118 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS and sunscreen formulations. Its efficiency as a UV absorber is due in part to its high molar absorptivity in various solvents and its ability to absorb UV radiation in the spectral range of 300 nm (in nonpolar solvents) to 316 nm (in polar solvents) without the extensive oxidation and intermolecular hydrogen bonding that PABA and some of its derivatives undergo. Recently, however, concern has been growing about the possible nitrosation of Padimate-O to a nitrosamine, 2-ethylhexyl p-(N-methyl-N-nitrosamino) benzoic acid (NMPABAO), in cosmetic products. In May 1988, the U.S. Food and Drug Administration (FDA) reported that a new nitrosamine was found in certain sunscreen products. Of the 17 sunscreen products tested, 14 were found to contain NMPABAO at levels varying from 60 to 1960 ppb (1). Animal studies have shown that nitrosamines are carcinogenic and may be mutagenic and teratogenic as well (2). More than 120 N-nitroso compounds have been examined for carcinogenic activity in animals, and approximately 80% of these have been found to be carcinogenic to some degree (2,3). NMPABAO was reported to be mutagenic in the Ames test using two strains of Salmonella typhimurium (4). However, in a more recent FDA study, NMPABAO was not found to be mutagenic in a battery of S. typhimurium studies (including the same assay in reference 4) or in the mouse lymphoma assay (5). The possible carcinogenicity of NMPABAO has not been evaluated in an animal bio- assay. We investigated the penetration and metabolism of NMPABAO in hairless guinea pig and human skin to evaluate the compound's safety. The viability of the skin was maintained in flow-through diffusion cells. The results of some of the experiments in which the viability of the skin was maintained were compared with those from exper- iments that used nonviable skin. The effects of two cosmetic vehicles (ethanol and a lotion) on the penetration and cutaneous metabolism of Padimate-O were also deter- mined. The nitrogen-nitrogen bond of nitrosamines is weak and can be broken by relatively low energy inputs, including energy from UV radiation (6). Therefore, the photodecompo- sition of NMPABAO in a cosmetic vehicle was investigated in preliminary studies (without skin) and in diffusion-cell skin absorption studies. MATERIALS AND METHODS [•4C]NMPABAO (specific activity 20.85 mCi/mmol, purity 98%) and [•4C]Padi- mate-O (specific activity 20.0 mCi/mmol, purity 86%) were synthesized by Research Triangle Institute, Research Triangle Park, NC. Padimate-O was further purified on Sep-Pak © silica cartridges (Waters Associates, Milford, MA) immediately before each experiment, giving a final purity of 96%. Absorption and metabolism experiments were conducted in vitro, using flow-through diffusion cells (7). The system was cleaned with a 70% ethanol solution to prevent interference due to bacterial metabolism. Most studies used skin from the hairless guinea pig (female, 3-6-months old) that was prepared with a Padgett dermatome (Padgett Instruments, Kansas City, MO) at a thickness of approximately 200 •m. Several studies were performed using surgical specimens of viable human skin obtained from abdominoplasty and dermatomed to 200 •m. Barrier integrity of the human specimens was verified by using [3H]water absorp- tion as a standard (8).

PADIMATE-O 119 Viability of skin was maintained in the diffusion cells for 24 h by using a HEPES- buffered Hank's balanced salt solution (HHBSS) with the addition of gentamycin sulfate and bovine serum albumin (9). In some experiments, sodium fluoride was added to the above receptor fluid to render the skin nonviable (10). All receptor fluids were sterilized before use by passage through a 0.2-lxm filter (Nalgene Company, Rochester, NY). A simple oil-in-water emulsion lotion was prepared for use as one of the vehicles in this study. The stearic acid (3%) and water (91%) emulsion contained 4% Padimate-O to simulate a sunscreen product and 1% propylene glycol as a humectant. Radiolabeled Padimate-O in both an ethanol and a lotion vehicle was applied to the skin (15 }xl/cm 2) at a chemical dose of 6.7 }xg/cm 2. Radiolabeled NMPABAO in the lotion was applied to the skin (15 [•l/cm 2) at a chemical dose of 7.0 }xg/cm 2. The skin was not occluded therefore, the vehicles were allowed to evaporate as they would during con- ditions of normal use. All skin surfaces were washed at 24 h with a 1% solution of soap (dishwashing detergent) and water to remove unabsorbed material. Test compound not removed from the skin by washing was considered to be absorbed and was added to the receptor fluid levels to determine total percutaneous absorption. Localization of compounds in skin was measured by stripping each skin section ten times with cellophane tape to remove the stratum corneum. Skin sections were then homog- enized in HHBSS solution by using a Polytron tissue homogenizer (Brinkmann Instru- ments, Westbury, NY). Small aliquots (0.2 ml) of the receptor fluid (collected at 6-h intervals), tape strips, and skin homogenates were analyzed for radioactivity by using a Beckman LS9000 scintillation counter (Beckman Instruments, Irving, CA). Parent compound and metabolites were extracted from the remaining receptor fluid and skin homogenates by using Sep-Pak C •g cartridges (Waters Associates, Milford, MA). All receptor fluids and skin homogenates were adjusted to pH 3.2-3.5 and filtered through the cartridges. Compounds bound to the Sep-Pak cartridges were eluted in 4 ml of acetone. The eluates containing parent compound and metabolites were applied to silica gel thin-layer chromatography (TLC) plates along with nonradiolabeled standards of Padi- mate-O (ICI Americas Inc., Wilmington, DE), dimethyl aminobenzoic acid (DMABA) (Aldrich Chemical Corp., Milwaukee, WI), NMPABAO, and p-(N-methyl-N-nitros- amino) benzoic acid (NMPABA) (Cosmetics Technology Branch, FDA, Washington, DC). TLC plates were developed with hexane:ethyl acetate (135'15) as the solvent system. Radioactivity on TLC plates was quantified by a Bioscan TLC plate scanner (Bioscan Inc., Washington, DC). Radioactivity in each spot was expressed as the percent of total radioactivity applied to the plate. Re-values corresponding to the radioactive peaks were compared with those of standard compounds. Initial photodecomposition studies were performed by spreading 300 mg of a commer- cial sunscreen product containing 800 ppb NMPABAO to a thickness of approximately 20 }xm over the bottom of a 20 x 150-ram petri dish. The petri dish containing the test portion was exposed to a Mutzhas UV light source with a Supuvasun 3000 filter (Mutzhas, Munich, Germany) at a distance of 207 cm from the light source. At this distance, UVA and UVB output readings averaged 6.7 x 10 -2 and 0.40 x 10 -2 mW/cm 2, respectively. Test portions were exposed for 0.0, 1.0, 1.5, and 2.0 rain, with two repetitions for each time period. For the 2.0-rain time interval, the UVB radiation dose was 0.48 mJ/cm 2. The human minimal erythema dose (MED) required to produce