Several pieces of EU Regulations and Guidance Documents (GDs) relevant for the safety assessment of industrial chemicals and cosmetic products are in place (as summarized in Table 1), which describe the information needed to assess potential environmental and human health-related adverse effects of industrial chemicals and cosmetic products.
Interface between REACH and the Cosmetic Products Regulation
A joint ECHA-Commission statement (ECHA 2014a) clarified the interface between REACH and the Cosmetic Products Regulation. According to that statement, the animal testing ban in the Cosmetic Products Regulation concerns the tests needed to prove safety of the cosmetic products on the ‘end users’ (e.g., consumers). The marketing ban of cosmetic products that have been tested on animals is triggered, if the results of a study on vertebrate animals, required pursuant to the information requirements set out in the REACH Regulation, are relied on in the cosmetic product safety report under the Cosmetic Products Regulation to demonstrate the safety for the end user of products containing the registered substance exclusively used in cosmetic products.Footnote 3
However, the risks arising from other sources of exposure than the end use of cosmetic products are not assessed under the Cosmetic Products Regulation. In particular, REACH requires the evaluation of the risks to workers and the environment if the substance is covered by this Regulation. Regarding the relationship between REACH and the Cosmetic Products Regulation, decisions of the board of appeal of ECHA have been taken and are currently challenged by NGOs and Cosmetics industry, as reported in case numbers A-009-2018Footnote 4 and A-010-2018,Footnote 5 which consider examples of substances exclusively used as an ingredient in cosmetic products.
Even if a substance is registered exclusively for cosmetic use, REACH animal testing requirements continue to apply, as a last resort, to assess the risk from exposure to workers and for all environmental endpoints. REACH does not contain an automatic exemption from the information requirements for registration if an ingredient is used as a substance in cosmetic products only. A registrant can benefit from an exemption only if he/she shows that the conditions for an adaptation (e.g., a waiver for the studies) are fulfilled. The animal testing ban under the Cosmetic Products Regulation, therefore does not prevent registrants from carrying out tests to comply with the information requirements of REACH. It is important to note though that tests on vertebrate animals should only be carried out as a very last resort, i.e., when no information which meets the information requirements is already available, and where no adaptation (e.g., where no alternative tests exist) can be applied. Moreover, it should also be noted that only the Court can provide a legally binding interpretation of Union law and more particularly on the relationship between the testing and marketing bans in the Cosmetic Products Regulation and the requirements of the REACH Regulation.
This manuscript focuses in particular on the regulatory requirements for relevant human-health related endpoints. More detailed information regarding these endpoints is reported in the next sections.
Skin corrosion and irritation and serious eye damage/eye irritation
Under CLP Regulation (2020f), the criteria for skin corrosive category and subcategories and skin irritation category are based on animal data; however, validated and accepted in vitro alternatives may also be used to help make classification decisions. The criteria for skin corrosion and irritation was updated to include criteria for the application of non-animal methods in the 8th revision of the GHS (UN-GHS 2019), and the CLP Regulation implementing GHS within the EU, will be revised accordingly.
Category 1 applies to corrosive substances, which can be further divided into three subcategories: category 1A, 1B and 1C, applied in the GHS and Packing Groups I, II and III applied in the UN Model Regulations for transport of dangerous goods. In the 21st revision of the Model Regulations (UN-TDG 2019) and in the 8th revision of GHS (UN-GHS 2019) the possibility for sub-classification based on in vitro data was introduced. Category 2 is attributed to irritant substances. Category 3 (mild skin irritation) is optional and is available for those authorities that want more than one skin irritation category (e.g., for classifying pesticides).
OECD GD 237 (OECD 2016a) describes waiving principles applicable to mammalian acute toxicity (oral, dermal and inhalation route), eye and skin irritation and skin sensitisation, intended for pesticides, but extendable also to other chemicals, formulations and biological materials. As specified in the OECD GD 237, “In the context of this document, acute toxicity studies refer to those assessing systemic toxicity as well as those assessing local irritation, corrosion or sensitisation”.
The Classification and Labelling (C and L) categories used are based on visually observable effects in rabbit skin following Draize skin corrosion and skin irritation test [EU test method B.4, equivalent to OECD TG 404 (OECD 2015c)]. However, as for skin corrosion/irritation, validated and accepted in vitro alternatives shall be used to make classification decisions (EC 2017d). This is also confirmed in the GHS Fig. 3.2.1, which reports tiered testing and evaluation of skin corrosion and irritation potential (see line 28f).
For serious eye damage/eye irritation, the classification system involves a tiered testing and evaluation scheme. The criteria themselves for irreversible or reversible eye effects are still based on animal data. On GHS level, the criteria for serious eye damage and eye irritation is currently under revision, and an updated text to include non-animal criteria is expected in the 9th revision of GHS in 2021. A substance or mixture classified as corrosive to skin is deemed to be classified for serious eye damage, to avoid any testing of corrosive substances for eye effects in vivo (ECHA 2017c).
Under REACH (2020g), for Annex VII and Annex VIII the assessment of skin irritation or skin corrosion using an in vitro test is foreseen. Regarding serious eye damage/eye irritation, the basic information requirement is an in vitro study, and a second in vitro study must be considered if the results from the first in vitro study do not allow a conclusive decision on classification for serious eye damage/eye irritation. Annex VIII foresees the assessment of skin irritation using the in vivo test only if the in vitro studies (under Points 8.1.1 and 8.1.2 of Annex VII) are not applicable, or their result(s) not adequate for classification and risk assessment. Same consideration is made for eye irritation. These amendments to Annexes VII and VIII relevant for skin corrosion/irritation and serious eye damage/eye irritation have been made in 2016 (EC 2016), considering the significant scientific progress in the development of alternative test methods for these endpoints. In particular, for both skin corrosion/skin irritation and serious eye damage/eye irritation, adequate information for the classification and risk assessment of a substance should be obtained in most cases solely on the basis of in vitro studies. For both these endpoints, in vivo studies may still be required in some cases for substances manufactured or imported in quantities of 10 tpy or more. Therefore, Points 8.1 and 8.2 of Annex VIII were amended so that the standard information requirements are now for the in vitro studies, while setting the conditions under which an in vivo study for skin irritation/corrosion and serious eye damage/eye irritation is still required.
Adopted in vitro OECD TGs and corresponding test methods indicated in Regulation 440/2008 (2019b) for skin corrosion/irritation and serious eye damage/eye irritation are reported in Table 2.
For cosmetic ingredients, skin corrosion/skin irritation and serious eye damage/eye irritation should be assessed using the adopted in vitro methods already specified in Regulation 440/2008 (2019b) (Table 2), together with in chemico/in silico [i.e., (Q)SAR]. Data obtained from the Draize rabbit test (EC B.4, OECD TG 404) should be provided when available if the test was performed before the animal testing ban, or if the data were obtained to be in compliance with other legislations (e.g., REACH). In SCCS/1602/18 (2018) it is further commented that currently available replacement alternatives for serious eye damage/irritation testing cannot identify any mild eye irritancy potential. Additionally, for eye irritation, no validated alternative method fully replacing the in vivo test (OECD TG 405, EC B.5) can be identified. Therefore, two separate decision trees for eye irritation were put forward: (i) a decision tree specific for hazard identification of the neat cosmetic ingredient (to classify irritant vs non-irritant, using physicochemical properties, read-across data, (Q)SAR results and in vitro eye irritation data); (ii) a decision tree for risk assessment of the neat ingredient in its final formulation(s) (i.e., formulation’s eye irritancy measured in one or more in vitro eye irritation test(s) vs measured irritancy of a benchmark control, including a confirmatory formulation test with human volunteers).
CLP (2020f) and REACH (2020g) do not specifically ask for photo-toxicity testing and/or labelling requirements. In the most recent SCCS Notes of Guidance (NoG), one in vitro test method, listed in Regulation 440/2008 (2019b) as test method B.41 In vitro 3T3 NRU Phototoxicity Test [equivalent to OECD TG 432 (OECD 2004c)] is indicated as a mandatory in vitro method to assess photo-induced toxicity, when in the exposure assessment (3.3 in NoG) under “functions and uses of cosmetic ingredients” (3.3.1 in NoG) of the dossier submitted, it is shown that exposure to sunlight is possible and the chemical structure indicates the possibility of UV absorption (aromatic groups, double bounds, etc.) and a UV spectrum shows UV absorption. If the UV spectrum does not show UV absorption, there cannot be photo-induced toxicity. As waving principles photo-toxicity tests should not be performed if the test material absorbs at wavelengths < 313 nm, and absorption at longer wavelengths is insufficient. For all UV-filters, (in Annex VI of Reg 1223/2009) the 3T3 NRU Phototoxicity test, comparing the cytotoxicity of a chemical tested in the presence and in the absence of exposure at a non-cytotoxic dose of ultraviolet/visible (UV/VIS) light (SCCS 2018), is mandatory (section 3.2.6 in NoG).
Apart from the 3T3 NRU PT [EC B.41, OECD TG 432 (OECD 2004c)], a reconstructed human skin model can be used as a second tier in particular in case of false positives in the 3T3 NRU PT to evaluate effects (checking for the solvents used), and the use of in chemico/in silico [i.e., (Q)SAR] is encouraged (SCCS 2018). While, to date, validated in vitro methods for the detection of photo-sensitisation are not yet available, chemicals showing photo-allergic properties are likely to give positive reactions in the 3T3 NRU PT test.
At present, no official guideline-based protocols for photo-irritation and photo-sensitisation testing in vivo have been evaluated (SCCS 2018).
To assess photo-mutagenicity/photo-clastogenicity, several assays have been adapted to a combined treatment of chemicals with UV–Visible light (EC 2003), including: (1) bacterial and yeast mutation assays, (2) tests for detecting clastogenicity, (3) tests for detecting gene mutations in mammalian cells, and (4) tests for detecting aneugenicity in mammalian cells in vitro. Other available tests are: the photo-Ames test, the photo HPRT/photo-mouse lymphoma assay, the photo-micronucleus test, the photo-chromosome aberration test and the photo-Comet assay (all to be evaluated on a case-by-case basis) (Brendler-Schwaab et al. 2004). In chemico/in silico methods are also indicated.
There is no requirement for photo-mutagenicity testing when the phototoxicity tests are negative, or if the compounds have a Molar Extinction Coefficient (MEC) below 1000 L mol−1 cm−1 (EFSA 2020).
According to CLP Regulation (2020f), hazard categories for germ cell mutagens are related to substances that may cause mutations in the germ cells of humans that can be transmitted to the progeny. Since human data are not available, the results obtained with mutagenicity or genotoxicity tests in vitro and in mammalian somatic and germ cells in vivo are used in classifying substances and mixtures within this hazard class. Category 1 (accounting for subcategories 1A and 1B) identifies substances known to induce heritable mutations (Cat 1A) or to be regarded as if they induce heritable mutations in the germ cells of humans (Cat 1B). Category 2 applies to substances that may induce heritable mutations in the germ cells, therefore causing concern for humans.
For a comprehensive coverage of the potential mutagenicity of a substance, information on gene mutations (base substitutions and deletions/additions), structural chromosome aberrations (breaks and rearrangements, defined as clastogenicity) and numerical chromosome aberrations (loss or gain of chromosomes, defined as aneuploidy) is required (EC 1223/2009) (EC 2020e; ECHA 2017b).
Under REACH (2020g), the assessment of mutagenicity follows a stepwise approach, which starts with a battery of in vitro tests, followed up by appropriate in vivo testing in case one or more of the in vitro tests are positive. The in vitro studies for mutagenicity include an in vitro gene mutation study in bacteria (Ames test), an in vitro cytogenicity study in mammalian cells (i.e., an in vitro chromosome aberration study or an in vitro micronucleus study) and, if both in vitro tests are negative, an in vitro gene mutation study in mammalian cells should be performed. If there is a positive result in any of the above in vitro studies and there are no results available from an appropriate in vivo study already, an appropriate follow-up in vivo study in somatic cells must be proposed by the registrant. In some cases, a second in vivo somatic cell test may be necessary depending on the quality and relevance of all available data. If there is a positive result from an in vivo somatic cell study, the potential for germ cell mutagenicity should be considered on the basis of all available data, including TK information (if available). Moreover, as for any other endpoint under REACH, the information required for a substance depends on its volume (tpy) of production or importation.
Several in vitro and in vivo test methods and OECD TGs for mutagenicity and genotoxicity are indicated in Regulation (EC) No 440/2008 (2019b), as summarised in Table 2.
To assess the potential for mutagenicity of a cosmetic substance (EC 1223/2009) (EC 2020e), two tests in particular are recommended: the Bacterial Reverse Mutation Test, Ames (OECD TG 471) (OECD 1997b), to assess gene mutations, and the In vitro Micronucleus Test (OECD TG 487) (OECD 2016o), to assess both clastogenicity and aneugenicity.
In cases where the bacterial reverse mutation test is not suited, as in the case of nanoparticles, a revised genotoxicity test battery, which includes in vitro mammalian cell mutagenicity and clastogenicity assessments, has been recommended (Elespuru et al. 2018).
If the results from both tests are clearly negative in adequately performed tests, it is very likely that the substance has no mutagenic potential. Likewise, if the results from both tests are clearly positive, it is very likely that the substance has mutagenic potential. In both cases, further testing is not necessary. If one of both tests is positive, the substance is considered an in vitro mutagen, and further in vitro testing is needed to exclude the potential mutagenicity of the substance under investigation. A toolbox for the evaluation in a Weight-of-Evidence (WoE) approach has been proposed in the SCCS/1602/18 (2018), which includes among others: the comet assay in mammalian cells, comet or micronucleus assay on 3D-reconstructed human skin, the Hen’s Egg test for Micronucleus Induction (HET-MN), mechanistic investigations (e.g., toxicogenomics) or internal exposure (TK), Reporter gene assays based on human, animal or bacterial cells (Pfuhler et al. 2020). For chemicals that are primarily associated with dermal exposure, the use of reconstructed human skin models has been explored and protocols have been developed for a reconstructed skin micronucleus test (RSMN) (Curren et al. 2006; Mun et al. 2009) and a RS Comet assay (i.e., 3D Skin Comet) (Reisinger et al. 2018) based on the best suited skin tissues (Curren et al. 2006; Pfuhler et al. 2011; Reisinger et al. 2018). The development of OECD test guidelines based on these tests is currently ongoing.
Acute systemic toxicity
In the Regulation (EC) No 1272/2008 (CLP) (2020f), acute toxicity hazard categories and acute toxicity estimates defining the respective categories are based on animal data, while categories for specific target organ toxicity after single exposure are based on evidence from humans and/or from experimental animals. Animal studies to assess adverse effects and LD50 or LC50 value of tested compounds (which may result from a single exposure, usually carried out with high doses of the test substance), are thought to allow determination or estimation of a range of severe acute toxic effects including mortality. Substances can be allocated to one of four toxicity categories based on acute toxicity by the oral, dermal or inhalation route according to the numeric criteria.
Under REACH (2020g), and as described in the ECHA Guidance (2017b), the assessment of acute systemic toxicity is among the standard information requirements for substances manufactured or imported into the EU in quantities of 1 tonne or more per year (tpy), and standard information requirements are specified in Annexes VII and VIII. Acute toxicity testing is not required if the substance is corrosive to the skin. In particular, as indicated under Annex VII (≥ 1 tpy), acute toxicity study(ies) via the oral route of exposure is(are) required, and waiving is allowed if a study on acute toxicity by the inhalation route is available. For substances manufactured or imported into the EU in quantities of ≥ 10 tpy (under Annex VIII), in addition to acute toxicity study(ies) via the oral route of exposure, information on at least one other route of exposure is requested, depending on the nature of the substance and the likely route of human exposure. As described in Column 2 of section 8.5.3 of Annex VIII, waiving of acute dermal toxicity testing is further allowed if: (i) the substance does not meet the criteria for classification for acute toxicity or STOT-SE (specific target organ toxicity-single exposure) by the oral route, and (ii) no systemic effects have been observed in in vivo studies with dermal exposure (e.g., skin irritation, skin sensitisation) or, in the absence of an in vivo study by the oral route, no systemic effects after dermal exposure are predicted on the basis of non-testing approaches [e.g., read across, (Q)SAR studies]. In line with this, WoE-based adaptation to the standard information requirement may be adopted for acute oral toxicity studies, particularly for substances to be registered at Annex VIII tonnage level and above (i.e., registrations at > 10 tpy), for which an oral sub-acute toxicity study (OECD TG 407) (OECD 2008a) or the combined repeated dose toxicity study with the reproduction/developmental toxicity screening test (OECD TG 422) (OECD 2016f) is required. This WoE adaptation proposed by ECHA (ECHA 2017b) applies to low toxicity substances (i.e., those that are not to be classified for acute oral toxicity). Further considerations regarding these adaptation rules are also discussed in Buesen et al. 2018; Gissi et al. 2017, 2018; Graepel et al. 2016.
According to the ECHA Guidance (2017b), derivation of LD50 or LC50 values is no longer considered essential. Indeed, some of the current standard acute systemic toxicity TGs [e.g., EU B.1 bis/OECD TG 420 (OECD 2002a) and OECD TG 433 (OECD 2018g)], use signs of non-lethal toxicity (rather than mortality). These test methods should be preferred as they present advantages over the other guidelines in terms of animal welfare.
Recommended test methods, as indicated in Regulation (EC) No 440/2008 (2019b), and corresponding OECD TGs for acute systemic toxicity are summarised in Table 2.
As per Regulation (EC) No 1223/2009 (Cosmetic Products Regulation) (2020e), acute systemic toxicity plays in practice a limited role for the cosmetics industry. Ingredients used in this sector essentially do not raise the risk of acute systemic toxicity and sufficient information is often available from repeated dose studies if conducted before 2013. Additionally, any possible impacts on the toxicological profile due to particle sizes, including nanomaterials, impurities of the substances and raw material used, and interaction of substances should be considered, and validated alternative non-animal methods applied.
According to the Notes of Guidance SCCS/1602/18 (2018), validated (animal-free) replacement methods for acute systemic toxicity are not available. However, data on acute systemic toxicity are not mandatory for assessing the safety of cosmetic ingredients for consumer uses. A WoE approach [e.g., data from chemical grouping/read-across, (Q)SAR, in vitro studies, or repeated dose toxicity studies] may be sufficient to drive conclusions on the safety of cosmetic products for acute systemic toxicity.
As already mentioned under "Skin corrosion and irritation and serious eye damage/eye irritation" section, OECD GD 237 opens the possibility to waive animal studies where the results of validated in vitro tests or alternative approaches are adequate to draw a conclusion regarding the classification of an acute hazard for a test chemical. These waiving principles are applicable to mammalian acute toxicity (oral, dermal and inhalation route), eye and skin irritation and skin sensitisation, and although they were mainly intended for pesticides, they can be extended to other chemicals, formulations and biological materials. The approaches outlined in OECD GD 237 should be used by regulatory jurisdictions as part of the WoE to determine the need for a mammalian acute toxicity study and establish appropriate classification and/or labelling.
Assessment of categories and subcategories for skin sensitisers under CLP (2020f) is done considering evidence derived from effects seen in humans and/or animal tests. Skin sensitisers are classified as Category 1. If data allow, optional subcategorisation of sensitisers into subcategories 1A (strong sensitisers) and 1B (other skin sensitisers) can be performed. As a general comment, when considered in the context of a WoE approach, evidence from animal studies is usually more reliable than evidence from human exposure, since the latter is usually derived under less controlled studies. Human evidence may derive from clinical experience, diagnostic patch testing, and other tests designed to confirm the absence of sensitisation potential under expected exposure conditions. Human tests for the purpose of hazard identification are not conducted in the EU because considered unethical.
REACH information requirements for skin sensitisation have been recently revised [Section 8.3 of Annex VII, as of May 2017 (EC 2017a)] and this information should come from: (i) in vitro/in chemico data addressing the three key events (KEs) described in the skin sensitisation Adverse Outcome Pathway (AOP) (i.e., molecular interaction with skin proteins, inflammatory response in keratinocytes, activation of dendritic cells) (Landesmann and Dumont 2012; OECD 2012); and (ii) an in vivo study, normally a Local Lymph Node Assay (LLNA) [described in OECD TG 429 (OECD 2010b)], in case the in vitro/in chemico studies are not applicable for the substance, or are not adequate for classification and risk assessment. In case a substance is considered a skin sensitiser, the revised REACH requirements also introduce the need to assess whether it can be presumed to have the potential to produce significant sensitisation in humans (i.e., GHS /CLP Cat. 1A).
The ECHA guidance document (ECHA 2017b) for this endpoint has been revised to inform about the recent adoption or revision of several EU test methods and/or OECD TGs for skin sensitisation. Additionally, information about the use of non-testing data has been updated to reflect ECHA’s current approach to dossier evaluation. The testing and assessment strategy for skin sensitisation has also been updated, and now it foresees the use of non-animal test methods addressing AOP KEs for generating adequate information. According to Annex VI, the registrant should gather and evaluate all existing available information before considering further testing. This includes structural considerations, physico-chemical properties, (Q)SAR, information from structurally similar substances, in vitro/in chemico data, animal studies, and human data. For classified substances, information on exposure, use and risk management measures should also be collected and evaluated to ensure that potential risks are identified and adequate risk management measures are taken.
The in vivo and in vitro test methods (and OECD TGs) for skin sensitisation (Regulation 440/2008 (2019b)) are summarised in Table 2. In particular, B.71: In vitro skin sensitisation assays (equivalent to OECD TG 442E) addresses the activation of dendritic cells, one KE in the AOP for skin sensitisation (Landesmann and Dumont 2012; OECD 2012), and provides three in vitro test methods addressing mechanisms under the same KE: (i) the human Cell Line Activation Test (or h-CLAT method), (ii) the U937 Cell Line Activation Test (or U-SENS), and (iii) the Interleukin-8 Reporter Gene Assay (or IL-8 Luc assay).
For testing of cosmetics ingredients, skin sensitisation is considered among the most relevant endpoints due to the high frequency of allergic reactions among the undesirable effects of cosmetic products. Notably, recent efforts have been made by the cosmetic industry to develop a non-animal, next generation risk assessment (NGRA) framework for the assessment of skin sensitisers (Gilmour et al. 2020).
Repeated dose toxicity
According to the CLP Regulation (2020f), categories for specific target organ-toxicity—repeated exposure are based on evidence from humans (although rarely available) and/or from in vivo laboratory animal studies. Under REACH, the standard information requirements for repeated dose toxicity are in vivo studies (in rats) of increasing minimum duration as the tonnage band increases. The oral route is the most common, but substance properties and the relevant exposure route for humans need to be taken into account. The standard information requirements on repeated dose toxicity are specified in REACH Annexes VIII-X. Information on a sub-acute (28-day) study is needed at Annex VIII (10–100 tpy) level. At the next tonnage band, a longer study, i.e., sub-chronic (90-day) study, is required. In addition, further studies may be needed at Annex levels IX and X to address concerns related to longer exposure duration, different route of administration and/or specific toxicological investigations, such as immunotoxicity or neurotoxicity. Long-term chronic toxicity studies may be needed based on human exposure considerations. In the context of REACH, the benchmark dose [BMD, defined as the dose corresponding to a ‘specific change in an adverse response compared to the response in unexposed subjects’ (Dakeishi et al. 2006)] may also be used, and species-specific information, e.g., on respiration rates and body weight, enable extrapolation between studies with different exposure routes.
Importantly, the ECHA Guidance (2017b) describes the use of an Integrated Testing Strategy (ITS) for repeated dose toxicity. In particular, testing for repeated dose toxicity is not required for chemicals produced at tonnage levels less than 10 tpy, whilst at higher production volumes, standard data requirements are increased with each tonnage.
As indicated in Regulation (EC) No 440/2008 (2019b), current standard test methods and corresponding OECD TGs are all in vivo studies (Table 2).
As outlined in both Regulation (EC) No 1223/2009 (Cosmetic Products Regulation) (2020e) and SCCS/1602/18 (2018), evaluation of systemic toxicity is a key element for cosmetic ingredients, which are repeatedly in contact with human skin and mucosa. If studies of only 28-day duration are available, a default assessment factor of three to extrapolate from subacute (28 days) to subchronic (90 days) toxicity may be used in the calculation of the Margin of Safety (MoS), as also applied under REACH (ECHA 2012). The inhalation route is only rarely used in repeated dose toxicity testing of cosmetic ingredients, unless a cosmetic product is intended to be used in an aerosolised, sprayable, or powdered form. If the dose regimen of a study was 5-day treatment per week, the derived dose-descriptor corrected by a factor of 5/7 is normally used. SCCS recognises that the BMD can be used as an alternative to the No Observed Adverse Effect Level (NOAEL) approach for deriving a Point of Departure (PoD), which is defined as the point on a toxicological dose–response curve corresponding to an estimated low effect level or no effect level (ChemSafetyPro 2018). The 28-day and 90-day oral toxicity tests in rodents are the most commonly used repeated dose toxicity tests. Preferably, studies of 90 days or more should be used in safety assessments. In a number of cases, dermal repeated dose toxicity studies are present among the submitted data for the cosmetic ingredients listed in Annexes III-VI of Cosmetic Products Regulation, as for example in the case of UV-filters.
Under CLP (2020f), hazard categories for carcinogens are largely based on human (if available) and/or animal evidence. Category 1 accounts for known or presumed human carcinogens on the basis of epidemiological and/or animal data. A substance may be further distinguished as category 1A (i.e., carcinogenic potential for humans, based on human evidence), or category 1B (i.e., presumed carcinogenic potential for humans, based on animal evidence). Category 2 is assigned to suspected human carcinogens, and this classification is done on the basis of evidence obtained from human and/or animal studies, which is not convincing enough to place the substance in Category 1A or 1B.
REACH (2020g) requires a carcinogenicity test for substances falling under Annex X (≥ 1000 tpy), in case: (i) of widespread dispersive use, or when there is evidence of frequent or long-term human exposure, and (ii) if the substance is classified for mutagenicity (germ cell mutagen category 3 under CLP, now category 2), or there is evidence from the repeated dose study(ies) that the substance is able to induce hyperplasia and/or pre-neoplastic lesions.
If the substance is classified as mutagen category 1A and 1B, the default presumption would be that a genotoxic mechanism for carcinogenicity is likely. In these cases, a carcinogenicity test will normally not be required, according to the standard information requirement (Annex X).
Proposals for conducting a carcinogenicity test should be made with regard to the potential risk to human health and with consideration of the actual or intended production and/or use pattern. However, REACH also requires that carcinogenic substances at all tonnage levels be identified as substances of high concern, taking into account information from all available relevant sources (non-human and human, non-testing and testing data), which can inform on hazard identification, underlying modes of action or carcinogenic potency. In addition, the classification and labelling as listed in Annex VI of CLP Regulation is legally binding and can trigger further assessment under REACH to decide if the substance should be formally identified as a substance of very high concern (SVHC) (Madia et al. 2016).
The ECHA Guidance (2017b) proposes a testing strategy entailing the following three steps for the assessment of carcinogenicity for substances at each of the tonnage levels specified in Annexes VII to X of REACH: (i) gather and assess all available test and non-test data from read-across and/or proper chemical category (chemical grouping) and suitable predictive models, and examine the WoE that relates to carcinogenicity; (ii) consider whether the standard information requirements are met; (iii) ensure that the information requirements of Annexes VII and VIII are met, and make proposals to conform to Annexes IX and X (whether further tests are needed to fulfil requirements under Annexes IX and X).
In case a carcinogenicity study needs to be conducted, a testing proposal needs to be submitted to the agency as specified in REACH. For substances at annex X, predictive techniques, such as chemical grouping and read-across, and the use of (Q)SARs may be supplemented with in vitro or alternative shorter-term in vivo studies to circumvent the need for a carcinogenicity study (ECHA 2017b).
Different sources of information may enable drawing inferences regarding the potential of a chemical to be carcinogenic to humans. In particular, non-human data, including non-testing data, testing data (both in vitro and animal), human data, and information on exposure, use and risk management should be considered (paragraph R.7.7.10, Information sources on carcinogenicity) (ECHA 2017b).
In the Regulation (EC) No 440/2008 (2019b), two in vivo tests are described: B.32. Carcinogenicity test [equivalent to OECD TG 451 (OECD 2018m)], and B.33. Combined chronic toxicity/Carcinogenicity test [equivalent to OECD TG 453 (OECD 2018o)], and one in vitro test: the B.21. In vitro Mammalian cell transformation test (see Table 2). At present, no validated (animal-free) replacement methods included in OECD TGs to study carcinogenicity are available.
As for industrial chemicals under REACH, also for cosmetics ingredients, genotoxicity information is the main driver for consideration of carcinogenicity.
Two OECD Guidance Documents (GDs) on in vitro Cell Transformation Assays (CTA) have been adopted: CTA in Syrian Hamster Embryo (SHE) cells performed at pH 6.7 and at pH 7.0 (OECD GD 214) (OECD 2015b), and CTA in Bhas 42 cell line (OECD GD 231) (OECD 2016b).
As suggested in the SCCS Notes of Guidance (SCCS 2018), a positive result in one of the in vitro genotoxicity tests may be indicative to consider a substance as a putative carcinogen. This indication may be further supported by a positive result in cell transformation assays. However so far, there are no specific requirements to obtain information on non-genotoxic carcinogenicity as such, and many non-genotoxic carcinogens may remain unidentified (Jacobs et al. 2016). According to the SCCS Notes of Guidance (SCCS 2018), also in vitro toxicogenomics can be used in a WoE approach, especially for the detection of non-genotoxic carcinogens. CTA in combination with other existing information and toxicogenomics approaches may be considered as part of integrated approaches to testing and assessment (IATA). Further information on the status of in vitro carcinogenicity testing can be found in (Adler et al. 2011; Jacobs et al. 2020; Madia et al. 2014, 2016; Worth et al. 2014).
Reproductive and developmental toxicity
CLP criteria for hazard categories for reproductive toxicants are either based on evidence from humans (rarely available) and/or data from animal studies (2020f). Category 1A is assigned to known human reproductive toxicants based on evidence in humans, and category 1B is assigned to chemicals that are presumed human reproductive toxicants based on data from animal studies. When there is mechanistic information that raises doubt about the relevance of the effects for humans, classification in Category 2, which identifies suspected human reproductive toxicants, may be considered more appropriate. Moreover, classification as a reproductive toxicant is made on the basis of a WoE assessment, i.e., all available information is considered together. This information may be derived from epidemiological studies and case reports in humans and specific reproduction studies in animals that investigate fertility, sexual function and developmental effects in offspring along with sub-chronic, chronic and special studies in animals that provide relevant information regarding toxicity to reproductive and related endocrine organs.
Under REACH (2020g), the reproductive toxicity of a substance is primarily assessed by means of three different studies: (i) a reproduction/developmental toxicity screening test (e.g., OECD TG 421/422), (ii) prenatal developmental toxicity studies in two species, and (iii) an extended one-generation reproductive toxicity study (EOGRTS). It should be considered that at Annex VII, none of these tests need to be provided, while at Annex VIII, a screening study is required as a minimum, with the proposal to consider performing a prenatal developmental toxicity study if there are any indications of concern for this endpoint from existing information. The EOGRTS would normally only be required at Annex X but could be triggered at lower tonnages (Annexes VIII or IX) on the basis of concerns of potential adverse effects from existing information. Theoretically, in exceptional cases, information from an EOGRTS in a second species or strain may be legally required at Annex X.
The EOGRTS [EC B.56, OECD TG 443 (OECD 2018l)] is now considered the information requirement for reproductive toxicity instead of the two-generation reproductive toxicity study [EC B.35, OECD TG 416 (OECD 2001)] based on an amendment from 2015 (Commission Regulation (EU) 2015/282) (EC 2015a). Although a two-generation reproductive toxicity study is accepted to cover the standard information requirement, instead of an EOGRTS, if initiated before March 13, 2015. EOGRTS offers a number of advantages in comparison to the two-generation reproductive toxicity study, as it assesses a greater number of animals of the first filial generation (F1) and addresses additional parameters, improving the sensitivity and level of information that can be obtained from the test, and may allow a reduction of the number of animals to be used (depending on the study design). The standard information requirement in Annexes IX and X should be limited to the basic configuration of EOGRTS (without extension to include an F2 generation).
Nevertheless, in certain specific cases, where justified, the registrant should be able to propose and ECHA should be able to request the performance of the F2 generation (e.g., on the basis of concerns for endocrine disruption), as well as the developmental neurotoxicity (DNT) and developmental immunotoxicity (DIT) cohorts. DNT and DIT are regarded as important and relevant developmental toxicity endpoints, which could be further investigated. However, analysing the DNT and DIT cohorts entails significant additional costs as well as subjecting animals to additional experiments. Currently, analysis of DIT and/or DNT cohorts is only requested subject to specific concern-driven triggers (see “Developmental neurotoxicity (DNT)” and “Immunotoxicity and developmental immunotoxicity (DIT)” sections).
In REACH, studies on reproductive and developmental toxicity are required from Annex VIII through Annex X, and the standard information requirements are cumulative (i.e., requirements at higher tonnage levels add to the information requirements at lower tonnage levels). If a substance is known to have an adverse effect on fertility, meeting the criteria for classification as Repr Cat 1A/1B, and the available data are adequate to support a robust risk assessment, then no further testing for sexual function and fertility will be necessary. However, testing for developmental toxicity must be considered. With regard to substances known to cause developmental toxicity and classified as Repr Cat 1A/1B, no further testing for developmental toxicity will be necessary, although testing for effects on fertility must be considered. In cases where there are serious concerns about the potential for adverse effects related to fertility or development, the registrant may propose an EOGRTS (Annex IX, Section 8.7.3) and/or a pre-natal developmental toxicity study (Annex IX, Section 8.7.2), as appropriate, instead of the screening study to address the concern(s). If there are no adverse effects leading to a concern for development, a pre-natal developmental toxicity study may not be used to fulfil the requirement for a reproductive screening study.
The ECHA Guidance (ECHA 2017b) further comments on the applicability of an ITS for reproductive toxicity, which is defined as an approach that combines one or more non-animal methods with animal studies to fulfil the information requirements, or could include only non-animal methods if they together covered all key aspects of reproductive toxicity. However, the use of non-animal methods should be assessed on a case-by-case manner, ensuring that the obtained results cover all of the key aspects of reproductive toxicity and are suitable for both risk assessment (e.g., derivation of NOAEL) and classification and labelling.
Table 2 summarises the test methods (Regulation 440/2008 (2019b)) and corresponding OECD TGs suitable to assess reproductive and developmental toxicity.
With regard to cosmetic ingredient safety assessment, the one or two-generation reproduction toxicity test (or the EOGRTS) were the most commonly performed in vivo reproductive toxicity studies before the animal testing ban.
Three alternative embryotoxicity-related methods are currently available: (1) the Whole Embryo Culture test (WEC), (2) the MicroMass test (MM), and (3) the Embryonic Stem cell Test (EST), which can all be used to identify strong embryotoxic substances (Balls and Hellsten 2002; Spielmann et al. 2006). At OECD level, a detailed review paper on “Pluripotent stem cell assays: Modalities and applications for predictive developmental toxicity” is currently under development.
Other in vitro methodologies, covering male and female fertility, implantation and pre- and postnatal development have been and are being developed, such as under ReProTect (http://www.reprotect.eu/) or the EURION cluster (https://eurion-cluster.eu/). It should be considered that, to date, validated animal-free methods accepted as a full replacement are not available (Adler et al. 2011; Worth et al. 2014), and that the available alternative methods are not able to mimic all of the various developmental stages, therefore a battery of tests will be needed. However, a more radical change towards next generation risk assessment may allow to move away from prediction of current toxicity classes to prediction of likely safe doses, as indicated in the OECD GD 275 (OECD 2017a).
Absorption, distribution, metabolism and excretion (ADME) and toxicokinetics (TK)
Information on the biological fate of a chemical in the body plays an important role in human safety assessment. While there are few explicit requirements in EU chemicals legislation for the generation of TK data (i.e., in vitro, in vivo measurements or computational predictions), the use of these data to support the assessment of systemic toxicity is widely recommended in regulatory guidance, although not consistently required in regulations (Bessems et al. 2015). For instance, ADME/TK information is required under the Biocidal Products (Regulation (EU) No 528/2012) (EC 2012) and Plant Protection Products [Regulation (EC) No 1107/2009 (EC 2009) and Commission Regulation (EU) No 283/2013 (EC 2013b)] (which are out of the scope of this document), and the EU Plant Protection regulation also requires the generation of human in vitro biotransformation data to compare with rodent data and studies. However, this is not the case for other regulations as briefly explained.
There are no CLP categories for TK and the CLP Regulation does not specifically require the assessment of ADME and TK (2020f). However, ADME and TK data may be used in a WoE approach to classify, lower the classification or abstain from classification for a particular toxicodynamic (TD) endpoint. For the classification of substances as carcinogens, all available information regarding the physicochemical, TK and TD properties of the substances, as well as information on structure activity relationships, should be taken into account to undertake classification.
Under REACH (2020g), TK studies in vivo are not required; however, all available information should be provided, including TK information. Importantly, human health hazard assessment shall consider ADME and TK of substances. Even though TK is not a toxicological endpoint and is not specifically required by REACH, the generation of TK information can help interpret data, assist testing strategy and study design, as well as category development, thus helping to optimise test designing. Furthermore, under REACH, TK data would be very useful for assessing read-across and categories, but as this is not a standard information requirement, that information is rarely available.
The ECHA Guidance (ECHA 2017b) reports many examples of recommendations on the use of TK data that would replace default assessment factors (e.g., Sections R.7.12 and R.8.4 in Chapters R.7.C and R.8, respectively). The guidance highlights that TK studies may be helpful in the evaluation and interpretation of repeated dose toxicity data (e.g., in relation to accumulation of a substance or its metabolites in certain tissues or organs), as well as in relation to mechanistic aspects of repeated dose toxicity and species differences. TK information can also assist in the selection of the dose levels. A very important observation is that TK and potential TD properties based on available data should be considered before undertaking animal tests. Understanding these properties will enable the design of appropriate protocols for the standard tests to be developed, especially with respect to tissue(s) to be investigated, the route of substance administration and the highest dose to be tested. If there is poor understanding of the systemic availability of a test substance, TK investigations or modelling may be necessary.
The three following test methods (and corresponding OECD TGs) for TK are indicated in Regulation 440/2008 (2019b): B.36. Toxicokinetics (in vivo) [equivalent to OECD TG 417 (OECD 2010a)], B.44. Skin absorption: In vivo method [equivalent to OECD TG 427 (OECD 2004a)], and B.45. Skin absorption: In vitro method [equivalent to OECD TG 428 (OECD 2004b)] (Table 2). These EU test methods and OECD TGs generate data for TK, and currently most of them are based on animal procedures as the traditional approach of obtaining whole-body TK parameters. However, by exploiting modern developments in predictive toxicology, there are increasing opportunities to generate human-relevant whole-body TK information using physiologically based kinetic (PBK) models (Paini et al. 2019).
These mathematical models, which represent the body as a set of interconnected compartments linked by blood flow, would enable not only the generation of TK data, but also the integration of human data generated by in silico and in vitro methods for ADME. The lack of standardisation of such methods hampers their regulatory acceptance and use (Bessems et al. 2015). However, there is an on-going international effort at OECD to promote the regulatory use of PBK models based on in silico and in vitro data and body physiological parameters (Sachana 2019).
In relation to cosmetic ingredients, information on TK parameters (e.g., human systemic and dermal exposure, and biotransformation) is recommended (EC 2020e). In particular, with regard to dermal/percutaneous absorption and in specific cases, data from in vivo studies that have been carried out before the animal testing ban, or data from in vitro biotransformation studies are required (SCCS 2018), to prove or to exclude certain adverse effects (e.g., EC B.44, 45; OECD TG 427, TG 428). For dermal absorption, it should be considered whether the formulation can affect compound bioavailability.
With regard to in vitro dermal absorption of cosmetic ingredients, some basic criteria have been provided when performing in vitro dermal absorption studies, along with rules to follow in case no dermal absorption studies are available (e.g., regarding the amounts to be applied and what to do in case the basic criteria have not been followed) (SCCS 2010).
For substances with very low dermal absorption and limited permeation (such as colourants or UV-filters with high molecular weight and low solubility), the epidermis may be excluded as a route of entry (WHO 2006). For nanomaterials, it should be ascertained whether the substance absorbed through the skin is in nanoparticle form or in a dissolved chemical state.
Besides the determination of TK parameters of the parent chemical, it is also essential to obtain accurate profiles of metabolites that could be more potent than the parent compound. Cells and cell fractions or organ specimens from human sources, although limited, are available, together with 3D cultures to preserve metabolic capacity and regulation of xenobiotic metabolising enzymes. Additionally, the use of-to-in vivo extrapolation (IVIVE) and PBK modelling is encouraged to translate external exposures into an internal (target) dose in the body and vice versa (Yoon et al. 2012). PBK models are increasingly being used to aid: (i) extrapolation within and between species (variability issues), (ii) route-to-route, (iii) dose extrapolation, and (iv) replacement of default assessment factors by more specific, substance-derived factors.