DETERMINATION OF RETINOIDS IN COSMETICS 495 Also, the standard solutions for all-trans-retinoic acid, all-trans-retinol, and all-trans- retinyl palmitate exposed to air and light at room temperature remained stable during the extraction time. However, to determine long-term cosmetic product stability, the 29 cosmetic products were stored under normal conditions (at room temperature in their original packaging) and re-analyzed after three to six months for all-trans retinol and all-trans retinyl palmi- tate. By comparing the results of previous analysis for each product with the subsequent result, it was possible to determine the extent to which the samples had become isomer- ized and/or oxidized or self-decayed over time. Most products showed some decomposi- tion to a varying degree. In addition, to determine the stability of standard retinoid solutions at 0.1 mg/ml, the stability of each standard was evaluated in terms of the effect of heat, light, and air with time. In the dark protected from air at 2°C, the standards generally lasted for about three months without signifi cant decomposition in the dark protected from air at room tem- perature, the standards lasted for approximately one month in the dark not protected from air at room temperature, the standards lasted for approximately one week exposed to light but protected from air, the standards lasted for several days and exposed to light and air, the standards decomposed within one day. The specifi c decomposition path of oxidation or isomerization and fi nal products differed depending on the retinoid and the specifi c conditions. However, in general, retinoic acid was most sensitive to light. Retinol was sensitive to air and light. Retinyl palmitate was most sensitive to heat. METHOD PERFORMANCE Twenty microliters of each standard solution was injected directly into the HPLC chro- matograph to determine the linearity of response. Calibration curves were constructed from a plot of peak area vs concentration for each retinoid. The calibration curves ob- tained were found to be linear, from 0.0003 mg/ml to 0.3 mg/ml. Regression correlation coeffi cients were greater than 0.995. The limit of detection (LOD), defi ned as three times the baseline noise (21), was different for each retinoid and ranged from 0.34 μg/g to 1.08 μg/g (average, 0.6 μg/g). Specifi c LODs for each retinoid analyzed were as fol- lows: 13-cis-retinoic acid (1.08 μg/g) all-trans-retinoic acid (0.63 μg/g) 13-cis-retinol (0.34 μg/g) all-trans-retinol (0.54 μg/g) all-trans-retinaldehyde (0.45 μg/g) and all- trans-retinyl palmitate (0.62 μg/g). The limit of quantifi cation (LOQ), defi ned as ten times the baseline noise (21), averaged 2.0 μg/g. Samples with retinoid concentrations exceeding the highest standards were appropriately diluted to allow use of calibration curves for quantitation. INTERFERENCE Sample extracts were not further purifi ed after the Celite extraction, to minimize degra- dation of the samples. At the detection wavelength of 330 nm, unidentifi ed peaks elut- ing at retention times near the retinoid analytes were rare. Extracts treated using a C18 cartridge or fi ltration through a 0.45-μm fi lter disk gave the same analytical results as

JOURNAL OF COSMETIC SCIENCE 496 extracts without further treatment. To confi rm the identity of analyte peaks, chromato- graphic peak identifi cations were confi rmed by retention time, peak width, and the UV spectrum of the chromatographic peak. RECOVERY STUDIES A tertiary solvent mixture was chosen that contained mutually miscible solvents: 1/3 (v/v) hexane, 1/3 (v/v) isopropanol, and 1/3 (v/v) ethyl acetate. This solvent mixture gave good recoveries of retinoids from a wide range of sample matrices. The solvent mixture was also compatible with direct injection onto the column without further sample ma- nipulation, which helped minimize the decomposition of the extracted retinoids. Table I shows results obtained for four different sample matrices spiked with four differ- ent levels of retinoic acid, retinol, and retinyl palmitate. Recoveries were generally 95% or greater. Since additional studies showed that increasing the volume of extractant did not lead to better recoveries, a single liquid/solid extraction using 10 ml of extractant was chosen for simplicity and rapidity. COSMETIC SKIN CARE PRODUCTS SURVEYED The 29 consumer cosmetics surveyed in this study included anti-wrinkle, anti-aging, skin renewal, line removal, skin-whitening, skin-moisturizing, and skin-cleansing prod- ucts, and were in the form of creams, lotions, complexes, and serums (see Table II for more information on each cosmetic product analyzed). Many of the products highlighted the presence of vitamin A and/or other vitamins in the formulation (e.g., “vitamin en- riched,” “a natural vitamin A wrinkle cream,” “retinol formula,” “a retinol facial treat- ment with multi-vitamins,” “refi ning night cream with 0.5% pure retinol,” “vitamin A cream extra strength,” “max retinol vitamin A,” and “double retinol wrinkle treatment”). Seventeen (59%) of the products included retinol, either in the product name or in the list of ingredients, four included retinyl palmitate, and six included both. One product found to contain retinoid did not indicate the presence of either retinol or retinyl palmi- tate in the list of ingredients. Twenty-fi ve (86%) of the products were labeled to contain anti-oxidants or stabilizers such as vitamin C, vitamin E, BHT, and di- or tetra-sodium EDTA to prevent oxidation or decomposition of the retinol or retinyl palmitate. In addition, 22 products (76%) were labeled to contain a sunscreen, recommended use with a sunscreen, or recommended night-time use only due to the possibility of increased photosensitivity. Finally, 18 prod- ucts (62%) provided product use warnings such as “for external use only,” “avoid contact with the eyes or mucous membranes,” “discontinue use if persistent irritation occurs,” or “keep out the reach of children.” SURVEY RESULTS The results of the analysis are given in Table III. Eighteen products (62%) were found to contain retinol. Six (21%) were found to contain retinyl palmitate, and fi ve (17%) were found to contain both retinol and retinyl palmitate. No products were found to contain