8 JOURNAL OF COSMETIC SCIENCE and gas chromatography. Morliere eta/, (11) have also demonstrated the E •, Z pho- toisomerization of octyl methoxycinnamate by steady-state and laser-flash photolysis in aqueous and organic solutions. With their spectrometric assays Deftandre and Lang (7) assessed the absorbance of octyl methoxycinnamate. The absorbance reflects the activity of UV filters. If the absorbance doesn't show significant variation, it means that either the filter doesn't degrade or the photoproducts formed (isomer or another molecule) are as efficient as the initial filter, which is apparently the case for octyl methoxycinnamate as demonstrated in their experiment. It seems to be also the case in our study: even if there is a decrease of filter amount, the new product formed can be an isomer also active against UV radiation. In order to verify this hypothesis, an i, vitro test (results not shown) was performed to assess the efficacy of the sunscreen after 20 and 40 MED irradiations, using a modification of the draft protocol from the Colipa to test sunscreen photostability (12,13). The losses of UVB and UVA protection were found to be less than 10%. Thus octyl methoxycinna- mate is photo-unstable, but the octyl methoxycinnamate photoproduct stays efficient against UV radiation, in accordance with the results of Deftandre and Lang (7). Kammeyer eta/. (14) also assessed the photostability of sunscreen agents ], vitro using spectrometric assays on stratum corneum sheets. After applying sunscreen agents on stratum corneum sheets, they measured the absorption spectra before and after irradia- tion. By comparing the two spectra obtained, they calculated the percentage of alteration of each filter. Several UV filters were classified according to this percentage of degra- dation. They found that octyl methoxycinnamate and 4-methylbenzilidene camphor were moderately photostable, with percentages of alteration of 25% and 18%, respec- tively. In the present experiment, the percentage of degradation of octyl methoxycin- namate is the highest compared to the other filters in the preparation, as in the Kam- meyer eta/. study, but the percentage of degradation of 4-methylbenzilidene camphor is not as important. This difference could come from the solvent in which they diluted the filter for application to the stratum corneum sheets. They used liquid paraffin (non- polar) for 4-methylbenzilidene camphor, whereas we used a mix of methanol/chloroform. Moreover, their experiments did take into accounts the irregularity of the stratum corneum that could have lead to non-homogeneous spreading of the sunscreen. These factors have consequences on the spectrophotometric assays. It is interesting to note that the nature of the cosmetic vehicle in which UV filters are introduced and the presence of other filters influenced their photostability: two ], vitro studies, one based on HPLC assays (13) and the other on transmission spectroscopy assays (12), have previously demonstrated that the association of octyl methoxycinnamate and butyldibenzoylmeth- ane was very photo-unstable. Marginean Lazar eta/. (8,15) studied the photochemical behavior of butylmethoxydiben- zoylmethane, 4-methylbenzilidene camphor, and octyl methoxycinnamate in different emulsion bases using an •, v•vo method as in the present study. They applied sunscreen preparations on the backs of several volunteers and exposed the treated sites at various suberythemal doses. They collected ten sequential strippings within the same area irradiated, and they assessed, after extraction of the filters by toluene, the amount of filter (in the first tape strip and in the next sequential nine tape strippings) by HPLC assays. They also conducted a study without irradiation to determine the amount of filters collected at the skin surface. When stripping was conducted without irradiation, the entire theoretical amount of filters was recovered in the sum of the first and nine

PHOTOSTABILITY OF UV FILTERS 9 sequential strippings. In the present study, the percentage of recovery (for all the filters) after 15 tape strippings is quite low: only 50% of the applied filters, because of the loss on the finger glove used for product application. Indeed, more than 45% of the UV filters were found on each finger glove. Therefore, including the amount of filter re- covered in the finger glove, almost all the applied filters were recovered. Marginean Lazar et at/. (15,16) worked also on the evaluation of the percutaneous pen- etration of UV filters. Using human skin samples set in Franz diffusion cells, they found that the penetration rate of UV filters was low and that no more than 5 % of the applied dose was recovered into the dermis after a 24-hour application period. Even if our experiment does not enable us to quantify the amount of filters that could have pen- etrated into the deeper part of the skin (because we only consider a portion of the stratum corneum), we can suppose, considering the total amount recovered (tapes + finger glove -- 95%), that this quantity is very low, as in the Marginean Lazar et at/. study. Moreover, when stripping was conducted after irradiation, they observed a gradual loss of filters, proportional to the UV-exposure dose. The present study was performed only after one dose exposure (40-MED) higher than the ones used in the Marginean Lazar et at/. study. However, both protocols showed a loss of octyl methoxycinnamate. The present Inethod enables us to compare the distribution profile of the UV filters in the stratum corneum according to the number of tape strippings, unlike the Marginean Lazar et at/. studies (8,15). Although the five filters were almost totally recovered without irradiation (taking into account the amount on the finger glove), 4-methylbenzilidene camphor and methylen bis benzotriazoyl tetramethylbutylphenol have different distri- bution profiles in the stratum corneum. Potard et al. (17) have used an HPLC analysis coupled with tape stripping for the quantification of various UV filters (including dioctylbutamido triazone and octyl methoxycinnamate) in various skin layers (stratum corneum, epidermis, dermis). This ex vivo study demonstrated that the quantity of filters collected by tape stripping decreased according to the number of strippings. In their experiment, the first six tape strippings contained more than 70% of the UV filters recovered in the stratum corneum. In the present experiment, after one hour of contact, more than 75% of the filters (for each filter) recovered were found in the first five adhesives. It means that the UV filters stay in the very upper layers of the stratum corneum, confirming other studies (16-19). CONCLUSION The Inethod combining tape stripping and HPLC assays used here gives results of the photochemical behavior of UV filters but also information on the distribution of these filters in the stratum corneum. The i, vivo results generated confirmed i, vitro results obtained on quartz plates. Moreover, this i, vivo Inethod is very close to what happens in real conditions of use, since the sunscreen is applied on the forearm and the interactions of the sunscreen agents with the stratum corneum can be studied. HPLC is more efficient than UV spectroscopy, as it enables one to detect photoproducts and offers quantitative results for the determination of the filters in strippings as opposed to qualitative data.