Introduction

In July 2010, the European Commission indicateda desire to consult the public on the five chaptersof a Draft Report on Alternative (Non-animal)Methods for Cosmetics Testing: Current Status andFuture Prospects — 2010, in order “to ensure thateach chapter correctly reflects the current state ofthe art and the prospects”. This draft report wasprepared by working groups of experts nominatedby the various stakeholders and chaired by theCommission’s Joint Research Centre (JRC), inorder to gain a broad and objective picture of thescientific/technical issues that relate to establish-ing alternative test methods for the five human

health(-related) effects falling under the 2013deadline for the marketing ban of the EU Cosmet-ics Directive. It was also intended that the draftreport would contain, where possible, a science-based estimate of the time necessary to achieve fullreplacement of animal testing for the respectiveendpoints.

Directive 76/768/EEC (the Cosmetics Directive)is a regulatory framework for the placing on themarket of cosmetic products, and subsequentamendments and associated directives have aimedto phase out the use of animal testing for thesepurposes. Besides the complete testing ban, a mar-keting ban has applied since 11 March 2009, for allhuman health(-related) effects, with the exception

ATLA 38, 345–353, 2010 345

Comment

A FRAME Response to the European CommissionConsultation on the Draft Report on Alternative (Non-animal) Methods for Cosmetics Testing: Current Statusand Future Prospects — 2010a

Michael Balls and Richard Clothier

Fund for the Replacement of Animals in Medical Experiments (FRAME), Nottingham, UK

Summary — This response on behalf of FRAME to the European Commission’s consultation on the fivechapters of the Draft Report on Alternative (Non-animal) Methods for Cosmetics Testing: Current Statusand Future Prospects — 2010, is via a Comment in ATLA, rather than via the template supplied by the Com-mission. This is principally so that a number of general points about cosmetic ingredient testing can bemade. It is concluded that the five draft chapters do not provide a credible basis for the Commission’sforthcoming report to the European Parliament and the European Council on the five cosmetic ingredientsafety issues for which the 7th Amendment to the Cosmetic Directive’s ban on animal testing was post-poned until 2013. This is mainly because there is insufficient focus in the draft chapters on the specificnature of cosmetic ingredients, their uses, their local effects and metabolism at their sites of application,and, in particular, on whether their possible absorption into the body would be likely to lead to their accu-mulation in target sites at levels approaching Thresholds of Toxicological Concern. Meanwhile, there con-tinues to be uncertainty about how the provisions of the Cosmetics Directive should be applied, given therequirements of the REACH system and directives concerned with the safety of other chemicals and prod-ucts.

Key words: alternative methods, animal tests, carcinogenicity, cosmetic ingredients, Cosmetics Directive,in silico, in vitro, REACH system, repeated dose testing, reproductive toxicity, skin sensitisation,toxicokinetics.

Address for correspondence: Michael Balls, FRAME, Russell & Burch House, 96–98 North SherwoodStreet, Nottingham NG1 4EE, UK.E-mail: michael.balls@btopenworld.com

aThe authors gratefully acknowledge their very useful discussions with Professor Horst Spielmann (Department of Biology, Chemistry and Pharmacy, Free University of Berlin, Berlin, Germany) during the preparation of this Comment.

of five issues, namely, repeated dose toxicity(including skin sensitisation and carcinogenicity),reproductive toxicity and toxicokinetics, for whichthe deadline was extended to 11 March 2013.

The consultation is related to a requirement thatthe Commission must inform the European Parlia-ment and the European Council “in case alterna-tive methods will not have been developed andvalidated by 2013 for the remaining endpoints thatare exempted from the Cosmetic Directive’s mar-keting ban until 2013”.

The introductory documents state that “The pub-lic is invited to comment on the five individualchapters of the draft report, each representing onehuman health(-related) effect. The focus of the con-sultation is to ensure that each chapter correctlyreflects the current state of the art and theprospects. As regards prospects, the experts evalu-ated how long it will take to develop and optimisesuch approaches up to a level that fulfils theECVAM criteria for entering prevalidation. Thetime required for the validation and regulatoryacceptance of these alternative approaches was notconsidered for the purpose of this exercise. In addi-tion, justified comments on the conclusion (i.e. thetimelines given to achieve full replacement) aresought.”

It is clearly indicated that “The Commission isinviting factual comments, which complement theinformation provided. Comments should include ref-erences and other substantiation where possible.”“Any comments and information on this public con-sultation should be submitted by using the providedcomment template form by mail, fax or e-mail, by 15October 2010 at the latest, to:European Commission, Health and ConsumersDirectorate-General (DG SANCO), Unit SANCO B2,Cosmetics and Medical Devices, B-1049 Brussels,Belgium (Fax: 00 32 (0) 2 296 64 67; E-mail: SANCO-COSMETICS-REPORT@ec.europa.eu)”.

The template requires the submitters of com-ments to indicate their names, affiliations andtypes of organisation, the submission date, andwhether the submission should/should not betreated as confidential. A separate template formshould be submitted for each of the five chapters,and the forms have four columns, for “commentnumber”, “line number”, “comment and rationale”,and “proposed revised text”.

The introductory documents, the five draft chap-ters, and the template are available at: http://ec.europa.eu/consumers/sectors/cosmetics/documents/public_consultation/index_en.htm

FRAME’s Submission

We have decided not to respond to the consultationby using the template system, but to convey ourviews via a Comment in ATLA, principally because

there are a number of important general points tobe made, which could not readily be made via the“line number/comment and rationale/proposedrevised text” template system. In addition, adetailed set of informed comments wouldinevitably involve many pages for each draft chap-ter, and it is not clear how the Commission or theauthors of the chapters would handle large num-bers of suggestions for “proposed revised texts”. Wealso doubt whether more than a handful of mem-bers of “the public” would have the competence orthe information necessary for providing assurancethat a chapter “correctly reflects the current stateof the art and the prospects”. People more cynicalthan ourselves would undoubtedly be tempted toconsider that the consultation has deliberatelybeen set up in this way, in order to discourage par-ticipation.

In our analysis of the draft chapters, we willfocus on what the Commission will need, if it is toprovide the European Parliament and the Euro-pean Council with information of the necessaryquality, which is specifically relevant to the safetyof cosmetic products, their ingredients, their pur-poses and the consequences of repeated exposureto them, in relation to the Cosmetics Directive,rather than on issues related to chemicals in gen-eral and to other products, which are regulated bythe REACH system and other directives.

In particular, it should be remembered that acosmetic product is defined as “any substance orpreparation intended to be placed in contact withthe various external parts of the human body (epi-dermis, hair system, nails, lips and external geni-tal organs) or with the teeth and the mucousmembranes of the oral cavity, with a view exclu-sively or mainly to cleaning them, perfuming them,changing their appearance and/or correcting bodyodours and/or protecting them or keeping them ingood condition” (1). By contrast, a medicinal prod-uct is defined as “a) any substance or combinationof substances presented as having properties fortreating or preventing disease in human beings; orb) any substance or combination of substanceswhich may be used in or administered to humanbeings, either with a view to restoring, correctingor modifying physiological functions by exerting apharmacological, immunological or metabolicaction, or to making a medical diagnosis” (1).

For the purposes of this Comment, we will goalong with the current practice of the Commissionand the European cosmetics industry, by avoidingthe term ‘cosmeceutical’, which refers to cosmeticproducts containing biologically active ingredientswhich are claimed to have medical or drug-likebenefits. Nevertheless, this issue will have to befaced up to at some time in the future, since, if theclaims increasingly being made about them aregenuine, these products would have to be treatedas medicinal products rather than cosmetics, in the

346 Comment

interests of avoiding damage to human health.A crucial question concerning cosmetic ingredi-

ents, is the extent to which they are absorbedacross the skin or other barriers which limit theentry of xenobiotics into the body (e.g. barriers inthe lungs, buccal cavity and vagina, which, in thiscase, are more important than the gastric or intes-tinal barriers). Metabolism is often referred to(especially in pointing out the limitations of invitro test systems), but it is metabolism in the skinand the other relevant barriers which may be ofparticular concern, since the importance of metab-olism by visceral organs, such as the liver and kid-neys, will depend on the site of exposure, the routeof absorption and transport to them. Surely, it isreasonable to assume that cosmetic products aredesigned so that significant amounts of their ingre-dients are not taken into the body, so as to exceedThresholds of Toxicological Concern (TTCs) andadversely affect the visceral organs or systems.

Another important criterion in evaluating thedraft chapters is the way in which they deal withthe limitations of animal models and of the currentOrganisation for Economic Co-operation andDevelopment (OECD) Health Effects Test Guide-lines (TGs), the relevance of which tends to be com-promised by species differences and theapplication of very high doses. How the data theyprovide can be interpreted rationally in terms ofpotential hazard, exposure and risk, and actualexperience, in humans, not forgetting the high lev-els of qualitative and quantitative variation inresponses to chemicals in the human population, isa vital component of the background against whichnon-animal methods and testing strategies forevaluating the safety of cosmetic ingredientsshould be developed, validated and applied.

Draft Chapter 1: Repeated Dose Toxicity

This 42-page draft chapter, by Stuart Creton, PilarPrieto, Alan Boobis, Wolfgang Dekant, Jos Klein-jans, Hannu Komulainen, Paolo Mazzatorta, AnnaPrice, Vera Rogiers, Greet Schoeters and MathieuVinken, focuses on the main available in vitro mod-els in relation to six of the most common targets fortoxicity (liver, kidney, central nervous system,lung, cardiovascular system, and haematopoieticsystem). It is concluded that complete replacementwill not be possible by 2013.

Various OECD TGs for repeated dose toxicityare discussed, and it is noted on line 136 that “the90-day oral toxicity in rats is the most commonlyconducted repeated dose toxicity study” for cos-metic ingredients, which is mostly used to derivethe No Observable Adverse Effect Level (NOAEL)for cosmetic ingredients. The NOAEL is used inthe calculation of the MoS (Margin of Safety)

and/or the Margin of Exposure (MoE) for cosmeticingredients (line 91).

There is no mention of OECD TGs 427 (2) and428 (3), on skin absorption in vivo and in vitro,respectively, and the assumption seems to be thatthe relevant dose is that which comes within thecompass of the animal tests, rather than theamount of a compound which is likely to reachthese six most common targets as a result ofrepeated exposure to the cosmetic product. Surely,it would be more useful to think in terms of a Mar-gin of Uptake (MoU), i.e. the difference betweenthe likely uptake of a cosmetic ingredient and thelevel in the body which would be a cause for con-cern (i.e. the TTC).

The skin seems to be regarded merely as a routeof entry into the body in general, as there is nomention of dermal toxicity per se — none of the sixtables on alternative methods for repeated dosetoxicity deal with the skin. It is also interesting tonote (line 598) that the liver is the main focus ofthe current COLIPA co-funded projects on non-ani-mal methods for assessing repeated dose toxicity.Nevertheless, much is being achieved in other tox-icological studies with reconstituted human skinequivalents, which might well prove to be suitablefor use in repeated dose studies.

Little attention is paid to the kinds of chemicalswhich are used as cosmetic ingredients, and which,on the whole, unlike pharmaceuticals, are notintended to be readily taken up into the blood-stream for distribution to internal organs. Indeed,the authors of this draft chapter would appear tobe more comfortable in dealing with issues relatedto the use and testing of pharmaceuticals. This isparticularly true of the section on the concept ofMode of Action (MoA) in the induction of adversehealth effects, which might be better applied in itsdevelopment stage to compounds likely to behighly toxic, rather than to cosmetic ingredients,which would overwhelmingly be expected to be tox-icologically inert at the levels to which users of cos-metics are exposed.

There is little information on how the cosmeticindustry deals with the issues raised by repeatedexposure to their products, and it is regrettable(line 677) that the Working Group received littleresponse to their enquiry via COLIPA as towhether companies would share their own experi-ences with illustrative examples. The paragraphoutlining the Unilever approach is encouraging(especially when considered with the 2009 reviewby Carmichael et al. [4], which is not listed in thisdraft report), but the short Nestlé section,although it does, at least, mention how exposure istaken into account, is more relevant to food prod-ucts than to cosmetics. Other companies, such asBeiersdorf, L’Oréal, and Procter and Gamble, haveinvested huge sums in the development of non-ani-mal methods — it would have been good to see

Comment 347

some input from them as well. In particular, itwould be interesting to ask the cosmetic industryfor information known to them concerning cos-metic ingredients which are taken into the body atlevels which could result in damage to the six tar-gets which are the focus of this draft chapter.

It should be noted, for example, that, if damageto the nervous system were to be caused, suchdamage would be more likely to affect the periph-eral nervous system than the central nervous sys-tem. There are in vitro methods for looking intothis kind of problem (e.g. 5, 6).

The draft chapter mentions some of the limita-tions of animal models, but rather than using thestructure of the placenta as an example of ananatomical difference between animal and humans(line 763), it would have been more relevant to cos-metic ingredient toxicity/safety to draw attentionto differences between animal and humans in thestructure of the skin, eyes and lungs.

Some regret is expressed that larger numbers ofanimals, sufficient to detect rare events, cannot beused, but it would, in any case, be very difficult toassess the relevance of such rare cases to thehuman situation. The same could be said of humanstudies, but there is undoubtedly a case for makinggreater use of human experience, including ethi-cally-conducted human trials, and of post-market-ing surveillance (other than for marketingpurposes).

We will make rather few comments on the sec-tions on non-animal methods, since their use couldonly be considered meaningfully against a suffi-cient understanding of cosmetics and their ingredi-ents.

There are a few references to in silico systems inthe draft chapter, and the focus is on the six com-mon targets and the Maximum RecommendedTherapeutic Dose (MRTD), which will tell theCommission very little of relevance to cosmeticingredients. This is a lost opportunity, since in sil-ico methods could be expected to have great valuein the future, as ways of predicting whether cos-metic ingredients would be likely to be metabolisedand/or have adverse effects at their sites of appli-cation, and/or be absorbed into the body in signifi-cant amounts relevant to TTCs.

It is concluded in the Executive Summary (line59) that none of the available in vitro methods aresuitable for quantitative risk assessment forrepeated dose toxicity. Since much of the focus ofthe draft chapter is on approaches or systems inthe course of development, that is an oversimplifi-cation. In fact, methods have been developed whichpermit repeated assays to be performed on thesame cell cultures, and this opens the possibility ofstudies on the extent of recovery from adverseeffects (7, 8). Squamous differentiation inbronchial epithelial cell cultures can be studied byusing a non-cytotoxic endpoint involving fluores-

cent cadaverine (9, 10). The possibility of repeatedtreatment of the same cell cultures has also beenexploited in studies on the repeated effects of radi-ation (11). These methods permit studies on recov-ery from damage, and on the effects of cosmeticingredients on recovery.

Overall, then, whatever its value in relation torepeated dosage and visceral target organ toxicity,this draft chapter will provide the Commissionwith little information of value concerning theapplication of the ban on the animal testing of cos-metic ingredients.

Draft Chapter 2: Skin Sensitisation

The conclusion in this 24-page draft chapter, byDavid Basketter, Silvia Casati, Klaus EjnerAndersen, Alexandre Angers-Loustau, AynurAptula, Ian Kimber, Reinhard Kreiling, Henk vanLoveren, Gavin Maxwell and Hanna Tähti, is thatno complete replacement of the in vivo methodswill be available by 2013, but that it should be pos-sible by 2017–19 “to make skin sensitisation riskassessment decisions by using a toolbox of non-ani-mal test methods for all cosmetic ingredients andexposure scenarios”.

Understandably, perhaps, since its area of cov-erage is more-tightly defined, this is a muchmore useful draft chapter. It demonstrates thatvery good effort is being focused on increasingour understanding of the induction and expres-sion of skin allergies, and indicates how this newknowledge can be applied to predicting the skinsensitiser potency of cosmetic ingredients. Thereis a good balance between the consideration of insilico tools and in vitro assays in the assessmentof key steps in the initiation of an allergicresponse.

The in silico approaches could be particularlyvaluable in predicting various effects of the struc-tures of chemicals, including haptenation (thecovalent binding of a chemical to a skin protein),upon which sensitisation potential depends, andmetabolism in the skin, as well as skin sensitis-ing potential itself. Knowing that a chemicalwould not bind to protein and that it could not bemetabolised by enzymes in the skin, would be ofgreat value.

Much is made of the value of the local lymphnode assay (LLNA) in predicting the results of thehuman repeated insult patch test (HRIPT), whichis somewhat surprising, since the evidence fromother toxicological studies is that species variationbetween animal models and humans limits thevalue of animal data for predicting human effects.That could be because the good correlations havemainly involved the most highly-potent sensitis-ers, where, perhaps, the power of the chemical inthe reaction overwhelms the variability in the

348 Comment

responding system which may be involved in thewide range of sensitivities to allergens in thehuman population. It is extremely unlikely thatsuch highly-potent sensitisers, which would affectthe human population in general, would be used ascosmetic ingredients.

The use of in vitro systems has great promise, asis reflected in this draft report. One recent study isworth mentioning here, which involved a compari-son of the human cell-line activation test (h-CLAT)and the LLNA, in which 100 chemicals were testedindependently by two laboratories (12). The testset included chemicals classified as extreme,strong, moderate and weak sensitisers, and non-sensitisers, in the LLNA. The correlation betweenthe results of the two tests was 84%, and reasonsfor the difference were discussed.

The draft chapter would benefit from an indica-tion of how the various approaches in use, or in thecourse of development, are, or could be, used in astepwise, decision-tree strategy for making deci-sions about the sensitising potential/safety of cos-metic ingredients. Also, the overall conclusionabout the time needed to achieve the replacementof animal tests may be unduly pessimistic. It is tobe hoped that this is not because there is satisfac-tion that the reduction and refinement aspects ofthe LLNA are sufficient to justify its continued usein the long term.

Draft Chapter 3: Carcinogenicity

This 38-page draft chapter, by Jan van Benthem,Susan Felter, Tuula Heinonen, Albrecht Poth,Rositsa Serafimova, Joost van Delft, Emilio Benfe-nati, Pascal Phrakonkham, Andrew Worth andRaffaella Corvi, is based on the belief that “the two-year bioassay in rodents is widely regarded as thegold-standard to evaluate cancer hazard andpotency”, and regrets that “this test is rarely doneon cosmetic ingredients because of expense, time,and animal welfare issues” (lines 43 and 91). Short-term in vivo and in vitro tests for genotoxicity areseen as inadequate substitutes, along with a 90-dayrepeated dose test for non-genotoxic carcinogens(line 94). In the Executive Summary (line 67), it isconcluded that “Taking into consideration the pres-ent state of the art of the non-animal methods, theexperts were unable to suggest a timeline for fullreplacement of animal tests currently needed tofully evaluate carcinogenic risks of chemicals.Although a timeline for full replacement cannot bedeveloped, clearly the timeline is expected toextend past 2013.” It is further concluded that “ani-mal testing bans under the 7th Amendment to theCosmetic Directive will have a profound effect onthe ability to evaluate and conduct a risk assess-ment for potential carcinogenicity of new cosmeticingredients” (lines 49, 97 and 1103).

It is amazing that such opinions could berecorded in 2010, in what should be an authorita-tive document, and this draft chapter could notpossibly be a reliable basis for the advice to begiven by services of the Commission to the Parlia-ment and the Council.

It is now widely felt that animal carcinogenic-ity data are not adequate to support the classifi-cation of a chemical as “probable humancarcinogen” or “probable non-carcinogen” (13,14). Indeed, the data from the bioassay in onerodent species (e.g. the rat) cannot be used to pre-dict what would be expected to happen in anotherrodent species (e.g. the mouse), so the lack ofapplicability of rodent data to humans is not sur-prising. This is because, in addition to significantphysiological differences between humans andlaboratory animals, the lifetime doses applied inthe rodent bioassay are very high, and are indeedmuch higher than could possibly be experiencedin humans, particularly in the case of cosmeticingredients.

There is a danger that some toxicologists wouldsettle for in silico and in vitro approaches whichcould provide the identifications and classificationscurrently provided by the rodent bioassay. Indeed,the in vitro cell transformation assays, some ofwhich have been in the course of development andevaluation for decades, are based on rodent cellcultures.

What is needed is a complete rethink, with theaim of developing methods for identifying potentialhuman carcinogens. Hopefully, these methodscould employ human material and/or human expe-rience, and could be based on understanding of keyevents in the carcinogenic process.

Meanwhile, particular attention should befocused on the likelihood that some cosmetic ingre-dients could be carcinogenic in humans, but thiswould require the paying of rigorous attention tothe nature of the chemicals themselves, theireffects at their sites of exposure, their uptake intothe body, and the levels at which they might betransported to internal body tissues. The TTC con-cept is very important here, and it is well sum-marised in the draft chapter (line 522). Insufficientattention is paid to the skin as a target organ forgenotoxins and carcinogens, though the use ofhuman reconstructed epidermis in vitro models forgenotoxicity testing in, for example, assays such asthe micronucleus assay and the Comet assay, isdiscussed (lines 701 and 731).

The need to focus on cosmetic ingredients them-selves should be clear. Most of them will beinnocuous in terms of all kinds of toxicity, butsome of them will be potent chemicals which willneed close attention. This latter group includesthe colourants, including hair dyes, UV filters(sunscreens) and preservatives, which are specifi-cally referred to in the Cosmetics Directive. It was

Comment 349

the positive results obtained with hair dye com-ponents in bacterial tests for mutagenicity (e.g.the Ames Test), which led to the emphasis ongenotoxicity testing in the application of theDirective in relation to cosmetic products ingredi-ents, and to the inclusion of so many genotoxicol-ogists among the experts who were directlyinvolved or consulted.

Little attention is paid to cosmetic ingredientsper se in the draft chapter. There is a brief discus-sion on hair dyes, but only to make the point thatthe in vitro systems used in a study failed to pre-dict the results of the rodent bioassay, so the ‘false-positive’ results they provided might have led tothe unnecessary abandonment of the use of 17 hairdyes. However, given the poor record of predictionsbased on the rodent bioassay, the in vitro resultsmight have been ‘true-positive’ for humans.

As in the case of draft Chapter 1, this draft chap-ter focuses too much attention on the tests them-selves, rather than on the chemicals that need tobe considered, how the body is exposed to them,and where and in what circumstances they mightreach a TTC. As a result, the statement that, if “anin vivo test is no longer possible, the safety of manypotential new cosmetic ingredients will not be ableto be substantiated” (line 1103) has no merit andshould not influence the application of the 7thAmendment to the Cosmetics Directive. If in vivotesting “is rarely done on cosmetic ingredientsbecause of expense, time, and animal welfareissues” (lines 44 and 92), what problems havearisen as a consequence of this, which now makethis a matter of such concern?

Draft Chapter 4: Toxicokinetics

The 73-page draft chapter, by Olavi Pelkonen, San-dra Coecke, Sofia Batista Leite, Ulrike Bernauer,Jos Bessems, Esther Brandon, Frederic Bois,Ursula Gundert-Remy, George Loizou, EmanuelaTestai and José-Manuel Zaldívar, is very interest-ing and encouraging, but it is so heavily based onexperience with pharmaceuticals that its relevanceto the Cosmetics Directive and 2013 is question-able. For example, Table 1 (five pages) on replace-ment methods for absorption and bioavailabilityrepeatedly states “developed for pharmaceuticals”,“not used for cosmetics” and “suitability for cosmet-ics should be evaluated”. Then Table 2 and therespective text sections (lines 685 to 857) onreplacement methods for distribution are mainlyfocused on partition between blood and tissues, theblood–brain barrier, the blood–placental barrierand the blood–testis barrier — important issues forpharmaceuticals, but not of primary concern forcosmetic ingredients unless they enter the body insignificant amounts which approach the TTC. Fig-ure 2 provides a valuable insight into future human

risk assessment with no animal assays, but again,it is likely to be more interesting to the pharma-ceutical industry than to the cosmetics industry.The same could be said about the sections onplasma protein binding (lines 634 and 658).

The sections on models for bioaccessibility (line505), absorption (line 534) and bioavailability(line 575) are more relevant, and OECD TG 428(3), on in vitro methods for determining dermalabsorption across animal and human skin, is dis-cussed, but the limitations of these methods areemphasised, as they may not provide suitableinput for more-conventional physiologically-based pharmacokinetic (PBPK) models (which,one might say, are more closely-related to oraluptake).

Table 3 and the respective text sections (lines860 to 1099) on metabolism are mainly concernedwith the metabolism of pharmaceuticals by thecytochrome p450 system.

The situation is best summed up by words fromthe draft chapter itself (line 1227): “Since only incases where a cosmetic ingredient is bioavailablefollowing dermal, oral, or inhalation exposure,[will] further tests on systemic and not just localtoxicity be necessary, additional efforts are neededto provide reliable alternative methods to assessthe bioavailability after oral and inhalation expo-sure. Several efforts have been undertaken toimprove the reliability [of] alternative test meth-ods available [for] assessing absorption via [the]gut and lungs, but still more work would berequired.” Surely, however, the key potential routeof entry for most cosmetic ingredients is via theskin, and this should have been given more atten-tion in this draft chapter.

This criticism is partly countered by the final,short section on “strategic considerations of riskassessment of cosmetic ingredients” (line 1421),and the decision-trees for absorption-based testing(Figure 6) and for (internal) exposure-based testing(Figure 7), which contrast well, at least in terms ofrelevance, with the schematic representation of aphysiologically-based toxicokinetic (PBTK) modelfor a woman (Figure 3). However, that the authorsof the draft chapter are more comfortable withproducts other than cosmetics re-emerges in theRecommendations section (line 1491), where theneed for better models of absorption via the lungand excretion via the kidney is emphasised, andthe efficiency of the conventional validation proce-dure is questioned, and an “expert consensus pro-cedure” is recommended to replace it. However, ifthese five draft chapters provided by experts forthe Commission are anything to go by, that wouldnot be a wise way forward. The validation processneed not, and should not, be conventional, butshould be scientifically rigorous, adaptable, trans-parent, and free of bias in support of vested inter-ests.

350 Comment

Draft Chapter 5: Reproductive Toxicity

In this 45-page draft chapter, by Sarah Adler,Thomas Broschard, Susanne Bremer, Mark Cronin,George Daston, Elise Grignard, Aldert Piersma,Guillermo Repetto and Michael Schwarz, relativelylittle attention is paid to specific issues related tocosmetic ingredients or products. What is provided isa general review which is perhaps of greater rele-vance to other chemicals and other products. This ismore or less admitted in the Executive Summary(line 59). However, even as a general review, thisdraft chapter is inadequate. We will return to thatpoint, after making some comments on some otheraspects of the draft chapter.

As is pointed out (line 127), for substances sub-mitted for inclusion in the positive lists of the Cos-metics Directive, a comprehensive dossier must beprovided for evaluation by the Scientific Commit-tee on Consumer Safety. Moreover, When consider-able oral intake is expected, or when dermalpenetration data suggest a significant systemicabsorption, information on toxicokinetics, carcino-genicity and reproductive toxicity “may become nec-essary” (line 132). However, there appears to belittle guidance on what studies may need to be con-ducted, on how oral uptake and dermal absorptionshould be estimated for cosmetic ingredients, or onwhat “considerable” or “significant” levels wouldmake testing “necessary”. Nevertheless, it ispointed out that a rat developmental toxicity studyaccording to OECD TG 414 (15) is usually consid-ered sufficient, although data from a one-genera-tion (TG 415) or a two-generation (TG 416) studymay be included in the manufacturer’s dossier(line 136).

An extensive inventory of current animal tests isprovided (lines 150 to 414), including the notoriousUterotrophic Bioassay (line 325) and the Hersh-berger Bioassay (line 356). It should be noted thata limit test can be performed with TG 414 (line173), and that an extended one-generation studycould replace the two-generation study in the nearfuture (line 392).

The section on alternative methods is compre-hensive, but uncritical. It includes the amphibianFETAX assay (line 457), which did not impress ina validation study, and zebra fish embryo (line438) and chick embryo (line 471) assays, which areunlikely to be necessary, if suitable mammalianstem cell-based tests (line 508), and especially,tests employing human material (lines 524 and543), become available. It is surprising that testsbased on human embryonic stem cells are notregarded more positively by the authors of thisdraft chapter, and it should be noted that the useof such cells in combination with metabolomicsshows great promise for predicting human devel-opmental toxicity (16).

Many other in vitro tests and biochemical tests,involving, for example, the placenta, male fertilityand endocrine disruption, are described, but theirrelevance to the safety of cosmetic ingredients isgiven little attention.

After a brief review of their current status, it isconcluded that “in silico approaches to predictreproductive toxicity are seldom robust and oftenhave poor predictive capacity” (line 924). A briefsection on QSARs and ADME related to reproduc-tive toxicity (line 870) is blatantly focused on phar-maceuticals and oral dosing, and the need for insilico systems for predicting the uptake of cosmeticingredients is not given the attention it surelydeserves.

The in vitro systems are criticised for represent-ing “only a very simplified picture of reality” (line973), but the limitations of the current strategiesbased on animal testing, are overlooked.

That final point is a crucial one, as the draft chap-ter does not recognise the warnings given by ThomasHartung (17), Hartung and Costanza Rovida (18)and others (e.g. 19, 20), concerning the cost of animaltesting and its inadequacy, which threaten disasterfor the REACH system, and, unavoidably, for anyother toxicity or safety evaluation which relies onthe current in vivo toxicity test procedures.

This is so important that we think it is worthincluding a long quotation from Hartung’s TenthFRAME Annual Lecture (17): “The thalidomidescandal had a tremendous impact on toxicology.Between 1956 and 1962, approximately 10,000children in Africa and Europe were born withsevere malformations, because their mothers hadtaken thalidomide during pregnancy. As a result,ever since then, there has been a strong emphasison testing to identify the teratogenic potentials ofsynthetic substances. Hence, more than 70% of thecosts and more than 80% of the animals requiredfor compliance with the EU REACH testing pro-gramme for new and existing chemicals will prob-ably be devoted to reproductive toxicity testing.Although experts estimate that only 2% of birthdefects are caused by chemicals (21), the belief thatwe must identify these, whatever it costs, is diffi-cult to overcome. Nevertheless, as we have shown(22), all this effort, cost and animal use might actu-ally result in the false-positive assignment ofreproductive toxic potential to about 60% of all thesubstances tested! This is because reproductivetoxic potential has a low prevalence among thethousands of chemicals to which the REACH sys-tem will apply, coupled with the low predictivevalue of the animal tests currently required by theregulators. Thus, the true positive reproductivetoxicants might not be identified, because of alarge number of false-positive results, and manyvaluable chemicals might be unnecessarily usedonly under strictly controlled conditions or notapproved for use at all!”

Comment 351

What is needed is a thorough and independentreview of the purpose of reproductive toxicitytesting and how it should be conducted, with a har-monised approach for chemicals and pharmaceuti-cals. The inescapable fact is that the predictivevalue of animal test data for humans is poor (23).The urgent need for new concepts has been put verypowerfully in a recent authoritative comment onThe way forward in reproductive/development toxi-city testing (24), which is not listed among the 105references attached to the draft chapter.

Until such a review has been conducted, and aset of constructive proposals have been presentedfor critical evaluation, we support the proposal byHartung and Rovida (18) for “a moratorium onreproductive-toxicity testing, or at least limitingtesting to the most suspicious substances, until theOECD guidelines [for an extended one-generationstudy] are completed and alternative strategies forscreening lots of chemicals are available”.

Meanwhile, since no convincing case has beenpresented for the reproductive toxicity testing ofcosmetic ingredients as a special case for applica-tion in addition to the testing conducted in compli-ance with the REACH system, such testing shouldbe banned forthwith, and should certainly not beconducted beyond 11 March 2013.

Conclusions

Without substantial revision, these five draft chap-ters cannot provide a credible basis for the Com-mission’s report to the European Parliament andthe European Council on the five cosmetic ingredi-ents testing issues for which the ban on animaltesting was postponed until 2013. Nor could theybe used to provide reliable estimates on the timeneeded to develop and validate replacement alter-native test methods.

This is principally because the draft reports ofthe five groups of experts are of dubious relevanceto any actions related to the Cosmetics Directive,because they are not sufficiently focused on issuesrelated to the nature of cosmetic ingredients, theiruses, their local effects and metabolism at theirsites of application, and their possible uptake andabsorption into the circulation, resulting in theirdistribution to visceral organs and tissues. With-out meaningful predictions of the likelihood ofuptake and, where uptake is likely, of the amountsthat would be distributed to various internal sites,discussions on repeated dose testing, toxicokinet-ics, and, in particular, carcinogenicity and repro-ductive toxicity, can have little or no value as abasis for deciding what tests should be conductedand what tests could safely be banned.

A crucial question is whether the uptake of a cos-metic ingredient could lead to a level in the bodywhich approached the TTCs for various possible

toxic effects. If, as would probably be the case inthe vast majority of instances, such TTCs would beunlikely ever to be approached, then testing for thetoxicity concerned, whether by in vivo, in silico orin vitro methods, could not be justified as neces-sary. The MoU concept, mentioned above, deservesserious consideration. Indeed, a different kind ofrisk assessment would be advisable, namely, anAbsorption Risk Assessment, i.e. the risk that aningredient and/or its metabolites might beabsorbed and accumulate at levels which wouldapproach TTCs and would therefore justify a Toxi-city Risk Assessment.

This whole exercise would have been much moreuseful, if the Commission had asked another work-ing group of experts to consider the uptake of cos-metic ingredients, and to produce a report beforethe other working groups began their delibera-tions.

It is especially disappointing that so little atten-tion is paid in the draft chapters to intelligent test-ing strategies and/or integrated testing strategies.Many schemes have been proposed, which arerelated to the toxicity of chemicals and to pharma-ceuticals, as evaluated in compliance with theREACH system legislation or directives other thanthe Cosmetics Directive. The leading cosmetic com-panies undoubtedly have such schemes for in-house use, which must take account of dermalpenetration or other routes of entry into the body.It would have been useful to have this reflectedmore fully in the draft chapters. It is particularlyimportant that such schemes show how in silicoand in vitro methods, along with the use of otherinformation not derived from animal experiments,could be used in complementary and mutually sup-portive ways, to show how human welfare couldbest be protected, without resort to compromisinganimal welfare.

It is not clear, from the introductory pages fromthe Commission, how the experts in the five work-ing groups were nominated, and how they wereselected. Apparently there was no open call forexperts to serve on the working groups. Thereremains a lack of transparency, since the expertsseem to have been “hand-selected”, and the result-ing evaluations are not unbiased. Of the 41experts, 6 were from Commission’s JRC, 7 fromindustry, 11 from research centres, 16 from acade-mia, with one private consultant. Their geographiclocations were as follows: Belgium (2), Denmark(1), Finland (4), France (1), Germany (8), Italy (8),Spain (1), Switzerland (1), The Netherlands (6),the UK (7), and the USA (2).

Nor is there any indication of what informationwas given by the Commission to the workinggroups, to guide them on how their discussions oncosmetic ingredients should be conducted, bearingin mind the requirements of other legislation, and,in particular, that related to the REACH system.

352 Comment

Decisions about the risk and safety of cosmeticproducts must surely depend on what is knownabout the chemicals they contain, most of whichare likely to be used in other kinds of products. TheCosmetics Directive cannot be meaningfullyapplied in a knowledge vacuum.

References

1. Anon. (2004). Guidance Document on the Demarca-tion between the Cosmetic Products Directive76/768 and the Medical Products Directive 2001/83as agreed between the Commission Services and theCompetent Authorities of the Member States. 11pp.Brussels, Belgium: European Commission.

2. OECD (2004). OECD Guidelines for the Testing ofChemicals. Section 4: Health Effects, No. 427. SkinAbsorption: In Vivo Method, 8pp. Paris, France:OECD. Available at: http://titania.sourceoecd.org/vl=7565018/cl=29/nw=1/rpsv/cw/vhosts/oecdjournals/1607310x/v1n4/contp1-1.htm (Accessed 30.09.10).

3. OECD (2004). OECD Guidelines for the Testing ofChemicals. Section 4: Health Effects, No. 428. SkinAbsorption: In Vitro Method, 8pp. Paris, France:OECD. Available at: http://titania.sourceoecd.org/vl=7565018/cl=29/nw=1/rpsv/cw/vhosts/oecdjournals/1607310x/v1n4/contp1-1.htm (Accessed 30.09.10).

4. Carmichael, P., Davies, M., Dent, M., Fentem, J.,Fletcher, S., Gilmour, N., MacKay, C., Maxwell, G.,Merolla, L., Pease, C., Reynolds, F. & Westmore-land, C. (2009). Non-animal approaches for con-sumer safety risk assessments: Unilever’s scientificresearch programme. ATLA 37, 595–610.

5. Suuronen, E.J., Nakamura, M., Watsky, M.A., Stys,P.K., Muller, L.J., Munger, R., Shinozaki, N. &Griffith, M. (2004). Innervated human cornealequivalents as in vitro models for nerve-target cellinteractions. FASEB Journal 18, 170–172.

6. Moore, P., Ogilvie, J., Horridge, E., Mellor, I.R. &Clothier, R.H. (2005). The development of an inner-vated epithelial barrier model using a humancorneal cell line and ND7/23 sensory neurons. Euro-pean Journal of Cell Biology 84, 581–592.

7. Clothier, R. & Sansom, R. (1996). Effects of surfac-tant retreatment in vitro: Changes in cell junctionsand in cell viability. ATLA 24, 859–865.

8. Clothier, R.H., Beed, M., Sansom, R. & Ward, R.(1997). An in vitro approach to the evaluation ofrepeat exposure in the prediction of toxicity. Toxi-cology in Vitro 11, 679–682.

9. Gray, A.C., Malton, J. & Clothier, R.H. (2004). Thedevelopment of a standardised protocol to measuresquamous differentiation in stratified epithelia, byusing the fluorescent cadaverine incorporationtechnique. ATLA 32, 91–100.

10. Gray, A.C., McLeod, J.C. & Clothier, R.H. (2007). Areview of in vitro modelling approaches to the iden-tification and modulation of squamous metaplasiain the human tracheobronchial epithelium. ATLA35, 493–504.

11. Clothier, R.H. & Bourne, N. (2003). Effects of THzexposure on human primary keratinocyte differen-tiation and viability. Journal of Biological Physics29, 179–185.

12. Ashikaga, T., Sakaguchi, H., Sono, S., Kosaka, N.,Ishikawa, M., Nukada, Y., Miyazawa, M., Ito, Y.,Nishiyama, N. & Itagaki, H. (2010). A comparativeevaluation of in vitro skin sensitisation tests: Thehuman cell-line activation test (h-CLAT) versus thelocal lymph node assay (LLNA). ATLA 38, 275–284.

13. Knight, A., Bailey, J. & Balcombe, J. (2006). Animalcarcinogenicity studies: 1. Poor human predictivity.ATLA 34, 19–27.

14. Knight, A., Bailey, J. & Balcombe, J. (2006). Animalcarcinogenicity studies: 2. Obstacles to extrapola-tion of data to humans. ATLA 34, 29–38.

15. OECD (2004). OECD Guidelines for the Testing ofChemicals. Section 4: Health Effects, No. 414. Prenatal Development Toxicity Study, 11pp. Paris,France: OECD. Available at: http://titania.sourceoecd. org/vl=7565018/cl=29/nw=1/rpsv/cw/vhosts/oecdjournals/1607310x/v1n4/contp1-1.htm(Accessed 30.09.10).

16. West, P.R., Weir, A.M., Smith, A.M., Donley, E.L.R.& Cezar, G.G. (2010). Predicting human develop-mental toxicity of pharmaceuticals using humanembryonic stem cells and metabolomics. Toxicology& Applied Pharmacology 247, 18–27.

17. Hartung, T. (2008). Toward a new toxicology — evo-lution or revolution. ATLA 36, 635–639.

18. Hartung, T. & Rovida, C. (2009). Chemical regula-tors have overreached. Nature, London 460,1080–1081.

19. Spielmann, H. & Vogel, R. (2006). REACH testingrequirements must not be driven by reproductivetoxicity testing in animals. ATLA 34, 365–366.

20. Combes, R. (2007). Reproductive toxicity testingunder the REACH System: Time for a paradigmshift. ATLA 35, 1–4.

21. Schaefer, C., Spielmann, H. & Vetter, K. (2006).Arzneiverordnung in Schwangerschaft und Stillzeit,781pp. Jena, Germany: Urban & Fischer.

22. Bremer, S., Pellizzer, C., Hoffmann, S., Seidle, T. &Hartung, T. (2007). The development of new con-cepts for reproductive toxicity applicable to largescale programmes. Current Pharmaceutical Design13, 3047–3058.

23. Bailey, J., Knight, A. & Balcombe, J. (2005). Thefuture of teratology is in vitro. Biogenic Amines 19,97–145.

24. Spielmann, H. (2009). The way forward in repro-ductive/development toxicity testing. ATLA 37,641–656.

Comment 353

After this Comment was completed, the Editoriala for this issue of ATLA was received from Horst Spielmann, AssociateEditor, Europe. It makes significant comments about the consultation and, in particular, about draft Chapter 5, onReproductive Toxicity.a Spielmann, H. (2010). The EU Commission’s Draft Report on Alternative (Non-animal) Methods for Cosmetics Testing.

Current Status and Future Prospects — 2010: A Missed Opportunity. ATLA 38, 339–343.