The basic principle of the in vitro transmittance method is the calculation of the monochromatic protection factor (mPF), at every given wavelength λ. The intensity of the UV radiation is reduced due to the presence of at least one UV filter in a sunscreen. This mPF(λ) is given by the inverse of the UV transmittance through the UV-absorbing film, 1/T(λ). As the spectral range relevant for the formation of erythema is between 290 and 400 nm, the mPF is averaged over this range. To obtain the SPF, this average must be weighted with the intensity of the light source, Ss(λ) and the erythema action spectrum, ser(λ), leading to the following Eq. (1) contributed by Sayre et al. [7]:

$$ } = \frac^ }}} (\lambda ) \cdot S_}} (\lambda )} }}^ }}} (\lambda ) \cdot S_}} (\lambda ) \cdot T(\lambda )} }}. $$

(1)

Data for Ss(λ) and ser(λ) are available [17, 18], but the transmittance T(λ) has to be determined for the respective sunscreen either by in vitro measurement or by simulation [19]. The latter is beyond the scope of this review. Ss(λ) is given in units of W/m2/nm, Ser(λ) without units. The product of Ss(λ) and Ser(λ) is called erythemal-effectiveness spectrum or erythemal-effective irradiance, Eer(λ), and again has units of W/m2/nm.

Sayre et al. used hairless mouse epidermis as substrate to compare the in vitro and in vivo SPF-results of seven commercial sunscreens. The authors concluded that their method was superior to any previously presented in vitro technique for testing sunscreen effectiveness [7]. It is important to note that the applied amount of sunscreen on both human skin and the mouse epidermis was 2 μl/cm2 (volume!/area). Only one of the seven sunscreens met the “ultra-protective” range of SPFs as categorized by the Food and Drug Administration (Table 1).

Table 1 Comparison of SPF obtained by human testing vs that obtained on hairless mouse epidermis (only the highest SPF value presented) [7]

In these first experiments on mouse skin, the in vivo and in vitro results did not differ significantly, but the in vitro results showed a large scatter that was even larger than the in vivo results. Out of curiosity, the in silico SPF value is given. If the UV-filter composition is known, this value can be calculated using one of the available tools [9, 10, 19]. In silico values are quite accurate, as can be seen in this case.

For about 25 years after Sayre’s initial conceptual work, there did not seem to be much effort to develop the in vitro method into a standard that could be widely adopted. A major problem appeared to be finding a suitable substrate that could mimic skin but was not as complex as using mouse skin.

In 2003, the trade organization Cosmetics Europe became active in this field after its predecessor COLIPA had already been involved in the in vivo method [20]. In a ring study with 10 commercial sunscreens, SPF values between 6 and 79 were compared with the in vitro values of six European test centers. Two of the ten products contained microtitanium dioxide (TiO2) in combination with ethylhexyl methoxycinnamate and benzophenone-3. The amount applied to poly(methyl methacrylate) sheets (PMMA) was 1.2 and 1.4 mg/cm2 (weight!/area). The conclusion was that the correlation was quite good at an application rate of 1.2 mg/cm2 and that a further interlaboratory test was required [21].

In 2004, the German Society for Cosmetic Chemistry (DGK) contributed with a comparative study in vitro/in vivo of 58 commercially available products with a declared SPF range of 4–60 [22]. They also used PMMA plates and varied the application amounts depending on the SPF value: 0.5, 1.0, 2.0 mg/cm2, which were adjusted especially for high SPF so that the absorption did not exceed a value of two (absorption < 2). In this study, mainly sunscreens with organic UV filters were tested; some sunscreens contained inorganic UV filters, TiO2 and/or ZnO, but these were not mentioned in the study. This study showed the complexity of the European sunscreen market and helped to find a standard substrate (PMMA plates) and a standard application rate (1.2 mg/cm2) for the future in vitro SPF method.

Figure 2 schematically illustrates the difference between sunscreens on the skin and sunscreens on PMMA sheets. On the skin, the SPF is determined where the thickness of the sunscreen film is thinnest and thus the UV transmission is highest and the local SPF is lowest, while the large amounts of sunscreen trapped in the V-shaped creases transmit very little UV radiation. The PMMA plates, on the other hand, do not have such deep valleys and are defined by their roughness. As can be seen schematically in Fig. 2, much less sunscreen is needed to achieve the same protective effect.

Fig. 2

Skin and plastic-plates with sunscreen (schematic). To cover the skin, a larger volume of sunscreen is required than on the PMMA plate. However, a higher density of a sunscreen leads to a lower volume, as the standards are weight- based (ISO 24444, ISO 24443)

By 2006, PMMA sheets were established as a substrate and the importance of the roughness of the chosen PMMA substrates, molded or sandblasted, was emphasized [11]. It was concluded that the roughness should be higher rather than lower for accuracy reasons, i.e., 6 μm rather than just 2 μm. This has been reflected in the UVA in vitro standards ISO 24443 since 2012 and also in the drafts of the new alternative SPF methods [15, 16].

In 2006, the European Commission recommendation introduced a labeling of the SPF and UVA protection and suggested that an in vitro method should be preferred [12]. Prior to this European recommendation, in vitro methods for measuring and labeling UVA protection already existed in Australia and the United Kingdom. At that time, however, there was no officially standardized SPF in vitro method.

In 2006, a new working group (WG 7, Sunscreen Test Methods) was set up as part of the technical committee ISO/TC 217 Cosmetics [13], which was founded in 1998, but the first standards were not expected for years of development work. Experts from sunscreen manufacturers, UV-filter suppliers and testing laboratories have since been actively contributing to the development.

In 2007 some research groups started to publish in vitro SPF results, e.g., the Coiffard/Couteau group at the University of Nantes, F, who developed their own in vitro method based on PMMA plates and a spectrophotometer with an integrating sphere [23,24,25,26]. They specifically tested mineral sunscreens and concluded that the SPF in vitro of ZnO-containing sunscreens is always lower than the in vivo value (label) [24]:

Titanium dioxide was thus seen to be much more effective than zinc oxide; indeed, no commercial form of zinc oxide tested can give a SPF higher than 10 at its maximum dose of use (25%).

In 2010, although the development work was intensified, the German cosmetic chemists concluded in their paper: In vitro sun protection factor: still a challenge with no final answer [27]. Based on the in vitro data measured in this multicenter study, the 20 authors acknowledged that no definitive in vitro SPF method can be presented, but pointed out which technical side effects may influence such a method, e.g., the quality of the spectrophotometer used, the amount of product applied, the pretreatment of the samples, the time and temperature of equilibration, the size of the measured area, the application process or the calculation based on standardized data. Finally, a reduction in standard deviations within individual laboratories for in vitro SPF tests was achieved, but no improvement in comparison between laboratories was achieved. The development of a valid and reliable SPF in vitro test therefore remained a challenge and further work was required to develop a satisfactory method.

In 2012, the UVA method ISO 24443 was published [28]. Although it is not explicitly an SPF in vitro method, it contains a standardized in vitro SPF, the so-called SPF0 value, and thus represents the first standardized in vitro SPF method, although it only serves as intermediate result to determine the UVA-PF. The substrate, the PMMA plates and the entire procedure for measuring this SPF0 are described in detail in ISO 24443 [28].

In 2015, the original ISO in vitro SPF project (“ISO 24445”), which had started in 2006, expired due to a lack of consensus among the national member organizations, which in turn confirmed the conclusion of the German scientists in 2010 that there is no final answer yet [27].

In 2016, Batzer et al. acknowledged that in vitro SPF measurement could not be established mainly due to poor reproducibility. However, with a new approach, the introduction of the “Dispersal Rate” (DR), they wanted to provide new suggestions for improved reproducibility and open a new space for discussion [14]. A total of 22 O/W emulsions from the sponsor’s product range (Beiersdorf AG) with proven in vivo SPF were measured in vitro following the international standard ISO 24443 on two different substrates (PMMA plates: WW5, sandblasted and HD6, molded). The ratio ‘in vitro SPFraw/in vivo SPF’ was calculated for each product. The composition of the products was analyzed for a parameter that correlates with this ratio. In addition, seven suitable calibration products were determined to transfer the in vitro SPFraw to the calibrated in vitro SPFcal.

In their general study, practically without ZnO sunscreens (composition not declared), Batzer et al. showed that the results for the SPFcal matched very well with SPFin vivo for 19-measured o/w emulsions on WW5 (sandblasted) plates. However, they found two products where the in vitro SPF was much lower than the in vivo SPF. The DR allowed them to identify a product characteristic parameter to predict too low measured in vitro SPFs on WW5 plates. The DR parameter mainly refers to the ratio of water to lipids in an emulsion. They found that products with few emollients and few emulsifiers are measured too low in vitro. There was much more variability on the molded-HD6 plates, they could not find similar results, although the mean value of the 22 products tested was closer to the in vivo value. The reason for the difference between the plates lies most likely in the surface structure, i.e., the better wettability of the sandblasted WW5 plates compared to the flat/smooth surface of the molded-HD6 plates. What we can learn from Batzer et al. for in vitro SPF testing is, that:

the type of PMMA plate matters;

the type of product matters;

calibration of product types improves reproducibility.

Batzer et al. presented a detailed realistic view of the state of in vitro SPF testing, as did Blum et al. for in vivo testing in 1946, 70 years earlier:

We demonstrate that, besides the known parameter, also the composition of the products should be considered for the interpretation of the in vitro SPF. Our findings could explain some multiple reported problems in correlation between in vitro and in vivo SPF, especially for higher SPFs.

Furthermore, in 2016, Miksa et al. presented a new approach to improve reproducibility, which is based on the automatic spreading of the sunscreen product to the PMMA plate by a robot and using two different types PMMA plates, molded (HD6) and sandblasted (SB6), to consider differences in the products [29, 30].

For application a positive-displacement pipette was used, 1.3 mg/cm2 of product to each HD6 plate (molded) and 1.2 mg/cm2 of the same product to each SB6 plate (sandblasted). In the development of the method, it was found that the use of an HD-SpreadMaster robot was critical in achieving a reproducible, standardized, homogeneous film distribution across sample plates [31]. In the meantime, this approach is under way to an international ISO standard [15].

In 2018 this method by Miksa et al. was adopted and promoted by Cosmetics Europe. A blinded ring-testing of this SPF in vitro method was published by Pissavini et al. [32]. Twenty four sunscreen products from Cosmetics Europe members were used (23 O/W and 1 W/O emulsion, 12× SPF < 29, 12× SPF > 30, 24× organic UV filters 8–27%, 10× inorganic UV filter 1–4%), resulting in an SPFin vitro vs SPFin vivo correlation as follows: y = 1.01x − 1.48, R2 = 0.87.

Two years later (2020), the same group of authors published a follow up study with an extra 76 sunscreen samples (total 100), among them 47 declared as “Mineral only” UV filters, mostly W/O emulsions, except for 3× O/W [33]. Unfortunately, the publication does not disclose the UV-filter composition of the sunscreens. “Mineral only” can mean TiO2 alone or ZnO alone or a mixture of TiO2/ZnO. For the 27 “Mineral only” with an SPFin vivo between 32 and 91, we can assume that they contain TiO2 to reach these high SPF values, but for the 20 “mineral only” sunscreens with SPFin vivo values between 9 and 29, we do not know. It would be interesting to learn how many of them are in the category “ZnO alone”, at higher concentration, e.g., 10–25%.

In 2018 Osterwalder et al. saw a need to publish an overview of the alternative methods, to give all available relevant methods a fair chance [34]. This overview later led to the industry consortium ALT-SPF [35]:

Over the last 2 to 3 decades alternative methods such as for example, in vitro transmission tests in silico calculation, or non-invasive in vivo testing (using negligible UV dose on humans) have emerged and been developed to a stage that makes them promising alternatives.

The desire to understand the precision behavior of the alternative methods demand sophisticated statistical tools to analyze and, in particular, to characterize them for a general public use. The accuracy, repeatability and reproducibility should be known to decide on the release of a method measuring protective abilities of sunscreens that are linked to skin-cancer prevention.

Comparing alternative SPF data against the gold standard requires a solid set of in vivo SPF data and thus represents a significant experimental effort that should not be repeated for each alternative method. This is how, in 2018, the main idea for the ALT-SPF project was born.

The results of the interlaboratory tests of the ALT-SPF consortium will be published separately for each alternative method in the table. For a deeper insight into the history and future of sun protection assessment methods, we recommend [36] (Table 2).

Table 2 Overview of alternative SPF methods in ALT-SPF consortium [35]