By Nicholas Ball, Eurotox Registered Toxicologist, The Dow Chemical Company
REACH Animal Testing Requirements
The European Commission states that the REACH regulation (EC 1907/2006) in Europe will improve the protection of human health and the environment through the better and earlier identification of the intrinsic properties of chemical substances. In delivering this goal, this regulation introduced for the first time a set of mandatory toxicological data requirements for every chemical manufactured, imported, or placed on the market at volumes of >1 tonne within the EU. The regulation uses the tonnage imported or manufactured per 'legal entity' as a kind of proxy for exposure; under the premise that the higher the tonnage made or imported, the greater the potential for exposure to humans and the environment, so there is a desire to know more about the intrinsic properties of these substances.
For a substance registered at >1000 tonnes per year the data requirements are extensive. For example, the human health toxicity requirements cover acute toxicity, local effects (irritation and sensitisation), genotoxicity, sub-acute and sub-chronic repeated dose toxicity, multi-generation reproductive toxicity, developmental toxicity in two species and if appropriate, carcinogenicity. Data on other endpoints such as neurotoxicity and immunotoxicity may also be submitted if relevant and/or available. Of all of these endpoints, validated in vitro tests are available only for the genotoxicity assessment and the initial assessment of skin and eye irritation. However, for the genotoxicity endpoint additional testing in vivo is necessary if there are any positive results in vitro. For skin and eye irritation, the REACH legal text requires the use of in vivo studies for substances registered at >10 tonnes unless the results of the in vitro studies were sufficient to allow classification. So, meeting all these data requirements would in principle require a significant amount of animal testing (Rovida & Hartung, 2009).
Given this demand for hazard information derived from animal studies, one of the challenges facing both the industry and the regulatory agencies (the European Chemicals Agency, ECHA, and the Member States competent authorities) is how to meet the information requirements while at the same time complying with the requirement to only conduct animal tests as a last resort. This mandate was incorporated into the REACH legal text to address the concerns of several stakeholders (for example, Sauer, 2004; Dandrea & Combes, 2003) that REACH could result in a significant increase in the number of animal tests conducted. It also triggered several projects to identify what could be done to meet the needs of the regulation while minimising the impact on animal testing (Höfer, et al., 2004; Grindon, et al., 2006, 2007, 2008a and b; Combes, et al., 2007, 2008a and b; Schaafsma et al., 2009). However, if this requirement to minimise animal testing is to be met it requires commitment from both the registrants and the regulatory agencies. As such, without simply adopting an unnecessarily conservative assessment of hazards, the registrants must make the best use of the various options to reduce the need for new animal studies, and the regulatory authorities must take into consideration the animal welfare aspects when reviewing dossiers and deciding on the need for additional studies. All of this must at the same time provide sufficient comfort that the hazards of substances are being identified and addressed as necessary so that human and environmental safety are not adversely impacted.
Opportunities for Reducing Animal Use
The REACH legal text provides several opportunities for meeting the needs of the regulation while minimising the use of new animal studies and ECHA have produced a 'Practical guide' (No.10) to assist registrants in doing so. Specifically, it is possible to adapt the standard testing requirements in order to make use of information that would allow a new animal study to be avoided without compromising the hazard characterisation process. For almost every data requirement there are specific circumstances where the required study may be considered as unnecessary and 'waived'. For example, if a substance is corrosive then one can waive the acute toxicity studies and skin sensitising potential study since these studies typically require high doses and would be confounded by the damage caused by corrosion. For studies addressing carcinogenicity, mutagenicity, reproductive and developmental toxicity there is the possibility to waive these if the substance is already classified as a category 1a or 1b carcinogen, mutagen, or reproductive toxicant according to the Classification, Labelling and Packaging Regulation (the EU implementation of the Globally Harmonized System for classification and labelling). Making use of these waiving arguments is straightforward due to the clear set of criteria that must be met. Registrants have typically made use of these waiving arguments, whenever possible, to avoid conducting unnecessary animal studies and as yet there is little evidence that such waivers have been rejected by ECHA (ECHA, 2012).
However, there are some other less specific ways to adapt the standard testing requirements that can be applied to the majority of data requirements. These are given in a separate annex of the REACH legal text (Annex XI) and cover three possible options: testing is technically not possible; substance-tailored exposure-driven testing; testing is not considered to be scientifically necessary (includes the possibility to use alternative tools such as QSAR, in vitro studies, and read across between analogues).
In discussing each of the possible ways to adapt the standard testing requirements it must be understood that outside of the minutes of Member State Committee meetings and progress reports from ECHA on the implementation of REACH there is no publically available registry of what has been accepted or rejected by ECHA. Nor is there a publically available 'Manual of Decisions' that can provide registrants with an insight into what is successful and what is not. It is recommended that these types of documents be made publically available to allow registrants to understand what meets the ECHA internal validation criteria for a robust justification for adaptations to standard testing requirements. However, even though this type of information is not publically available yet, the minutes of the Member States Committee meetings highlight a number of interesting cases that have been discussed and give an initial impression of how successful adaptations have been.
Not Technically Feasible
Considering each of the possibilities to adapt the testing requirements, demonstrating a test is not technically possible is probably the most straight-forward. In order to make use of this approach the registrant must provide a convincing argument that a test cannot be conducted for technical reasons. Many arguments are built around the physical chemical properties of the substance. For example, if a chemical is flammable in contact with water or explosive then one could argue that any toxicological or ecotoxicological study would be inherently dangerous for the technicians and inhumane for the test organisms. It also follows that the reactivity of the substance would limit the potential for human or environmental exposure that could result in an adverse effect. However, arguing that a substance cannot be tested in an in vivo test system due to its corrosivity is not as straight forward. For instance, if the substance is corrosive due to a high pH but is used in formulations with a neutral pH, it could be argued the predominant exposure would be to a neutral material which may have additional intrinsic toxicological properties. Therefore, the registrant should consider the feasibility and relevance of conducting the test with a neutralised material before attempting to argue the study is not technically feasible. When utilising the 'not technically feasible' approach for waiving studies, it is therefore important to consider other aspects such as the potential for exposure (and its relevance) to humans and the environment and the possibility to modify the test system before simply waiving the study.
This leads rather nicely on to the possibility to modify the data requirements by developing 'Substance-tailored exposure-driven testing'. As stated earlier, the data requirements set out in the REACH regulation are linked to the annual manufacturing and importing volume in tonnes, which serves as a proxy for exposure. In some cases the simplistic use of tonnage as an indicator of exposure is inappropriate, for example, if you have a high volume substance but the actual exposure to the environment or human population is very low due to the way the substance is used or handled. In these circumstances it can be argued that the required level of knowledge about the intrinsic hazardous properties can be reduced. This reduction in data requirements is illustrated by the case of a Strictly Controlled Isolated Intermediate. If the registrant can provide sufficient evidence that a chemical intermediate is rigorously contained throughout its life cycle, such that human and environmental exposure is highly unlikely, they can register the chemical intermediate with a minimal dataset. For a >1000 tonnes per year intermediate, this approach reduces the data requirements to those of a 1-10 tonnes substance. For substances that do not meet the definition of an intermediate, it is also possible to propose waiving of studies by arguing that the substance is handled under strictly controlled conditions. Within the REACH legal text there are criteria that must be met if this approach is taken and in many cases this requires substantial documentation to prove that the substance is rigorously contained throughout its entire life cycle. This approach does not require the registrant to perform an exposure or risk assessment to demonstrate the absence of exposure, relying instead on the documented description of how the substance is rigorously contained throughout its life cycle. To date, there have not been any cases where this approach has been discussed at the Member State Committee meeting and so it is not possible to state whether this option has been utilised successfully. However as more dossier evaluations are performed by ECHA it is likely that cases where this approach has been taken will be challenged and precedents will be set regarding the nature of documentation required to support it.
Alternatively, it is possible to perform an exposure and risk assessment to support the claim that there is no significant exposure. This approach requires some toxicological data so that the registrant can set a derived no effect level (DNEL) for the purposes of the risk assessment. Unfortunately, although there is some guidance available on what is necessary, it is certainly not a straightforward approach. Specifically, if a registrant considers waiving a 90-day study, the annex XI of the REACH legal text states in a footnote that a risk assessment utilising a DNEL generated from a 28-day study would not be acceptable. Similarly, if a developmental toxicity study or a multi-generation toxicity study is to be waived, then a DNEL taken from a reproductive toxicity screening study (OECD test guideline 421 or 422) would not be sufficient. This makes it somewhat difficult to determine what data can be used in setting the DNELs for this approach. This is evident in one case where exposure based waiving was used and subsequently presented to the Member States Committee for discussion. The registrant argued to waive a reproductive screening study based no significant exposure and provided an exposure assessment to support this. In the minutes of the meeting, the committee stated that the exact definition of the terms 'significant exposure' and 'well below DNEL' in the REACH text were missing and that the guidance does not explain these concepts. It appears from the minutes that a decision on what is considered acceptable was not made, however this did not preclude the conclusion that in this case the arguments to use exposure based waiving were insufficient. As such it appears that while there may not be a conclusive definition of what is 'significant exposure' or 'well below the DNEL', the outcome of this case appears to set a precedent for what does not meet the definition. This is just one case, and although there is likely to be more clarity on this issue in the future, it does illustrate the challenges posed in utilising an exposure based waiving approach today. It is recommended that ECHA and the Member States Committee attempt to reach a consensus on these key terms as soon as possible to allow registrants to build robust arguments in support of this approach, subsequently increasing the success rate of this approach in minimizing the use of animal studies.
The final possibility for adapting the standard data requirements is based on the conclusion that the testing is 'Scientifically unjustified'. A registrant may argue that a test is not necessary because other data can be used to address the data requirements. As such, any new study would be considered as scientifically unjustified because it will not add to the hazard characterisation process. The justification can be prepared using a weight of evidence approach, the use of read across, the use of quantitative structure activity models (QSAR), or available in vitro methods. Unfortunately the utility of the latter two tools as 'stand alone' approaches is limited given that the recent ECHA report on REACH states that QSARs and in vitro tools cannot be used as one-to-one replacements for the larger, more animal intensive studies such as repeated dose and reproductive/developmental toxicity. They can, however, be used as part of a weight of evidence approach, or to reinforce the grouping of substances into a category (ECHA, 2012).
The weight of evidence approach and the grouping of substances into a category, therefore, represent the main opportunities for arguing a study is scientifically unjustified. One can use these approaches on their own or, more effectively, in combination. The goal being to convince the regulatory community that the available data allows the robust hazard characterisation required under REACH.
The use of read across and weight of evidence are not new approaches. These techniques have been used successfully for many years in the US EPA and OECD High Production Volume chemical screening programs (OECD HPV), the European Existing Substances risk assessment program, and European decisions on harmonised classification and labelling of substances. As such there is a significant history of how to successfully define and describe chemical categories and how to use data within the category to characterise the hazards of the members. Although in the past the assessment of categories was relatively 'ad-hoc', guidance on how to build and report categories was developed at the end of the 1990's (Rosenkranz and Cunningham, 2001) and utilised in the OECD HPV program. The latest version (OECD, 2007) is currently under revision to include updates to the use of QSAR tools and 'adverse outcome pathways' in building categories. This 2007 OECD HPV guidance document on categories was essentially adopted as the guidance on how to utilise categories and read across under REACH. As such there is consistency between the use of categories under REACH and under the OECD HPV program.
Given this consistency one would expect that the use of categories and read across in REACH registration dossiers would be relatively straight forward, particularly where the categories have been previously reviewed and accepted by the US EPA or OECD HPV programs. However experience with REACH is showing that the previous acceptance of a read across within a category under the OECD HPV program does not guarantee its acceptance under REACH. An example of this is a case where the final decision from ECHA and the Member State Committee is currently the subject of an Appeal.
The case in question involved the grouping of two substances in order to use data from one to address the data requirements (repeated dose and developmental toxicity) of the other. The basis for the read across was one substance being rapidly metabolised in vivo (through hydrolysis of an ester bond) releasing two metabolites; acetic acid, and the read across substance, i.e. the parent substance was the acetate of the metabolite. It was argued that systemic exposure would be predominantly to the metabolite and therefore the toxicity data on the metabolite could be used to assess the sub-chronic and developmental toxicity of the parent molecule. Although the use of read across between these substances had been accepted under the OECD HPV program, ECHA and the Member State Committee came to a different conclusion. They considered that the use of read across from the metabolite to the parent compound for the developmental toxicity endpoint left sufficient residual uncertainty and therefore requested a new developmental toxicity study on the parent compound. However, they concluded that there was sufficient weight of evidence to allow/support the use of read across for the 90-day study endpoint. The justification for rejecting the read across argument for the developmental study was that the rate of hydrolysis, as available from an in vitro study on the parent compound, was not sufficiently rapid to preclude systemic exposure to the parent compound, which may have additional hazardous properties not predictable from the data on the metabolite. For the 90-day study, they concluded that 28-day repeated dose toxicity data available for both the parent and the metabolite were consistent, providing sufficient weight of evidence to address the endpoint without the need for an additional study.
While the decision may be debatable, it certainly pushes the balance of addressing the remaining uncertainty and avoiding animal testing towards doing more testing. In the above case, one of the critical points when assessing the read across was that the metabolism data alone was not sufficient. It therefore became necessary to assess the toxicological consistency of the parent and metabolite by comparing the available shorter term or screening studies. For the repeated dose endpoint there was a 28-day study that could be compared. However, for developmental toxicity there was no reproductive or developmental screening data on the parent compound. The Committee therefore concluded that they could not exclude the possibility of a difference between parent and metabolite, thus requesting that the developmental toxicity study be performed.
The decision on the above case also demonstrates that acceptance of read across for one endpoint does not guarantee acceptance for another. This is in line with the guidance; however, it presents an additional challenge to registrants to ensure that their arguments supporting read across are sufficiently robust for each endpoint 'on their own'.
In addition, the read across for developmental toxicity was rejected even though it had been previously accepted under the OECD HPV program where many of the EU member states are represented. Although one could argue that there are differences between the aims of the OECD HPV program and REACH, both address the hazard characterisation of the substances reviewed and both rely on essentially the same guidance document to assist registrants in preparing a category. If ECHA and the Member State Committee will not accept read across arguments already accepted under programs like the OECD HPV program, then registrants are left with significant uncertainty about what else is needed over and above the requirements of the OECD HPV program to meets the needs of ECHA, while relying on a guidance document that is applicable to both programs.
This case was the first to be discussed in the Member State Committee and highlighted that, similar to the exposure based waiving issue, the authorities did not yet have a consensus on what was acceptable as justification supporting read across. Again, this decision to reject the read across sets a precedent regarding what is not acceptable even if it does not shed light on what is. It also triggered additional activities within ECHA to build a framework for assessing read across, the so called 'Read Across Assessment Framework' (RAAF). This RAAF is currently in preparation and will guide ECHA staff and the Member State Committee through the future assessment of read across arguments presented in REACH dossiers. It is noteworthy that the development of RAAF so far has not involved stakeholders external to ECHA and it is unclear if there will be the opportunity to provide input prior to it being finalized. Since the above case was discussed at the Member State Committee there have been several other cases discussed and decisions on the acceptability of read across, waiving arguments, and testing proposals have been made. In many cases, the outcome of the discussions is that new studies are required, although, there have been some occasions where the committee has recommended assessing the potential for using read across to similar substances prior to testing.
In light of the decisions being made today on the dossiers for the large tonnage substances, one important issue for the future is how the use of read across can be used for the lower tonnage substances. Consider a registration for a 10-100 tonne substance where the only repeated dose and reproductive toxicity studies required are a 28-day study and a reproductive developmental screening study (these can also be combined into a single study - OECD 422). In this situation there are no 'lower tier' or shorter studies that could provide support for a read across argument for repeated dose or reproductive toxicity endpoints. Therefore, it is difficult to predict how ECHA and the Member State Committee will address read across between these lower volume substances given the potential 'residual uncertainty' that would be inherent to this approach under these circumstances.
Responsibility of Registrants for Minimising Animal Testing
Although the above assessment may give the perception that ECHA and the Member State Committee can be conservative in their assessment of the use of alternative approaches, they are certainly not responsible for every failed attempt to use them. It should be recognised that waiving studies can appear to have significant cost and time benefits to registrants faced with a large number of substances to register and a relatively short time in which to do it. Preparing a waiving argument that allows a registrant to avoid the conduct of a 2-generation reproductive toxicity study, a 90-day study, and a developmental toxicity study can save close to one million dollars in testing costs (in addition to approximately 5000 animals). As such, it can be seen (perhaps inappropriately) as a quick and cheap way to register a substance. This has led to many examples of dossiers submitted in 2010 where justifications for read across, waivers, etc. are less than robust. For example, in their recent report on REACH (ECHA, 2012) ECHA has commented that many of the dossiers examined to date that have used read across do not provide the necessary robust support for it, and in some cases are limited to single sentence arguments such as 'the substances are structurally similar and therefore read across is used'. Therefore, it is clear that not all registrants have approached the use of adaptations to test requirements in the same robust manner and may have used them inappropriately or without sufficient justification.
Although the potential financial benefit of utilising approaches like read across or exposure based waiving are difficult to ignore, it is important that registrants understand that by failing to invest in robust arguments to defend the use of these tools, there is the possibility that their use will be rejected and the tests demanded. This has several significant consequences, firstly the alternative approaches may have been valid if robustly supported and so the rejection leads to a potentially unnecessary test. Secondly, REACH requires registrants to have legitimate access to the data they need, and therefore must purchase this access from data owners. Compiling a category may therefore require a registrant to pay for access to many additional studies or conduct 'bridging studies' in order to provide all the necessary data in their dossier. This may not be as expensive as running an entire new set of studies, but it can be far from cheap. Therefore, if ECHA were to subsequently reject the read across and request studies, the registrants are left in the position that access to data has been purchased and yet they must now pay more for new studies. This can be hard to swallow and may deter future attempts to use tools like read across. Finally, and perhaps more importantly, there is a reduction in the credibility of the registrants where poor argumentation is submitted in support of waivers such that future assessments of dossiers may be viewed in a more negative light.
Conclusions and Recommendations
Overall it could be argued that to some extent the aims of the REACH legislation are difficult to reconcile. On the one hand it demands a significant set of toxicity studies, while on the other hand mandates that registrants do everything they can to avoid conducting new studies. To add to this situation is the lack of clarity regarding what is an acceptable argument to adapt the standard testing requirements. It is therefore interesting that the authors of an analysis of the recent ECHA report on the progress on REACH (Spielmann, et al., 2011) conclude that first wave of registrations in 2010 appears to have been successful in delivering dossiers while at the same time minimising the use of new animal studies. As indicated above, although some may believe this to be the case, it is clear that as dossiers are reviewed by ECHA and member state competent authorities not all attempts to avoid new animal testing are proving successful and the initial conclusions regarding how successful REACH has been in this regard are likely to change, a concern echoed by BUAV (BUAV 2011).As such, the general feeling within the Industry is that the decisions on waiving of studies and the use of read across are being addressed rather conservatively and perhaps inconsistently, with more dossiers failing than succeeding. Conversely, the apparent opinion of the regulatory community is that Industry has not done enough to justify their use of alternative approaches to meet the data requirements of REACH. It should be understood, however, that we are still at the start of the journey with many new regulatory processes being followed for the first time. As more experience is gained within both the regulatory and industry community, it is hoped that a greater understanding of what should be required to support the use of alternative approaches under REACH will allow more robust dossiers to be compiled and fewer requests for new animal studies, thus delivering on the goal to increase the level of human and environmental protection without resorting to excessive animal testing.
Some recommendations to improve the outlook for making use of adaptations to the testing requirements are:
- A published manual of decisions to allow registrants to understand what is acceptable and what is not
- Consensus within the regulatory community on what constitutes a robust argument for waiving a study based on 'exposure considerations'
- What is 'Significant exposure' and how low is 'Well below the DNEL'?
- Publication of the RAAF, and openness to review and refinement by stakeholders outside of ECHA
- A review of what data are needed to support the use of read across and how this is affected by the lower data requirements for lower volume substances
- More commitment and greater investment by registrants to make adaptations to standard testing requirements as robust as possible
©2012 Nicholas Ball