In vitro assay of high-spf sunscreens

j. Soc. Cosmet. Chem., 48, 289-295 (November/December 1997) In vitro assay of high-spf sunscreens R. P. STOKES and B. L. DIFFEY, Regional Medical Physics Department, Dryburn Hospital, Durham DH1 5TW, UK. Accepted for publication December 1, 1997. Synopsis In vitro spectral transmission measurements using excised human epidermis as the substrate were used to determine the photoprotection provided by physical and organic chemical sunscreens encompassing a wide range of sun protection factor (SPF). The measured SPFs were in good agreement with the quoted SPFs of the products. This in vitro technique using human epidermis could prove reliable for evaluating the SPF of high-protection sunscreens for which in vivo assay is problematic due to the impractically long irradiation times to achieve erythema on sunscreen-protected skin. We also compared our calculated SPFs, which assumed a natural sunlight spectrum, with those that would have been obtained assuming a xenon-arc solar simulator spectrum. We found that for products with relatively low UV-A absorption, the use of a solar ß simulator for in vivo testing overestimates the SPF that would be expected in sunlight. INTRODUCTION The photoprotection provided by a sunscreen product is assessed in terms of its sun protection factor (SPF). Sunscreen SPFs are generally measured by in vivo assay and defined as the ratio of the ultraviolet (UV) dose required to cause minimal erythema in protected skin to that required for unprotected skin. Internationally agreed procedures (1,2) define protected skin as that to which a 2-mg/cm 2 layer of sunscreen has been applied. In vivo assay is problematic for high-protection sunscreens (SPF > 25) because of the impractically long UV irradiation times and variability of results (3). Consequently, it would be particularly desirable to use a reliable in vitro assay for these products, whereby the transmission of UV radiation is measured first through a substrate and then through the substrate with applied sunscreen. The ratio of UV transmission without sunscreen to that with sunscreen gives a measure of photoprotection. A wide range of substrates has been used for in vitro assay, including wool, pig skin, hairless mouse epidermis, human epidermis, human stratum corneum, synthetic skin casts, and surgical tape (4). Unfortunately, most substrates are least reliable when products offering high protection are assayed. The substratexpected to give results closesto in vivo assay is human epidermis. In the study reported here, we have applied physical sunscreens, with varying concentrations of the same active ingredient, and organic chemical sunscreens to excised human epidermis, and using a spectral transmission technique (5), measured the SPF at an 289

290 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS application thickness of 2 mg/cm 2. Our aims were to determine whether the technique was a reliable way of measuring high SPF (>25) products and to evaluate the difference between SPFs measured in vivo using a xenon arc solar simulator and those expected in natural sunlight. MATERIALS AND METHODS ISOLATION OF EPIDERMIS Skin was taken from the underside of the female breast during the operation of breast reduction. It was obtained by a process known as de-epidermalization, the principle of which is to remove the epidermis and epidermal appendages while leaving the deepest layers of the dermis in situ. The skin (approximately 10 x 4 cm) was received within one day of surgical operation, and these strips were cut into squares of approximately 4 cm x 4 cm. The samples of skin were placed in a water bath at 60øC for 45 seconds (6). On removal from the water bath, the epidermis was gently separated from the dermis by careful peeling. Epidermal sheets were stored in physiological saline at 4øC until required, which was normally within five days. Sheets of epidermis can be stored at 4øC for several weeks without loss of barrier function (6). SUNSCREEN PRODUCTS Five physical sunscreen products were used, each containing titanium dioxide as the sole active ingredient at concentrations of 4.4%, 6.9%, 7.8%, 8.6%, and 12.0%, respectively. The first four products were commercially available and had quoted SPFs of 8, 15, 25, and 35. The product with the highest concentration was not yet available commercially but was expected to have an in vivo SPF of 35 or higher. These sunscreens are identified as P8, P15, P25, P35, and P35+, the numbers denoting nominal SPF. Five organic chemical sunscreens were used and these are identified as C5, C15, C20, C30, and C50, the numbers again denoting quoted SPF. The active ingredients contained within these sunscreens were as follows: C5: Butyl methoxydibenzoylmethane and methylbenzylidene camphor. C15: Butyl methoxydibenzoylmethane, methylbenzylidene camphor, and octyl salicyl- ate. C25: Butyl methoxydibenzoylmethane, methylbenzylidene camphor, octyl methoxycinnamate, and titanium dioxide. C30: Butyl methoxydibenzoylmethane, methylbenzylidene camphor, octyl salicylate, and titanium dioxide. C50: Octocrylene, octyl methoxycinnamate, octyl salicylate, and oxybenzone. EXPERIMENTAL TECHNIQUE A piece of epidermis (2 x 2 cm) was placed over a circular aperture of diameter 1.5 cm cut into an aluminium holder. The holder was positioned so that the circular aperture was directly over the teflon input optics of an Opttonic model 742 spectroradiometer controlled by a Hewlett Packard HP85 microcomputer. Radiation from a 75 W xenon arc lamp (filtered by a Schott UG5 filter) was directed onto the epidermis via a light

HIGH-SPF SUNSCREENS 291 guide, and the photocurrent was recorded from 290 to 400 nm in steps of 5 nm. A micropipette was then used to dispense an amount of sunscreen equivalent to 2 mg/cm 2 onto the epidermis. The sunscreen was "spotted" at several positions on the epidermis, and a light, circular rubbing motion with a gloved finger was used to give as uniform a layer as possible. The sunscreen was allowed to dry for 20 minutes, and the photocurrent was again measured in 5-nm steps between 290 and 400 nm. For each wavelength, the ratio of the photocurrent recorded before and after application of the sunscreen was calculated; this gave the monochromatic protection factors, PF0 ), which were used in the following expression (5) to give the SPF: 400 Z EOQ½0Q 290 SPF - 400 [ 1 ] 290 where E0 ) is the spectral irradiance of sunlight expected for a clear sky at noon in midsummer for a latitude of 40øN (solar altitude 70ø), and ½0 ) is the effectiveness of radiation of wavelength ) nm in producing delayed erythema in human skin (7). For the physical sunscreens containing 4.4, 6.9, 7.8, and 8.6% TiO2, measurements were made on four samples of epidermis from one subject (epidermis A) and a single epidermal sample from each of six other subjects (epidermis B, C, D, E, F, and G). In the case of the sunscreen containing 12% TiO2, measurements were not made on epidermal sample A. For the organic chemical sunscreens, measurements were made on a single epidermal sample from each of six subjects (epidermis H, I, J, K, L, and M). RESULTS AND DISCUSSION Figure 1 shows absorption spectra normalized to equal area for the five physical sunscreens. It can be seen that the spectra have similar shapes, the small differences between spectra probably being due to different size distributions of the TiO2 particles within the sunscreens (8,9). Figure 2 shows the corresponding absorption spectra for the five organic chemical sunscreens. It can be seen that the spectra for C5, C! 5, C25, and C30 are similar, all four products having the active ingredients methylbenzylidene camphor and butyl methoxydibenzoylmethane in common. However, the spectrum for C50 shows that this sunscreen has relatively little absorption in the long UV-A region. Table I summarizes the SPFs measured on each of the five physical sunscreens. For the products P8, P15, P25, and P35, the variance in SPFs measured on four samples of epidermis from a single volunteer (A) was compared with that obtained from SPFs measured on epidermis from the six other volunteers (B-G). In every case there was no significant difference (p > 0.05; Stairnov test). We conclude from this analysis that the variability in the measured SPFs is primarily associated with the experimental technique (most probably the application of sunscreen to the epidermis), rather than to variability in epidermal architecture between subjects. The mean SPFs for a 2-mg/cm 2 application thickness of the physical sunscreens are given in the final column of Table I and, with the exception of product P35, are in close

292 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS 0.075-0.060 o.o45 0.030 0.015 0.000 290 300 310 320 330 340 350 360 370 380 390 400 Wavelength, nm Figure 1. The absorption spectra of physical sunscreen products normalized to equal area. agreement with those claimed by the manufacturer. Furthermore, the coefficients of variation for each product are generally smaller than those expected from in vivo assay of products of similar SPF (3), indicating that in vitro assay using excised human epidermis as a substrate is an extremely reliable technique. Therefore, while an in vivo SPF was not available for P35 +, we can infer from Table I that we expecthe product to have an SPF of around 36 at an application thickness of 2 mg/cm 2. It should also be noted that the SPFs we obtained for each product increased with increasing TiO 2 concentration (p < 0.0001; Spearman coefficient of rank correlation). In particular, P35 +, which contained 12% TiO2, offered significantly higher protection (SPF 36) than P35, which contained 8.6% TiO 2 (SPF 23) (p -- 0.02; Mann-Whitney U test). Table II gives the SPFs measured on epidermis from six volunteers (H-M) for the five organic chemical sunscreens. It can again be seen that, with the exception of product C50, the mean SPFs are in close agreement with those claimed by the manufacturer. The fact that the SPF of C50 is significantly less than 50 is not surprising since this product offered relatively little protection against UV-A radiation. UV-A contributes between 15 % and 25 % of the erythemal dose from sunlight, depending on latitude, season, time of day, and atmosphericonditions (10), and hence in the extreme case of a sunscreen that absorbs no UV-A radiation, the maximum SPF that can be achieved is only 6, irrespective of the concentration of UV-B absorbers. It is unlikely that a sunscreen offering as low a UV-A protection as C50 could, in practice, provide an SPF as high as 50. One reason for the discrepancy between the SPF we measured for C50 and that claimed by the manufacturer is the difference between the ultraviolet spectrum of natural sun-

HIGH-SPF SUNSCREENS 293 0.09-0.075-0.060-0.045-0.030-0.015 c5 c15... c25 - - - c30... c50 0 i I I i i I I I i l 290 300 310 320 330 340 350 360 370 380 390 400 Wavelength, nm Figure 2. The absorption spectra of organic chemical sunscreen products normalized to equal area. Table SPFs Obtained From Each Physical Sunscreen Quoted Sunscreen SPF A1 A2 A3 A4 B C D E F G Mean + SD P8 8 7.0 10.2 8.9 8.9 11.2 11.3 7.4 9.3 8.6 9.9 9.3 + 1.4 P15 15 15.3 15.7 14.0 13.9 14.3 9.8 13.9 13.4 12.7 11.3 13.4 _+ 1.8 P25 25 24.4 27.9 17.7 23.4 23.2 26.0 17.6 26.1 22.1 19.0 22.7 _+ 3.6 P35 35 23.4 20.7 22.1 31.4 22.6 21.0 18.1 21.1 26.4 24.9 23.2 + 3.7 P35+ >35 a -- 41.3 43.1 18.3 38.7 36.9 35.0 35.6-+8.9 Al 4 representhe measurements on four samples of epidermis from subject A, and B-G represent measurements on epidermis from six other subjects. a Not evaluated with epidermis from subject A. I light and that of a xenon arc solar simulator used for the in vivo testing of sunscreens. The in vitro method for determining SPFs described in this paper assumes a solar spectrum that represents the spectral irradiance expected at noon, on a clear midsummer's day, at latitude 40øN. In vivo testing of sunscreens, on the other hand, employs a xenon arc filtered by WG320 and UG11 optical filters. The purpose of the UG11 filter is to remove visible radiation, but it also attenuates the longer UV-A wavelengths (11). Hence, UV radiation from a solar simulator is deficient in UV-A1 (340-400 nm) relative to natural sunlight, and SPFs measured by in vivo testing would be expected to differ from those expected in natural sunlight. In the case of a sunscreen with strong UV-B absorption but weak UV-A absorption, the deficiency in UV-A absorption will be

294 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS Table II SPFs Obtained From Each Organic Chemical Sunscreen Quoted Sunscreen SPF H I J K L M Mean + SD C5 5 6.2 6.0 6.8 6.6 6.8 5.9 6.4 + 0.4 C15 15 16.1 19.0 17.9 18.2 14.7 17.6 17.3 + 1.6 C25 25 19.6 22.2 22.1 25.7 26.8 29.2 24.3 _+ 3.6 C30 30 20.7 27.8 39.7 27.4 28.1 22.0 27.6 _+ 6.7 C50 50 22.6 31.1 20.9 24.2 28.5 27.0 25.7 + 3.8 H-M represent measurements on epidermis from six subjects. compensated for by the relatively low levels of UV-A from the solar simulator, and the sunscreen will therefore appear to have a higher SPF than would be obtained in natural sunlight. In order to evaluate the importance of the above effect, we recalculated our SPFs using the COLIPA xenon-arc solar simulator spectrum (1) in Equation 1. The results are shown in Table Ill, and it can be seen that for most of the sunscreens there is little difference between the calculated SPFs obtained using a natural solar spectrum and a xenon-arc solar simulator spectrum. This is not surprising since the majority of the sunscreens studied provided broad-spectrum protection. However, for product C50, which offers relatively little UV-A protection, the SPF calculated using the solar simulator spectrum is significantly higher than that calculated using the solar spectrum. We infer from these data that the use of solar simulators for i, vivo measurements of products with a high ratio of UV-B to UV-A absorption will overestimate the protection provided against natural sunlight. In conclusion, i. vitro determination of SPF using excised human epidermis is a quick and reliable alternative to i. vivo measurement for sunscreens expected to have high photoprotection, particularly since it yields SPFs more representative of natural sunlight for products that do not provide broad-spectrum protection. Table III Comparison of SPFs Calculated Assuming a Natural Solar Spectrum (clear sky at noon in midsummer at a latitude of 40øN) and the COLIPA Standard Xenon-Arc Solar Simulator Spectrum Sunscreen SPF calculated assuming natural solar spectrum SPF calculated assuming xenonarc solar simulator spectrum P8 9.3 10.0 P15 13.4 14.0 P25 22.7 24.9 P35 23.2 25.6 P35+ 35.6 43.5 C5 6.4 6.6 C15 17.3 19.0 C25 24.3 27.3 C30 27.6 32.0 C50 25.7 35.9

HIGH-SPF SUNSCREENS 295 ACKNOWLEDGMENTS This study was funded by the Department of Health. The views expressed are those of the authors and not necessarily those of the Department of Health. REFERENCES (1) COLIPA Sun Protection Factor Method, European Cosmetic Toiletry and Perfumery Association (COLIPA), Brussels, Belgium, October 1994. (2) Department of Health and Human Services, FDA, USA, Sunscreen drug products for over the counter use: Tentative final monograph: proposed rule. Federal Register, 58(90), 28194-28302 (1993). (3) J. Ferguson, "European Guidelines (COLIPA) for Evaluation of Sun Protection Factors," in Sunscreens: Development, Evaluation, and Regulatory Aspects, 2nd ed., N.J. Lowe, N. A. Shaath, and M. A. Pathak, Eds. (Marcel Dekker, New York, 1997), pp. 513-525. (4) B. L. Diffey, "Indices of Protection From In Vitro Assay of Sunscreens," in Sunscreens: Development, Evaluation, and Regulatory Aspects, 2nd ed., N.J. Lowe, N. A. Shaath, and M. A. Pathak, Eds. (Marcel Dekker, New York, 1997), pp. 589-600. (5) B. L. Diffey and J. Robson, A new substrate to measure sunscreen protection factors throughouthe ultraviolet spectrum, J. Soc. Cosmet. Chem., 40, 127-133 (1989). (6) H. Schaefer and T. E. Redelmeier, Skin Barrier: Principles of Percutaneous Absorption (Karger, Basel, 1996), p. 133. (7) A. F. McKinlay and B. L. Diffey, "A Reference Action Spectrum for Ultraviolet-Induced Erythema in Human Skin," in Human Exposure to Ultraviolet Radiation: Risks and Regulations, W. F. Passchief and B. F. M. Bosnjakovic, Eds. (Elsevier, Amsterdam, 1987), pp. 83-87. (8) J. L. Robb, L. A. Simpson, and D. F. Tunstall, Scattering and absorption of UV radiation by sunscreens containing fine particles and pigmentary titanium dioxide. D.C.I. Mag., 32-40 (March 1994). (9) M. W. Anderson, J.P. Hewitt, and S. R. Spruce, "Broad-Spectrum Physical Sunscreens: Titanium Dioxide and Zinc Oxide," in Sunscreens: Development, Evaluation, and Regulatory Aspects, 2nd ed., N.J. Lowe, N. A. Shaath, and M. A. Pathak, Eds. (Marcel Dekker, New York, 1997), pp. 353-397. (10) M.A. Pathak, "Photoprotection Against Harmful Effects of Solar UVB and UVA Radiation: An Update," in Sunscreens: Development, Evaluation, and Regulatory Aspects, 2nd ed., N.J. Lowe, N.A. Shaath, and M. A. Pathak, Eds. (Marcel Dekker, New York, 1997), pp. 59-79. (11) R. M. Sayre and P. P. Agin, A method for the determination of UVA protection for normal skin. J. Am. Acad. Dermatol., 23, 429-440 (1990).