Last reviewed 14 June 2016

In this special report, Dr Mel Cooke of Alchemy Compliance Limited discusses the issues of animal welfare and non-animal testing in the determination of chemical hazards for risk assessments. However, he concludes that there is much more work to be done at the international level in making available new regulatory testing methodologies that are not animal-based.


The principle legislation in the EU for determining chemical hazard, risk assessment and communication in the supply chain are as follows.

  • EC Regulation 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), as amended. See (

  • EC Regulation 1272/2008 on classification, labelling and packaging of substances and mixtures (CLP), as amended. See (

Without knowledge of chemical hazard, it is impossible to protect human health and the environment from deleterious effects.

The REACH Regulation mandates a process for compiling information on the intrinsic hazardous properties of all chemical substances in commercial circulation above 1t per year per manufacturer or importer. An industry consortium comprising all chemical companies involved in manufacturing or importing a particular substance must normally produce a single registration for that substance. The information required for the REACH registration depends on the tonnage at which the substance is supplied. The information has to be submitted to European Chemicals Agency (ECHA) and eventually published as a freely available web resource.

According to the CLP Regulation, the main use for the hazard data produced in the REACH process is hazard classification. The CLP Regulation contains criteria for differentiation into:

  • physico-chemical hazard classes (eg flammable liquid, oxidising solid, explosive)

  • toxicological hazard classes (eg acute toxicity, irritation, carcinogenicity)

  • ecological hazard classes (eg dangerous to the aquatic environment).

Usually, categorisation is then broken down into different levels of the hazard. Each hazard class/category has associated pictograms, signal words, hazard statements and precautionary statements that are communicated in the supply chain through labelling and safety data sheets (SDSs).

Laboratory animals have complex biological systems that make them useful surrogates for determining potential hazardous effects in humans. Animal testing was long considered the “gold standard” for generating information useful for risk assessment and classification of the chemical.

However, animal welfare groups have raised ethical questions, particularly on the rights of humans to inflict suffering on other sentient creatures, and there are significant differences in the toxicological responses of laboratory animals and humans that have to be taken into account during chemical risk assessment. There is, therefore, a tension between ensuring human and environmental safety, and maintaining animal welfare.

REACH and CLP refer to Directive 2010/63/EU of the European Parliament and of the Council of 22 September 2010 on the Protection of Animals Used for Scientific Purposes that aims to minimise the use and suffering of animals based on the principle of the three Rs.

  1. Replace the use of animals with alternative techniques or avoid the use of animals altogether.

  2. Reduce the number of animals used to a minimum by obtaining required information from fewer animals or more information from the same number of animals.

  3. Refine experimental procedures to reduce suffering as much as possible.

This report focuses on the requirements within the REACH and CLP Regulations for generating information for chemical safety assessment and classification, and, in particular, on “alternative” strategies that do not involve animal testing.

REACH and animal testing

The main aim of the REACH Regulation is the protection of man and the environment, which is achieved through provisions to better understand the hazardous properties of chemical substances.

Testing of mixtures on vertebrate animals is rarely justified, as the properties of mixtures are assessed through their ingredient hazards and concentrations.

REACH specifies a tonnage-dependent strategy for generating data on substances. REACH Annexes VII to X cover physico-chemical, toxicological and ecotoxicological properties, often referred to as “endpoints”, which must be included in a registration dossier to be submitted to the ECHA.

Typically, a chemical substance supplied at between 10t and 100t per annum will require the toxicological and ecological properties indicated in Table 1 below, which are potentially investigated by animal testing. The welfare provisions within the REACH Regulation refer to vertebrate animals only.

Table 1: Typical information requirements for REACH registration for substances supplied at 10–100t per annum that potentially require animal testing (see REACH Annex VIII)


Animal testing method

Annex VII or VIII ref.

Skin corrosion/irritation


Annex VIII, 8.1

Eye irritation


Annex VIII, 8.2

Skin sensitisation

B.6 or B.42

Annex VII, 8.3

Acute toxicity: oral


Annex VII, 8.5.1

Acute toxicity: inhalation (gases, volatile liquids, dusty solids)


Annex VIII, 8.5.2

Acute toxicity: dermal


Annex VIII, 8.5.3

Repeated-dose toxicity: short-term (28 d), most appropriate route (usually oral)


Annex VIII, 8.6.1

Reproductive toxicity: screening test

OECD 421 or 422

Annex VIII, 8.7.1


Aquatic toxicity: short-term toxicity in fish


Annex VIII, 9.1.3

There is a 2018 deadline in the REACH Regulation for consortia to provide registration dossiers for older “phase-in” substances between 10t and 100t per annum. ECHA expects anywhere between 25,000 and 50,000 substance registrations, so the potential for very large animal use is clear.

The testing methodologies are very rigorous so that results can be considered robust and reliable. The test methods are developed by the Organisation for Economic Co-operation and Development (OECD), and so are internationally recognised.

In EU legislation, the adopted test methods are collected in EC Regulation 440/2008 laying down test methods pursuant to EC Regulation 1907/2006 (REACH), which is amended periodically to include new test methods ( The validation and legal adoption of new test methods are necessarily a long process, but non-animal tests are being fast-tracked for regulatory acceptance.

The European Centre for the Validation of Alternative Methods (ECVAM) of the Commission’s Joint Research Centre (JRC) plays an important role in the scientific assessment and validation of alternative test methods.

For both REACH and CLP, any new toxicological or ecotoxicological tests have to be done under a strict quality control regime called Good Laboratory Practice (GLP; 2004/10/EC), or equivalent, and with due respect for the animal welfare for vertebrate animals. Such testing is usually outsourced to a contract research laboratory, which must apply minimum standards for housing and care, and have a rigorous assessment and justification of suffering caused to the animals.

The REACH Regulation, Article 25, explicitly states that testing on vertebrate animals for the purposes of this regulation shall be undertaken only as a last resort and further explanation is given in Recital 37.

“Implementation of this Regulation should be based on the use of alternative test methods, suitable for the assessment of health and environmental hazards of chemicals, wherever possible. The use of animals should be avoided by recourse to alternative methods validated by the Commission or international bodies, or recognised by the Commission or the Agency as appropriate to meet the information requirements under this Regulation.”

The CLP Regulation has similar provisions.

Therefore, when investigating a particular hazardous property (eg eye irritation, skin sensitisation), the registrant must first consider all other options before commissioning an animal test. These options are described below.

Procedures with REACH for reduction of animal tests

Procedures with REACH help reduce the overall use of animals in testing.

Data sharing

REACH minimises the overall use of animals for testing by obligating companies owning data on a particular substance, which has been generated through testing on vertebrate animals, to share it with all other registrants for the substance.

REACH obligates potential registrants to participate in a consortium (or so-called Substance Information Exchange Forum (SIEF)) to provide a single registration for each chemical substance. This principle of “one substance, one registration” was recently re-enforced in Regulation 2016/9 (see that obliges ECHA to ensure that all registrants of the same substance are part of the same registration. Companies providing data to the SIEF receive compensation for their data from other members of the SIEF.

REACH Article 30:

“Before testing is carried out in order to meet the information requirements for the purposes of registration, a SIEF participant shall inquire whether a relevant study is available by communicating within his SIEF. If a relevant study involving tests on vertebrate animals is available within the SIEF, a participant of that SIEF shall request that study.”

The owner of the data and the new potential registrant are obliged to “make every effort” to reach an agreement on the sharing of the information.

Similarly, if a potential registrant learns from ECHA (following an Article 26 enquiry) that a previous registrant has animal test data for a particular substance, then the two parties are obliged to share that data.

Article 26.3:

“Studies involving vertebrate animals shall not be repeated.”

If the studies that ECHA hold are over 12 years old, then they are not protected and the new registrant can use the data without compensating the original owner.

Test plan scrutiny

Furthermore, for high production volume chemicals (>100t per annum) that require testing according to Annexes IX and X of REACH, the lead registrant, on behalf of all registrants in a joint submission, has to submit a testing proposal to ECHA for testing on vertebrate animals. The testing proposals are made public on the ECHA website and are scrutinised and commented on by interested parties, often non-government organisations (NGOs) with an interest in animal welfare. The registrant has to await ECHA’s decision on the proposal before proceeding with animal tests.

Integrated testing strategies

Rather than go straight to the animal test for a particular hazard endpoint, sometimes a testing strategy is required, involving sequential steps in testing. The main strategies are given in adaptations, as given in the REACH testing annexes (Column 2 of Annexes VII to X or on the more general conditions given in Annex XI of the regulation).

A good example is the testing strategy for the assessment of skin irritation, which comprises several steps, as given in Figure 1 below.

A complete guide to testing strategies is dependent on the properties of the substance and is beyond the scope of this article, but registrants should consult the detailed guidance on the ECHA website or get expert advice.

Figure 1: Integrated testing strategy for determining skin irritation, demonstrating animal testing as “last resort”.

Alternative methods for assessing hazard

Structurally related substances (read-across and grouping)

Animal testing on a substance can be avoided if the target substance, for which there is no information for a particular endpoint and for which a structurally similar chemical, the source substance, is identified for which there is adequate data. This data from the source substance can be “read-across” to the target substance by analogy.

This approach is mandated in the REACH Regulation (Article 13).

“In particular for human toxicity, information shall be generated whenever possible by means other than vertebrate animal tests, through the use of alternative methods, for example, in vitro methods or qualitative or quantitative structure-activity relationship models or from information from structurally related substances (grouping or read-across).”

Justification for the read-across is required. The source and target substances should have proven purity/impurity profiles and have similar physico-chemical, toxicological and ecotoxicological properties in the available data for both substances.

The concept of read-across can be strengthened if one considers a group (also known as a category) of structurally related substances, rather than a target substance and a single source. Since there are more members, the read-across is based on more data points and so is more robust. In this approach, registrants are recommended to include as many structural analogues as possible to demonstrate similarity and trends in the properties of the substances making up the group. Instead of testing every property of every chemical substance, properties of some members of the group can be estimated with an overall reduction in the use of animal testing.

Some typical category types are given below.

  • Homologous series of organic substances, such as alcohols or carboxylic acids, which structures differ only in the number of carbon atoms in the chain (eg hexanol, heptanol, octanol, etc). This is especially useful if the unknown substance lies in the middle of the sequence so that the classification can be interpolated from the properties of other members of the group.

  • Organic salts with differing only in the metal cation, where the hazardous properties are associated with the organic part (eg sodium oxalate, potassium oxalate, calcium oxalate).

  • Substances that have only one functional group that dominates the toxicity profile, such as isocyanates or nickel compounds.

  • The unknown substance rapidly produces the same metabolites in vivo as a substance with known classification, as sometimes occurs for simple esters.

It is fundamentally more difficult for the authorities to assess the quality and robustness of read-across and grouping approaches, and ECHA is systematically assessing read-across in current registration dossier evaluations. This activity called The Read-Across Assessment Framework (RAAF), published by ECHA, provides a framework and guidance for evaluation of a proposed read-across approach.

Qualitative or quantitative structure-activity relationship models

An animal test can be avoided if a hazardous property of a substance can be predicted using structural group analysis or computer modelling. Qualitative or quantitative structure-activity relationship (QSAR) varies from simple calculations to structural alerts to computational or “in silico” predictions, to complex “expert” databases that predict specific toxicological properties. This approach is specifically allowed in the REACH Regulation.

Recital 38:

“The generation of information by alternative means offering equivalence to prescribed tests and test methods should also be allowed, for example, when this information comes from valid qualitative or quantitative structure-activity models or from structurally related substances.”

Since QSAR encompasses a variety of approaches, it is not possible to comment generally on its accuracy and robustness. The registrant is tasked with validating the results, usually through assessment of chemicals similar to the target chemical, but with known hazardous properties and providing justification of the approach in the registration dossier. QSAR is very useful as supporting data in a weight of evidence (WOE) approach (below).

A number of chemicals can be classified on the basis that they contain a structural feature that confers a particular hazard. Most notable of these is the classification of isocyanate-containing substances as inhalation sensitisers. Other structural features have been associated with adverse effects and the United States Environmental Protection Agency (USEPA) provides a useful guide (TSCA New Chemicals Program (NCP): Chemical Categories, Office of Pollution, Prevention, and Toxics (OPPT) USEPA, August 2010.

SARs that are more sophisticated have been developed using computational techniques that examine a chemical structure for features that might be an “alert” for particular properties.

The Danish EPA has used a number SARs to produce an advisory list for classification of dangerous substances (

A popular and freely available programme, Ecological Structure Activity Relationships (ECOSAR), is useful for predicting physico-chemical and ecotoxicological properties, and is a free download available from the USEPA website.

DEREK is a commercial software produced by Lhasa Limited, which is a useful tool for predicting toxicity, particularly skin sensitisation and mutagenicity.

It is tempting to use predictive programmes such as “black boxes” to provide quick and cheap (compared to testing) results. However, REACH guidance emphasises the need to have validated QSAR that has been based on reliable information and can be demonstrated to work for the particular type of substance in question. Detailed guidance is available for the use of QSAR in REACH Guidance on Information Requirements and Chemical Safety Assessment, Chapter R.6: QSARs and Grouping of Chemicals.

In vitro testing

In vitro (literally “in glass” but referring to processes taking place in a test tube, culture dish or elsewhere outside a living organism) experimentation covers a wide range of testing procedures. Some in vitro tests are fully validated and included in the OECD and EU testing methodology, and so have equivalent standing as the respective animal tests provided that the substance falls within the applicability domain of the test. For those methods included in REACH testing at 10–100t per annum (ie for the 2018 deadline for phase-in substances), see Table 2 below. These include methods that should be used as part of an integrated testing strategy (see above).

Table 2: Validated EU methods applicable to REACH testing at 10–100t per annum (see Annex VIII of REACH).


In vitro testing method

Annex VII or VIII ref.

In vitro skin corrosion test


Annex VII, 8.1

In vitro skin irritation test


Annex VII, 8.1

In vitro eye irritation test

B.47, 48

Annex VII, 8.2

Mutagenicity: in vitro gene mutation study in bacteria


Annex VII, 8.4.1

Mutagenicity: in vitro cytogenicity study in mammalian cells or in vitro micronucleus study

B.10 or B.49

Annex VIII, 8.4.2

Mutagenicity: in vitro gene mutation study in mammalian cells


Annex VIII, 8.4.3

There are a number of less-established approaches. These methods can be used only as supportive information for REACH registration, but can be used as part of a WOE (below). Such information can also be useful for classification purposes, but generally requires expert judgment on its applicability and robustness.

For the future, so-called “adverse outcome pathways (AOPs)” may be able to predict hazardous properties by using computation and in vitro techniques to model key biochemical events in the interaction of chemicals with living systems. Such an approach could remove the need to use animals and also allow quick screening of substances for a particular hazard (

Human data (eg epidemiological data or data from human trials)

Any epidemiological data or human data (eg from patch testing of cosmetics) should be taken into consideration. The evidence should be of reliable quality and this may be difficult to ascertain. Such evidence is usually used only when positive to classify a product, rather than negating a classification that would otherwise apply. However, where reliable data is available, it takes precedence over animal data (CLP Regulation, Annex I,

Some online databases, notably the Hazardous Substances Database (HSDB), provided as part of the online Toxnet by the US National Institute of Health (NIH), contain human health effects for chemical substances. In theory, such data can be used for classification purposes, but it is often difficult to assess the reliability. Information about the purity/impurity profile of the test substance, critical for hazard assessment, is often lacking.

WOE, including literature

If the quality of the data is sufficiently robust, it is possible to use a WOE approach and use the literature data in place of the studies for registration and for purposes of classification.

The information should be assessed for quality and reliability, particularly that the test materials and methods are documented. See ECHA Guidance on Information Requirements and Chemical Safety Assessment: Chapter R.4: Evaluation of Available Information.

Such information is often missing from literature reports. However, if several sources state a similar value for a hazard, then credence can be given to the result as a part of a WOE approach. See How to Report Weight of Evidence: Practical Guide 2, ECHA, March 2010. There are many good, accessible resources freely available online.

The OECD has published Screening Information Datasets (SIDS). The SIDS are similar to REACH Annex VIII (10–100t) data requirements. ChemIDplus Advanced ( includes several databases, including the HSDB, CCRIS, and ITER. The main advantage of ChemIDplus is the flexibility in the search criteria and, in particular, that it allows searching by chemical structure.

TOXNET is a collection of databases maintained by the US National Library of Medicine (NLM) that contains information on hazardous chemicals, toxic releases and environmental health.

Registry of Toxic Effects of Chemical Substances (RTECS) is a subscription service that contains data on 160,000 chemicals in the following areas.

  • Primary irritation.

  • Mutagenicity.

  • Reproductive effects.

  • Carcinogenicity.

  • Acute toxicity.

  • Repeated-dose toxicity.

Handbooks tend to be more limited in the number of chemicals listed, but can provide useful information for commodity chemicals. The following handbooks are often cited as reliable compilations.

  • The Merck Index: An Encyclopedia of Chemicals, Drugs, and Biologicals, 15th Edition, M J O’Niel, Royal Society of Chemistry, April 2013.

  • Sax’s Dangerous Properties of Industrial Materials, 12th Edition, R J Lewis, Sr, John Wiley & Sons, Inc, August 2012.

  • Handbook of Chemistry and Physics, 93rd Edition, W M Haynes, Editor-in-Chief, CRC Press, June 2012.

  • Patty’s Toxicology, 6th Edition, E Bingham, PhD, John Wiley & Sons, Inc, September 2012.

There are of course hundreds of other potential sources of chemical data. The above resources are likely to give useful results for most common industrial chemicals. ECHA has produced Guidance on Information Requirements and Chemical Safety Assessment: Chapter R.3: Information Gathering on literature searching for chemical hazard data that give additional sources of data.

Animal tests can be avoided if there is a WOE which points to the likely properties of a substance. This approach may be applied if there is sufficient information from several independent sources leading to a conclusion that a substance does or does not have a particular dangerous property, while the information from each single source alone is regarded insufficient to support this assertion.

This approach is described in the REACH Regulation (REACH Annex XI, 1.2).

“Where sufficient weight of evidence for the presence or absence of a particular dangerous property is available:

  • further testing on vertebrate animals for that property shall be omitted,

  • further testing not involving vertebrate animals may be omitted.”

All available information bearing on the determination of hazard is considered together, such as the results of suitable in vitro tests, animal data, information from grouping, read-across, QSAR, human experiences such as occupational data and data from accident databases, epidemiological and clinical studies and well-documented case reports and observations. The quality and consistency of the data is given appropriate weight, and both positive and negative results are considered in a single WOE determination.


Chemical companies that produce or import chemicals have legal obligations to demonstrate that uses of their products are safe for humans and the environment. This requires them to test their products to determine the potential hazards of ingredients. Animals have complex biological systems that can be used in such testing under strictly controlled conditions to help in hazard determination.

Animal welfare is a key factor in legislation. Chemical companies are obligated to use animals only as a last resort. Available data on vertebrate animals must be shared. Other methods for hazard determination should be used first and include the use of existing data from literature, similar substances, SARs, and non-animal testing. For REACH registration, the hazard data needs to be of high quality, but sometimes a registrant can use a WOE from poorer data if it indicates the same result.

The chemical industry has made tremendous progress in embracing the 3Rs principles of animal welfare. This has occurred through NGOs and animal rights activists influencing public opinion and the regulators. However, there is much more work to be done at the international level in making available new regulatory testing methodologies that are not animal-based, particularly for replacing long-term animal studies as are required to assess repeated-dose toxicity, carcinogenicity, mutagenicity and toxicity for reproduction.