of vitamins directly in the cosmetic formulations (O/W and W/O emulsions) without preliminary sample treatment other than dilution (see Section 3.2). 3.2. Validation of analytical methods and analysis of donor compositions We tested accuracy, precision and robustness on standard solutions containing three levels of each target vitamin. The content of each vitamin was analyzed for three non-consecutive days (15 replicate analyses): accuracy and precision were evaluated from mean percent recovery and % RSD as detailed in Table 2. Overall the performance of the analytical method was judged adequate to the subsequent kinetic investigation. Only with the two lower concentrations of vitamin C recoveries were lower than expected and tended to decrease in the analysis performed on subsequent days. Indeed the inter-day RSD was significantly larger than intra-day. We attributed it to air-oxidation of ascorbate in the standard solutions, which could not be completely suppressed by DTT (vide infra). To overcome this source of error, care was taken to analyze receptor samples containing ascorbate immediately upon withdrawal (or within 1 h) during the kinetic investigation. Robustness was tested by performing two parallel sets (n = 5, intra-day) of analyses on three-levels standard vitamins solutions (containing 0.4 mg/mL DTT) using our method (setting 1) or by changing sensitive parameters like temperature, flow rate, and eluent composition by a reasonable amount (setting 2: temperature, −5 ◦ C; flow rate, −20%; and [TFA], −20%; see Section 2.3.2). Significantly different (p < 0.05) recovery values are marked in Table 2 with an asterisk. Only in 4 out of 12 samples % recovery was significantly different and differences were always within ±5%, indicating a good robustness on the method. To test the selectivity, the same analytical methods was used to determine the actual content of vitamins in each donor formu-
3. Results and discussion 3.1. HPLC analysis In order to be suited to the kinetic investigation the analysis protocol had to be sufficiently quick to keep pace with hourly sampling and avoid column changing. We employed a Synergi Hydro-RP C18 column (with polar end-capping), which was suited to both polar and apolar analytes, and preferred separate methods for lipid-soluble (LM) and water-soluble (WM) vitamins, with run times as short as 13 min and 17 min (Fig. 1) for lipid- and water-soluble vitamins respectively. According to the experimental design the diffusion kinetics of lipid- and water-soluble vitamins could be studied separately (with sampling frequency as short as 20–30 min) or, if needed, combined in a sequence WM → LM on the same column, with an overall analysis time of 45 min, including column conditioning. Despite their simplicity, these methods offered excellent resolution (selectivity coefficient ␣ was 2.87 and 2.20 for water- and lipid-soluble vitamins respectively), linearity and retention time repeatability (% RSDs were: 4.62; 2.69; 1.83 and 1.25 for vitamin C, B6, A and E respectively, in 20 replicate analyses), and
Fig. 1. Typical chromatograms obtained by HPLC-DAD analysis of lipid-soluble (A and B) and water-soluble (C and D) vitamins with methods described in Section 2.2.2. The detection wavelengths are A = 320 nm, B = 290 nm, C = 290 nm, D = 245 nm; and number correspond to 1 = retinol acetate, 2 = ␣-tocopherol, 3 = pyridoxine hydrochloride, 4 = ascorbic acid.
464
S. Gabbanini et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 461–467
Table 2 Validation of the analytical methods for the analysis of vitamins in the receptor solution. An asterisk indicates statistically different results (p < 0.05; two-sided) between settings (see text). Analyte
Nominal concentration [mg/mL]
Accuracy: average % recovery (n = 15)
Precision: intra-day (n = 5)/inter-day (n = 15) % RSD
Robustness: average % recovery ± % RSD (n = 5) Setting 1/setting 2
Ascorbic acid Level 1 Level 2 Level 3
0.0020 0.0080 0.0800
85.8 92.7 105.3
2.58/10.51 1.31/7.86 0.18/3.53
93.4 ± 1.7/96.1 ± 1.0* 96.6 ± 1.1/96.3 ± 0.8 104.1 ± 0.3/101.6 ± 0.8*
Pyridoxine Level 1 Level 2 Level 3
0.0021 0.0084 0.0840
102.2 104.0 107.0
1.79/4.56 0.45/0.76 0.15/1.50
101.4 ± 1.2/99.7 ± 0.9 95.2 ± 1.1/96.3 ± 0.8 101.3 ± 0.2/106.1 ± 0.4*
Retinol acetate Level 1 Level 2 Level 3
0.0040 0.0200 0.0400
97.1 101.9 103.1
0.88/5.17 0.25/3.16 0.54/2.48
96.8 ± 0.6/97.0 ± 0.5 100.2 ± 0.2/99.8 ± 0.3 102.5 ± 0.3/103.6 ± 0.8
␣-Tocopherol Level 1 Level 2 Level 3
0.0044 0.0220 0.0440
108.6 101.6 103.8
4.16/6.87 1.75/2.73 0.82/1.52
102.4 ± 1.7/103.1 ± 1.0 101.9 ± 1.2/102.3 ± 0.8 100.2 ± 0.3/103.6 ± 0.8*
Table 3 HPLC-DAD replicate analyses (intra-day, n = 5) of the vitamins incorporated aqueous solutions and O/W or W/O emulsions used as donor in the diffusion experiments. Peak purity test indicates the method selectivity (max score = 999.9). Analysis methods and sample treatment are reported in Sections 2.2.2 and 2.3.3 respectively. Emulsion/vitamins
Analyte
Average peak purity ± SD
Average content [mg/mL] ± SD
W/O/water-soluble
Ascorbic acid Pyridoxine
Average Rt [min] ± SD 2.33 ± 0.01 6.68 ± 0.08
999.0 ± 0.1 998.6 ± 0.8
16.48 ± 0.20 17.73 ± 0.08
O/W/water-soluble
Ascorbic acid Pyridoxine
2.33 ± 0.01 6.60 ± 0.07
999.0 ± 0.2 998.4 ± 0.5
16.16 ± 0.21 18.09 ± 0.10
W/O/lipid-soluble
Retinol acetate ␣-Tocopherol
5.06 ± 0.12 11.11 ± 0.07
987.4 ± 16.3 973.2 ± 32.8
18.07 ± 0.20 33.26 ± 0.11
O/W/lipid-soluble
Retinol acetate ␣-Tocopherol
5.05 ± 0.11 11.09 ± 0.07
977.3 ± 26.1 970.3 ± 26.9
17.62 ± 0.12 32.10 ± 0.13
Aqueous solution
Ascorbic acid Pyridoxine
2.47 ± 0.30 6.46 ± 0.07
993.6 ± 10.3 996.9 ± 3.1
17.04 ± 0.08 16.90 ± 0.11
lation prepared for the subsequent kinetic investigation (O/W and W/O emulsions), after dilution in methanol or methanol/water and filtration (see Section 2.3.3). Overall no significant interference was suffered from the matrix or the many components in the emulsions, and DAD peak purity was >940 (94%) for any vitamin in any composition. This served also to provide analytical data for CD values to be used in the kinetic investigation. Results are collected in Table 3. 3.3. Vitamins’ stability in donor/receptor media Oxidative degradation of ascorbic acid in aqueous solution has been reported to proceed with apparent zero-order kinetics with recovery (linearly) dropping from 100% to 1% in less than 24 h at 25 ◦ C and pH 6.9 [23]. From the reported data it is possible to extract an empirical equation describing % recovery decay of the form: % recovery = 100 − kt, with k = −8.23 (%/h). To maintain ascorbic acid in the reduced form following the report by Shephard et al. [22], we added 0.4 mg/mL DTT to any sample or emulsion, however some progressive loss of ascorbate recovery was observed during method validation (standards solutions were kept at 4 ◦ C for 1–2 days between analyses at 30 ◦ C) despite the presence of DTT. To investigate the stability of all tested vitamins under our experimental settings we analyzed at time intervals donor (O/W and W/O emulsions) and receptor (pH 7.4) standard preparations containing 0.4 mg/mL DTT and either water-soluble or lipid-soluble vitamins stored at 4 ◦ C in the dark or at 34 ◦ C in a thermostatted bath (to reproduce the condition used for diffusion kinetics) for 48 h. While pyridoxine and ␣-tocopherol were perfectly stable, both ascorbate
and retinol acetate were significantly degraded in receptor solution at 34 ◦ C, degradation following apparent zero-order kinetics (Appendix A, Fig. S3), as previously described. The measured rate constants for degradation were k = 0.68 (%/h) and k = 0.86 (%/h) for ascorbic acid and retinol acetate, corresponding to recoveries of 68% and 59% respectively at 48 h. The rate of ascorbic acid degradation was decreased by over 10 times by DTT (despite the higher pH and temperature, by comparison with literature data [23]) however it was not completely suppressed. At 4 ◦ C in receptor solution in the dark only ascorbic acid showed some (slower) degradation with apparent k = 0.17 (%/h) corresponding to 92% recovery at 48 h. The stability of retinol acetate under these conditions suggests that photo-degradation rather that oxidation is responsible for its decay. Only ascorbic acid showed instability in the donor formulations at 34 ◦ C with decay rate constants of 0.21 and 0.32 (%/h) in W/O and O/W emulsions respectively, confirming previous observations on the dependence of ascorbic acid degradation on the medium [24]. All analytes were instead stable in the W/O and O/W emulsions at 4 ◦ C for 48 h. Degradation data of ascorbic acid and retinol acetate in receptor at 34 ◦ C were used to correct the results from diffusion kinetics. 3.4. Diffusion through synthetic membranes We operated under “infinite dose” conditions and avoid significant decrease of analytes’ concentration in the donor during the diffusion experiment. The trans-membrane diffusion profile was determined by HPLC analysis of the receptor medium, sampling
S. Gabbanini et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 461–467
Fig. 2. Typical kinetic plots of the cumulative amount Q of vitamins released from O/W emulsion through polyethersulfone membranes under static setting. Plots are obtained from one single experiment each for water-soluble and lipid-soluble vitamins. Legend: 䊉 = pyridoxine, = ascorbic acid, = ␣-tocopherol, = retinol acetate.
from each diffusion cell at regular intervals. Under the steady-state approximation, the diffusion process occurs in accordance with Fick’s first law (Eq. (1)). dQ = PS (CD − CR ) dt
(1)
Here PS (cm/h) is the skin permeability coefficient and CD and CR (g/cm3 ) are the drug concentrations in the donor and receptor chambers, while Q is the cumulative amount diffused through a unit of skin surface (g/cm2 ). Integration of Eq. (1) gives: Q = PS (CD − CR )t
(2)
Fig. 2 shows that with polyethersulfone membranes the quantity Q varied linearly with time, in agreement with Eq. (2) assuming that CD CR (infinite dose condition). Some deviation from linear behaviour was observed during the diffusion of water-soluble vitamins from O/W emulsion due to the very rapid diffusion, being faster than re-equilibration of vitamins in the viscous donor emulsion. The Flux J (g/cm2 /h) calculated from the slope of Q vs time during the first part of the kinetic study, afforded the skin permeability coefficient (Eq. (3)). PS =
J CD
(3)
465
Results obtained with polyethersulfone membranes are collected in Table 4. In general water-soluble pyridoxine hydrochloride (B6 ) and ascorbate (C) were released much more rapidly and effectively than lipid-soluble retinyl acetate (A) and ␣-tocopherol (E), qualitatively paralleling their solubility in the aqueous receptor fluid, which appeared to be the main driving force of the diffusion process. Interestingly the release of vitamins depended on the type of cosmetic formulation. Formulations having water in the external phase (O/W emulsions or lotions), released vitamins significantly more rapidly than W/O emulsion, although the actual contribution of the formulation depended on the physicochemical properties of the analyte. O/W emulsions were as effective as lotions in the release of water-soluble vitamins and approximately three times as effective as W/O emulsions, regardless of the structure of the vitamin. Conversely the permeability coefficient of retinyl acetate was only marginally increased (+20%) on changing from W/O to O/W emulsions. Somewhat surprising was instead the relative increase (five times) recorded for ␣-tocopherol. These findings can be rationalized considering the overall diffusion process: donor → membrane → aqueous receptor: water-soluble vitamins clearly benefit from an aqueous donor (or external phase) where they are more soluble maximizing the diffusion gradient (CD − CR ). Conversely with the very apolar ␣-tocopherol (log P = 12.2) having oil in the external phase (W/O emulsions) brings to very unfavourable partition (donor/receptor), which is not balanced by the larger concentration gradient. With less apolar retinyl acetate (log P = 7.9) the two factors, diffusion gradient and partition, almost compensate each other, bringing to only minor changes in Ps value. These findings suggest that polyethersulfone membranes, although being more permeable than more sophisticate models of human epidermis (vide infra), would provide a convenient and economical model to study the intrinsic release properties of different formulations. Polycarbonate membranes (0.01 m pore size) under identical experimental settings offered almost no barrier to the diffusion of vitamins. With aqueous donor the diffusion was faster than we could reliably monitor and brought to equilibration of the donor and receptor compartments (CD = CR ) within the experiment time. This result confirms that polycarbonate membranes are not a valuable model to study the release of active ingredients from cosmetic formulations. 3.5. Diffusion through reconstructed human epidermis With RHE membranes experiments were conducted as described for synthetic membranes (see Section 3.4). Due to the slower permeation encountered with RHE, the kinetic study was extended for 10 h with a follow-up the next day up to 24 h.
Table 4 Donor concentration (CD ), Lag-time (Lt ), flux (J) and permeability coefficient (PS ) through polyethersulfone membranes and RHE membranes of vitamins released from cosmetic formulations at 34 ◦ Ca Donor concentrations (CD ) is reported in Table 3. Polyethersulfone membranes b
RHE membranes
J × 10 [mg/cm /h] 3
2
PS × 10 [cm/h]
Lt [h]b
J × 103 [mg/cm2 /h]
PS × 105 [cm/h]
0.35 0.25 0.02 0.037
3.86 ± 0.32 7.54 ± 1.05 4.20 ± 1.03 ∼∞
43.52 ± 3.87 28.42 ± 2.52 0.050 ± 0.009 ∼0
264.1 ± 23.5 160.3 ± 14.2 0.28 ± 0.05 ∼0
0.84 2.55 0.02 0.04
1.05 ± 0.12 2.10 ± 0.24 6.15 ± 1.38 >21
141.42 ± 6.92 85.04 ± 6.64 0.063 ± 0.019 ∼0
875.1 ± 42.8 470.1 ± 36.7 0.36 ± 0.11 ∼0
57.82 ± 2.46 58.94 ± 2.20
0.95 ± 0.14 1.62 ± 0.18
145.11 ± 7.72 51.65 ± 4.34
851.6 ± 45.3 305.6 ± 25.7
3
Formula
Analyte
Lt [h]
W/O emulsion
Ascorbic acid Pyridoxine Retinol acetate ␣-Tocopherol
∼0 ∼0 0.31 ± 0.23 4.12 ± 1.34
309.8 390.0 5.18 1.33
± ± ± ±
5.6 4.2 0.32 1.22
19.31 22.86 0.29 0.040
± ± ± ±
O/W emulsion
Ascorbic acid Pyridoxine Retinol acetate ␣-Tocopherol
∼0 ∼0 0.45 ± 0.15 0.34 ± 0.26
966.1 1061.10 6.09 9.74
± ± ± ±
13.2 44.55 0.41 1.21
61.69 60.81 0.35 0.30
± ± ± ±
Aqueous solution
Ascorbic acid Pyridoxine
∼0 ∼0
a b
Mean ± SD (n = 3). Estimated from extrapolation of the diffusion curve to Q = 0.
984.4 ± 41.8 996.0 ± 37.2
466
S. Gabbanini et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 461–467
Fig. 3. Kinetic plots of the cumulative amount Q of vitamins released from O/W emulsion through RHE membranes under static setting. Plots are averaged from three experiments each for water-soluble (A) and lipid-soluble (B) vitamins. The large error bars account mainly for variation in lag-time. Legend: = ascorbic acid, = pyridoxine, = retinol acetate, 䊉 = ␣-tocopherol.
Typical kinetic plots are depicted in Fig. 3, showing reasonably linear diffusion profiles as previously discussed for synthetic membranes, which suggests that permeation occurs mostly by passive, diffusion-like, transport processes. With RHE, significant deviation from linear behaviour was observed only for ascorbate in the final portion of the diffusion experiment, conceivably due to airoxidation of the vitamin in the receptor solution. Large lag-times were always observed for the permeation through RHE, particularly for lipid-soluble vitamins, possibly due to the high affinity for the Stratum Corneum (SC) and phospholipids’ bi-layers in epidermis, as previously reported both for RHE and native epidermis. [1,2] Indeed ␣-tocopherol gave some detectable diffusion in the receptor only after 22 h, resulting in no measurable human skin permeability under our experimental settings. Conversely, we attributed the shorter lag-times necessary to establish the diffusion of ascorbic acid and pyridoxine to the predominance of the intercellular diffusion route for these small water-soluble molecules. [25] Results are summarized in Table 4. In general RHE was much more selective than synthetic membranes and permeability coefficients were up to two orders of magnitude lower than those recorded with polyethersulfone membranes. Particularly ascorbic acid and pyridoxine had Ps values ∼10 and ∼20 times lower on RHE, while for retinol acetate the release was ∼100 times slower. The high selectivity of RHE suggests it is a much more valuable model to mimic the behaviour in vivo than synthetic membranes. As mentioned, for ␣-tocopherol the permeability was decreased so much to prevent measurement with our approach. To a qualitative level the role of the formulation in determining the release through RHE was similar to that observed with polyethersulfone. This very relevant point has been addressed with contradicting outcomes in the literature: while some investigations using RHE or animal skin failed detecting a clear role of the formulations in the trans-epidermal absorption of tocopherols, [26] others are in line with our findings. [27] 4. Conclusions In the present study a sensitive, accurate and reproducible analytical method is described to investigate the kinetics of diffusion of lipid- and water-soluble vitamins contained in cosmetic formulations, through Reconstructed Human Epidermis or synthetic polyethersulfone membranes in vitro. The described method and instrumental setting are sufficiently simple and cost-effective to
be suited to the quality-efficacy assessment of cosmetic formulations. Due to the limited dispersion of experimental kinetic results (good repeatability), the use of RHE favourably compares with the larger variability normally encountered with native human epidermis, while overcoming the problems of limited availability. Although polyethersulfone membranes are less selective barriers to the diffusion of vitamins than RHE, they can be a convenient model to evaluate the intrinsic release ability of the formulation before testing on RHE. Conversely polycarbonate membranes offer no significant barrier to the release of active molecules from cosmetic formulations. Acknowledgments This work was financially supported by BeC srl (Forlì, Italy) and MIUR (Rome, Italy). The authors are grateful to Prof. Lorenzo Rodriguez (University of Bologna) for helpful discussion. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jpba.2010.01.029. References [1] C. Antille, C. Tran, O. Sorg, J.-H. Saurat, Penetration and metabolism of topical retinoids in ex vivo organ-cultured full-thickness human skin explants, Skin Pharmacol. Physiol. 17 (2004) 124–128. [2] A. Mavron, V. Raufast, D. Redoules, Skin absorption and metabolism of a new vitamin E prodrug, ␦-tocopherol-glucoside: in vitro evaluation in human skin models, J. Control. Rel. 100 (2004) 221–231. [3] EU regulation 76/768/EEC, February 2003. [4] A. Kligman, The future of cosmeceuticals: an interview with Albert Kligman, MD, Dermatol. Surg. 31 (2005) 890–891. [5] D. Kligman, Cosmeceuticals, Dermatol. Clin. 18 (2000) 609–615. [6] C.R. Thornfeldt, Cosmeceuticals: separating fact from vodoo science, Skinmed 4 (2005) 214–220. [7] (a) OECD, Skin Absorption: In Vitro Method, Test Guideline 428, 2004.; (b) OECD, Guidance Notes for the Estimation of Dermal Absorption Values, 2008.; (c) OECD Guidance Document for the Conduct of Skin Absorption Studies Number 28, 2004. [8] (a) Colipa Guidelines for Percutaneous Absorption/Penetration, 1997.; (b) Colipa Guidelines for the Evaluation of the Efficacy of Cosmetic Products, Revised Version, May 2008.; (c) Colipa ID 7: Serious about Cosmetics—Serious about Alternative Methods. Updated Version, May 2007. [9] M.K. Das, A. Bhattacharya, S.K. Ghosal, Effect of different terpene-containing essential oils on percutaneous absorption of trazodone hydrochloride through mouse epidermis, Drug Deliv. 13 (2006) 425–431.
S. Gabbanini et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 461–467 [10] A. Mittal, U.V.S. Sara, A. Ali, M. Aquil, The effect of penetration enhancers on permeation kinetics of nitrendipine in two different skin models, Biol. Pharm. Bull. 31 (2008) 1766–1772. [11] S. Schreiber, A. Mahmoud, A. Vuia, M.K. Rübbelke, E. Schmidt, M. Schaller, H. Kandárová, A. Haberland, U.F. Schäfer, U. Bock, H.C. Korting, M. Liebsch, M. Schäfer-Korting, Reconstructed epidermis versus human and animal skin in skin absorption studies, Toxicol. In Vitro 19 (2005) 813–822. [12] D. Howes, R. Guy, J. Hadgraft, J. Heylings, U. Hoeck, F. Kemper, H. Maibach, J. Marty, H. Merk, J. Parra, D. Rekkas, I. Rondelli, H. Schaefer, U. Tauber, N. Verbiese, Methods for assessing percutaneous absorption, in: ECVAM Workshop Report 13, ATLA 24 (1996) 81–106. [13] J. Reichling, U. Landvatter, H. Wagner, K.-H. Kostka, U.F. Schaefer, In vitro studies on release and human skin permeation of Australian tea tree oil (TTO) from topical formulations, Eur. J. Pharm. Biopharm. 64 (2006) 222–228. [14] K. Cal, M. Sopala, Ex vivo skin absorption of terpenes from Vicks VapoRub® ointment, Med. Sci. Monit. 14 (2008) PI19–PI23. [15] F. Netzlaff, C.-M. Lehr, P.W. Wertz, U.F. Schaefer, The human epidermis models EpiSkin, SkinEthic and EpiDerm: an evaluation of morphology and their suitability for testing phototoxicity, irritancy, corrosivity, and substance transport, Eur. J. Pharm. Biopharm. 60 (2005) 167–178. [16] M.D. Ynsa, E. Gontier, A. Mavon, P. Moretto, M. Rosdy, Comparative study between reconstructed and native human epidermis using nuclear microscopy, Nucl. Instr. Meth. Phys. Res. B 249 (2006) 710–714. [17] M. Schaefer-Korting, U. Bock, W. Diembeck, H.-J. Duesing, A. Gamer, E. HaltnerUkomadu, C. Hoffmann, M. Kaca, H. Kamp, S. Kersen, M. Kietzmann, H.C. Korting, H.-U. Kraechter, C.-M. Lehr, M. Liebsch, A. Mehling, C. MuellerGoymann, F. Netzlaff, F. Niedorf, M.K. Ruebbelke, U. Schaefer, E. Schmidt, S. Schreiber, H. Spielmann, A. Vuia, M. Weimer, The use of reconstructed human epidermis for skin absorption testing: results of the validation study, ATLA 36 (2008) 161–187.
467
[18] M. Schaefer-Korting, A. Mahmoud, S.L. Borgia, B. Brueggener, B. Kleuser, S. Schreiber, W. Mehnert, Reconstructed epidermis and full-thickness skin for absorption testing: influence of the vehicles used on steroid permeation, ATLA 36 (2008) 441–452. [19] S. Gabbanini, E. Lucchi, M. Carli, E. Berlini, A. Minghetti, L. Valgimigli, In vitro evaluation of the permeation through reconstructed human epidermis of essentials oils from cosmetic formulations, J. Pharm. Biomed. Anal. 50 (2009) 370–376. [20] D. Pabla, H. Zia, A comparative permeation/release study of different testosterone gel formulations, Drug Deliv. 14 (2007) 389–396. [21] Y. Frum, G.M. Eccleston, V.M. Meidan, Evidence that drug flux across synthetic membranes is described by normally distributed permeability coefficients, Eur. J. Pharm. Biopharm. 67 (2007) 434–439. [22] A.B. Shephard, S.C. Nichols, A. Braithwaite, Moisture induced solid phase degradation of l-ascorbic acid: Part 1: a kinetic study using tristimulus colorimetry and a quantitative HPLC assay, Talanta 48 (1999) 585–593. [23] M.-L. Chang, C.-M. Chang, Simultaneous HPLC determination of hydrophilic whitening agents in cosmetic products, J. Pharm. Biomed. Anal. 33 (2003) 617–626. [24] R. Austria, A. Semenzato, A. Bettero, Stability of vitamin C derivatives in solution and topical Formulations, J. Pharm. Biomed. Anal. 15 (1997) 795–801. [25] B.W. Barry, Dermatological Formulations Percutaneous Absorption, Marcel Dekker Inc., New York, 1983. [26] M.S. Roberts, K.A. Walters (Eds.), Dermal toxicity assessment, 2nd edition, Informa Healthcare USA Inc., New York, 2008. [27] F. Dreher, F. Fouchard, C. Patouillet, M. Andrian, J.T. Simonnet, F. Benech-Kieffer, Comparison of cutaneous bioavailability of cosmetic preparations containing caffeine or ␣-tocopherol applied on human skin models or human skin ex vivo at finite doses, Skin Pharmacol. Appl. Skin Physiol. 1 (Suppl. 15) (2002) 40–58.