Directorate-General for Internal Market, Industry, Entrepreneurship and SME's
Consumer, Environmental and Health Technologies
Brussels, 29 October 2015
19th Meeting of Competent Authorities for REACH and CLP (CARACAL)
12 – 13 November 2015
Room Robert Schuman, BERLAYMONT
Brussels Concerns: Skin sensitisation – discussion on the need for quantitative potency information sufficient to allow sub-categorisation according to CLP Agenda Point: 13 Action Requested: The CARACAL are invited to take note of this document for discussion.
Written comments on this document should be sent by 11 December 2015 to email@example.com and firstname.lastname@example.org.
Contact allergy to chemicals is a recognised health problem for workers, professional and consumers. For skin sensitising substances and mixtures containing them, CLP rules ensure that warnings are provided that the substance or mixture may cause an allergic skin reaction, as well as information about the identity of an allergenic substance in order to protect individuals with a known allergy to this substance. REACH also requires for most substances a chemical safety assessment, and the application of appropriate measures to adequately control the identified risks.
During considerations about the use of the recently developed in vitro test methods for skin sensitisation, questions were raised about their compatibility with the CLP and REACH requirements. In particular it was debated whether data need to be sufficient for sub-categorisation and setting of SCLs in order to fulfil the information requirement under REACH. This document aims to provide the necessary background and describe the issues identified in order to initiate further discussion on the use of alternative approaches, be it in vitro tests or non-testing approaches, for skin sensitisation under the frameworks of REACH and CLP. The aim of this discussion is to define an appropriate balance between the objectives of these Regulations to protect human health (e.g., by providing appropriate hazard information to consumers and workers via labelling) and animal welfare (by promoting the use of non-animal testing approaches). It also aims to guide future work on the technical level (e.g. ECHA's guidance development) as regards the level of information that is required to be provided under REACH.
Annex I, section 184.108.40.206.1. of the CLP Regulation sets the following hazard categories for skin sensitisers: a general category 1 (sensitisers) which exists in parallel to the sub-categories 1A (strong/extreme skin sensitisers) and 1B (moderate skin sensitisers), based on sensitising potency. It prescribes the classification of substances identified as skin sensitisers in Category 1 where available data are not sufficient for sub-categorisation. For cases where data are sufficient a potency assessment allows the allocation of skin sensitisers into sub-category 1A or sub-category 1B.
The generic concentration limit (GCL) for classification of mixtures is lower for skin sensitisers Cat. 1 A (0.1%) than for Cat. 1 and Cat. 1B (1.0%). According to the CLP guidance, for extreme sensitisers an SCL lower than the generic concentration limit (GCL) for sub-category 1A, may be set (0.001% or individual values based on reliable data).
Mixtures containing skin sensitisers at or above the generic or specific concentration limits have to be labelled with H317: May cause an allergic skin reaction.
In order to protect individuals diagnosed with contact allergy, labelling of mixtures with EU H208 (‘Contains (name of sensitising substance). May produce an allergic reaction’) is required if a mixture contains a sensitising substance at a concentration of at least one tenth of the generic/specific classification limit, even if the mixture is not classified for skin sensitisation.
REACH information requirement for skin sensitisers
REACH requires information on skin sensitisation for all substances subject to registration, irrespective of the tonnage band.
REACH Annex VII, point 8.3 sets the standard information requirement for skin sensitisation. According to the current text, if none of the waiving conditions in column 2 is met, and an adaptation according to Annex XI cannot be made, an in vivo test is required. The LLNA is being specified as the first–choice method (without differentiating between the three available variants, i.e. OECD TG 429, TG 442A, TG442B), while other in vivo methods should only be used in exceptional circumstances and require justification. Adaptation possibilities according to Annex XI include the use of existing data, weight of evidence, (Q)SAR, in vitro methods, grouping and read-across.
The proposed revision of Annex VII, point 8.3 would add an additional possibility in column 2 to waive the in vivo study in the case that sufficient information adequate for classification and/or risk assessment is available from non-animal approaches.
In relation to risk assessment for skin sensitisers, the ECHA guidance1 provides that extreme and strong sensitisers (Cat. 1A) should be strictly contained and dermal contact avoided. In cases where potency categorisation of a sensitising substance is not possible (Cat. 1), Risk Management Measures and Operating Conditions equivalent to the ones applicable to Cat. 1A skin sensitisers should be considered.
Status of alternative methods for skin sensitisation
In recent years, significant progress has been made in the development of alternative methods to assess skin sensitisation. Based on an OECD-agreed Adverse Outcome Pathway (AOP), a number of tests have been developed, each addressing one of the first three (out of four) key events in the AOP. Three of these methods, the Direct Peptide Reactivity Assay (DPRA), the ARE-Nrf2 Luciferase Test Method (KeratinosensTM) and the human Cell Line Activation Test (h-CLAT), have successfully undergone validation by the European Centre for the Validation of Alternative Methods (EURL-ECVAM). DPRA and the ARE-Nrf2 Luciferase assay have reached international acceptance as an OECD test guideline2, while the draft h-CLAT test guideline is still under discussion and expected to be adopted by OECD in 2016.
The OECD in vitro test guidelines indicate that the two methods can be used to support discrimination between skin sensitizers (i.e. GHS Category 1) and non-sensitizers within the context of an integrated approach to testing and assessment, yet they do not allow, on their own, sub-categorisation. Further the test guidelines state that the tests may potentially contribute to the assessment of skin potency in the context of an IATA, but that further work, preferably based on human data, is required to determine how they can inform potency assessment.
On the EU level, the process of taking up DPRA and the ARE-Nrf2 Luciferase assay in the test method regulation has been initiated, and the 7th ATP containing these methods is expected to be adopted and published in Q3 2016.
A draft Commission regulation revising the information requirements in REACH Annexes VII and VIII for several endpoints includes skin sensitisation and proposes to include a waiving possibility for the in vivo test if sufficient information that is adequate for classification and/or risk assessment is available from non-animal approaches. This draft Commission Regulation has been discussed by the REACH Committee in the meeting of 21/22 October 20153 and a vote on this proposal is planned for the REACH Committee meeting in December 2015.
In light of the upcoming 2018 registration deadline for low-tonnage substances and ECHA’s intention to complete registration related guidance updates as far as possible by mid-2016, ECHA has initiated the process of updating the endpoint-specific guidance (section R.7.3) for skin and respiratory sensitisation to account for the recent developments and to provide guidance on the use of the alternative methods for skin sensitisation. The meeting of the Partner Expert Group (PEG) for this guidance update has taken place on 6 October 2015.
In the context of this discussion, it should be noted that non-testing alternative approaches such as read-across, weight of evidence and (Q)SAR have found widespread use in the registration dossiers already submitted under REACH. According to the 117(3) report of 2014, only about 50% of the submitted endpoint study reports for skin sensitisation used an in vivo study, the rest was mostly covered by read across (25%), weight of evidence (16%) and other alternative approaches (3%). In around 6% of the total entries, the registrants have proposed to omit the in vivo study.
Issues under discussion
In the context of the ongoing initiatives to promote the uptake of alternative testing methods for skin sensitisation for regulatory purposes, several questions, which are listed below, have been raised concerning the consequences for the classification and labelling of skin sensitisers, the level of protection of workers/consumers and the compatibility of alternative approaches (including, but not only concerning, the new in vitro methods) with CLP and REACH rules.
Does CLP allow the use of in vitro methods for skin sensitisation?
Several recitals and articles in CLP concern the relevant data to be considered, the choice of new tests, should they be necessary for the purpose of classification and labelling, and the use of alternatives to animal testing.
Article 5(1) of CLP on the identification of relevant available information refers to data generated with methods in the TMR or internationally recognised/validated methods (OECD), but also to any information generated in accordance with section 1 of Annex XI of REACH (which includes use of existing data, weight of evidence, (Q)SAR, in vitro methods, grouping and read-across).
Article 7 (1) of CLP stipulates that new tests on animals for the purpose of the CLP Regulation, shall be undertaken only where no alternatives which provide adequate reliability and quality, are possible. In addition, Recital 27 states that "Future criteria […] should under no circumstances lead to the use of animal tests where alternative tests are adequate for the purpose of classification and labelling".
Article 8 of CLP on the generation of new information for the purpose of the CLP regulation sets the obligation to exhaust all other means of generating information including by applying the rules provided for in REACH Annex XI, section 1 before performing new tests.
Article 9(1) of CLP on the evaluation of hazard information for substances and mixtures stipulates that the identified information shall be evaluated by applying the criteria for classification in Annex I. In case the criteria cannot be applied directly to available information, Art. 9(3) states that "manufacturers, importers and downstream users shall carry out an evaluation by carrying out a weight of evidence determination using expert judgement…"
The current classification criteria for skin sensitisers described in Annex I, section 3.4 of CLP are based on human and animal data. No criteria for classification based on in vitro methods or other non-animal approaches are given in the section on skin sensitisation Notably, Table 3.4.2. specifically refers to evidence in humans or positive results from an animal test, but does not mention the use of in vitro methods for the classification as skin sensitisers. The use of "non-standard tests" is only mentioned in section 220.127.116.11.4.3, in the context of evaluation of human and animal data, and for the case that none of the criteria for classification on this basis are met – in this case, the presence of two pieces of other evidence, including positive data from non-standard tests, may indicate a need to nevertheless consider classification.
Overall, based on the general provisions of CLP listed above, the Commission services are of the opinion that the use of alternative approaches, including the use of in vitro test methods, is already currently possible under CLP by applying the weight of evidence assessment according to Art. 9(3). Nevertheless, to take account of recent developments, it would be desirable to develop GHS criteria for the classification of skin sensitisersi based on in vitro methods.
Appropriateness of the Generic Concentration Limit (GCL) for Category 1 skin sensitisers
During the discussion on the introduction of in vitro methods for skin sensitisation the problem was pinpointed that the GCL of Cat.1 skin sensitisers is equivalent to that of Cat.1B sensitisers. It was questioned whether it is appropriate to apply a less stringent GCL in the absence of data, thus providing a disincentive for registrants to generate potency information. An initiative to modify this provision could be considered in order to oblige registrants to apply a more stringent GCL in the absence of appropriate potency data.
Does the Standard Information Requirement under REACH include potency information? Does such data have to be adequate for sub-categorisation and setting of SCLs, where appropriate?
REACH, through Articles 13 and 25, contains the obligation that information on human toxicity shall be generated whenever possible by means other than vertebrate animal tests, through the use of alternative methods, and that testing on vertebrate animals for the purpose of this regulation shall be undertaken only as a last resort. Recital 38 states that the generation of information by alternative means offering equivalence to prescribed tests and test methods should also be allowed.
Annex VII, point 8.3 defines an in vivo test for skin sensitisation as the standard information requirement for this endpoint. It further states that the LLNA is the first-choice method for in vivo testing. The use of other in vivo methods (typically using guinea pigs) requires specific justification. The most-used version of the LLNA using radioactive labelling (EU TM B.42, OECD TG 429) provides an EC3 value which is the effective concentration of the test substance required to produce a threefold increase in lymphocyte proliferation in draining lymph nodes. The EC3 gives an indication of the sensitising potency of a substance, although the assay has not been validated for this purpose. Annex I, Tables 3.4.3. and 3.4.4 of CLP provides specific criteria to classify substances in sub-categories 1A and 1B based on EC3 values derived from the LLNA according to EU TM B.42/OECD 429 and on data from guinea pig tests. The CLP guidance describes how SCLs may be set based on potency data derived from these in vivo studies .It should be noted that two variants of the LLNA exist which use non-radioactive labelling methods (EU TM B.50/B.51, OECD TG 442A/442B). These variants use different threshold values for the identification of sensitisers that the radioactive method, and no criteria for potency-grading have been established for them.
Recital 38 in REACH states that “The generation of information by alternative means offering equivalence to prescribed tests and test methods should also be allowed” which may be taken to indicate that an alternative approach should not render a lower level of information. The adaptation rules in Annex XI, as well as the proposed revised text for REACH Annex VII, point 8.3 require data derived from alternative approaches to be "adequate for classification", without giving further indications how to interpret such adequacy for endpoints where classification subcategories exist in parallel with a general category. In such a case it may be considered that the level of required information is defined by the test that constitutes the standard information requirement for in vivo testing, i.e. the LLNA.
On the other hand, it has to be considered that a binding requirement for data to provide potency information sufficient for sub-categorisation and SCL-setting where appropriate would apply not only to dossiers to be submitted with the newly developed in vitro tests, but also to all other dossiers using adaptation possibilities given by Annex XI. Such a requirement has up to now never been formulated by either the Commission services or by ECHA in their guidance to registrants and would have wide-ranging implications if it were introduced now. An analysis of dossiers submitted under the first two registration deadlines showed that ca 50% of endpoint study records contained adaptations4 according to Annex XI (mainly WoE and read-across). An absolute requirement for data to provide “equal” information as the current standard information requirement, the LLNA, and thereby be sufficient for sub-categorisation and SCL-setting where appropriate may largely prevent the use of Annex XI adaptations for dossiers now under development, and result, if applied retrospectively, in non-compliance of a high number of dossiers already submitted to ECHA under the believe that sub-categorisation is not mandatory. The respective requirements stemming from REACH and CLP , on the one hand to protect human health by generating and evaluating data on skin sensitisation, and, on the other hand, to use alternatives to animal testing wherever possible, therefore need to be carefully balanced.
Balancing the protection of human health and animal welfare
In case it would be established that REACH sets a mandatory requirement that data derived from alternative approaches give sufficient potency information allowing sub-categorisation and SCL setting, where appropriate, the possibility to use such data would currently be limited to consistently negative results, or require an elaborate weight-of-evidence assessment to allow scientific justification of an potency estimate. The large majority of registrants would chose to perform an LLNA, which is the established method giving results which can be directly applied for a classification decision according to the established criteria. With such an approach, strong and extreme sensitisers could be identified and mixtures containing them can be classified and labelled accordingly. It needs to be recognised, however, that, as any toxicological test, the LLNA is not failproof, and its limitations have been recognised5. Most importantly, especially the capability of the LLNA to reliably identify strong sensitisers as defined by CLP seems to be limited. Thus reliance on LLNA data does not ensure that all classification decisions based on it would be adequate and sufficiently protective.
On the other hand, if data derived from alternative approaches would be always accepted if adequate for hazard identification, but not providing any potency information necessary for sub-categorisation, this would maximise the use of alternatives by registrants and thus animal welfare benefits (even though at the current point in time many registrants may still rely on LLNA as the experience with and availability of in vitro test methods are limited and interpretation of the results requires extensive expert knowledge,, while the LLNA provides clear-cut and easily usable results and is recognised as the standard that provides adequate and protective classification). Such an approach renouncing potency data to a large extent would lead to a certain number (depending on the extent of use) of possibly unidentified strong and extreme sensitisers and consequently to non-classification of some mixtures containing sensitisers at a relevant concentration and to a lack of appropriate warnings and information on the labels.
Recognising that balancing the two goals of human health protection and the promotion of alternatives to animal testing is challenging and that possibilities of doing that have probably not been exhausted, ECHA outlined a possible option in the draft update of its Guidance. In the draft Guidance sent to the Partner Expert Group (PEG), the registrants were advised to start the testing and assessment strategy with in vitro tests, and in case sub-categorisation cannot be concluded, based on all available information, sub-category 1A is recommended as a default. Also conditions under which LLNA should be done (as a last resort) were described in the draft guidance. This option would allow the use of the in vitro test methods in most cases, while applying a potentially over-cautious approach that would ensure appropriate protection of workers and consumers. However, at the current state of science, it may not reliably identify extreme sensitisers and quantify their potency for SCL setting in the absence of human experience. Another drawback of this approach is that it will lead to conservative (over-) classification in some case. On the other hand, as the potency prediction models based on in vitro test results are likely to become available in near future, the problem may be temporary (A number of companies are already using in-house approaches for risk assessment, and a project by Cosmetics Europe to develop an approach for potency prediction is expected to be finalised within 6 to 12 month).
However, the PEG did not agree on the recommendation to use Cat. 1A as a default as the CLP Regulation allows the use of the general Cat. 1. Instead it was agreed to recommend in the guidance to address skin sensitisation potency by all possible means and to include a specific Appendix on how to potentially address it (to be developed together with ECHA, MS and Industry).
As there is currently no widely approved approach to assess the skin sensitisation potency by means of in vitro test methods with or without additional information, at this point in time no firm recommendations can be provided, however promising approaches will be described in the guidance document.
How could potency/sub-categorisation be addressed in the absence of human and animal data?
Clear-cut criteria for sub-categorisation in CLP only exist for human and for most animal data. However, the determination of potency of skin sensitisers by alternative methods currently receives a lot of attention in scientific research, driven by the needs of the Cosmetics sector where skin sensitisation is one of the key endpoints for risk assessment and animal tests are not possible any more. A number of approaches and analyses in this respect have been published in the scientific literature6. Although a commonly agreed approach to quantitatively predict potency is not yet available, e.g. reactivity information from in vitro tests, structural alerts or sensitisation potency of related substances may provide indications of a high sensitisation potency and allow a weight of evidence assessment whether or not a substance is likely to be a strong or extreme sensitiser.
In the absence of criteria for sub-categorisation and SCL setting based on data from alternative approaches in CLP and the corresponding ECHA Guidance on the application of the CLP criteria, the ongoing update of the Guidance on information requirements for the specific endpoint of sensitisation, offers the possibility to introduce recommendations/expectations on how and to which extent the issue of sub-categorisation needs to be addressed by registrants using alternative approaches, considering also criteria for classification and which actions should be taken in case reasons for concern are found.
ECHA had proposed a precautionary approach for classification if no confident decision on sub-categorisation can be taken based on all available evidence, i.e. Cat. 1A classification as default, and setting of a specific concentration limit of 0.1%. However such an approach was not supported by the PEG in the meeting on 6 October 2015 on the draft guidance for skin sensitisation. Unless this approach was nevertheless supported by CARACAL and stakeholders, possible alternative approaches to be recommended in the guidance could be e.g.
- recommend addressing potency by all possible means when using in vitro approaches
- provide minimum requirements for the evidence that needs to be gathered and considered in order to identify or exclude concern that a substance could be a strong or extreme sensitizer
- recommend classification as Cat. 1A or setting of a concentration limit of 0.1% if there are indications that a substance could be a strong or extreme sensitizer, while in other cases Cat. 1 classification would apply- recommend additional testing if no confident decision on sub-categorisation can be taken, including in vivo testing as a last resort
- recommend additional testing (e.g. LLNA) if concerns for high sensitisation potency are found, in order to be able to identify extreme sensitisers and set appropriate specific concentration limits.
In view of the above, it is obvious that in the application of alternative testing methods, before a commonly agreed approach for the quantitative estimation of the skin sensitising potency of substances has been developed, a balance has to be struck between the joint objectives of CLP and REACH to protect human health and to reduce animal testing to the necessary minimum. ECHA's guidance should reflect this balance while providing registrant with the clear approach how to efficiently generate reliable data for this endpoint.
However, instead of introducing an explicit and absolute requirement for quantitative information on the potency of a skin sensitiser, it may be possible to develop a refined approach that sets more flexible requirements to ensure that registrants gather sufficient evidence that would allow a decision whether or not there are grounds to suspect that a substance has a high potency. Depending on such a concern, further investigations or a precautionary classification may be warranted.
The Commission services would like to consult MS and stakeholders on their views regarding the above points with a view to develop a balanced and generally accepted approach to the use of alternative testing and non-testing methods for skin sensitisation and the exact level/quality of information required to fulfil the REACH information requirements for this endpoint. The outcome of this consultation will feed into the further development of ECHA's updated guidance for this endpoint in ECHA.
2 OECD (2015). TG 442C, In Chemico Skin Sensitisation - Direct Peptide Reactivity Assay (DPRA) OECD (2015). T G 442D, In Vitro Skin Sensitisation - ARE-Nrf2 Luciferase Test Method
3 Questions on sub-categorisation of skin sensitisers were raised by several MS in this context, however, in light of the planned discussion in the present CARACAL, this issue was not discussed in-depth.
4 ECHA (2014). The Use of Alternatives to Testing on Animals for the REACH Regulation. Second Report under Article 117(3) of the REACH Regulation
5 Anderson et al. (2011). The LLNA: a brief Review of Recent Advances and Limitations. Journal of Allergy, Vol 2011, Article ID 424203, http://dx.doi.org/10.1155/2011/424203ICCVAM (2011). Test Method Evaluation Report: Usefulness and Limitations of the Murine Local Lymph Node Assay for Potency Categorization of Chemicals Causing Allergic Contact Dermatitis in Humans. NIH Publication Number 11-7709, National Toxicology Program, Research Triangle Park, NC 27709, http://iccvam.niehs.nih.gov/docs/immunotox_docs/LLNA-pot/TMER.pdf
Basketter et al. (2009). Nothing is perfect, not even the local lymph node assay: a commentary and the implications for REACH. Contact Dermatitis 60:65-69
6 Jaworska et al. (2013). Bayesian integrated testing strategy to assess skin sensitisation potency: from theory to practice. Journal of Applied Toxicology DOI 10.1002/jat.2869Reisinger et al. (2015). Systematic evaluation of non-animal test methods for skin sensitisation safety assessment. Toxicology in Vitro 29:259-270
Roberts et al. (2015). Principles for identification of high potency category chemicals for which the dermal sensitisation threshold (DST) approach should not be applied. Regulatory Toxicology and Pharmacology 72:683-693
Teunis et al. (2014). International ring trial of the Epidermal Equivalent Sensitiser Potency Assay: Reproducability and Predictive Capacity. Altex 31:251-268