in vitro assessment of sunscreen photostability: the effect of radiation source, sunscreen...

In vitro assessment of sunscreen photostability:the effect of radiation source, sunscreenapplication thickness and substrateR . S T O K E SRegional Medical Physics Department, Dryburn Hospital, Durham DH1 5TW, U.K.

B . D I F F E Y *Regional Medical Physics Department, Newcastle General Hospital, Newcastle, NE4 6BE, U.K.

Received 22 June 1998Accepted 17 November 1998

Keywords: human epidermis; photostability; sunscreens

Synopsis

The photostabilities of four sunscreen products have been assessed in vitro by applying sunscreen to a substrateand measuring the spectral transmission prior to, and after exposure to a source of ultraviolet (UV) radiation.Results were independent of whether an application thickness of 1 or 2 mg/cm2 was used, and whether the UVsource was natural sunlight or a xenon arc solar-simulator. There were significant differences, however, betweenresults obtained on a roughened quartz substrate and those obtained on excised human epidermis. It is unlikelythat any substrate will give an exact representation of the in vivo situation and, indeed, both quartz and excisedhuman epidermis have disadvantages associated with their use. However, the ranking of the four products interms of their photostability was the same for both substrates. This implies that transmission spectroscopy, witheither a quartz or a human epidermis substrate, can be used successfully to compare the photostabilities ofdifferent sunscreen products.

Resume

On a evalue la photostabilite de quatre ecrans solaires in vitro en appliquant l’ecran solaire sur un substrat et enmesurant la transmission spectrale avant, et apres l’exposition a une source de rayonnement ultraviolet (U.V.).Les resultats sont independants de l’application d’une epaisseur de 1 ou 2 mg/cm2, et du fait que la sourced’U.V. soit la lumiere solaire naturelle ou un arc de simulation solaire au xenon. On observe de differencessignificatives, cependant, entre les resultats obtenus sur un substrat de quartz rugueux et ceux obtenus sur del’epiderme humain entaille. Il est peu probable qu’un quelconque substrat donne une representation exacte dela situation in vivo et, en fait, aussi bien le quartz que l’epiderme humain entaille presentent des inconvenientsassocies a leur utilisation. Cependant, le classement des quatre produits en termes de photostabilite est le memepour les deux substrats. Ceci signifie que la spectroscopie par transmission, soit avec un substrat en epidermehumain soit avec du quartz, peut etre utilisee avec succes pour comparer la photostabilite de differents ecranssolaires.

Introduction

The active molecules within a sunscreen formulation can dissipate absorbed ultraviolet(UV) energy by, for example, re-emitting the energy in the form of heat, fluorescence or

* To whom correspondence should be addressed; Tel: 144–191–2731577; Fax: 144–191–2260970

0142–5463 © 1999 International Journal of Cosmetic Science

International Journal of Cosmetic Science 21: 341–351 (1999)

phosphorescence; transferring the energy to neighbouring molecules; undergoing reversi-ble isomerization; or undergoing irreversible decomposition. These reactions can result ina loss in the sunscreen’s efficiency and may also yield allergic or toxic by-products. Thisis clearly undesirable and as a result sunscreen manufacturers try to design products whichare photostable.

The photostability of sunscreen products is usually assessed in vitro, either bytransmission spectroscopy [1–5] or separation techniques such as high-performance liquidchromatography (HPLC) [3–6] or gas chromatography (GC) [7]. It is, in fact, recom-mended that both approaches be used in parallel since transmission spectroscopy can beaffected by optical artifacts such as UV absorbing photoproducts, whilst separationtechniques do not quantify the effects of any photo-induced changes in terms of asunscreen’s photoprotection. In the case of transmission spectroscopy, sunscreen isgenerally applied to a substrate and its spectral transmission measured prior to, and afterexposure to a UV source. Several different substrates have been used in this type ofanalysis including excised human stratum corneum [1], roughened quartz plates [2,3],glass plates [4], reconstructed human epidermis [5] and excised human epidermis [5].There have also been differences in the type of UV source and the sunscreen applicationthickness used. The aim of the present study is to examine how transmission measurementsof photostability are affected by choice of substrate, UV source and application thickness.We have taken four commercially available sunscreen products and exposed them tonatural sunlight and solar simulator radiation using roughened quartz and excised human-epidermis substrates, and application thicknesses of 1 and 2 mg/cm2.

Materials and methods

Sunscreen products

Four commercially available sunscreen products were used in this study, and these wereidentified as A, B, C and D. Only product A had a ‘photostable’ claim on its packaging.The nominal SPFs quoted on the packaging and the active ingredients of the four productsare as follows:

Product A (SPF 15) Octocrylene, titanium dioxide, butyl methoxydibenzoylmethane andterephthalylidene dicamphor sulfonic acid.

Product B (SPF 25) Octyl methoxycinnamate, butyl methoxydibenzoylmethane, benzophe-none-3.

Product C (SPF 25) Octyl methoxycinnamate, butyl methoxydibenzoylmethane,4-methylbenzylidene camphor and titanium dioxide.

Product D (SPF 15) Octyl methoxycinnamate, butyl methoxydibenzoylmethane andtitanium dioxide.

Substrates

Excised human-epidermis Skin was taken from the underside of the female breast duringthe operation of breast reduction. It was obtained by a process known as de-epidermalisation, the principle of which is to remove the epidermis and epidermal

342 Stokes and Diffey

appendages whilst leaving the deepest layers of the dermis in situ. The skin (approximately10 3 4 cm) was received within 1 day of surgical operation, and was cut into squares ofapproximately 4 3 4 cm. The samples of skin were placed in a water bath at 60°C for 45sec [8]. On removal from the water bath, the epidermis was gently separated from thedermis by careful peeling. Epidermal sheets were stored in physiological saline at 4°C untilrequired, which was normally within 5 days. Sheets of epidermis can be stored at 4°C forseveral weeks without loss of barrier function [8].

Quartz plates One surface of two quartz plates (50 3 50 mm) was roughened using aground-diamond glass scorer in order to facilitate the application of sunscreen. Each platewas pressed down on the surface of the glass scorer and moved in a figure of eight pattern(in order to avoid circular or linear grooves) for approximately 2 min.

Sources of UV radiation

Natural sunlight Samples were irradiated at Durham, UK (latitude 55°N). They wereplaced in direct sunlight between 11:00 am and 3:00 pm on days between the months ofJune and August 1997.

Solar simulator The solar simulator comprised a 900 W xenon arc lamp filtered by UG5(1 mm) and WG320 (1 mm) optical glass filters. Samples were positioned 12 cm from theexit port of the solar simulator where the UV irradiance was similar to that received by thesamples exposed to natural sunlight. An Optronic model 742 spectroradiometer was usedto measure the spectral irradiance of the solar-simulator output and it was found that theUVB and UVA intensities were 0.11 mW/cm2 and 5.5 mW/cm2, respectively.

Experimental technique

The substrate was positioned directly over the teflon input optics of an Optronic model 742spectroradiometer controlled by a Hewlett-Packard HP85 microcomputer. Radiation froma 75 W xenon arc lamp (filtered by a Schott UG5 colour glass filter) was directed onto thesubstrate via a light guide, and the photocurrent recorded from 290 to 400 nm in steps of5 nm. A micropipette was then used to dispense the required amount of sunscreen onto thesubstrate. The sunscreen was spotted at several positions on the substrate and rubbed (witha latex-gloved finger which was not pre-saturated with sunscreen) in a circular pattern forabout 10 sec to give as uniform a layer as possible. The sunscreen was allowed to dry for20 min and the photocurrent again measured in 5 nm steps between 290 and 400 nm. Foreach wavelength, the ratio of the photocurrent recorded before and after application of thesunscreen was calculated; this gave the monochromatic protection factors, PF(λ), whichwere used in the following expression [9] to give the SPF:

SPF 5

O400

290E(l)ε(l)

O400

290E(l)ε(l)/PF(l)

where E(λ) is the spectral irradiance of sunlight expected for a clear sky at noon inmidsummer for a latitude of 40°N (solar altitude 70°), and ε(λ) is the effectiveness of

343In vitro assessment of sunscreen photostability

radiation of wavelength λ nm in producing erythema in human skin [10]. The substratewas then positioned in either direct sunlight, or in front of the exit port of the solarsimulator. During irradiation, the UV dose-rate was measured at 1 minute intervals usinga UV photodiode probe interfaced with a Psion model LZ64 data-logger, previouslycalibrated with the Optronic spectroradiometer. The temperature was monitored using a K-type thermocouple to check that it remained below 35°C. At regular intervals, the substratewas removed from its irradiation position and the SPF determined as described above.Sunscreen was also applied to a second substrate which acted as a control. The control wasstored in the dark, but maintained at the same temperature as the irradiated substrate andits SPF determined at the same times. For a given sunscreen, three measurements weremade for each permutation of substrate (quartz and epidermis), application thickness (1and 2 mg/cm2), and UV source (natural sunlight and solar simulator).

Results and discussion

Tables Ia through d give the mean SPFs obtained for products A, B, C and D under eachset of experimental conditions. Because of the changing UV irradiance in natural sunlight,samples were removed for SPF determination at times which did not correspond exactly toexposures of 6, 12 and 18 SED.1 Consequently, we interpolated between measurementsusing a cubic spline fit to estimate the SPFs after UV exposures of 6, 12 and 18 SED. [Theambient diurnal erythemal radiation on an unshaded, horizontal surface under clear skiesat 55°N is about 35 SED in mid-summer, and 1 SED in mid-winter]. The SPFs obtainedfor a 2 mg/cm2 layer of sunscreen on epidermis are in close agreement with the nominalSPFs on the sunscreen packaging. This was expected as we have previously demonstratedthat a human-epidermis substrate gives SPFs which are in close agreement with thoseobtained by in vivo assay [11,12]. For a given application thickness, the results for a quartzsubstrate are significantly higher than those for epidermis. This is primarily becauseroughened quartz does not have the same surface topology as the excised epidermis andtherefore the distribution of sunscreen is different. Also, a quartz substrate does not allowfor any skin-sunscreen interaction.

In order to investigate whether irradiation affected the photoprotection of the fourproducts, normalised graphs of SPF versus UV exposure were plotted for each serialmeasurement on a given sample. The areas under the graphs for irradiated and controlsamples were then compared using an unpaired t-test [2]. It should be noted that thestatistical power of the analysis is low because experimental and time constraints meantthat only three samples could be used for each set of experimental conditions. Onepidermis, the SPFs of products A and B were not significantly altered (P . 0.3; unpairedt-test) by irradiation. On quartz, however, the SPF of product A decreased by 10–20% aftera UV exposure of 18 SED, while the SPF of product B decreased by approximately 50%.This dependence of photochemical behaviour on substrate type is discussed in more detaillater. Products C and D exhibited a marked reduction in SPF (P , 0.01; unpaired t-test)under all conditions of substrate, UV source and sunscreen thickness.

1 Standard Erythema Dose (SED) is a measure of sunburning radiation: a minimal erythema on theunacclimatised skin of subjects who always burn and never tan would require an exposure of about 1.5 SED,whereas in people who tan easily and rarely burn, an exposure of about 6 SED would be needed to cause aminimal erythema.


Table Ia SPFs obtained for product A under each set of experimental conditions (Mean 6 SD).

UV exposure (SED)Sunscreen thickness

Substrate (mg/cm2) Irradiation source 0 6 12 18

Roughened quartz1

Solar simulator 15.3 ± 2.1 13.6 ± 2.2 13.3 ± 2.0 12.6 ± 2.1Sunlight 13.4 ± 0.6 12.0 ± 0.7 12.3 ± 0.6 11.8 ± 0.8

2Solar simulator 35 ± 4 32 ± 4 31 ± 3 31 ± 3Sunlight 38 ± 9 30 ± 5 30 ± 6 29 ± 5

Human epidermis1


2Solar simulator 14.5 ± 1.5 17.2 ± 1.4 17.3 ± 1.4 17.6 ± 1.0Sunlight 18.5 ± 4.1 21.8 ± 3.3 22.4 ± 4.1 22.2 ± 2.9

Table Ib SPFs obtained for product B under each set of experimental conditions (Mean 6 SD).



Roughened quartz1



Human epidermis1



345In vitro

assessment of sunscreen photostability

Table Ic SPFs obtained for product C under each set of experimental conditions (Mean 6 SD).



Roughened quartz1



Human epidermis1



Table Id SPFs obtained for product D under each set of experimental conditions (Mean 6 SD).



Roughened quartz1



Human epidermis1



346Stokes and D

iffey

To investigate whether irradiation affected the absorption spectra of the products, weplotted normalised graphs of UVA:UVB absorbance ratio [13] versus UV exposure foreach serial measurement on a given sample. The areas under graphs for irradiated andcontrol samples were compared using an unpaired t-test. Irradiation did not significantlyalter (P . 0.15; unpaired t-test) the UVA:UVB absorbance ratio of product A onepidermis. On quartz, however, irradiation caused a reduction in the UVA:UVBabsorbance ratio. This reduction was associated with a loss in UVA absorbance, but therewas also some loss in UVB absorbance. Irradiation caused a decrease (P , 0.001) in theUVA:UVB absorbance ratio of product B on both epidermis and quartz. On epidermis, thisdecrease was due to a loss in UVA absorbance which in itself was not sufficient to causea significant change in SPF which is primarily a measure of UVB protection. On quartz,however, product B lost both UVB and UVA absorbance resulting in the observeddecrease in SPF. Products C and D exhibited a marked reduction in UVA absorbance(P , 0.002; unpaired t-test) under all conditions of substrate, UV source and applicationthickness (Fig. 1).

The decrease in UVA absorbance exhibited by all four products is probably due to a lossin butyl methoxydibenzoylmethane, a UVA filter which is known to be photo-unstable.Roscher et al. [7], for example, irradiated a solution of butyl methoxydibenzoylmethane incyclohexane for 100 h using a 450 W mercury vapour lamp and found that there was totaldecomposition of the butyl methoxydibenzoylmethane yielding products such as tert-butylbenzene, p-tert-butylbenzoic acid and p-methoxybenzoic acid. In conditions morerepresentative of those in which sunscreens are used in practice, Deflandre and Lang [3]

Figure 1. Absorption spectra of product C obtained prior to, and after a UV exposure of6 SED; product was applied to a quartz substrate.


found that butyl methoxydibenzoylmethane was photo-unstable when incorporated in aparticular non-ionic emulsion and irradiated with a solar simulator. In addition, Diffey etal. [2], using transmission spectroscopy with a quartz substrate, and Forestier et al. [6],using HPLC and a polymethylmethacrylate substrate, found that butyl methoxydibenzoyl-methane can be very photo-unstable in certain sunscreen formulations. One means ofimproving its stability is to use it in combination with certain other UV filters such as4-methylbenzylidene camphor [2,5,6,14], terephthalylidene dicamphor sulfonic acid [2,6]and octocrylene [14]; it is believed that butyl methoxydibenzoylmethane is able to transferabsorbed UV energy to these compounds instead of undergoing alternative reactions whichresult in a loss of sunscreening efficiency. The reasonably good photostability of productA can therefore be attributed to the inclusion of octocrylene and terephthalylidenedicamphor sulfonic acid. It should be noted, however, that product C exhibits a markedloss in UVA absorbance despite containing 4-methylbenzylidene camphor.

There was no significant difference (P . 0.1; unpaired t-test) between the resultsobtained for samples irradiated with natural sunlight and those irradiated with the solarsimulator. The latter incorporates a UG5 filter and is therefore deficient in UVA andvisible radiation compared to natural sunlight. Our results imply that this deficiency in thesolar simulator spectrum does not significantly affect photostability measurements on thefour sunscreen products studied.

There was no significant difference (P . 0.1; unpaired t-test) between results obtainedfor application thicknesses of 1 and 2 mg/cm2. To a first approximation, one may expectthe effect of any photodegradation to be less for the thicker layer of sunscreen.Gonzenbach [14] considered a layer of sunscreen to be made up of several overlyingsheets. Initially, most photons incident on the sunscreen will be absorbed in the upper sheetand any photodegradation will occur first in this sheet. As a result, the upper sheet willgradually become transparent to UV photons and photodegradation will begin to occur inthe underlying sheet. The sheets of sunscreen will therefore lose their protection one byone. A thicker layer of sunscreen comprises a greater number of sheets and hence therewill be some sheets providing full protection for a longer period of time. The relative lossof protection would therefore be less for a thicker layer than a thinner layer. One reasonwhy our results do not support this theory may be the uneven surface topologies of thequartz and epidermis substrates. As a result, applied sunscreen will not form a uniformlayer but will instead concentrate in the sulci. Even at an application thickness of 2 mg/cm2

some of the ‘ridges’ will be covered, or mininally covered, by sunscreen and the loss ofprotection will be more rapid than expected. There were significant differences (P , 0.05;unpaired t-test) between the results obtained for roughened quartz and human epidermissubstrates (Fig. 2). These differences are probably due to a combination of thefollowing:

1. De-emulsification of the sunscreens. The four sunscreens studied consisted of emulsionsof oil in water, the active sunscreen molecules residing in the oil phase. On applicationto a substrate the emulsion breaks and the water evaporates to leave behind an oil filmcontaining the sunscreen molecules. During this process, the SPF of the product canchange significantly [15]. The exact nature of the de-emulsification process andtherefore any changes in SPF depend upon the type of substrate. Further research isneeded to quantify these changes (and identify their causes) on quartz and epidermis.


2. Absorption of sunscreens. Absorption of a sunscreen into a substrate results insignificant changes in its SPF. The amount of absorption and therefore the change inSPF depend upon the type of substrate [16].

3. Thermal effects. These include thermally-induced changes in the sunscreen chemicalsand evaporation of the sunscreen, both of which could be affected by substrate, anddrying out of the epidermis. As the epidermis dries out, its UV transmission is modified[17] and as a result there will be changes in the observed SPF. Drying of the epidermisalso alters its physical structure and, in particular, makes the surface topology moreeven. This may affect the distribution of the applied sunscreen and also therefore itsperformance.

4. Photochemical effects. Sunscreen molecules may dissipate absorbed UV energydifferently depending on their surrounding environment [18]. In the case of anepidermal substrate, for example, some sunscreen molecules may be able to transferabsorbed energy to secondary molecules within the epidermis. This pathway would notbe available on quartz and therefore the molecules would have to dissipate energy by adifferent means which may affect the sunscreen’s efficiency.

We corrected our data for (1), (2) and (3) by normalizing against the control sampleswhich had been stored at the same temperature as the irradiated samples but in the dark.The results for quartz and epidermis substrates were then re-compared and it was foundthat there was improved agreement implying that sunscreen de-emulsification and thermaleffects do account for some of the difference in results between quartz and epidermis.

Figure 2. Normalized SPF versus UV exposure for products A, B, C and D on quartz and epidermissubstrates.


However, there were still significant differences (P , 0.05; unpaired t-test) between theresults obtained on the two substrates, most notably for products B and D. This impliesthat the photochemical behaviour and therefore the photostability of sunscreens can varydepending on the type of substrate to which they are applied.

Conclusions

The results of this present study suggest that serious consideration should be given to thechoice of substrate for assessing the photostability of sunscreens. It is unlikely that anysubstrate will give results which are in exact agreement with the in vivo situation, andindeed both substrates used in the present study have disadvantages associated with theiruse. Roughened quartz, for example, does not take into account any sunscreen-skininteraction: sunscreen formulations do not necessarily de-emulsify on quartz in the sameway as they do on skin, while sunscreen molecules may dissipate absorbed UV energy bydifferent mechanisms on quartz than on skin. Excised human epidermis dries out duringirradiation causing changes in its structure and absorption spectrum. These changes can betaken into account by comparing irradiated samples to control samples stored in the dark,but it is still not clear whether the chemical behaviour of sunscreens on the dryingepidermis is the same as that on human skin in vivo. In addition, excised human epidermisis difficult to obtain. The ranking of products in terms of their photostabilities was the samefor both substrates (i.e. Product A . Product B . Product C . Product D). This indicatesthat although no substrate may give quantitatively the same results as would be obtainedin vivo, the technique can be used to compare the photostabilities of different sunscreenproducts. Additional information may be obtained using other techniques such as chromato-graphy and the in vivo stripping method described by Marginean Lazar et al. [5].

Acknowledgements

This study was funded by the Department of Health. The views expressed are those of theauthors and not necessarily those of the Department of Health.

References

1. Kammeyer, A., Westerhof, W., Bolhuis, P.A., Ris, A.J., Hische, E.A. The spectral stability ofseveral sunscreening agents on stratum corneum sheets. Int. J. Cosmet. Sci. 9, 125–136(1987).

2. Diffey, B.L., Stokes, R.P., Forestier, S., Mazilier, C., Rougier, A. Suncare product photo-stability: a key parameter for a more realistic in vitro efficacy evaluation. Eur. J. Dermatol. 7,226–228 (1997).

3. DeFlandre, A. and Lang, G. Photostability assessment of sunscreens. Benzylidene camphor anddibenzoylmethane derivatives. Int. J. Cosmet. Sci. 10, 53–62 (1988).

4. Berset, G., Gonzenbach, H., Christ, R., Martin, R., DeFlandre, A., Mascotto, R.E., Jolley,J.D.R., Lowell, W., Pelzer, R., Stiehm, T. Proposed protocol for determination of photostabilityPart I: Cosmetic UV filters. Int. J. Cosmet. Sci. 18, 167–177 (1996).

5. Marginean Lazar, G., Fructus, A.E., Baillet, A., Bocquet, J.L., Thomas, P., Marty, J.P.Sunscreens’ photochemical behaviour: in vivo evaluation by the stripping method. Int. J.Cosmet. Sci. 19, 87–101 (1997).


6. Forestier, S., Mazilier, C., Rougier, A. Suncare product photostability: a key parameter for amore realistic in vitro efficacy evaluation. Part II: chromatographic analysis. 19th WorldCongress of Dermatology, Sydney (1997).

7. Roscher, N.M., Lindemann, M.K.O., Kong, S.B., Cho, C.G., Jiang, P. Photodecomposition ofseveral compounds commonly used as sunscreen agents. J. Photochem. Photobiol. A: Chem 80,417–421 (1994).

8. Schaefer, H., Redelmeier, T.E. (1996) Skin barrier: Principles of Percutaneous Absorption.Karger, Basel p. 133 (1996).

9. Diffey, B.L., Robson, J. (1989) A new substrate to measure sunscreen protection factorsthroughout the ultraviolet spectrum. J. Soc. Cosmet. Chem. 40, 127–133 (1994).

10. McKinlay, A.F., Diffey, B.L. A reference action spectrum for ultraviolet-induced erythema inhuman skin. In: Human Exposure to Ultraviolet Radiation: Risks and Regulations (Eds. W.F.Passchier and B.F.M. Bosnjakovic). Amsterdam: Elsevier, 1987, 83–87.

11. Stokes, R.P., Diffey, B.L. (1997). In vitro assay of high-SPF sunscreens. J. Soc. Cosmet. Chem.48, 289–295 (1997).

12. Stokes, R.P., Diffey, B.L. How well are sunscreen users protected? Photodermatol Photo-immunol Photomed. 13, 186–188 (1997).

13. Diffey, B.L. Indices of protection from in vitro assay of sunscreens. In: Sunscreens:Development, Evaluation and Regulatory Aspects (Eds. N.J. Lowe, N.A. Shaath and M.A.Pathak). Marcel Dekker, New York, 1996, 589–600.

14. Gonzenbach, H.U., Pittet, G.H. Photostability, a must? International Conference BroadSpectrum Protection: ‘The Issues and Status’, London (1997).

15. Sottery, J.P., Jaramillo, J.H. Delivering enhanced UVA and UVB protection: a unified theory ofsunscreen performance on skin. Proceedings of the 19th IFSCC Congress, Sydney (1996).

16. Sottery, J.P. The current status and future of in vitro SPF testing. International ConferenceBroad Spectrum Protection: ‘The Issues and Status’, London (1997).

17. Bruls, W.A.G. Optical properties of human epidermis. PhD Thesis, University of Utrecht, TheNetherlands.

18. Rieger, M.M. Photostability of cosmetic ingredients on the skin. Cosmet. Toil. 112, 65–72(1997).


in vitro assessment of sunscreen photostability: the effect of radiation source, sunscreen...

Documents