Exposure of the human body to sunlight and UV _A/UV _B-light and being tanned became fashionable, particularly during summer and holidays. However, with the rise in sunlight exposure an impressive increase in the number of skin cancer cases occurred (English et al., 1997, Whiteman and Green 1999). A direct link between the carcinogenic action and sunlight radiation has been made and the International Agency for Research on Cancer (IARC) has classified solar radiation, UV _A and UV _B, as human carcinogens (IARC, 1992).

National and international health authorities have urged the public to take protective measures, among these to use sunscreens. IARC has recently evaluated the cancer-preventive activity of sunscreens (IARC, 2000). They conclude that the use of sunscreens reduces the risk of sunburn in humans and probably prevents squamous-cell carcinoma of the skin. No conclusion could be drawn about cancer preventive activity against basal-cell carcinoma and cutaneous melanoma. However, use of sunscreens can extend the duration of intentional sun exposure and such an extension may increase the risk for cutaneous melanoma.

Sunscreens today contain one or several UV-B filters, often enriched with UV-A filters. UV-filters usually are synthetic organic chemicals but may also be inorganic in nature. Their safety for human use is regulated by national and international bodies. In the EU, before a new UV-filter is allowed on the market, a stringent toxicological safety evaluation is carried out and only in the case of a safe toxicological profile and a margin of safety of at least 100, the molecule can be approved by the SCCNFP for human use (Notes of Guidance for testing of cosmetic ingredients for their safety evaluation SCCNFP/0321/00 Final).

Comparable safety approval procedures exist in the USA and Japan.

2. Terms of Reference

The SCCNFP has been asked to evaluate the possible estrogenic effects of organic UV filters used in cosmetic products and to respond to the following questions :

* Could the SCCNFP provide a critical analysis of the article " In vitro and in vivo estrogenicity of UV screens" by Margret Schlumpf et al?

* More generally, does the SCCNFP consider that organic UV filters used in cosmetic sunscreen products have any estrogenic effects which have the potential to affect human health?

3. Expert review

Introduction

There is growing concern regarding possible harmful consequences of exposure to xenobiotic compounds that are capable of modulating or disrupting the endocrine system. This concern for endocrine disrupting chemicals is directed at both wildlife and humans. It should be noted that although there are associations between endocrine disrupting chemicals, so far investigated, and human health disturbance, a causative role of these chemicals in diseases and abnormalities related to an endocrine disturbance has not been established. The Scientific Committee on Toxicity, Ecotoxicity and the Environment (CSTEE) has recently discussed toxicological test guidelines and testing strategies and has concluded that reliance on in vitro assays for predicting in vivo endocrine disrupter effects may generate false-negative as well as false-positive results. Thus, the development of in vitro pre-screening test methods is not recommended. The current enhancement by OECD of the existing Repeated Dose 28-day Oral Toxicity Study in Rodents (406) and the Two-Generation Reproduction Toxicity (416) tests has high priority support [CSTEE, 1999]. Recently, an article by Schlumpf et al (2001) suggested that several UV screens show estrogenic activity. They used an in vitro test with the MCF-7 breast cancer cell line and an in vivo rat uterotrophic assay. The investigations of this Swiss group on the safety of UV-filters have alerted the popular press and the public. Controversial interpretations of the results initiated vigorous debates about the safety of sunscreens. More particularly estrogenic properties were attributed to a number of UV-filters including benzophenone-3, homosalate, 4-methyl-benzylidene camphor, octyl-methoxycinnamate and octyl-dimethyl-PABA, even suggestions towards carcinogenic effects were made.

On 27/4/2001, an ad hoc working group of the SCCNFP has discussed the issue and published a preliminary report on the scientific review of the investigations of Schlumpf et al. (2001). Serious concerns were expressed as to the validity of the results published.

As a follow-up, a final expert opinion of the SCCNFP is given below, consisting of :

(1) a scientific review of the investigations of Schlumpf et al. (2001),
(2) a scientific review of the investigations carried out by the industry concerned,
(3) a risk assessment and margin of safety according to the EU-procedure for UV-filters,
(4) data on human exposure to environmental and dietary estrogens,
(5) conclusions.

3.1. Scientific Review of the Results of Schlumpf et al. (2001)

3.1.1. Study protocol and results

The in vitro and in vivo estrogenicity of 5 UV _B-filters, benzophenone-3 (Bp-3), homosalate (HMS), 4-methyl-benzylidene camphor (4-MBC), octyl-methoxycinnamate (OMC) and octyl-dimethyl-PABA (OD-PABA) and 1 UVA-filter, butyl-methoxydibenzoylmethane (B-MDM) were studied.

3.1.1.1. In vitro study

A general screening assay (E-screen) with a human breast cancer cell line, MCF-7 cells, was carried out. A positive test was based upon the binding of the test compound with the estrogen receptor leading to cell proliferation.

As a positive control, 17 b -estradiol, was used and it was, as expected, positive in the assay. The 5 UV-B filters were found to be positive in the assay and caused cell proliferation. The UV-A filter gave a negative result. EC ₅₀ values for 17 b -estradiol, Bp-3, 4-MBC, OMC, OD-PABA and HMS were found to be 1.22 pM, 3.73 µM, 3.02 µM, 2.37 µM, 2.63 µM and 1.56 µM, respectively .

The results were supported by the expression of the estrogen-dependent pS ₂ protein and by an inhibition of effects with the anti-estrogen ICI 182,780.

3.1.1.2. In vivo tests

A uterotrophic assay was carried out using two different exposure routes, namely

- oral exposure of young Long-Evans rats to the 6 UV-filters from day 21 of life until day 24 of life, with ethinylestradiol serving as a positive control;

- dermal exposure of hairless Nu rats to 4-MBC from day 21 of life until day 26 of life by immersion of the animals in warm olive oil containing 2.5%, 5% and 7.5% of 4-MBC, respectively.

For the oral exposure, a dose-dependent increase of uterine weights was observed for 4-MBC and OMC, a slighter response was seen for Bp-3, but no maximal effect was seen as was the case for the positive control.

ED ₅₀ values were found to be 0.818 µg/kg/day, 1000-1500 mg/kg/day, 309 mg/kg/day and 934 mg/kg/day, for the positive control, Bp-3, 4-MBC and OMC, respectively. OD-PABA, HMS and B-MDM were found to be inactive.

For the dermal exposure assay, 4-MBC exhibited a dose-dependent increase in uterine weight, with a significant effect at a concentration of 5% and 7.5% in olive oil.

3.1.2. Comments by SCCNFP

3.1.2.1. In vitro study

- The potency of the positive control is in the order of picomoles; the in vitro potency of the UV-filters tested lays in the range of micromoles, which means a difference of 1 million units. The in vitro potency of the UV-filters is thus importantly lower than the one observed for 17 b -estradiol. Probably a lot of industrial chemicals would show some in vitro estrogenic effects when this type of comparisons is taken seriously.

- It should be emphasized here that in vitro assays can only demonstrate whether UV-filters bind on the estrogen receptor or not, but they do not provide evidence whether the compounds have estrogenic activity or not. In vitro assays are therefore screening tests useful in setting priorities for further in vivo testing. The CSTEE committee clearly stated in its report on endocrine disrupters (1999) that utilising in vitro data for predicting in vivo endocrine disrupter effects may generate false negative as well as false positive results and that major emphasis should therefore be put on in vivo assays. Claiming that 5 UV-filters have estrogenic properties based on an in vitro test is premature.

The in vitro ranking for the UV-filters going from Bp3, 4-MBC, OMC, OD-PABA to HMS, did not correspond with the in vivo results. Indeed, in the latter test 4-MBC was most active, followed by OMC and Bp-3. The most active UV-filter in vitro displayed only a weak activity in vivo. In addition OD-PABA and HMS were found to be inactive. Only precise toxicokinetic data can link the in vitro and in vivo data, a conclusion that was also reached by the authors.

3.1.2.2. In vivo studies

- The OECD draft protocol on the rodent uterotrophic assay, was issued on April 21, 2000. The protocol used by the Swiss group dates from before that time and therefore shows some important deviations. Moreover, GLP conditions have not been applied.

Deviations from the current OECD guideline proposal :

· the choice of the rat strains is unusual and not explained.

· the exposure period of the rats runs until the 26 ^th day of life, which is too close to the onset of puberty.

· the dermal exposure conditions are inappropriate : dipping pups into olive oil is not a standard procedure and the galenic form to deliver the UV-filter, namely a solution in warm olive oil, is not reflecting in use conditions : indeed, today sunscreens are formulated as poorly penetrating o/w-emulsions.

· the calculation of the absorbed dose via dermal exposure is unclear and oral intake by the animals cannot be excluded.

- The potency of the positive control, ethinylestradiol, is in the order of 1 µg/kg/day; the potency of the UV-filters tested lays in the range of 100 to 1000 mg/kg/day; which means a difference of 100.000 to 1 million units. The in vivo potency of the UV-filters is thus importantly lower than the one observed for the control hormone. Furthermore, 3 of the 6 UV-filters have no measurable potency at all.

- The uterotrophic assay can only serve a limited function as a test for in vivo identification of chemicals with estrogenic activity. The uterotropic assay is a short-term high-dose test.

3.2. Scientific Review of the Investigations carried out by Industry

3.2.1. Submission of 4-methyl-benzylidene camphor (Colipa code S60)

Two uterotrophic assays have been carried out in immature female Sprague Dawley rats [ Crl : CD (SD) BR rats ] either by the subcutaneous route (subcutaneous injections) either by the oral route (gavage). Animals were dosed on 3 consecutive days, day 19-20-21 of life; the positive control was 17-ethinylestradiol. The protocol was according to the OECD guidelines proposal and the tests were carried out under GLP conditions.

Six groups were tested :

- 0, 10, 100,1000 mg/kg/day S60 in corn oil
- 0.3 µg/kg/day of control in ethanol
- 1 µg/kg/day of control in ethanol

Subcutaneous study : A statistically significant lower mean body weight gain was observed in the 1000 mg/kg/day S60-treated group during the interval day 19-day 20, compared to the vehicle control group.

Oral study : A statistically significant lower mean body weight gain was observed in the 100 mg/kg/day and the 1000 mg/kg/day S60-treated group during the interval days 19-20 and day 20-21, respectively, together with a lower mean daily food consumption.

For both studies, the positive control induced a significant dose-related increase of the uterus weight (both as wet uterus and blotted uterus). S60 did not induce a significant increase of the uterus weight at 10, 100 and 1000 mg/kg/day.

3.2.2. Submission of octylmethoxycinnamate (Colipa code S28)

A uterotrophic assay of OMC was carried out using female immature Wistar rats [ CRL : WI(GLX/BRL/HAN) IGS BR ] by the oral route (gavage) for 3 consecutive days. The positive control was DES-SP (diethylstilbestrol dipropionate).

Four groups were tested :

- 0, 250, 1000 mg/kg/day S28 in olive oil,
- 5 µg/kg/day of positive control in olive oil.

The protocol deviated from the OECD guideline proposal, but was carried out under GLP conditions. A statistically significant lower mean body weight gain was observed in the 250 mg/kg/day S28-treated group during the interval day 2-day 3; the same was true for the 1000 mg/kg/day S28-treated group during the interval day 0-day 3.

The positive control induced a significant uterotrophic effect and showed histopathologic changes in the uterus.

S28 did not induce a uterotrophic effect and no histopathologic changes could be shown in the uteri concerned.

3.2.3. Submission of benzophenone-3 (Colipa code S38)

A uterotrophic assay of Bp-3 was carried out using female immature Wistar rats [ Chbb : THOM, SPF ] of 22 days old. The compound was given by the oral route (gavage) for 4 consecutive days. The positive control was DES-SP.

Four groups were tested :

- 0, 500, 1000 mg/kg/day S38 in sesame oil,
- 5 µg/kg/day of control in sesame oil.

Appropriate control groups were included. The protocol deviated from the OECD guideline proposal, but the test was performed under GLP-conditions. Dosing was carried out until day 26, which is too close to the onset of puberty. A statistically significant lower body weight gain was observed in the 1000 mg/kg/day S38-treated group during the interval day 0-day 1.

The positive control induced a significant increase in uterine weight (absolute and relative). S38 did not promote growth of the uterus and therefore does not exhibit estrogenic activity.

Industry made 4 uterotrophic studies available on UV-filters of which 2 were performed with 4-MBC (subcutaneous and oral administration), one with Bp-3 (oral route) and one with OMC (oral route). The results of the 4 studies show no evidence for any uterotrophic response and doses up to 1000 mg/kg/day were used for the 3 UV-filters concerned. The animals were female immature Sprague-Dawley or Wistar rats treated for 3 to 4 consecutive days.

Only the 4-MBC study is strictly carried out according to the OECD guideline proposal and no evidence for uterotrophic activity could be seen, although 4-MBC was found to be the most active UV-filter in the study of Schlumpf et al. (2001), showing a significant increase in uterine weight at a dose of 119 mg/kg/day. In the same study Bp-3 had a weak effect at a dose of 1.500 mg/kg/day but in the study conducted by industry no uterotrophic effect could be detected at dosing of 1000 mg/kg/day. The dosing at 1.500 mg/kg/day is higher than the top dose present in the OECD guideline proposal and should therefore be seen as a negative result.

Thus as far as Bp-3 is concerned, the results of Schlumpf et al and industry are in line. A negative finding for Bp-3 was also reported earlier by Baker et al (2000) during the poster session in the SOT meeting in Philadelphia. For Bp-3 it was already known that about 1% of the Bp-3 dose in rats is metabolised to p-hydroxy-benzophenone, which might exhibit an estrogenic effect (Hayden et al.1997, Felix et al.1998, Stocklinski et al.1980, Nakagawa et al 2001).

A clear discrepancy, however, exists between the negative results obtained by industry and the positive ones of Schlumpf et al. (2001), in particular for 4-MBC. Differences in strain of the animal may be an important factor. This was highlighted in the expert report of Bolt et. al [2001, in press], which discusses the significant differences in toxicokinetics of p-tert-octylphenol found in different rat strains (Certa et al. 1996, Upmeier et al. 2000).

Also dosing of the animals was different : oral administration of UV-filters was performed by gavage in the industrial studies and by mixing the chemicals in the food by the Swiss group.

3.3. Risk Assessment and Margin of Safety

3.3.1. Some general considerations

On a general basis, the MoS is calculated by dividing the lowest No Observable Adverse Effect Level (NOAEL) of a compound by its Systemic Exposure Dose (SED) during normal foreseeable use. If the MoS exceeds 100, the compound is regarded as safe for use.

The question can be raised whether an additional safety factor should be introduced for children with regard to the use of sunscreens.

Based on a number of previous publications [Schaefer and Riedelmayer 1996, Marzulli and Maibach 1984, Jiang et al. 1999 and Weltfriend et al. 1996], Nohynek and Schaefer [2001] recently concluded that there is no reason to assume that children should be more susceptible to potential adverse effects of topically applied sunscreens than adults. This conclusion was based upon the findings that

- there is no significant difference between the skin of children and adults regarding the penetration of topically applied substances;

- Skin permeability to externally applied substances remains relatively constant throughout life; this particularly being confirmed for sunscreens;

- The skin of children is not more susceptible to local irritant effects of topically applied substances than adult skin.

Thus the major difference between adults and children, relevant for risk assessment of a topically applied substance, is the larger body surface / body weight ratio of children, when compared with that of adults. As a consequence, the relative systemic exposure of children to a topically applied substance may be somewhat (about 1.4 times) higher than that of a typical adult.

Based on this relatively small difference between the systemic exposure of adults and children, no additional safety factor is introduced for children.

3.3.2. Risk assessment and margin of safety for 4-MBC

3.3.2.1. According to SCCNFP notes of guidance

The safety of 4-MBC has been reviewed by the SCCNFP (XXIV/1377/96 rev.1/98) in 1998. It was concluded that tests for skin irritation, sensitisation, phototoxicity, photosensitisation, photocontact allergy, mutagenicity and photomutagenicity were negative. Percutaneous absorption was estimated to be 1.9%. Teratogenicity tests were negative and via dermal application no stimulating effect on the thyroid function could be seen as was suggested by subchronic oral tests.

The Margin of Safety [ MoS = NOAEL / SED] was found to be 110, which is acceptable.

3.3.2.2 According to data of Schlumpf et al. (2001)

The NOEL (estrogenic activity) of 4-MBC in the uterotrophic assay in immature Long-Evans rats published by Schlumpf et al (2001) is 66 mg/kg/day.

SED = 0.23 mg/kg/day

" Screening MoS" = 66 mg/kg/day / 0.23 mg/kg/day = 289, which is higher than 100 and would consequently be acceptable.

Important to notice is the calculation of a " Screening MoS", since :

· the exposure time can hardly be called long-term or chronic, but is clearly short-term, which is of great significance for the use of the deduced "NOEL"-value.
A 2-generation study would be able to generate the real NOEL value necessary for the calculation of the MoS related to reproduction toxicity. However, this particular type of study should only be considered as a last resort because of the large number of animals required to perform it correctly.

· as long as the relevance for humans of positive results in a uterotrophic assay is not known, it remains questionable whether it is correct to use a safety margin for hormonal activity.

3.3.3. Risk assessment and margin of safety for OMC

3.3.3.1. According to SCCNFP notes of guidance

The safety of OMC has been reviewed by the SCC (SPC/1037/93, S28) in 1993. It was concluded that the compound has a low acute toxicity. OMC is not irritating or sensitising in animals, but can be very rarely responsible for allergic contact dermatitis in man.

Mutagenicity, photomutagenicity and photoclastogenicity tests were negative. The teratogenic activity has a NOAEL of more than 500 mg/kg bw/day, which was the highest dose tested.

The percutaneous absorption was estimated to be 2%, a figure derived from experiments in human and animal skin in vitro, plus the results of an in vivo human study via oral uptake.

The MoS was calculated to be 750, which is acceptable.

3.3.3.2. According to data of Schlumpf et al (2001)

The NOEL (estrogenic activity) of OMC in the uterotrophic assay in immature Long-Evans rats (Schlumpf et al. 2001) is 522 mg/kg/day.

SED = 0.6 mg/kg/day

" Screening MoS" = 522/0.6 = 870, which would be acceptable (>100).

3.3.4. Risk assessment and margin of safety of Bp-3

3.3.4.1. According to SCCNFP notes of guidance

Bp-3 was taken up in the Annex of the UV-filters before the activities of the SCC started. Therefore, Bp-3 has not undergone the standard safety procedure review by the SCCNFP.
It is advised to ask the industry for a complete toxicological dossier on Bp-3 so that a full risk assessment and a calculation of the margin of safety can be performed according to the SCCNFP standards.

According to the Final Report on the Safety Assessment of Benzophenones-1, -3, -4, -5, -9 and -11) (Cosmetic Ingredient Review 2000), the risk assessment can provisionally be performed as follows :

LD ₅₀ oral rat > 2000 mg/kg
LD ₅₀ dermal rabbit > 16.000 mg/kg
NOEL (27d oral, rat) = 1% in diet
NOEL (90d oral, rat) = 0.1% in diet [effects (90d oral, rat) noticed at 0.5% to 1% in diet]

Bp-3 is not irritating to skin and eyes of rat, not photosensitising, not phototoxic in guinea pigs and rabbits, not sensitising in guinea pigs and not mutagenic in the Ames-test. Bp-3 is not irritating or sensitising on human skin, although some cases of positive patch tests have been seen in humans.

From these data a NOEL (90d, oral rat) can be estimated :

food uptake of adult rat is » 10g/100g bw
no effect when Bp-3 0.1% in diet » 1g/kg/day (diet)
» 100 mg/kg/day in rat
thus the estimated NOEL is 100 mg/kg/day.

The percutaneous absorption is not known; therefore the real SED cannot be calculated. However, assuming a percutaneous absorption in the order of 1% (Nohynek and Schaefer, 2001), a SED = 0.3 mg/kg can be calculated.

MoS = 333, acceptable (>100)

Note that this is an estimation.

3.3.4.2. According to Schlumpf et al (2001)

NOEL (estrogenic activity) of Bp-3 in the uterotrophic assay in immature Long-Evans rats published by Schlumpf et al (2001) is 937 mg/kg/day. Since the SED is not known, the MoS cannot be calculated. Assuming the SED = 0.3 mg/kg/day,

" Screening MoS" = NOEL/SED = 937/0.3 = 3123 >100, would be acceptable.

3.4. Data on Human Exposure to Environmental and Dietary Estrogens

In table 1 recently published data are shown with respect to the potency of xenoestrogens in the rodent uterotrophic assay.

Table 1 : Potency of xenoestrogens in the rodent uterotropic assay.

^a in alfalfa and leguminosae
^b soybeans
^c soybean-containing bread sold in UK health food stores (Ashby and Tinwell, 1998).
^d from Casanova et al. 1999.

From these data it appears that UV-filters have an extremely small relative potency in comparison with ethinylestradiol (1 to 1 million) but also a low relative potency in comparison with dietary estrogens.

When the mass balance of human exposure to environmental and dietary estrogenic compounds was estimated, data as represented in table 2 were found (Safe 1995).

It is known that several food formulas contain soy products. Soy contains phytoestrogens including genistein and daidzein. The total phytoestrogen content of infant food formulas for instance, represents 135 ± 5 µg/g total genistein and daidzein (Irvine et al. 1998) and may go up as high as 600µg/g (Zimmerli et al. 1997).

From these data it seems that the estrogenic potency of UV-filters is several orders of magnitude lower as compared with that of natural dietary estrogens.

During the last years, it became evident that a variety of different mechanisms of endocrine disruption exist for different compounds [CSTEE, 1999]. For pragmatic reasons this will not be further discussed here.

Table 2 : Estimated mass balance of human exposure to environmental and dietary estrogens (Safe 1995).

3.5. Conclusions

(i) A number of important technical shortcomings in the study of Schlumpf et al. were detected. This was mentioned in the first part of this report as well as in a preliminary report of an ad hoc working Party.

(ii) Industry has performed a further uterotrophic assay in rats for 4-MBC according to the OECD guideline proposal and under GLP conditions; no evidence was found for an increase in uterus weight (part 2 of this report).

(iii) Industry has provided data on uterotrophic assays in rats for Bp-3 and OMC (not according to OECD guidelines, but under GLP conditions), in which no positive uterotrophic effect could be detected for both UV-filters (part 2 of this report).

(iv) The margins of safety for 4-MBC and OMC, calculated according to the SCCNFP, using NOAELs obtained from subchronic animal studies, are higher than 100.
The estimated margin of safety for Bp-3 is higher than 100 (part 3 of this report).

(v) The calculated " Screening MoS", in which the experimental non-estrogenic-effect-level data, obtained by Schlumpf et al. (2001), are used in the official MoS calculations of the EU, are found to be higher than 100 for 4-MBC, OMC and Bp-3 (part 3 of this report).

However, the data presented by Schlumpf et al (2001) are unsuitable for long-term risk assessment. Use of the data of Schlumpf et al will only give a rough approximation of the possible risk. A two-generation reproduction toxicity test might possibly generate more accurate data.

(vi) The activity of the UV-filters found in the study of Schlumpf et al. (2001) is very low in comparison with exposure to "estrogenic" substances in food (flavonoids) and hormonal therapy (birth control pill, morning after pill, post-menopausal therapy) (part 4 of this report).

With the information available above, the SCCNFP concludes that there is no need for regulatory actions to protect the consumer with regard to potential estrogenic effects of the UV-filters studied :

- Although positive data were observed for the UV-filters HMS and OD-PABA in the in vitro assay (Schlumpf et al., 2001), this in vitro test with MCF-7 cells remains only a screening. The uterotrophic assay ( in vivo test) was negative in the Schlumpf study.

- The UV-A filter B-MDM was negative in both the in vitro and in vivo assay.

- Bp-3 showed to be positive in the in vitro assay by Schlumpf et al. (2001) and weakly positive in the in vivo test. The SCCNFP recommends that industry is asked to submit a complete toxicological dossier on Bp-3 in order to perform a full risk assessment and to calculate a final MoS for Bp-3 according to the EU standards for UV-filters.

As UV-filters are an effective tool to protect humans from excessive exposure to sunlight, a known carcinogen, their use is recommended by the SCCNFP.

3.6. Opinion (answer to the questions)

Answer to the question ' Could the SCCNFP provide a critical analysis of the article "In vitro and in vivo estrogenicity of UV screens" by Margret Schlumpf et al?' :

The article of " In vitro and in vivo estrogenicity of the UV screens" by M. Schlumpf et al. has been critically analysed and the comments of the SCCNFP can be summarised as follows :

In vitro study :

- The potency of the positive control is in the order of picomoles; the in vitro potency of the UV-filters tested lays in the range of micromoles, which means a difference of 1 million units. The in vitro potency of the UV-filters is thus importantly lower than the one observed for 17 b -estradiol. Probably a lot of industrial chemicals would show some in vitro estrogenic effects when this type of comparisons is taken seriously.

- It should be emphasised here that in vitro assays can only demonstrate whether UV-filters bind on the estrogen receptor or not, do not provide evidence whether the compounds have estrogenic activity or not. In vitro assays are therefore screening tests useful in setting priorities for further in vivo testing. The CSTEE committee clearly stated in its report on endocrine disrupters (1999) that utilising in vitro data for predicting in vivo endocrine disrupter effects may generate false negative as well as false positive results and that major emphasis should therefore be put on in vivo assays. Claiming that 5 UV-filters have estrogenic properties based on an in vitro test is premature.

The in vitro ranking for the UV-filters going from Bp3, 4-MBC, OMC, OD-PABA to HMS, did not correspond with the in vivo results. Indeed, in the latter test 4-MBC was most active, followed by OMC and Bp-3. The most active UV-filter in vitro displayed only a weak activity in vivo. In addition OD-PABA and HMS were found to be inactive. Only precise toxicokinetic data can link the in vitro and in vivo data, a conclusion that was also reached by the authors.

I n vivo study :

The OECD draft protocol on the rodent uterotrophic assay, was issued on April 21, 2000. The protocol used by the Swiss group dates from before that time and therefore shows some important deviations. Moreover, GLP conditions have not been applied.

Deviations from the current OECD guideline proposal :

* the choice of the rat strains is unusual and not explained

* the exposure period of the rats runs until the 26 ^th day of life, which is too close to the onset of puberty

* the dermal exposure conditions are inappropriate: dipping pups into olive oil is not a standard procedure and the galenic form to deliver the UV-filter, namely a solution in warm olive oil, is not reflecting in use conditions: indeed, today sunscreens are formulated as poorly penetrating o/w-emulsions.

* the calculation of the absorbed dose via dermal exposure is unclear and oral intake by the animals cannot be excluded.

* The potency of the positive control, ethinylestradiol, is in the order of 1 m g/kg/day; the potency of the UV-filters tested lays in the range of 100 to 1000 mg/kg/day; which means a difference of 100.000 to 1 million units. The in vivo potency of the UV-filters is thus importantly lower than the one observed for the control hormone. Furthermore, 3 of the 6 UV-filters have no measurable potency at all.

* The uterotrophic assay can only serve a limited function, as a test for in vivo identification of chemicals with estrogenic activity. The uterotrophic assay is a short- term high-dose test.

The SCCNFP came to the conclusion that a number of important technical and scientific shortcomings are present in the study of M. Schlumpf et al.

Answer to the question ' More generally, does the SCCNFP consider that organic UV filters used in cosmetic sunscreen products have any estrogenic effects which have the potential to affect human health?' :

Based on the actual scientific knowledge, the SCCNFP is of the opinion that the organic UV-filters used in cosmetic sunscreen products, allowed in the EU market today, have no estrogenic effects that could potentially affect human health.

3.7. References

- Ashby, J. and Tinwell, H. Oestrogenic activity of burgen bread to female rats. - Hum Exp Toxicol 1998; 17: 598-599.
- Baker, V.A., Jones, P.A. and Lea, L.J. Benzophenone sunscreen agents assessment of oestrogenic activity. Poster at the 39 ^th SOT meeting Philadelphia, US. March 19-23, 2000.
- [Abstract: Baker, V.A., Jones, P.A. and Lea, L.J. Toxicology 2000; 148: 74]
- Baker, V.A., Jones, P.A. and Lea, L.J. Assessment of the oestrogenic activity of benzophenone sunscreens agents [Abstract]. Toxicologist 2000; 54: 262.
- Bolt, H.M., Guhe, C. and Degen G.H. Comments on : "In vitro and in vivo estrogenicity of UV-screens."
Environ Health Perspect 2001, in press.
- Casanova, M., You, L., Gaido, K.W., Archibeque-Engle, S., Janszen, D.B. and Heck, H.A. Development effects of dietary phytoestrogens in Sprague-Dawley rats and interactions of genistein and daidzein with rat estrogen receptors alpha and beta in vitro. Toxicol Sci 1999; 51: 236-244.
- Certa, H., Fedtke, N., Wiegand, H.J., Müller, A.M.F. and Bolt, H.M. Toxicokinetics of p-tert-octylphenol in male Wistar rats. Arch Toxicol 1996; 71: 112-122.
- CSTEE opinion on human and wildlife health effects of endocrine disrupting chemicals, with emphasis on wildlife and on ecotoxicology test methods. Report of the Scientific Committee on Toxicity, Ecotoxicity and the Environment (CSTEE), 8 ^th CSTEE Plenary Meeting, Brussels, 4 March 1999.
- English, D.R., Amstrong, B.K., Kricker, A. and Fleming, C. Sunlight and cancer. Cancer Causes Control 1997; 8: 271-283.
- Felix, T., Hall, B.J. and Brodbelt, J.S. Determination of benzophenone-3 and metabolites in water and human urine by solid-phase microextraction and quadrupole ion trap GC-MS. Anal Chim Acta 1998; 371: 195-203.
- Hayden, C.G.J., Roberts, M.S., Benson, H.A.E. Systemic absorption of sunscreen after topical application. Lancet 1997; 350: 863-864.
- IARC Monograph on solar and ultraviolet radiation. International Agency for Research on Cancer, World Health Organisation, Lyon, France. 1992, vol.55.
- IARC Handbooks of Cancer Prevention. Sunscreens. Lyon, France. 2001, vol 5.
- Irvine, C.H., Fitzpatrick, M.G. and Alexander, S.L. Phytoestrogens in soy-based infant foods: concentrations, daily intake and possible biological effects. Proc Soc Exp Biol Med 1998; 217: 247-253.
- Jiang, R., Roberts, M.S., Collins, D.M. and Benson H.A. Absorption of sunscreens across human skin: An evaluation of commercial products for children and adults. Br J Clin Pharmacol 1999; 48(4), 635-637.
- Marzulli, F.N. and Maibach, H.I. Permeability and reactivity of skin related to age. J Soc Cosmet Chem 1984; 35, 95-102.
- Nakagawa, Y., Suzuki, T., Tayama, S. Metabolism and toxicity of benzophenone in isolated rat hepatocytes and estrogenic activity of its metabolites in MCF-7 cells. Toxicology 2001 (in press).
- Nohynek, G.J. and Schaefer, H. Benefit and risk of organic ultraviolet filters. Regul Toxicol Pharmacol 2001; 33, 1-15.
- Safe, S.H. Environmental and dietary estrogens and human health: is there a problem? - Environ Health Perspect 1995; 103: 346-351.
- SCCNFP Scientific Committee on Cosmetic Products and Non-food Products intended for Consumers Preliminary observations of an ad-hoc Working Party of the SCCNFP. In vitro and in vivo estrogenicity of UV screens. 27 April 2001.
- Schaefer, H. and Riedelmayer, T.E. Skin barrier: Principles of percutaneous absorption. 1996; pp 118-128, Karger, Basel, Switzerland.
- Schlumpf, M., Cotton, B., Conscience, M., Haller, V. Steinmann, B. and Lichtensteiger, W. In vitro and in vivo estrogenicity of UV screens. Environ Health Perspect 2001; 109: 3: 239-244.
- Stocklinski, A.W., Ware, O.B. and Oberst, T.H. Benzophenone metabolism. I. Isolation of p-hydroxyphenone from rat urine. Life Sci 1980; 26: 365-369.
- Upmeier, A., Degen, G.H., Schumacher, U.S., Certa, H. and Bolt, H.M. Toxicokinetics of p-tert-octylphenol in female DA/Han rats after single i.v. and oral administration. Arch Toxicol 1999; 73: 217-222.
- Whiteman, D.C. and Green, A.C. Melanoma and sun exposure: Where are we now? Intern. J. Dermatol 1999; 38: 481-489.
- Weltfriend, S., Bason, M., Lammitausta, K. and Maibach, H.I. Irritant Dermatitis (irritation). In Dermatotoxicology (F.N. Marzulli and H.I. Maibach, Eds.), pp 97-100. Taylor and Francis, London.
- Zimmerli, B. and Schlatter, J. Vorkommen und Bedeutung der Isoflavone Daizein und Genistein in der Säuglingsnahrung. Mitt Gebiete Lebensm 1997; 88: 219-232.