Environmental Research 135 (2014) 156–164

Contents lists available at ScienceDirect

Environmental Research journal homepage: www.elsevier.com/locate/envres

Optimizing the aquatic toxicity assessment under REACH through an integrated testing strategy (ITS) Anna Lombardo a, Alessandra Roncaglioni a, Emilio Benfenati a,n, Monika Nendza b, Helmut Segner c, Sonja Jeram d, Eduard Pauné e, Gerrit Schüürmann f,g a IRCCS – Istituto di Ricerche Farmacologiche “Mario Negri”, Department of Environmental Health Science, Laboratory of Environmental Chemistry and Toxicology, Via G. La Masa 19, 20156 Milan, Italy b Analytisches Laboratorium, Bahnhofstr. 1, 24816 Luhnstedt, Germany c Centre for Fish and Wildlife Health, University of Bern, Postbox 8466, CH-3001 Bern, Switzerland d National Institute of Public Health, Trubarjeva 2, SI-1000 Ljubljana, Slovenia e SIMPPLE S.L., Cr Joan Maragall 1 1r, Catalonia, 43003 Tarragona, Spain f Helmholtz Centre for Environmental Research – UFZ, Department of Ecological Chemistry, Permoserstraße 15, 04318 Leipzig, Germany g Institute for Organic Chemistry, Technical University Bergakademie Freiberg, Leipziger Strasse 29, 09596 Freiberg, Germany

art ic l e i nf o

a b s t r a c t

Article history: Received 26 February 2014 Received in revised form 31 July 2014 Accepted 1 September 2014

To satisfy REACH requirements a high number of data on chemical of interest should be supplied to the European Chemicals Agency. To organize the various kinds of information and help the registrants to choose the best strategy to obtain the needed information limiting at the minimum the use of animal testing, integrated testing strategies (ITSs) schemes can be used. The present work deals with regulatory data requirements for assessing the hazards of chemicals to the aquatic pelagic environment. We present an ITS scheme for organizing and using the complex existing data available for aquatic toxicity assessment. An ITS to optimize the choice of the correct prediction strategy for aquatic pelagic toxicity is described. All existing information (like physico-chemical information), and all the alternative methods (like in silico, in vitro or the acute-to-chronic ratio) are considered. Moreover the weight of evidence approach to combine the available data is included. & 2014 Elsevier Inc. All rights reserved.

Keywords: ITS REACH Aquatic toxicity Alternative methods Hazard assessment

1. Introduction To protect human health and environment, the European REACH (registration, evaluation, authorization and restriction of chemicals) regulation entered into force in June 2007. REACH requires that all the substances produced or imported in Europe above 1 t/y should be registered. The registrants need to supply

Abbreviations: ACR, acute-to-chronic ratios; AF, assessment factor; BCF, bioconcentration factor; BOD5, 5-days biochemical oxygen demand; C&L, classification and labeling; CLP, classification, labeling and packaging; COD, chemical oxygen demand; CSA, chemical safety assessment; EBW, exposure-based waiving; EC10, 10% effect concentration; EC50, 50% effect concentration; ECHA, European Chemicals Agency; ErC50, EC50 in terms of reduction of growth rate; ITS, integrated testing strategy; LC50, 50% lethal concentration; LOEC, lowest observed effect concentration; NOEC, no observed effect concentration; PEC, predicted environmental concentration; PBT, persistent, bioaccumulative and toxic; PNEC, predicted no effect concentration; REACH, European regulation on registration, evaluation, authorization and restriction of chemicals; TD50, disappearance time of 50% of the initial amount of substance; vPvB, very persistent and very bioaccumulative; WoE, weight of evidence n Corresponding author. Fax: þ39 02 39014735. E-mail address: [email protected] (E. Benfenati). http://dx.doi.org/10.1016/j.envres.2014.09.002 0013-9351/& 2014 Elsevier Inc. All rights reserved.

physico-chemical, toxicological, ecotoxicological and environmental information about the substances depending on the tonnage level (REACH Annexes VII–X). Moreover, ECHA (European Chemical Agency) requests from the registrants to perform, depending on the tonnage level, a chemical safety assessment (CSA) to identify and describe the conditions under which the manufacturing and use of a substance is considered to be safe (ECHA, 2009). CSA goes through three steps: hazard assessment, exposure assessment and risk characterization. Our work is confined to the hazard assessment that requires the available information on the substance and its uses. Important aspects are (1) an evaluation of the potential of a substance to cause adverse effects to human health and the environment to derive threshold levels, for example the predicted no effect concentration (PNEC), and (2) an assessment of persistent, bioaccumulative and toxic (PBT) and very persistent, very bioaccumulative (vPvB) properties of the substances. The criteria for the PBT/vPvB assessment are reported in the new REACH Annex XIII (adopted by means of Commission Regulation (EU) No 253/2011 of 15 March 2011). In addition, the registrants should perform the classification and labeling (C&L) of the chemicals as dangerous (or not) on the basis of the criteria given in the CLP (classification, labeling and packaging) directive

A. Lombardo et al. / Environmental Research 135 (2014) 156–164

157

Table 1 Review of the cost and the animal needed in testing for REACH are reported. For fish also the number of estimated tests is given. Guideline

No. of tests (estimation)

Cost per test (€)

No. of animals per test

Source/Ref

Algae Short and long term

OECD 201

4510 4500

Fleischer, (2007) Deliverable 1 of Antares EU project, 2010 parts 4 and 5a

Invertebrates Short term

OECD 202

Long term

OECD 211

3742 2800 13,426 15,000

Fleischer, (2007) Deliverable 1 of Antares EU project, 2010 parts 4 and 5a Fleischer, (2007) Deliverable 1 of Antares EU project, 2010 parts 4 and 5a

Fish Short term

OECD 203

79,707 4193 3200

Long term

OECD 210

OECD 212 OECD 215

a b

7–42 (14)b 42

93,843 26,254 34,000 10,238 7200 16,426 15,500

300–420 (400)b 600 none 120

Rovida and Hartung, (2009) van der Jagt et al., 2004 Deliverable 1 of Antares EU project, Rovida and Hartung, (2009) van der Jagt et al., (2004) Deliverable 1 of Antares EU project, Fleischer, (2007) Deliverable 1 of Antares EU project, Fleischer, (2007) Deliverable 1 of Antares EU project,

2010 parts 4 and 5a

2010 parts 4 and 5a 2010 parts 4 and 5a 2010 parts 4 and 5a

http://www.antares-life.eu/index.php?sec=home&pg=activitiesresults. min–max (normal).

(CLP, 2008). The information needed for C&L and CSA are partly overlapping, but the CSA is not required at the lower tonnage levels. The CLP directive entered into force in 2009 and intends to align the existing EU legislation with the United Nations Globally Harmonized System (GHS). The REACH and CLP information requirements involve an intensive use of animal testing if conventional hazard testing methods are continued to be used. Authority (2010) counted that for a single substance, with no pre-existing data, and no attempt to minimize animal testing, registration could require over 5000 animals, assuming little or no avian testing. Without innovative testing strategies, REACH would cause prohibitive consumption of test animals and also excessive expenses and, furthermore, exceed the given time frame. To prevent animal testing on vertebrates as much as possible, REACH calls to use all available data and alternative methods (ECHA, 2010; REACH, 2006) also according to the Directive 2010/63/EU on protection of animals used for scientific purposes. The ECHA guidance (ECHA, 2012a) defines aquatic pelagic toxicity as the property of a substance to be detrimental to freshwater and marine organisms living in the water column. It is assumed that the aquatic toxicity for pelagic organisms is mainly related to the exposure of a substance in water bodies. Generally, to represent the biota in this compartment, a simplified food chain including three trophic levels is employed for hazard assessment: algae or aquatic plants as primary producers, invertebrates (in particular Daphnia sp.) as primary consumers, and fish as secondary consumers. REACH annexes (VII–X) list the testing requirements from the three trophic levels for chemicals with different production volumes. To satisfy the regulatory requirements, the registrants have to deal with a variety of endpoints, data and information requirements. This can be confusing and expensive. Indeed, in some cases the data necessary to satisfy the requirements of the REACH annexes are different from the data required by C&L or to perform the PBT/vPvB assessment and the CSA (e.g. for poorly water soluble substances REACH annexes require long term toxicity data whereas for C&L short term toxicity data are also required). Table 1 reports some estimation of costs, number of animals needed per test and also the total number of fish necessary to evaluate aquatic pelagic toxicity under the REACH registration. It becomes evident that if full testing is performed for all the

compounds in the registration process, the expenses for the registrants and the number of animals used will be too high to be acceptable. The ways forward are integrated testing strategies (ITSs). A scientifically accepted definition of ITS still does not exist (Hartung et al., 2013). However, any ITS aims to optimize the testing of chemicals with regard to costs (money, time, animal welfare, etc.) in relation to the information gain to come to a robust conclusion about a particular hazard of a chemical. In this paper, we present an ITS for the aquatic pelagic toxicity in the form of a flow chart to combine different options of testing and non testing methods. Our ITS can guide registrants to pool information from different sources in a weight of evidence (WoE) approach to obtain an efficient hazard assessment as required by REACH. Some ITSs for aquatic toxicity already have been proposed in the ECHA guidance (ECHA, 2012a, 2012b). These schemes address the individual REACH requirements for C&L, CSA and PBT assessments separately, but they do not consider the complete spectrum of the assessments as a whole. It is our objective to build a combined ITS for the different aquatic toxicity assessment demands under REACH based on a wide range of building blocks in terms of available data and methods. In this paper we present an ITS for aquatic pelagic toxicity, developed within the EU Integrated Project OSIRIS.1 The ITS addresses the full range of REACH requirements (information necessary to satisfy the annexes and to perform C&L, PBT and CSA) using in silico, in vitro and in vivo methods. According to Annex XI of the REACH regulation (REACH, 2006), different methods to reduce animal testing are considered: the use of existing data (physico-chemical, human, ecotoxicological and environmental data), WoE evaluations, in silico methods (like (Q) SAR ((quantitative) structure-activity relationship) or read-across) and in vitro methods.

2. Material and methods The ITS presented here is based on the ECHA Guidance for the implementation of REACH (ECHA, 2012a, 2012b), the ECHA Guidance on the Application of CLP

1 OSIRIS. EU Project, contract no. GOCE-CT-2007-037017, 2007-2011. OSIRIS project website: http://www.osiris-reach.eu/. OSIRIS web tool website: http:// osiris.simpple.com/OSIRIS-ITS/welcome.do.

158

A. Lombardo et al. / Environmental Research 135 (2014) 156–164

Table 2 Summary of the REACH requirements for aquatic pelagic toxicity. In bold the data not required at lower tonnages are indicated. ErC50 stands for a 50% reduction of algal growth rate, EC50 for the 50% effect concentration, LC50 for 50% lethal concentration and NOEC for no observed effect concentration. t/y

Requirements Algae/aquatic plants

Invertebrates

Fish

Degradation

1–10

REACH Annex VII C&L

EC50a

–

Ready biodegradability

EC50 þ NOEC or rapid ErC50 þ NOEC or rapid degradation degradation studies þbioaccumulation data studies þ bioaccumulation data (or log Kow data) (or log Kow data) – – – – ErC50 EC50a

LC50 þ NOEC or rapid degradation studies þbioaccumulation data (or log Kow data) – – LC50a

Rapid degradationb

ErC50 þ NOEC or rapid degradation studies þbioaccumulation data (or log Kow data) NOEC

LC50 þ NOEC or rapid degradation studies þ bioaccumulation data (or log Kow data) NOECe

10–100

PBTc CSA (PNEC)c REACH Annex VIII C&L

PBTc c

4100

ErC50

CSA (PNEC) REACH Annex IX

NOEC/ErC50 ErC50

C&L

d

EC50þ NOEC or rapid degradation studies þ bioaccumulation data (or log Kow data) NOECe d

d

NOEC/EC50 EC50 or NOEC (depending on CSA requirements)a

NOEC/LC50 LC50 or NOEC (depending on CSA requirements) a

PBTc

ErC50 þ NOEC or rapid degradation studies þbioaccumulation data (or log Kow data) NOEC

EC50þ NOEC or rapid degradation studies þ bioaccumulation data (or log Kow data) NOECe

LC50 þ NOEC or rapid degradation studies þ bioaccumulation data (or log Kow data) NOEC e

CSA (PNEC)c

NOEC/ErC50d

NOEC/EC50d

NOEC/L50d

– – Ready Biodegradability þ hydrolysis studies Rapid degradationb

Half life for soil, water (marin and fresh), sediment (marine and fresh) – Ready Biodegradability þ hydrolysis studies þ biotic degradation tests (depending on CSA requirements) Rapid degradationb

Half life for soil, water (marin and fresh), sediment (marine and fresh) –

a

Long term values shall be considered in case of poorly soluble substance. Rapid degradation includes different kind of measurements like ready biodegradability, BOD5/COD (5-days biochemical oxygen demand/chemical oxygen demand) or other convincing scientific evidence (e.g. hydrolysis of photochemical degradation). c These are general requirements. Specific requirements should be evaluated on a case by case bases. d For CSA, the PNEC values should be calculated on the available data. The long term values are preferred (ECHA, 2008). e If it is the most sensitive species. b

criteria (2012) and the REACH annexes (VII–X). The requirements are summarized to help the reading of the ITS (see also Table 2). On the basis of these requirements, the proposed ITS is described in Section 3.

indicates the need to investigate further the effects on aquatic organisms (ECHA, 2012a). 2.4. Substances produced above 1000 t/y (reference REACH annex: X)

2.1. Substances registered below 10 t/y (reference REACH annex: VII) No additional data is required for aquatic toxicity. Short term toxicity on invertebrates and growth inhibition on algae or aquatic plants are required unless mitigating factors (e.g. the substance is highly insoluble in water or the substance is unlikely to cross biological membranes) are justifiable or, for invertebrates, long-term aquatic toxicity studies are already available (ECHA, 2012a) or preferred (i.e. if the substance is poorly soluble in water; REACH Annex VII). At these tonnages, PBT and CSA are not mandatory but C&L requires at least short term toxicity data for the three trophic levels. If long term toxicity data are not available, short term toxicity data for the three trophic levels can be used together with information about degradation and bioaccumulation to evaluate the chronic toxicity class of the compound. If information for all the three trophic levels are not complete, the assessment can be performed on the basis of the available data.

2.5. The ITS In this work we present an ITS to combine these requirements with the suggestions of Annex XI (see Introduction Section 1) and available information to evaluate if existing data satisfy the information requirements and, in the case of data gaps, guide the selection of (non)testing methods to produce the necessary information most efficiently regarding number of animals, costs and time. The ITS was produced balancing the REACH requirements and the available methods and information. We started from the ITSs described in the ECHA guidance (2012a and b), trying to combine them and to add new strategies and methodologies to fill the data and information gaps.

2.2. Substances registered between 10 and 100 t/y (reference REACH annex: VIII) In addition to the requirements for the substances registered below 10 t/y, short term toxicity testing on fish should be performed unless mitigating factors are justifiable or long term toxicity data are already available or preferred (i.e. if the substance is poorly soluble in water; ECHA, 2012a). Above 10 t/y also PBT assessment and CSA are needed in addition to C&L. To perform C&L the same data as below 10 t/y are necessary: at least short term toxicity data on all the trophic levels. The new REACH Annex XIII (Commission Regulation (EU) 253/2011) describes how to assess if a compound is PBT or vPvB. The assessment is based on long term toxicity data, whereas short term toxicity data can only be used as screening information. About CSA, if further effects on aquatic organisms have to be investigated, on the basis of short term data, long term testing shall be considered. 2.3. Substances produced between 100 and 1000 t/y (reference REACH annex: IX) In addition to the requirements for the substances produced between 10 and 100 t/y, long term toxicity testing on invertebrates and fish is required if the CSA

3. Results and discussion In the following paragraphs, we describe the ITS and discuss the reasoning underlying the various steps and branching points of the flow chart (see Fig. 1 for the conceptual scheme of the ITS and Appendix A for the complete ITS). Whenever information on toxicity is required (short and long term), we suggest to follow six questions: 1. Which data are existing? 2. Which physico-chemical and environmental data are required and existing (e.g. water solubility and degradation)? 3. In case of lacking experimental data, can they be generated using in silico methods ((Q)SAR and read-across)?

A. Lombardo et al. / Environmental Research 135 (2014) 156–164

159

Fig. 1. The conceptual scheme of the ITS for the assessment of aquatic pelagic toxicity.

4. Can lacking toxicity data be predicted using acute-to-chronic ratios (ACR) (both to generate chronic values from existing acute toxicity data or vice versa)? 5. Can in vitro methods provide toxicity information (unfortunately, this possibility is currently only a theoretical option because in vitro methods are not yet validated)? 6. Is it still necessary to perform in vivo tests (they should be used only as last resort and generally require authorization)? When reliable short term toxicity data are not available but long term toxicity data are available and sufficient for REACH annexes requirements, PBT/vPvB assessment and CSA may be based on long term data (using the ACR) or in silico methods. For the C&L (that needs short term toxicity values), testing methods (both in vitro and in vivo) are not suggested because no new tests should be performed to satisfy C&L requirements. The proposed strategy is based on the REACH Annex XI suggestions but it is also integrated with new approaches, like the use of ACR. The ITS (see Appendix A for the complete flow chart), described in more detail in the following paragraphs, starts with the evaluation of existing data (i.e. degradation time) and mitigating factors. If available, this information allows deciding if the compound needs an aquatic toxicity evaluation or could be considered “safe” for the aquatic pelagic environment without further analysis. After this initial evaluation, we suggest retrieving the data necessary for the C&L and to satisfy the REACH requirements, i.e. toxicity values for the three trophic levels depending on the tonnages and the water solubility. This information can be also used to perform a preliminary CSA, but, before to proceed with the CSA refinement (that needs long term toxicity data), we suggest

evaluating the PBT/vPvB criteria because they can be used for the risk identification that is necessary to decide if the CSA refinement is necessary. Moreover the information retrieved for the PBT/vPvB assessment can be used for the CSA refinement. 3.1. General requirements The first step of the ITS goes through the analysis of degradation time and possible mitigating factors. The analysis of degradation time is suggested in the ECHA guidance (2012a) whereas the mitigating factors are reported in the REACH Annexes VII–X. For both we suggest some methods and strategies on the basis of the available methods. 3.1.1. Degradation time The aquatic toxicity assessment may be performed on the parent compound or on the degradation products (that could be more toxic). If the substance degrades (in terms of disappearance time, DT50) in less than 1 h the breakdown products should be analyzed instead of the parental; the parent compound should be evaluated if it degrades in more than 3 days. If the degradation time is between 1 h and 3 days, a case by case decision has to be made (ECHA, 2012a). We furthermore suggest considering at this point the possibility of pseudo-persistence, i.e. compounds may be rapidly degraded but due to continuous release into the environment their environmental levels may be high. The issue of pseudopersistence may be relevant for CSA and PBT assessments, but not the other REACH requirements since it combines information on hazard and exposure, thus risk. Information about degradation

160

A. Lombardo et al. / Environmental Research 135 (2014) 156–164

rates are due at any tonnage: below 10 t/y ready biodegradability studies, between 10 and 100 t/y ready biodegradability and hydrolysis studies, over 100 t/y should be added also further biotic degradation testing (e.g. simulation testing on ultimate degradation in surface water) depending on CSA requirements. So information on the degradation time is usually available or can be obtained using in silico methods. The suggested laboratory tests are the OECD 111 (2004), the OECD 301 (1992), the OECD 303 (2001), the OECD 308 (2002) and the OECD 309 (2004), whereas the examples of in silico methods are ChemProp (http://www.ufz. de/index.php?en=10684), EPISuite HydroWin or BioWin (http:// www.epa.gov/oppt/exposure/pubs/episuite.htm). 3.1.2. Mitigating factors The mitigating factors reported in the REACH annexes (that are supposed to indicate that aquatic toxicity is unlikely to occur) are applied if, for instance, a substance is highly insoluble in water or unable to cross biological membranes. This block is qualitative in the present ITS scheme, and so the user should decide case by case if a mitigating factor is justifiable or not, using a WoE approach. As yet, robust cut off limits for the mitigating factors are not established. Indeed, ECHA states that no scientific basis exists to define molecular characteristics linked to the ability of a substance to cross biological membranes (ECHA, 2012a) and no thresholds are available to identify substances with very low solubility in water. Despite these difficulties, some tools or methods are suggested. Among the in silico methods (in particular to evaluate the solubility of the substance) we suggest ChemProp (http:// www.ufz.de/index.php?en=10684) or a combination of CLOGP (a commercial software to estimate log Kow), ACD/pKa (a commercial software to calculate pKa), and TPSA (for topological polar surface area estimation) to estimate the permeability of compounds (Nakao et al., 2009). In vitro methods are suggested to estimate uptake (exploratory level; e.g. PAMPA (Kwon and Escher, 2008) and in vitro epithelia (Masungi et al., 2004, Wood et al., 2002)). 3.2. C&L and REACH annexes If the registrant wants to proceed with the substance assessment, algae or aquatic plants growth inhibition values (ErC50, 50% effect concentration in terms of reduction of growth rate) are necessary. Since the reference test for algae (OECD 201, 2011) and aquatic plants (OECD 211, 2012) allows to obtain both short (ErC50) and long term (LOEC (lowest observed effect concentration) or NOEC (no observed effect level), statistically determined) toxicity values at the same time, we suggest to obtain both values. These values can be used, at any tonnage, to satisfy REACH annexes requirements, C&L, PBT assessment and CSA. The next steps depend on the tonnage of production/ importation of the chemical as well as on water solubility. This information is fundamental to decide if short or long term data should be preferred. Indeed, for poorly soluble compounds (i.e. water solubilityo1 mg/l; ECHA, 2012a) we suggest obtaining long term toxicity values also at low tonnages. Indeed, REACH annexes report that long term testing shall be considered for poorly water soluble substances and if long term toxicity data are available, no short term toxicity data are necessary. Therefore, at this level, it is important to evaluate the water solubility of the substance. This property is required for all chemicals (unless specific mitigating factors are applicable). If no experimental data are available, the following in silico tools are suggested: Alogps 2.1 (http://www.vcclab.org/lab/alogps/), EpisuiteWsKow (http://www.epa.gov/oppt/exposure/pubs/episuite.htm), Episuite-WATERNT (www.epa.gov/oppt/exposure/pubs/EPISuitedl. htm) or ChemProp (http://www.ufz.de/index.php?en=10684). The reference tests are the OECD 105 (1995) or the EU method A.6, 2008.

In this ITS four cases were identified. Case 1. Substance poorly soluble in water and produced/imported below 10 t/y. In addition to toxicity data on algae/aquatic plants, toxicity data for invertebrates are needed. Long term toxicity data are preferred for poorly water soluble substances. If reliable long and short term toxicity data are not available the registrant can obtain long term toxicity data using in silico, in vitro or in vivo methods (in this order as explained above). If long term toxicity data are available, they satisfy REACH annexes requirements but they are not sufficient for C&L that also need short term toxicity data (according to CLP legislation). Since no new tests should be done for C&L and no information for PBT/vPvB assessment and CSA are needed below 10 t/y, we suggest to verify (a) if the compound is already classified according to older legislation (i.e. the Directive 67/548/EEC or 1999/45/EC) amended and repealed by CLP Directive or (b) if the compound can be classified on the basis of the algae/aquatic plants data (i.e. short term toxicity on algae/ aquatic plantso1 mg/l). Indeed, it is sufficient that the compound is very toxic (i.e. category acute 1 (E(L)C50o1 mg/l) or category chronic 1 (NOECo0.1 mg/l)) for one of the three trophic levels representing the aquatic environment to be considered very toxic for the entire aquatic environment. This information is not sufficient to determine the M-factor (the multiplying factor, that should be set for all the compounds in category Acute 1 or Chronic 1), that needs toxicity data (acute and chronic) on the most sensitive species. To obtain the necessary information, we suggest to derive the short term toxicity data from the long term toxicity data (i.e. ACR, Hoff et al., 2010) or to use QSAR or read-across methods. Case 2. Substance soluble in water and produced/imported below 10 t/y. Only short term toxicity data on invertebrates are required in addition to algae/aquatic plant data by REACH annexes, C&L can be performed using the available data and no information for PBT/ vPvB assessment and CSA is needed (because of the low tonnage of production/importation). For this reason, if reliable short term toxicity data on invertebrates are available, no other data are needed. If they are not available, before obtaining short term toxicity data on invertebrates (using always first in silico methods, then in vitro and only as last resort in vivo tests) we suggest verifying if long term toxicity data on invertebrates are available. In this case they are sufficient to satisfy REACH annexes requirements (REACH annexes report that if long term toxicity data are available, the short term ones are not due) but not for C&L, so we suggest using only non-testing methods like ACR (Hoff et al., 2010) or in silico methods to fill the data gaps. Case 3. Substance soluble in water and produced/imported above 10 t/y. In this case short term toxicity data for both invertebrates and fish should be provided to satisfy REACH annexes and C&L requirements. These data can also be used to perform a preliminary CSA and the PBT/ vPvB assessment. If short term toxicity data for invertebrates are not available we suggest deriving them from the long term test results (if already available and possible) or to obtain them using alternative methods where possible. If long term (but not short term) toxicity data for fish are available, they are sufficient for REACH annexes requirements but for C&L the short term toxicity values are necessary (but no new test should be performed for CLP, as already explained). For this reason in the scheme it is indicated to verify (a) if the substance is already classified (according to the old legislation, the Directive 67/ 548/EEC or the 1999/45/EC, amended and repealed by CLP Directive) or (b) if it is possible to classify it using the available data (i.e. at least one of the short term toxicity data with algae/aquatic plants or invertebrates is below 1 mg/l). As already explained, this is sufficient for the classification into an acute or a chronic category but not for the determination of the M-factor. In any case, the use of ACR (Hoff et al., 2010) or in silico methods is suggested to fill the data gap.

A. Lombardo et al. / Environmental Research 135 (2014) 156–164

If no toxicity data for fish are available the scheme indicates a procedure, based on WoE approach, to choose the best short term test. To apply this procedure, two conditions should be satisfied: (1) the substance can be classified in Verhaar class 1 (non-polar narcotic compounds) or 2 (polar narcotic compounds) (Verhaar et al., 1992) and (2) experimental short term toxicity data for algae/aquatic plants and/or invertebrates are available. In this case, the registrant can use logP-based QSARs to calculate acute toxicity values for algae/aquatic plants and/or invertebrates. Indeed, logPbased equations are considered good for Verhaar class 1 and 2 (Nendza et al., 2014). The next step of this procedure is the check of the appropriateness of these equations for the analyzed compound. To do this, we suggest verifying if experimental and predicted values (with logP-based QSARs) for algae/aquatic plants and/or invertebrates are in agreement. In this case QSAR or readacross methods are suggested to derive the short term toxicity data for fish; otherwise an integrated use of in silico and in vitro methods can be used. If they give reliable results, the obtained information can be used to perform the C&L; otherwise in vivo methods are necessary. If experimental short term toxicity data for algae/aquatic plants and/or invertebrates are not available we suggest using an integrated value from in silico and in vitro, if it is reliable, otherwise in vivo test. The use of integrated values from in silico and in vitro methods is suggested to increase the reliability of the prediction since the use of different methods (if in agreement) gives a higher safety. The preferred in vivo test is the OECD 203 – fish acute toxicity test (OECD, 1992) that includes a limit test (that uses less animal and time). It is performed using as exposure concentration the lowest EC50 from short term tests on algae/aquatic plants and invertebrates, as reported in the OECD 126 Guidance – short guidance on the threshold approach for acute fish toxicity (OECD, 2010). According to this guideline, we suggest performing the limit test at the lower LC50 concentration of previous tests on algae and invertebrates and only if mortality is observed to perform the full test in case the results of step down procedure would not be conclusive. Instead of the OECD 203 that is conducted on adult fish, the OECD 236 (2013) – fish embryo acute toxicity (FET) test could be performed. Case 4. Substance poorly soluble in water and produced/imported above 10 t/y. Long term toxicity data for invertebrates and fish shall be considered when a substance is poorly soluble in water. In this scheme, if long term toxicity data for invertebrates (and/or fish) are available, they can be used, otherwise we suggest verifying if reliable short term toxicity data on invertebrates (and/or fish) are available. In this case they can be used to perform C&L and to satisfy REACH annexes requirements. If they are not available, long term toxicity data should be obtained trying preferably alternative methods first. To perform C&L, the short term toxicity data can be derived from the long term (ACR, Hoff et al., 2010) or obtained with in silico methods. As for Case 3, all the available values can be used to perform a preliminary CSA and the PBT/vPvB assessment. Table 3 summarizes some methods (in silico, in vitro and in vivo) to obtain short or long term toxicity values for algae/ aquatic plants, invertebrates and fish. Some software programs to classify the compounds according to the Verhaar classification scheme are also reported. Only freely available and commonly used methods were selected on the basis of Nendza et al. (2013). 3.3. Preliminary CSA At this point the registrant can also calculate the PNEC (predicted no effect concentration) and the PEC (predicted environmental concentration) for the CSA using the available information.

161

Only if a risk is identified (see paragraph Section 3.5) the CSA should be refined using long term toxicity data on the most sensitive species of invertebrates or fish. Since PBT/vPvB is one of the steps to identify risk, before to proceed with the CSA refinement, the scheme proposes to verify the PBT assessment. 3.4. Identification of T criterion for PBT assessment According to the new REACH Annex XIII (Commission Regulation (EU) No 253/2011), a substance is considered to fulfill the T criterion when the NOEC or 10% effect concentration (EC10) for marine or freshwater organisms is less than 0.01 mg/l. Two other criteria are reported in new REACH Annex XIII that are: 1. The substance meets the criteria for classification as carcinogenic (category 1A or 1B), germ cell mutagenic (category 1A or 1B), or toxic for reproduction (category 1A, 1B, or 2) according to CLP Directive and 2. there is other evidence of long term toxicity, as identified by the substance meeting the criteria for classification: specific target organ toxicity after repeated exposure (STOT RE category 1 or 2) according to CLP Regulation); however these are used only to avoid unnecessary tests (when they are already available). On these bases, if the substance is already classified as H340 (may cause genetic defects), H350 (may cause cancer), H360 (may damage fertility or the unborn child), H372 (causes damage to organs through prolonged or repeated exposure) or H373 (may cause damage to organs through prolonged or repeated exposure), it can be considered T, so the user should proceed with the risk identification. In the other cases, if the NOEC (for all the trophic levels) is available and greater than 0.01 mg/l the substance can be classified as not toxic and the user should verify if the risk identification criteria are satisfied, otherwise the T criterion should be further analyzed. As reported in ECHA Guidance (ECHA, 2012b), if a chronic test is not technically feasible (e.g. if log Kow46) the NOECsediment can be used to derive the NOECwater (see ECHA, 2012b for details). Accordingly to ECHA guidance (2012b), to reduce animal testing in the PBT assessment, the T criterion should be the last to examine. So, the scheme suggests verifying first if the compound is P and B. If not, the compound cannot be PBT so the user should proceed with the risk identification, while if the compound is P and B, the T assessment is necessary. Accordingly with ECHA guidance (2012a and b), to perform the T evaluation, it is suggested verifying, on the bases of the short term toxicity data which is the most sensitive species among invertebrates and fish (to be considered the most sensitive species, fish should be more sensitive than invertebrates by at least a factor of 10), and to retrieve long term toxicity data only for this trophic level. Since for algae/aquatic plants the OECD 201 (2011)– freshwater alga and cyanobacteria, growth inhibition test give both short and long term toxicity data, NOEC for algae/aquatic plants are not specifically included in the scheme but they should be considered in the PNEC calculation. 3.5. Risk identification Once the PBT assessment is finished, the user should proceed with the risk identification to verify if the CSA refinement is necessary or not. To identify a risk from CSA, in the ECHA Guidance (2012a) two conditions are reported: 1) PEC/PNEC 41 (where PEC are the predicted environmental concentration) or

162

Table 3 Suggested methods for aquatic toxicity (the list is not exhaustive). In silico

In vitro

In vivo/laboratory

Other

Short term toxicity on algae/aquatic plants

EPISuite – Ecosar (http://www.epa.gov/oppt/exposure/pubs/episuite.htm), OECD QSAR Toolbox (http://www.oecd.org/document/23/0,3746,en_2649_34379_33957015_1_1_1_ 1,00.html)

–

OECD 201 (2006), OECD 221 (2006)

-

Short term toxicity on invertebrates

EPISuite – Ecosar (http://www.epa.gov/oppt/exposure/pubs/episuite.htmwww.epa.gov/ oppt/exposure/pubs/episuitedl.htm), ChemProp (http://www.ufz.de/index.php? en=10684), T.E.S.T – Daphnia magna LC50 (48 h) (http://www.epa.gov/nrmrl/std/cppb/ qsar/), OECD QSAR Toolbox (http://www.oecd.org/document/23/0,3746,en_2649_34379_ 33957015_1_1_1_1,00.html)

–

OECD 202 (2004)

ACR (Hoff et al., 2010)

Short term toxicity on fish

Fish embryo acute EPISuite – Ecosar (http://www.epa.gov/oppt/exposure/pubs/episuite.htm), ChemProp (http://www.ufz.de/index.php?en=10684), T.E.S.T. – Fathead minnow LC50 (96hr) (http:// toxicity (FET) test (OECD 236, 2013), www.epa.gov/nrmrl/std/cppb/qsar/), Lazar – EPA v4b Fathead Minnow Acute Toxicity LC50_mmol (http://lazar.in-silico.de/predict), Demetra – Acute toxicity for Rainbow Trout (Oncorhynchus mykiss): LC50 96-h exposure (http://www.demetra-tox.net), OECD QSAR Toolbox (http://www.oecd.org/document/23/0,3746,en_2649_34379_33957015_1_1_1_ 1,00.html)

OECD 203 (1992)

Step-down approach for PNEC (Jeram et al. 2005), ACR (Hoff et al., 2010)

Long term toxicity on invertebrates

EPISuite – Ecosar (http://www.epa.gov/oppt/exposure/pubs/episuite.htm), OECD QSAR Toolbox (http://www.oecd.org/document/23/0,3746,en_2649_34379_33957015_1_1_1_ 1,00.html)

–

OECD 211 (2008)

ACR (Hoff et al., 2010)

Long term toxicity on fish

EPISuite – Ecosar (http://www.epa.gov/oppt/exposure/pubs/episuite.htm), OECD QSAR Toolbox (http://www.oecd.org/document/23/0,3746,en_2649_34379_33957015_1_1_1_ 1,00.html)

–

OECD 210 (2013), OECD 212 (1998), OECD 215 (2000)

ACR (Hoff et al., 2010)

Verhaar classification

ChemProp (http://www.ufz.de/index.php?en=10684), Toxtree – Verhaar scheme; Verhaar – scheme (Modified) (http://ihcp.jrc.ec.europa.eu/our_labs/predictive_toxicology/qsar_tools/ toxtree), OECD QSAR Toolbox (e.g. acute aquatic toxicity MOA by OASIS, acute aquatic toxicity classification by Verhaar – modified, acute toxicity classification by ECOSAR) (http://www.oecd.org/document/23/0,3746,en_2649_34379_33957015_1_1_1_ 1,00.html)

–

–

Rapid degradation

EPISuite – Biowin (http://www.epa.gov/oppt/exposure/pubs/episuite.htm), VEGA – Ready – biodegradability, (http://www.vega-qsar.eu/download.html), Toxtree – Start biodegradability (http://ihcp.jrc.ec.europa.eu/our_labs/predictive_toxicology/qsar_tools/ toxtree), OECD QSAR Toolbox (http://www.oecd.org/document/23/0,3746,en_2649_ 34379_33957015_1_1_1_1,00.html)

OECD 301 methods A-F (1992), OECD 306 (marine water), OECD 310 (2006), OECD 309 (2004), OECD 302 (), OECD 303 (2001), OECD 111 (2004), OECD XXX (1997)

–

Water solubility

EPISuite – Wskow and WATERNT (http://www.epa.gov/oppt/exposure/pubs/episuite.htm), – OECD QSAR Toolbox (http://www.oecd.org/document/23/0,3746,en_2649_34379_ 33957015_1_1_1_1,00.html)

OECD 105 (1995); EU methods A.6 (2008)

–

A. Lombardo et al. / Environmental Research 135 (2014) 156–164

Endpoint

A. Lombardo et al. / Environmental Research 135 (2014) 156–164

2) log Kow 43 (or BCF 4100) and PEClocal (or PECregional) 4 1/100th of the water solubility where BCF means bioconcentration factor (i.e. the ratio between the substance concentration in the fish and in the water) that is used to evaluate the bioaccumulation of the substance. If one of these conditions is verified, the CSA refinement is necessary; otherwise the aquatic toxicity assessment has reached a conclusion. Compounds classified as P and B (both T and non T) meet a part of the second criterion so in the scheme is indicated only to retrieve the PEC (from exposure assessment) and to verify the second part of the criterion: PEClocal (or PECregional) 41/100th of the water solubility. 3.6. CSA refinement When risks are identified, a CSA based on short term toxicity data is no more sufficient and a CSA refinement should be done (using long term toxicity data). If they are not available for both invertebrates and fish, the use of the EBW (exposure-based waiving, Marquart et al., 2012) decision scheme is suggested. It is implemented in the OSIRIS web tool (http://osiris.simpple. com/OSIRIS-ITS/its/welcome.do). The web tool is freely available and implements Integrated Testing Strategies (ITS) for REACH for several endpoints. ITS are implemented as interactive decision schemes that make use of the information provided by the user, either in vivo or in vitro assay results, in silico method results or other information. Assay results can be imported into the database from IUCLID5 files, so the user does not need to manually input them into the tool. The decision scheme in the web tool asks questions to the user and asks for more data when required, guiding the user in the process. The decision scheme is fully documented, and the user can get the details of the decisions made. The web tool has been developed with the Java programming language using the Spring framework, and stores its data using a MySql database. The decision schemes themselves have been implemented using jBPM, a tool to model and execute complex data-driven workflows. It allows evaluating if the exposure of the aquatic compartment can be excluded and so the CSA refinement is not necessary. Otherwise, accordingly with ECHA guidance (2012a) long term toxicity data on the most sensitive species of invertebrates or fish are necessary. The PNEC should be recalculated on the bases of the most sensitive species long term data. If the sensitivities between fish and invertebrates do not differ by at least a factor of 10, long term data for invertebrates should be used and only if PEC/PNEC Z1 calculated using long term data for invertebrates and an assessment factor (AF) of 50, also long term toxicity data for fish are necessary to complete the CSA. 3.7. Discussion To help the registrant obtaining the required information some freely available tools are suggested in Table 3. The list of methods (in silico, in vitro or in vivo/laboratory) is, however, not exhaustive. Moreover, not all the methods reported are equally reliable (e.g. alternative methods on short term toxicity are generally better than methods for long term toxicity). The user should also consider that the suitability of a model is substance-dependent (i.e. a method could be perfect for a chemical and wrong for another). The respective non-testing methods and models are discussed in great detail in Nendza et al. (2013). Our efforts were directed at simplifying the testing strategy in a straightforward manner. Still, expert judgment remains fundamental to use this ITS. In some points a case by case analysis is required, such as for the degradation products assessment. Other

163

issues like the evaluation of the mitigating factors, the agreement between predicted and experimental values for algae/aquatic plants and/or invertebrates, the PNEC calculation or WoE analysis need specific expertise. In other cases, some tools can be of help, such as the ToxRtool spread sheet (ECVAM, http://ecvam.jrc.ec. europa.eu/) to evaluate the reliability of a value. Also the in silico methods, that sometimes are freely available and appear easy to use, need expert evaluation of the validity of the predicted values, given the model satisfies the OECD principles (OECD 69, 2007). As well, only an expert can evaluate if an in vivo test can be (technically) performed or if a value resulting from a non-testing method could be used (e.g. in case of highly lipophilic substances). The EBW, that could be a good way to verify if the substance could be dangerous for the aquatic environment, is introduced only at later stages of the scheme (in the CSA refinement). This choice is due to the fact that it is a complex methodology and needs many (test) data. In the scheme we try to cope with conflicting requirements. Often, REACH asks for short term toxicity data, but for PBT or CSA long term data are required. If only long term data are available to satisfy C&L requirements, we suggest ways to obtain the missing short term data without performing further tests, but using nontesting methods.

4. Conclusions An ITS scheme was developed to optimize the choice of the best tests and non-tests to perform the evaluation of the aquatic pelagic toxicity. It combines the existing methods and the available information with the REACH requirements including the data needed to satisfy REACH annexes, the C&L, the PBT assessment and the CSA. The ITS allows to generate robust data for chemical hazard assessment while reducing animal testing, and to integrate in a well-managed way existing information with estimates from alternative methods. The ITS described in this study is an extension of a previous version (developed before the last modification to ECHA guidelines (ECHA, 2012, 2012a,2012b) that is freely available through the OSIRIS web tool.

Acknowledgments This research was financially supported by the European Union OSIRIS Project, European Commission, FP6 Contract no. GOCE-CT2007-037017.

Appendix A. Supporting information Supplementary data associated with this paper can be found in the online version at http://dx.doi.org/10.1016/j.envres.2014.09. 002. References CLP: Regulation (EC No 1272)/2008 of the European Parliament and of the Council of 16 December 2008 on classification, labelling and packaging of substances and mixtures, amending and repealing Directives 67/548/EEC and 1999/45/EC, and amending Regulation (EC) No 1907/2006. Council Directive of 27 June 1967 on the approximation of laws, regulations and administrative provisions relating to the classification, packaging and labelling of dangerous substances (67/548/EEC). Commission Regulation (EU) No 253/2011 of 15 March 2011 amending Regulation (EC) No 1907/2006 of the European Parliament and of the Council on the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) as regards Annex XIII. Directive 1999/45/EC of the European Parliament and of the Council of 31 May 1999 concerning the approximation of the laws, regulations and administrative

164

A. Lombardo et al. / Environmental Research 135 (2014) 156–164

provisions of the Member States relating to the classification, packaging and labelling of dangerous preparations. 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. ECHA, 2008. Guidance on Information Requirements and Chemical Safety Assessment Chapter R.10: characterization of dose [concentration]-response for environment, May 2008. ECHA, 2009. Guidance in a Nutshell – Chemical Safety Assessment, September 2009. ECHA, 2010. Practical Guide 10: How to Avoid Unnecessary Testing On Animals, June 2010. ECHA, 2012. Guidance on the Application of the CLP Criteria, November 2012. ECHA, 2012a. Guidance on Information Requirements And Chemical Safety Assessment, Chapter R.7b: Endpoint Specific Guidance, November 2012. ECHA, 2012b. Guidance on Information Requirements and Chemical Safety Assessment, Chapter R.11: PBT Assessment, November 2012. EU method A.6, Council Regulation (EC) No 440/2008 of 30 May 2008 laying down test methods pursuant to Regulation (EC) No 1907/2006 of the European Parliament and of the Council on the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH). Fleischer, M., 2007. Testing cost and testing capacity according to the REACH requirements – results of a survey of independent and corporate GLP laboratories in the EU and Switzerland. J. Bus. Chem. 4, 96–114. Hartung, T., Luechtefeld, T., Maertens, A., Kleensang, A., 2013. Food for thought: integrated testing strategies for safety assessments. Altex 31, 3–18. Hoff, D., Lehmann, W., Pease, A., Raimondo, S., Russom, C., Steeger, T., 2010. Predicting the Toxicities of Chemicals to Aquatic Animal Species. EPA, U.S.. Jeram, S., Riego Sintes, J., Halder, M., Baraibar Fentanes, J., Sokull-Klüttgen, B., Hutchinson, T., 2005. A strategy to reduce the use of fish in acute ecotoxicity testing of new chemical substances notified in the European Union. Regul. Toxicol. Pharmacol. 42, 218–224. Kwon, J.H., Escher, B.I., 2008. A modified parallel artificial membrane permeability assay for evaluating bioconcentration of highly lipophilic chemicals in fish. Environ. Sci. Technol. 42, 1787–1793. Marquart, H., Meijster, T., Van de Bovenkamp, M., Ter Burg, W., Spaan, S., Van Engelen, J., 2012. A structured approach to Exposure Based Waiving of human health endpoints under REACH developed in the OSIRIS project. Regul. Toxicol. Pharmacol. 62, 231–240. Masungi, C., Borremans, C., Willems, B., Mensch, J., van Dijck, A., Augustijns, P., Brewster, M.E., Noppe, M., 2004. Usefulness of a novel Caco-2 cell perfusion system. I. In vitro prediction of the absorption potential of passively diffused compounds. J. Pharm. Sci. 93, 2507–2521. Nakao, K., Fujikawa, M., Shimizu, R., Akamatsu, M., 2009. QSAR application for the prediction of compound permeability with in silico descriptors in practical use. J. Comput. Aided Mol. Des. 23, 309–319. Nendza, M., Müller, M., Wenzel, A., 2014. Discriminating toxicant classes by mode of action: 4. Baseline and excess toxicity. SAR QSAR Environ. Sci. 25, 393–405. Nendza, M., Gabbert, S., Kühne, R., Lombardo, A., Roncaglioni, A., Benfenati, E., Benigni, R., Bossa, C., Strempel, S., Scheringer, M., Fernández, A., Rallo, R., Giralt, F., Dimitrov, S., Mekenyan, O., Bringezu, F., Schüürmann, G., 2013. A comparative survey of chemistry-driven in silico methods to identify hazardous substances under REACH. Regul. Toxicol. Pharmacol. 66 (3), 301–314. OECD, 2010. Short guidance on the threshold approach for short term fish toxicity, ENV/JM/MONO (20010)17. OECD series on testing assessment No. 126, Organisation for Economic Co-operation and Development, Paris, France. OECD 105, 1995. Water Solubility. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 105, Paris, France. OECD 111, 2004. Hydrolysis as a function of pH. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 111, Paris, France. OECD 201, 2011. Freshwater Alga and Cyanobacteria, Growth Inhibition Test. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 201, Paris, France.

OECD 202, 2004. Daphnia sp., acute immobilisation test. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 202, Paris, France. OECD 203, 1992. Fish, short term toxicity test. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 203, Paris, France. OECD 210, 2013. Fish, Early-life Stage Toxicity Test. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 210, Paris, France. OECD 211, 2012. Daphnia magna Reproduction Test. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 211, Paris, France. OECD 212, 1998. Fish, Short-term Toxicity Test on Embryo and Sac-fry Stages. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 212, Paris, France. OECD 215, 2000. Fish, Juvenile Growth Test. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 215, Paris, France. OECD 221, 2006. Lemna sp. Growth Inhibition Test. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 221 Paris, France. OECD 236, 2013. Fish Embryo Acute Toxicity (FET) Test. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 236, Paris, France. OECD 301, 1992. Ready Biodegradability. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 301, Paris, France. OECD 303, 2001. Simulation Test – Aerobic Sewage Treatment: 303 A: Activated Sludge Units – 303 B: biofilms. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 303, Paris, France. OECD 308, 2002. Aerobic and Anaerobic Transformation in Aquatic Sediment Systems. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 308, Paris, France. OECD 309, 2004. Aerobic Mineralisation in Surface Water – Simulation Biodegradation Test. Organisation for Economic Cooperation and Development (OECD), OECD Guideline for the Testing of Chemicals No. 309, Paris, France. OECD 69, 2007. Guidance document on the validation of (quantitative) structureactivity relationship [(Q)SAR] models. Organisation for Economic Cooperation and Development (OECD), OECD Environment health and safety publications, series on testing and assessment No. 69, Paris, France. REACH: Regulation (EC) No 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REAC H), establishing a European Chemicals Agency, amending Directive 1999/45/EC and repealing Council Regulation (EEC) No 793/93 and Commission Regulation (EC) No 1488/94 as well as Council Directive 76/769/EEC and Commission Directives 91/155/EEC, 93/67/EEC, 93/105/EC and 2000/21/EC. Rovida, C., Hartung, T., 2009. Re-evaluation of animal numbers and costs for in vivo tests to accomplish REACH legislation requirements for chemicals – a report by the transatlantic think tank for toxicology (t4). Altex 26, 187–208. UK REACH Competent Authority Information Leaflet Number 18 – Minimisation of Animal Testing, January 2010. Van der Jagt, K., Munn, S., Torslov, J., de Bruijnk, J., 2004. Alternative approaches can reduce the use of test animals under REACH. Addendum to the report Assessment of additional testing needs under REACH Effects of (Q)SAR, risk based testing and voluntary industry initiatives. Verhaar, H.J.M., van Leeuwen, C.J., Hermens, J.L.M., 1992. Classifying environmental pollutants. 1: Structure–activity relationships for prediction of aquatic toxicity. Chemosphere 25, 471–491. Wood, C.M., Kelly, S.P., Zhou, B., Fletcher, M., O’Donell, M., Eletti, B., Paert, P., 2002. Cultured gill epithelia as models for freshwater gills. Biochim. Biophys. Acta Biomembr. 1566, 72–83.