Chances and limits of an improved method to assess water resistance of cosmetic sunscreen products in vitro on polymethylmethacrylate plates

Synopsis

While sun protection factor (SPF) and UVA protection are the most important determinants of a cosmetic sunscreen product, water resistance is the third important feature. The Colipa in vivo method is the internationally accepted standard method to assess water resistance. It is time-consuming and expensive. A screening method to quickly predict water resistance properties on low cost therefore is a specific request of product developers. Several in vitro screening methods are published but the predictive power of all these methods is limited. In this paper, we describe an adaptation of the in vitro UVA protection method of Colipa for a water resistance screening. Although the method is quick and most parts are standardized and approved by Colipa, the results were not in advantage of other published screening methods. Taking into account, the scatter of the results, seven of 16 sunscreen products that were developed as water resistant formulations could be unambiguously estimated to be water resistant by the in vivo water resistance screening method on five subjects while nine failed. In 10 of these 16 results, the in vitro SPF-based method was in accordance with in vivo findings, whereas in six cases, the method was not predicting correctly. Based on these results, the authors recommend to use the in vitro screening methods to pre-select from candidates which cannot all be tested in vivo. The pre-selected products can be screened in the Colipa in vivo water resistance method with a reduced number of volunteers (usually 5) to predict water resistance. In case, the water resistance estimated in such an in vivo screening is found at about 65% or higher the study can be deemed successful and completed with further subjects to fulfil the Colipa requirements.

Résumé

Alors que le facteur de protection solaire (FPS) et la protection contre les UVA sont les déterminants les plus importants d'un produit cosmétique, la résistance à l'eau d'une crème solaire en est la troisième fonction importante. La méthode Colipa in vivo est la méthode standard internationalement acceptée pour évaluer la résistance à l'eau. Elle est longue et coûteuse. Un procédé de criblage pour rapidement prédire les propriétés de résistance à l'eau, à un coût faible, est donc une demande spécifique des développeurs de produits. Plusieurs méthodes de criblage in vitro sont publiées, mais la valeur prédictive de toutes ces méthodes est limitée. Dans cet article, nous décrivons une adaptation de la méthode de protection UVA in vitro du Colipa pour une évaluation de la résistance à l'eau. Bien que la méthode soit rapide et la plupart des pièces soient standardisées et approuvées par Colipa, les résultats n'étaient pas avantageuse par rapport aux autres méthodes de criblage publiées. Tenant compte de la dispersion des résultats, 7 des 16 produits de protection solaire résistantes à l'eau pouvaient sans ambiguïté être estimés résister à l'eau par la méthode de criblage in vivo sur 5 sujets, tandis que 9 ont échoué. Dans 10 de ces 16 résultats, la méthode basée sur le FPS in vitro était conforme aux conclusions de l'in vivo alors que dans 6 cas, la méthode n'a pas prédit correctement. Sur la base de ces résultats, les auteurs recommandent d'utiliser la méthode de criblage in vitro pour une présélection des candidats qui ne peuvent pas tous être testés in vivo. Les produits présélectionnés peuvent être testés par la méthode du Colipa in vivo avec un nombre réduit de volontaires (généralement 5) pour prédire la résistance à l'eau. Dans le cas où la résistance à l'eau estimée dans un tel criblage in vivo se trouve à environ 65% ou plus, l'étude peut être considérée comme réussie et complétée par d'autres sujets pour répondre aux exigences Colipa.

Introduction

Already in 1979, Greiter et al. [1] published an in vitro method to measure the water resistance of sunscreens. They used sheep wool and pigskin as a replacement for human skin, treated it with the sunscreen products and immersed it in water. As product developers were regularly looking for improved cost-effective and quick laboratory methods to estimate the water resistance of their sunscreen formulations, further methods were developed and published. To replace living human skin, very different materials were used, as biopsied human epidermis [2] or non biological materials as TRANSPORE tape [3] or transparent plastic plates of Polymethylmethacrylate (PMMA) [3, 4]. Hagens et al. [5] developed a method on living human skin without UV-irradiation. They applied drops of water to the human volar forearm after sunscreen application and measured the contact angle of the water drop. A high contact angle indicates a hydrophobic and water repellent surface. The higher the contact angle was after a sunscreen product application, the higher was the chance that the tested sunscreen was water resistant. Good correlation of in vivo and in vitro data was shown in three publications, which all came out in 2007 [3-5]. Despite of the promising correlations in all three methods, a scatter in data was found which cannot be neglected. It remained unclear in how many cases a developer of a sunscreen can correctly estimate water resistance or lack of water resistance for a newly developed product. The goal of Pissavini et al. [4] was not to estimate in vivo results by their in vitro test. So, they did not provide the predictive power of their method. Hagens et al. [5] estimated the predictive value. They found a negative predictive value of 37.5% for their method and a positive predictive value of 100%. This means that test products with a high contact angle could be identified as water resistant in all cases, whereas for products with a low contact angle, there was a high uncertainty if they were water resistant or not in the in vivo method.

From all published methods available, the conclusion is that the predictive value of these methods is by far not high enough to replace the Colipa in vivo method.

The aim of this work is to improve water resistance testing in vitro by combination of parts of the Colipa in vivo method for water resistance testing [6] with the Colipa UVA in vitro method [7]. As a substrate to replace living skin, we used roughened PMMA plates as described in the Colipa UVA in vitro method. We measured the transmission of the product treated plates before and after water exposure by use of the spectro radiometric method as described in the Colipa UVA in vitro test. The water exposure was adopted as closely as possible to the in vivo procedures in a whirlpool as described in the Colipa in vivo water resistance method. In this study, we used the same location of plates in the pool were the test fields of the human volunteers were positioned in the in vivo part of the study (see Fig. 1). Sixteen prospectively water resistant sunscreen products were tested with the Colipa in vivo water resistance method as well as with our improved in vitro method. Their sun protection factor (SPF) was found in the range of 12–72 and the in vivo water resistance was found in the range of 43–97%. The in vitro determination of water resistance was performed in two ways. The first variant included the complete procedure of the Colipa in vitro UVA test with pre-irradiation of the plates and evaluation of the results on the basis of the in vitro SPF results. In the second variant, results were based on the in vitro UVA protection factors as we were interested if the water resistance based on UVA protection goes in line with the water resistance based on UVB- or erythema measurements.

Figure 1. In vitro evaluation of the water resistance of sunscreen products. The slides are not visible as they are attached to the two holders facing the back wall of the whirlpool.

Materials and methods

The determination of the UV protection in vitro was performed according to the Colipa Guideline 2009 [7]. In a first step, disposable PMMA slides (5 × 5 cm Plexiglas XT colourless 24770 UVD, one side sandblasted with Corundum, particle size 90–150, Sa 2 microns; Schönberg Plexiglas, Hamburg, Germany) were weighed before application of the test products. An amount of 0.75 mg cm⁻² of each test product was applied spotwise on the roughened side of the slides. To achieve a sufficient homogeneous distribution, test products were distributed as a thin film using a pre-saturated fingertip. A slight pressure was applied over the whole area for <30 s, and then a higher pressure was exerted for another 20–30 s. The coated slides were weighted again and left in the dark at ambient temperature (below 40°C) for equilibration for at least 15 min. Afterwards, the slides were re-weighed to determine the amount of product after the waiting period. Additional weighing of the dry residue of products was used as a further control of the correct application dose because application by weight of quickly evaporating products can be difficult.

Six slides per test product were prepared. Three slides were treated with the watering procedure and afterwards spectroradiometric measurements were performed according the Colipa guideline [7] as described below. Three further slides were measured due to the same procedure but leaving out the watering procedure.

For evaluation of the water resistance of sunscreen products, the product-coated slides were placed into a whirlpool with water circulation (temperature: 29 ± 2°C) to fully submerse them for 20 min. The slides were placed to the position where the test areas on the back of the volunteers are positioned in the in vivo water resistance studies (see Fig. 1). By this, the water exposure and movement was comparable to the in vivo studies. After the first watering period, the slides were allowed to air-dry for 15 min. Subsequently, the product-coated slides were immersed in the whirlpool for a second period of 20 min followed by drying on a heated plate of 37 ± 2°C for 15 min or until complete drying. No water drops on the test slides at the end of this last drying period had to be visible. If water droplets remained on the slides, drying time was extended.

The measurement and irradiation of the six slides were performed according to the Colipa UVA protection factor measurement and with devices qualified for the Colipa method [7, 8]. Nine points on every slide were measured with the spectroradiometer. After the first spectroradiometric measurements, an irradiation of the product-coated slides followed. This irradiation of the slides was performed only once and before the final measurement to cover possible loss of protection due to photo instability of the sunscreen filters. Comparable to the in vivo water resistance measurement which is performed on two test areas (the area to determine dry SPF and the area for wet SPF detection), the in vitro water resistance measurements were also performed on two slides.

The irradiation time was determined in accordance with Colipa guideline [7], and the irradiation was performed using the solar simulator Long-arc Atlas SuntestTM insolator, type CPS+, filtered with its original UV short cut-off filter (Ref: 56052371) and IR dichroic mirror (Ref: 56052059) (Atlas MTT GmbH, Linsengericht/Altenhasslau, Germany) as UV light source.

For spectroradiometric measurements of the slides UV transmission from 290 to 400 nm, the UV spectral radiometer Tristan exUV-U-H (m.u.t. GmbH, Wedel, Germany) equipped with a LAX Cute Xenon UV Light Source (Asahi Spectra U.S.A. Inc., Torrance, CA, U.S.A.) was utilized.

Blank measurements were performed using glycerine-coated slides as described in the Colipa guideline [7].

Analysis of data was also performed in accordance with the Colipa Guideline [7]. The UVAPF and UVBPF (SPF in vitro) of a product were calculated as the mean of the 27 single results as derived from the individual plates.

The percentage of Water Resistance Retention (% WRR) concerning the SPF and the UVA data was calculated according to the following formulae:

• Calculation of in vitro UVB water resistance (WRR):
• • 
• with UVBdry = In vitro UVB protection factor (in vitro SPF) without water immersion (dry).
• UVBwet = In vitro UVB protection factor (in vitro SPF) after water immersion (wet).

• Calculation of in vitro UVA water resistance (WRR):
• • 
• with UVAdry = In vitro UVA protection factor without water immersion (dry).
• UVAwet = In vitro UVA protection factor after water immersion (wet).

For both methods, 95% confidence limits were calculated from the 27 single water resistance results per plate.

The in vivo determination of SPF was performed according to the International SPF Test Method of 2006 [9] and the Colipa Guideline for evaluating water resistance of 2005 [6]. The dry and wet SPF was determined on the back of five human subjects (screening on a reduced case number) using an appropriate sun simulator. The sun simulator unit used for this study was a 300 W Multiport (SOLAR Light, Philadelphia, PA, U.S.A.). The UV spectrum of the device is regularly calibrated and complied with the International SPF Test Method of 2006 [9].

The skin type of each subject was estimated on the back of the subjects with a tri-stimulus Chromameter (Chromameter CR 300; Minolta, Langenhagen, Germany) on the first study day. To estimate the unprotected minimal erythemal dose (MED), the individual typological angle (ITA°) was calculated and used as a predictor of the subjects' MED.

An amount of 2 mg cm⁻² of the respective test product was applied quickly in small droplets with a syringe all over the test area. Spreading was performed by gently rubbing with a saturated finger-cot, first with rotating followed by crosswise movements to achieve a final uniform coating on the skin. The spreading time was between 20 and 50 s.

Untreated and product treated areas were irradiated with increasing amounts of UV-doses. Between 15 and 30 min after product application, the irradiation of the test areas was started. On each test area, six small circular spots of approximately 1 cm² were irradiated with increasing doses with an increment of 1.25 or 1.12. The estimated MED dose was irradiated on the untreated control area as the fourth step of the six doses of irradiation. The treated areas were irradiated with the doses of the untreated area multiplied by the estimated SPF of the test products. Visual rating of skin erythema was performed 16–24 h after irradiation by a technician blinded to the application scheme.

On the following day, water exposure was started between 15 and 30 min after product application. The subjects stayed in a whirlpool with water circulation (temperature: 29 ± 2°C) for 20 min. After a waiting period of 15 min, a second water exposure of 20 min was performed. After a further waiting period of 15 min or until complete drying of the test fields, the treated areas and a further untreated area were irradiated, as described above. Sixteen to 24 h after irradiation the visual reading of skin erythema was performed by a trained technician.

• Calculation of in vivo water resistance (WRR):
• • 
• with SPFdry = SPF in vivo without water immersion (dry).
• SPFwet = SPF in vivo after water immersion (wet).

For the in vivo screening method, 95% confidence limits of water resistance were calculated from the five single water resistance results.

Results

Table 1 shows the water resistance values received by the in vitro determination of the UVB and UVA protection factors after water immersion and for the in vivo determination of water resistance due to the Colipa method including the 95% confidence intervals.

Table 1. In vivo sun protection factor (SPF) and mean percentage of water resistance (WRR) of sunscreen products as determined in vivo as well as in vitro based on UVB and UVA protection and 95% confidence intervals

Product	In vivo SPF dry	In vivo SPF % WRR	95% CI (%)	in vitro UVB % WRR	95% CI (%)	in vitro UVA % WRR	95% CI (%)
1	35	52	15.9	86	14.0	61	2.0
2	31	96	19.3	44	8.7	56	1.0
3	42	67	9.5	39	3.5	55	2.2
4	54	97	15.2	114	17.4	51	2.3
5	17	37	11.2	80	4.8	62	2.6
6	31	65	21.3	49	6.7	34	1.8
7	57	43	11.2	36	3.8	75	5.2
8	72	77	18.3	102	14.6	44	2.5
9	17	70	11.6	75	6.9	66	4.6
10	12	96	18.5	60	10.4	28	2.7
11	26	56	16.8	38	3.2	19	1.2
12	34	46	18.3	49	5.1	24	1.1
13	29	52	14.4	70	9.5	23	1.4
14	26	68	18.9	109	16.8	39	5.8
15	41	44	22.8	60	11.6	50	2.8
16	37	55	7.4	103	23.3	69	2.9

Figure 2 shows the UVB water resistance measured in vitro (water resistance based on in vitro SPF) compared to the vivo water resistance. A slight correlation (R = 0.201) was found. A test product can be labelled as water resistant according to the Colipa method [6] in case the lower 95% confidence limit of water resistance lies above 50%. According to the in vivo water resistance, seven test products unambiguously showed water resistance and nine products failed. Ten of these 16 products were estimated correctly and unambiguously to be water resistant or not based on the SPF in vitro method. (Two of the cases were counted as unambiguous were the lower confidence limit just minimally crossed the 50% water resistance line). Four of the test products that were not water resistant in vivo were found to be water resistant in the in vitro water resistance method (overestimation). On the other hand, in two cases of unambiguous water resistance in vivo, the in vitro SPF method showed clearly no water resistance (underestimation).

Figure 2. Correlation of in vitro water resistance based on SPF, and in vivo water resistance of 16 sunscreen products shown with 95% confidence intervals.

Regarding the water resistance based on the in vitro UVA protection factor (see Fig. 3), no correlation at all was found with the in vivo water resistance method. Water resistance based on in vitro UVA protection reveals a much lower scatter compared to the in vivo and in vitro SPF method. This can be explained by the in vivo SPF based correction procedure of the UVA in vitro method (7). However, four of the eight products unambiguously classified in vivo were predicted clearly wrong by the in vitro UVA protection method.

Figure 3. Correlation of in vitro water resistance based on UVA protection factor, and in vivo water resistance of 16 sunscreen products shown with 95% confidence intervals.

Discussion

The in vitro water resistance method based on the in vitro SPF showed only a weak correlation to the in vivo water resistance results. On the first view, this looks worse when compared to the cited published methods. But it has to be considered that in this study, we included only test products that were expected to be water resistant. Therefore, the range of expected water resistance results between 0% and about 40% is missing compared with the other methods where also products known to be not water resistant and products with very low water resistance potential were investigated. Only 13% of the test products were unambiguously underestimated and 25% overestimated by our SPF in vitro method. A clear advantage or disadvantage of our in vitro SPF method compared with the in vitro method of Pissavini et al. [4] or Hagens et al. [5] cannot be stated. For a valid comparison, a larger number of tested products would be needed.

The results of the in vitro water resistance based on UVA protection factors as shown in Fig. 3 are completely uncorrelated to in vivo data and have no predictive value for in vivo water resistance. In some cases, products which were water resistant in vivo and in vitro based on SPF showed only little UVA water resistance. As it can be seen in Table 1, product 8 was water resistant in vivo and in SPF in vitro assessment but showed no water resistance based on the in vitro UVA protection factors. Possibly, these are cases where UVA filters are stronger washed off than UVB filters. As seen for product 7 in Table 1, the opposite is also possible. This product failed water resistance in vivo and in UVB in vitro assessments, while the water resistance based on UVA protection was 75%.

There is an important methodological problem in the estimation of water resistance for the in vivo screening on five subjects as well as for the method based on in vitro SPF. Both methods have considerable high standard deviations. In vivo the variability was higher for the water resistance test as compared to the dry SPF test. This is due to the fact that two variable values (SPF wet and dry) are used to calculate the water resistance value. In a screening, on five volunteers test products often need a mean water resistance of more than 65% to meet the water resistance requirements with certainty. The confidence limits of the in vitro SPF water resistance procedure are comparably high. As a consequence in both methods candidates fail because of ambiguous results.

A second problem is the lack of good correlation between in vivo and in vitro SPF-based water resistance. It is improbable that the procedures of product application and spectral measurements were insufficient because we followed the well-elaborated Colipa UVA in vitro method in all steps. We used an approved spectroradiometer and participated successfully in a round robin study to optimize the product application on PMMA plates (8). We assume that the substrate (roughened PMMA plates) is only poorly matching the properties of the living skin, and therefore, the result has no higher predictive value.

Conclusion

The existing in vitro methods are of help to select the most promising candidates in case not all developed formulations can be tested in the in vivo method. However, all the reviewed in vitro methods as well as ours failed the attempt of a well predictive method. They all bear the risk that in some cases, truly water resistant products will be sorted out or a product with promising in vitro results fails in the in vivo study. Therefore, it can be recommended to include an in vivo screening after selecting candidates with in vitro methods. A reduced number of volunteers can be used to predict the water resistance of the sunscreens. In cases of a scatter as typically found in our trial five volunteers are sufficient for such an in vivo screening. When 65% of water resistance or more is found success can be estimated and further volunteers can be included in the study to complete it and fulfil the requirements of the Colipa method.

Acknowledgement

The study was funded by proDERM.