APPLICATION OF ALTERNATIVE METHODS IN THE REGULATORY ASSESSMENT OF CHEMICAL SAFETY RELATED TO HUMAN EYE IRRITATION & SEVERE EYE IRRITATION

Transcription

1 R/RT(2009)3/2 APPLICATION OF ALTERNATIVE METHODS IN THE REGULATORY ASSESSMENT OF CHEMICAL SAFETY RELATED TO HUMAN EYE IRRITATION & SEVERE EYE IRRITATION CURRENT STATUS AND FUTURE PROSPECTS Developed for and sponsored by the Swiss Federal Office of Public Health FOPH by Dr Chantra Eskes on behalf of Orange House Partnership npo March 2010 Kampendaal 83, B-1653 Dworp (Brussels) Belgium tel:

2 APPLICATION OF ALTERNATIVE METHODS IN THE REGULATORY ASSESSMENT OF CHEMICAL SAFETY RELATED TO HUMAN EYE IRRITATION & SEVERE IRRITATION CURRENT STATUS AND FUTURE PROSPECTS Dr. Chantra Eskes March 2010 The Orange House Partnership vzw, Kampendaal 83, 1653 Dworp, Brussel Sponsored by : Swiss Federal Office of Public Health Division of Chemical Products Bern, Switzerland Contract n / / -23

3 (this page is left intentionally blank) p 2 out of 84

4 LIST OF CONTENTS 1 REGULATORY CONTEXT IN THE EUROPEAN UNION DEFINITIONS REGULATORY REQUIREMENTS FOR EYE IRRITATION TESTING THE NEED FOR ALTERNATIVE TEST METHODS REGULATORY IN VIVO AND ALTERNATIVE TEST METHODS SEQUENTIAL TESTING STRATEGIES Non-animal Bottom-Up and Top-Down Approaches for Eye Irritation Testing Integrated Testing Strategies for Eye Irritation Testing CLASSIFICATION SYSTEMS IN VIVO ACUTE EYE IRRITATION TESTING MECHANISMS OF EYE IRRITATION AND SEVERE EYE IRRITATION METHOD DESCRIPTION ACCORDING TO OECD TG LIMITATION OF THE DRAIZE EYE RABBIT TEST Test material exposure Variability of the Draize eye irritation test Inter-species differences and Predictive Capacity of the Draize eye irritation test Ethical issues CLASSIFICATION AND LABELLING THE UN GLOBALLY HARMONISED SYSTEM (GHS) FOR CLASSIFICATION & LABELLING (UN, 2003, 2009) THE NEW EU CLP CLASSIFICATION SYSTEM (EC, 2008B) THE EU DSD CLASSIFICATION SYSTEM (EC, 2001) COMPARISON OF CLASSIFICATION SYSTEMS OTHER CLASSIFICATION SYSTEMS The (Modified) Maximum Average Scores The EPA Classification System IN VITRO ALTERNATIVE METHODS FOR EYE IRRITATION TESTING INTRODUCTION VALIDATED ASSAYS Organotypic assays Cytotoxicity- and cell function-based assays THE BOVINE CORNEAL OPACITY AND PERMEABILITY ASSAY (BCOP): SEVERE EYE IRRITATION & NON- CLASSIFICATION Principles of the test Preparation of the corneas according to OECD TG Method description Proficiency testing according to OECD TG 437 (severe eye irritants) Known applicability and limitations THE ISOLATED CHICKEN EYE TEST (ICE): SEVERE EYE IRRITATION Principles of the test Preparation of the eyes according to OECD TG Method description according to OECD TG Proficiency testing according to OECD TG 438 (severe eye irritation) Known applicability and limitations THE CYTOSENSOR MICROPHYSIOMETER ASSAY (CM): SEVERE EYE IRRITATION & NON-CLASSIFICATION Principles of the test Method description according to Invittox Protocol Proficiency testing Known applicability and limitations THE FLUORESCEIN LEAKAGE ASSAY (FL): SEVERE EYE IRRITATION p 3 out of 84

5 Principles of the test Method description according to Invittox Protocol Proficiency testing Known applicability and limitations IN VITRO - IN VIVO EYE IRRITATION TEST METHODS COMPARISON OTHER ALTERNATIVES METHODS FOR EYE IRRITATION TESTING THE LOW VOLUME EYE TEST (LVET) (Q)SARS FOR EYE IRRITATION ON GOING VALIDATION STUDIES AND RESEARCH PROGRAMS CURRENT VALIDATION AND EVALUATION STUDIES In vitro testing strategies for regulatory eye irritation testing Reconstructed human tissue models Ocular Irritection and Slug Mucosal Irritation assay Anti-Microbials and Cleaning Products (AMCP) ONGOING RESEARCH AND DEVELOPMENT Improvements of validated assays Development of new in vitro assays and biomarkers FUTURE PROSPECTS AND RECOMMENDATIONS TO ACHIEVE ANIMAL REPLACEMENT REFERENCES LIST OF ABBREVIATIONS AMCP BCOP BfR BfR-DSS BGA/BMBF Cat. Cat. 1 Cat. 2 Cat. 2A Cat. 2B CM cm 2 COLIPA CPSC CTFA CV DMEM DOI EC EC/HO ECETOC ECHA ECVAM EMEM EPA ESAC EU EU CLP Anti-microbial and Cleaning Products Bovine Corneal Opacity and Permeability test German Federal Institute for Risk Assessment BfR Decision Support System German Federal Health Office / Department of Research and Technology Category Irreversible effects on the eyes / serious damage to eyes Reversible effects on the eyes / irritating to eyes Irritating to eyes Mildly irritating to eyes (optional) Cytosensor Microphysiometer assay Centimeter square European trade association for the cosmetic, toiletry and perfumery industry US Consumer Product Safety Commission US Cosmetics Toiletry and Fragrance Association Coefficient of Variation Dulbecco s Modified Eagle s Medium Depth of Injury European Communities European Commission / UK Home Office European Centre for Ecotoxicology and Toxicology of Chemicals European Chemicals Agency European Centre for the Validation of Alternative Methods Eagle s Minimal Essential Medium US Environmental Protection Agency ECVAM s Scientific Advisory Committee European Union EU Regulation 1272/2008 on the Classification, Labelling and Packaging of Substances and Mixtures p 4 out of 84

6 EU DSD EU DPD FDA FHSA FL FL 20 GHS HBSS HCE HET-CAM I ICCVAM ICE IRAG IRE IVIS LVET MDCK MHW/JCIA min mg ml (M)MAS MRD 50 MTT NC NICEATM nm NRR OD OECD PCOP PMNs PorCORA (Q)SAR R36 R41 RBC REACH RhT SAR sec SI SMI TG UN US UV/VIS VMG w/v l EU Dangerous Substances Directive 67/548/EEC EU Dangerous Preparation Directive 199/45/EC US Food and Drug Administration US Federal Hazardous Substances Act Fluorescein Leakage assay Concentration of test material causing 20% of fluorescein leakage (FL assay) Globally Harmonized System for hazard classification and labeling Hank s Balanced Salt Solution Human Corneal Epithelium model Hen s egg test on the Chorio-Allantoic Membrane assay Irritant US Interagency Coordinating Committee on Validation of Alternative Methods Isolated Chicken Eye test Interagency Regulatory Alternatives Group Isolated Rabbit Eye test In Vitro Irritancy Score (BCOP test) Low Volume Eye Test Madin Darby Canine Kidney cell line (FL assay) Japanese Ministry of Health and Welfare/ Japanese Cosmetics and Toiletries Association Minutes Milligram Milliliter (Modified) Maximum Average Scores Dose of test material which induces 50% decrease in Metabolic Rate (CM assay) 3-(4,5-Dimethyl-2-thiazol-2-yl)-2,5-diphenyl tetrazolium bromide, Thiazolyl blue; EINECS number , CAS number Non-classified US National Toxicology Program Interagency Center for the Evaluation of Alternative Toxicological Methods Nanometer Neutral Red Release assay Optical Density Organisation for Economic Co-operation and Development Porcine Corneal Opacity and Permeability test Polymorphonuclear leukocytes Porcine Corneal Ocular Reversibility Assay (Quantitative) Structure-Activity Relationship Irritating to eyes Risk of serious damage to eyes Red Blood Cell haemolysis test EU Regulation 1907/2006 on the Registration, Evaluation, Authorisation and restriction of Chemicals Reconstructed human Tissues Structure-Activity Relationship second Severe Irritant Slug Mucosal Irritation assay Test Guideline United Nations United States Ultra-violet / Visible Validation Management Group Weight/volume Microliter p 5 out of 84

7 1 Regulatory context in the European Union 1.1. Definitions Eye irritation is generally defined within the regulatory context as the production of changes in the eye following the application of a test substance to the anterior surface of the eye, which are fully reversible within 21 days of application (OECD, 2002; UN, 2003, 2009, EC 2008a,b). Eye corrosion (OECD, 2002 and EC, 2008a) or serious eye damage/irritation (EC, 2008b and UN, 2003, 2009) is generally defined within the regulatory context as the production of tissue damage in the eye, or serious physical decay of vision, following application of a test substance to the anterior surface of the eye, which is not fully reversible within 21 days of application. More recently ocular corrosives were also defined as substances that cause irreversible tissue damage to the eye, and ocular severe irritants as substances causing serious eye damage/irritation as described in the former sentence (OECD, 2009a,b). For the purposes of this document, the following terms will be employed: - eye irritation will refer to eye irritation effects as defined above, - severe eye irritation will refer to eye corrosion, ocular corrosives, serious eye damage/irritation and ocular severe irritants as defined above. - eye irritation testing will refer to the test that allows identifying eye irritation and severe eye irritation effects as defined here above Regulatory requirements for eye irritation testing The European Union chemicals policy adopted in 2006 for the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) establishes standard information requirements that need to be submitted for the registration and evaluation of chemicals. Such information requirements are specified in details in the REACH Annexes VI to XI (EC, 2006). According to Annex VI, the registrant should gather and evaluate all available information before considering further testing. These include physico-chemical properties, (Quantitative) Structure-Activity Relationship ((Q)SAR), grouping, in vitro data, animal studies, and human data (for details see chapter 1.5.2). Information on exposure, use and risk management measures should also be collected and evaluated. If these data are inadequate for hazard and risk assessment, further testing should be carried out in accordance with the requirements of REACH Annexes VII and VIII, which are based on the tonnage levels of the manufactured or imported chemicals. The standard toxicological information requirements for substances manufactured or imported in quantities between one tonne and 10 tonnes are laid down in Annex VII. If new testing data are necessary, these must be derived from in vitro methods only. Annex VII does not foresee in vivo eye irritation testing (which allows identification of eye irritation and severe eye irritation). The standard information required at this tonnage level for eye irritation testing can be satisfied by following three consecutive steps: (1) an assessment of the available human and animal data, (2) an assessment of the acid or alkaline reserve, (3) in vitro study for eye irritation. Specific adaptations are given that specify when step 3 does not need to be conducted. These are if: 1) the available information indicates that the criteria are met for classification as corrosive to the skin or irritating to eyes, or 2) the substance is flammable in air at room temperature. For substances manufactured or imported in quantities of 10 tonnes, the toxicological information requirements are laid down in Annex VIII. Such information is additional to that required in annex VII, and requires in vivo eye irritation testing (which includes identification of eye irritation and severe eye irritation). The following specific rules are given that allow deviating from the standard regime: p 6 out of 84

8 - the substance is classified as irritating to eyes with risk of serious damage to eyes; or - the substance is classified as corrosive to the skin and provided that the registrant classified the substance as eye irritant; or - the substance is a strong acid (ph<2) or base (ph>11.5); or - the substance is flammable in air at room temperature. Importantly, Annex VI also states that new tests on vertebrates shall only be conducted or proposed as a last resort when all other data sources have been exhausted. In particular it states that the in vivo testing requirement of Annex VIII can be adapted by the rules laid down in Annex XI allowing to avoid unnecessary animal testing. Annex XI establishes amongst others, the conditions in which the standard testing may not be scientifically necessary. In vitro test methods fall within this category. Annex XI states that if the results obtained from the use of an in vitro methods do not indicate a certain dangerous properties, a confirmatory test according to annex VII To X may be waived if the following conditions are met: 1. results are derived from an in vitro method whose scientific validity has been established by a validation study, according to internationally agreed principles 2. results are adequate for the purpose of classification and labelling and/or risk assessment, and, 3. adequate and reliable documentation of the applied method is provided. In addition, it states that results obtained from a suitable in vitro method may be used to indicate the presence of a certain dangerous property, or may be important in relation to a mechanistic understanding which may be important for the assessment. Suitable in vitro methods means sufficiently well developed according to internationally agreed test development criteria (e.g., criteria from the European Centre for the Validation of Alternative Methods, ECVAM, for the entry of a test into the prevalidation process). However, depending on the potential risk, immediate or proposed confirmation (based on the tonnage levels) may be necessary requiring tests beyond the information foreseen in Annexes VII to X. In order to apply the information testing requirements as laid down in REACH, the European Chemicals Agency has issued an Endpoint Specific Guidance on the REACH Information Requirements and Chemical Safety Assessment (ECHA, 2008a). For eye irritation testing (which allows identification of eye irritation and severe eye irritation), a testing strategy is proposed to be followed, including a step-wise approach that takes into consideration information from the skin corrosion test, the physico-chemical properties of the test substance, existing human, animal dermal toxicity, QSAR and in vitro data on the test material, a weight-of-evidence evaluation, the generation of new in vitro data and only as a last resort, the generation of new in vivo testing (for details see chapter 1.5.2) The need for alternative test methods At the European Union (EU) level, not only REACH but also the cosmetics legislation has accelerated the need for alternative methods to toxicological testing. As mentioned above, within REACH, in vitro testing is required as standard information for substances marketed in volumes between 1 and 10 tonnes per year (EC, 2006). Such requirement could lead to testing of up to 20,000 existing chemicals using in vitro methods. Moreover REACH regulation whereas 1 and article 1 promote alternative methods for safety testing. Article 25 states that animal testing must be used as a last resort, which encourages the exploitation of useful alternative methods. Article 13 states, that information on hazards (regarding positive results) and risks may be generated by suitable alternative methods that have not yet been taken up as official regulatory test methods, upon the condition that such methods fulfil the requirements of Annex XI (e.g., ECVAM criteria for the entry of a test into the prevalidation process). If such methods are moreover validated, they may be used for identifying positives as well as negatives (EC, 2006). The 7 th amendment to the EU Cosmetics Directive (Directive 2003/15/EC) went further and prohibited animal testing of finished products from 2004 and of ingredients from The animal testing ban is p 7 out of 84

9 reinforced by marketing bans of cosmetics tested on animals from 2004 (finished products), 2009 (acute effects) or 2013 (repeated-dose toxicity, toxicokinetics, reproductive toxicity; EC, 2003). In addition, the European Union Directive 86/689 on animal protection also promotes the use of alternative methods. It states that an experiment shall not be performed if another scientifically satisfactory method of obtaining the result sought, not entailing the use of an animal, is reasonably and practicably available (EC, 1986). As a consequence, there is a strong need for in vitro alternative methods to fulfil these current regulatory requirements within the European Union. In particular, the European Chemicals Agency has issued a guidance for the evaluation of available information for REACH, in which it states that there are two ways for using data from in vitro studies: 1) information from validated in vitro tests: may be used to determine whether a substance has or not dangerous properties, allowing to fully or partly replace an animal test. In that case one of the criteria for acceptance is the adequacy of the information generated using such test(s) for the purpose of classification and labelling and/or risk assessment, 2) information derived from suitable in vitro methods: can be used for determining the presence of a certain dangerous property, adapting the standard testing regime as set out in annex XI. Finally, information from in vitro tests may be also used to provide with mechanistic insights (ECHA, 2008b). As such, the scientific validation of such in vitro methods certify their level of relevance and reliability to be used in the regulatory framework for detecting both positive and negative results, as full replacement, partial replacement, reduction or refinement of the animal testing. The area of eye irritation testing, together with skin corrosion and irritation, represents one of the most advanced areas for the validation of alternative test methods. Perhaps because of the cruelty and the high public concern of such testing procedures, eye irritation has been a pioneer field in which major efforts were undertaken as early as the 1980 s and 1990 s to develop, evaluate and validate in vitro methods to replace the Draize Eye Irritation test (Eskes et al., 2005; Wilhelmus, 2001). The present document describes in details the most advanced in vitro methods for regulatory use, and gives an overview of the current ongoing activities to overcome the remaining gaps in order to achieve full replacement of the animal test Regulatory in vivo and alternative test methods The conventional test for the eye irritant and severe irritant potential of chemicals is the rabbit eye test, which was developed by Draize et al. (1944) and which has become the international standard assay for acute ocular toxicity (OECD TG 405, 2002; EC B.5, 2008a). The test material is applied to the conjunctival sac of the eye of the animal, and subsequent physiological responses are classified by careful visual examination of the cornea, iris and conjunctiva, and given a numerical score (for details see chapter 2). A variety of different scoring systems for assessing the extent of injury to the corneal, the iridial and the conjunctival compartments are currently applied in different regulations, ranging from maximum single tissue scores to averaged weighed sum scores for all three tissues (for details see chapter 3). To reduce and/or replace the Draize rabbit eye test with in vitro test methods, the use of testing strategies is generally recommended, due to the fact that the range of criteria for injury and inflammation covered by the Draize rabbit eye test is unlikely to be covered by a single in vitro test. In 2005, a Workshop organised by the European Centre for the validation of Alternative Methods (ECVAM) has identified and proposed a testing strategy to be used for regulatory purposes to replace or reduce the animal test (Scott et al, 2010). The proposed strategy combines the strengths of particular in vitro assays to address required ranges of irritation potential and/or chemical classes. It proposes, based on expected irritancy of the test substance, the use of one of the two following tiered testing approaches before progression of further in vitro testing (for details see chapter 1.5.1): p 8 out of 84

10 - a Bottom-Up approach, in which testing begins with test methods that can accurately identify test materials that do not require classification for eye irritation, i.e., able to distinguish nonclassified test materials from irritant and severe irritant ones, - or a Top-Down approach, in which testing begins with test methods that can accurately identify eye corrosives and severe irritants from the irritant and non-classified ones. These two tiered testing approaches have served as the basis for the validation efforts undertaken since then for eye irritation testing in Europe and in the United States. Alternative methods able to identify severe eye irritants, and therefore useful to initiate a Top-Down Approach, have been validated in 2007 and adopted by the OECD in These are the in vitro / exvivo organotypic models Bovine Corneal Opacity and Permeability test (BCOP), and the Isolated Chicken eye test (ICE) as the Test Guidelines (TG) 437 and 438 (OECD, 2009a,b). Two additional cell-function based in vitro assays have been recently endorsed as scientific valid by the ECVAM s Scientific Advisory Committee (ESAC) for the identification of severe irritants, the Cytosensor Microphysiometer and Fluorescein Leakage (ESAC, 2009a). In addition, in July 2009 a cell-function based assay, the Cytosensor Microphysiometer, has been endorsed as scientifically valid for the first time to identify non-classified test materials from irritant and severe irritant ones, and therefore useful to initiate a Bottom-Up Approach (ESAC, 2009a). However, this assay was found to be valid only for the limited applicability domain, of water-soluble surfactants and surfactant-containing mixtures. Current validation efforts seem to have identified also the organotypic BCOP assay as a promising in vitro assay for distinguishing non classified test materials from irritant and severe irritant ones, however its validity statement is not yet available to the public domain (ICCVAM, 2009a,b). A summary of the available validated and accepted alternatives for eye irritation testing and their applicability domains are given in table 1. These validated in vitro assays represent partial replacement of the traditional in vivo test to be used in tiered testing schemes as proposed by Scott and co-authors (2010) and integrated into the more comprehensive strategies as recommended by the OECD (2002), the Globally Harmonised System for hazard classification and labelling (UN, 2003, 2009), and the REACH Endpoint Specific Guidance (ECHA, 2008a). Table 1. Validated and accepted in vitro methods for eye irritation testing, their purposes, status, application and limitations. Strategy Purpose Test Method Status Application and Limitations Top-Down Approach Identification of severe eye irritation Positive results lead to severe irritant classification. Bovine Cornea Opacity Test Isolated chicken eye test Validated and adopted by OECD (TG 437) Validated and adopted by OECD (TG 438) Positive results obtained with alcohols or ketones should be interpreted cautiously due to high false positive rates for those chemical classes. Positive results obtained with alcohols should be interpreted cautiously due to risk of overprediction linked to high false positive rates for this chemical class. Negative results require further testing. Cytosensor Microphysiometer Validated in 2009 Applicable to water soluble substances and mixtures Fluorescein Leakage Validated in 2009 Applicable to water soluble substances and mixtures Bottom-Up Approach Identification of nonclassified test materials Negative results lead to no classification. Positive results require further testing. Cytosensor Microphysiometer Bovine Cornea Opacity Test Validated in 2009 Peer reviewed in 2009 Applicable to water soluble surfactants and surfactant containing mixtures. To be confirmed once evaluation of validity is finalised. p 9 out of 84

11 1.5. Sequential testing strategies Several integrated testing strategies have been proposed in the passed for eye irritation testing, and have since then, been taken up into the regulatory requirements at the United Nations, OECD and European Union levels (DeSilva et al., 1997; GHS, 2003, 2009; OECD, 2002; EC, 2008a). Moreover, non-animal testing strategies combining specific in vitro assays to replace or reduce animal testing for regulatory purposes have been proposed by Scott and co-authors (2010). The present chapter will discuss in a first step the tiered testing approaches based on in vitro assays, and in a second step, the more comprehensive integrated testing strategies which include in vitro, human, (Q)SAR and in vivo data as recommended by the OECD, the UN GHS and the REACH Endpoint Specific Guidance Non-animal Bottom-Up and Top-Down Approaches for Eye Irritation Testing It is generally acknowledged that the range of criteria for injury and inflammation covered by the Draize rabbit eye test is unlikely to be covered by a single in vitro test. Among the recommendations made to progress validation efforts of in vitro alternatives in view to replace or reduce the Draize rabbit eye test, an important one was to make use of testing strategies that utilises the strengths of individual in vitro test methods to address required ranges of irritation potential and/or chemical classes (Eskes et al., 2005). For that purpose, ECVAM organised in 2005 a workshop to identify potential test strategies based on the current uses and applicability of in vitro methods for eye irritation testing (Scott et al., 2010). The meeting involved around 30 participants from industry, regulatory bodies, contract laboratories, academia and animal welfare groups. Participants were requested to nominate test methods for eye irritation testing based on their applicability, to provide supportive in vitro and in vivo data, and to identify potential test strategies to assess eye irritation and severe eye irritation effects based on their experiences and uses of the test methods. Two promising tiered testing approaches have been identified concurrently by the two break-out groups of participants. Based on the expected irritancy of the test materials as identified by their physico-chemical properties, one of the two approaches would be followed (figure 1): - a Bottom-Up which starts with using in vitro test methods that can accurately identify nonirritants; - or a Top-Down which starts with using in vitro test methods that can accurately identify severe irritants. Progression of further in vitro testing would depend on the outcome of the first testing and would follow the strategies as summarized in figure 1. The difficulty in predicting the middle category of irritancy (e.g. R36, GHS Categories 2A and 2B) was recognized. Irrespective of the starting point, the approach would identify non-irritants and severe irritants, leaving all others to the mild/moderate irritant categories (GHS 2/R36). p 10 out of 84

12 Existing information e.g., physico-chemical properties Bottom Up Approach Identify non irritants Top Down Approach Identify severe irritants In vitro test 1 Non irritant (no classification) In vitro test 1 Severe irritant (R41/Cat.1) In vitro test 2 Severe irritant (R41/Cat.1) In vitro test 2 Non irritant (no class.) Irritant (R36/Cat.2) or confirmatory test Irritant (R36/Cat.2) or confirmatory test Figure 1: Bottom-Up and Top-Down In Vitro Strategy Approach for Eye Irritation Testing. If the test material is expected to be a no-low eye irritant the Bottom-Up approach is initiated. Conversely, the Top-Down approach is initiated if the test material is expected to be a severe eye irritant. Validated methods would be used in a two-step procedure to determine if a test material is a non-classified or severe irritant (EU R41/GHS cat.1). A default EU R36 / GHS cat. 2 classification could be assigned if neither a non-classified or severe irritancy assignment were made. In order to identify the most suitable assays that will serve as the building blocks for such approaches, the performances and applicability domains of individual alternative methods were since then, or are currently being evaluated by retrospective and prospective validation studies. The evaluation and construction of testing strategies combining in the most optimal way the different validated methods is currently being carried out by using data mining techniques and statistical tools in a joint effort between ECVAM and COLIPA, the European cosmetics association. The aim is to determine the most suitable strategies for the classification of test substances according to their irritation potential, as well as to identify the most promising strategies with higher impact in reducing animal testing, and the higher benefits and lower potential costs linked to testing and incorrect predictions. Once available, it is foreseen that such strategies will be challenged with a new set of test chemicals prior to be forwarded for regulatory acceptance (Zuang et al., submitted) Integrated Testing Strategies for Eye Irritation Testing The sequential testing strategy proposed in the OECD TG 405 (2002) and in the EU test method B.5 (2008a) is based in a stepwise order on: existing human and animal data, (Q)SAR, ph considerations, the use of validated and accepted in vitro tests, and as a last step, the in vivo rabbit eye test when all the other tiers have produced negative results (see figure 2). In this strategy the in vitro assays are recommended to identify corrosives and irritants, but not non-classified substances. However, it has been recognised that the middle category of irritancy (e.g. R36, GHS Categories 2A and 2B) ought to be more difficult to discriminate from the two other categories (Scott et al., 2010). The UN GHS proposed strategy is similar to the one proposed by the OECD TG 405 (UN, 2003, 2009; OECD, 2002), with the only difference being that in step 4 other information on skin corrosion are recommended to be used instead of systemic toxicity via the dermal route, which could lead to the classification of ocular corrosive and severe irritant. p 11 out of 84

13 1a. Existing human and/or animal data on eyes NC I SI 1b. Existing human and/or animal data on skin I SI 2a. SAR for eye corrosion / irritation I SI 2b. SAR for skin corrosion SI 3. ph & buffering capacity if relevant ph 2 or 11.5 (& high buffering capacity) eye corrosive SI 4. Systemic toxicity via dermal route if highly toxic no eye testing needed 5. Validated and accepted in vitro or ex vivo test for eye corrosion / severe eye irritation SI 6. Validated and accepted in vitro or ex vivo test for eye irritation I 7. In vivo skin irritation/corrosion (OECD TG 404), if SI corrosive or severe irritant response assume corrosive to eyes 7. In vivo rabbit eye test using 1 animal SI 8. Confirmatory test using 1or 2 further animals NC I SI Figure 2. Summary steps of the testing strategy for eye irritation as recommended in OECD TG 405 (2002 adapted). NC: Non Classified; I: Eye Irritant; SI: Severe eye irritant In the European Union, the Endpoint Specific Guidance to REACH recommends the use of a similar sequential strategy for eye irritation testing as shown in figure 3 (ECHA, 2008a). If the building blocks correspond to the ones recommended in the OECD TG 405, the test strategy introduces some new elements: - the use of broader physico-chemical properties, - the use of existing in vitro data, - the use of weight-of-evidence analysis of all existing and relevant data, - the use of validated and accepted in vitro methods for the identification of non irritants in addition to the identification of irritants and severe irritants, so that the in vivo test might be avoided, Moreover, it allows the use of non-validated in vitro methods for the identification of positive results, although confirmation may be necessary depending on potential risk as defined in annex XI of REACH (EC, 2006). p 12 out of 84

14 0. Information on skin corrosion SI 1. Existing data on Physico-chemical properties I SI 2. Existing human data I SI 3. Existing animal data from eye irritation studies NC I SI 4. Existing data on acute dermal toxicity 5. Existing (Q)SAR data and read-accross I SI 6. Existing in vitro data NC I SI 7. Weight-of-evidence analysis NC I SI 8. New in vitro / ex vivo test for eye irritation testing NC I SI 10. New in vivo test for irritation NC I SI Figure 3. Summary steps of the testing strategy for Eye Irritation as recommended in the Endpoint Specific Guidance to REACH (ECHA, 2008a adapted). NC: Non Classified; I: Eye Irritant; SI: Severe eye irritant A further integrated decision-tree testing strategy has been proposed for eye irritation with respect to the requirements of the REACH regulation by Grindon and co-authors (2008). In this test strategy a sequential strategy makes use of existing data on eye irritation and severe eye irritation, ph, physicochemical properties, in silico predictions, in vitro tests for eye irritation and severe irritation, a weightof-evidence evaluation of all data, and as a last resort an in vivo test based on the Low Volume Eye Test, which is a refinement to the traditional Draize rabbit eye test. Finally, specific test strategies have been proposed for the hazard and risk assessment of cosmetic ingredients by McNamee and co-authors (2009). Here again the use of weight-of-evidence analysis is proposed to evaluate all available data such as physicochemical properties, literature, animal, in vitro, human, read-across, SAR and skin corrosivity testing. Such evaluation may lead to the severe eye irritation hazard determination. If such cannot be predicted, either a severe eye irritation hazard may be accepted as a worst case scenario, or an in vitro test for severe eye irritancy is used. If a positive result is obtained the material shall be considered severe irritant, whereas if a negative result is obtained a second in vitro test for irritancy may be used which may lead either to the irritant or nonirritant hazard prediction. No in vivo and human testing are suggested for hazard assessment. However, for risk assessment in which in vitro methods are suggested to be used to determine the irritation potential of diluted test material or formulations, a confirmatory test in humans is suggested to be considered for the testing of non-irritating concentrations under use exposure conditions. p 13 out of 84

15 1.6. Classification Systems At the EU level, the classification and labelling systems used have been defined in the past by: 1) for substances, by the Dangerous Substances Directive (EU DSD; EC, 2001), and 2) for mixtures, by the Dangerous Preparation Directive (EU DPD; EC, 1999). However, a Globally Harmonised System (GHS) for classification and labelling has been proposed in 2003 by the United Nations member countries (UN, 2003, 2009). Since then many countries, several international organizations as well as United Nations programmes and specialized agencies concerned with chemical safety in the field of transport or the environment, occupational health and safety, pesticide management and prevention and treatment of poisoning, are in the process of implementing the GHS. The European Union has introduced the GHS classification and labelling system since 2008 within the Regulation on the Classification, Labelling and Packaging of Substances and Mixtures (EU CLP; EC, 2008b). As a consequence the EU DSD and EU DPD classification systems are currently being progressively replaced by the novel EU CLP classification system. The timelines foreseen for the progressive implementation of the EU CLP are described below and summarized in figure 4. Until 1 December 2010 Substances and mixtures shall be classified, labelled and packaged in accordance with EU DSD and EU DPD, respectively. They may also be classified, labelled and packaged in accordance with EU CLP. In that case they shall not be labelled and packaged according to EU DSD or EU DPD. When a substance or mixture is classified, labelled and packaged according to EU CLP the classification information according to both systems shall be provided in the Safety Data Sheet. From 1 December 2010 to 1 June 2015 Substances shall be classified, labelled and packaged in accordance with EU CLP, but also classified in accordance with EU DSD in order to allow these classifications to be used in the classifications of mixtures. Classifications in accordance with both systems shall be included in Safety Data Sheets, but classifications in accordance with EU DSD shall not appear on the label. Mixtures shall be classified, labelled and packaged in accordance with EU DPD. They may also be classified, labelled and packaged in accordance with EU CLP. In that case they shall not be labelled and packaged according to EU DPD. When a mixture is classified, labelled and packaged according to EU CLP the classification information according to both systems shall be provided in the Safety Data Sheet. From 1 June 2015 Both substances and mixtures shall be classified, labelled and packaged in accordance with EU CLP. EU DSD and EU DPD are repealed from 1 June 2015 and classification according to these directives is not allowed. However, substances classified, labelled and packaged in accordance with EU DSD and already placed on the market ( on the shelves ) before 1 December 2010, and mixtures classified, labelled and packaged in accordance with EU DPD and already placed on the market ( on the shelves ) before 1 June 2015, do not have to be relabelled and repackaged in accordance with EU CLP until 1 December 2012 and 1 June 2017, respectively. p 14 out of 84

16 st Dec 1st Dec 1st June 1stJune Substances & Mixtures: EU DSD & DPD (or EU CLP) SDS* : EU DSD or DPD (and EU CLP if applied) Substances: EU CLP (& DSD but no label) SDS*: EU CLP & DSD Relabeling of Substances C&L EU DSD and placed on market before 1 Dec 2010 Mixtures: EU DPD (or EU CLP) SDS*: DPD (and EU CLP if applied) Substances & Mixtures: EU CLP only Relabeling of Mixtures C&L DPD and placed on market before 1 June 2015 Figure 4. Schematic view of the progressive implementation in the EU of the new EU CLP classification system, and phasing out of the EU DSD and DPD. *SDS: Safety Data Sheet The scores and in vivo observations used for classifying substances for eye irritation or severe eye irritation according to the UN GHS, EU DSD and EU CLP are given in Chapter 3. p 15 out of 84

17 2 In vivo Acute Eye Irritation Testing The standard method for determining whether a substance is capable of producing ocular irritation or severe eye irritation is the rabbit eye test developed by Draize and co-workers (1944). In the original test nine rabbits were used. The treated eyes of two groups of three rabbits each were washed with 20 ml water 2 or 4 seconds after instillation of the material. In the third group of three remaining rabbits, the treated eyes remained unwashed. This test was initially developed to evaluate products that come into contact with the eye and ocular adnexa such as ophtamological preparations and cosmetics. It was incorporated in 1964 in the United States (US) into the Food and Drug Administration (FDA) regulations dealing with pharmaceuticals and cosmetics, and then in 1972 also into the guidelines of the Consumer Product Safety Commission (CPSC) dealing with household products. The test was subsequently modified to use fewer animals and was specified by the US Federal Hazardous Substances Act (FHSA) which deals with agricultural and industrial chemicals. In 1978, the US Environmental Protection Agency (EPA) proposed a testing guideline for ocular irritation testing on pesticides. Finally the OECD test guidelines program have recommended since 1981 slightly modified versions of the test (ILSI, 1996; Wilhelmus, 2001). Currently, depending on the regulatory agency, the number of rabbits required for a study of ocular irritation can vary. The current OECD, EU and EPA regulations require the testing in three animals. Indeed, Springer and co-workers (1993) have shown that reducing the sample size from six to three animals preserved very good accuracy. The authors recommended using three animals either in a one-stage or in a two-stages approach by using first two animals before the need for testing in one more animal. However, some U.S. agencies (e.g., CPSC, FDA) still require at least six rabbits to be examined to classify the effects produced by a test substance (CPSC, 2003). The washing procedures also vary depending on the regulatory context. The current EPA and CPSC guidelines (as well as the previous version of the OECD TG 405) do not allow washing of the eyes before 24 hours post-treatment. In its current version, the OECD TG 405 as well as the EU B.5 test method, both make an exception for solids, which may be removed with saline or water after 1 hour posttreatment (OECD, 2002; EC, 2008a; EPA 1998). To minimize pain and suffering of rabbits exposed to potentially corrosive agents, OECD and EU test guidelines also suggest the use of a sequential testing. If a test substance is anticipated to produce a severe effect (e.g., corrosive effect), a test in a single rabbit may be conducted. If a severe effect is observed in this rabbit, further testing does not need to be conducted. In cases where no severe effects are observed, two additional rabbits can be examined to classify the ocular effects produced by the test substance (OECD, 2002, EC, 2008a). Furthermore, both guidelines recommend the use of a tiered testing strategy as descried in chapter Mechanisms of eye irritation and severe eye irritation The ocular system is a complex and specialized organ responsible for the vision and for the detection, localization and analyses of light. Anatomically it comprises an exterior compartment with the cornea and sclera, an intermediate compartment with the anterior (iris and ciliary body) and posterior (choroid) parts, and an internal compartment with the sensory part of the eye, the retina (figure 5). p 16 out of 84

18 Figure 5: Schematic diagram of the human eye. The external layer of the eyes, which may first enter in contact with external potential noxious agents, is represented by the cornea. It is a transparent organ that covers the iris, pupil and anterior chamber and can be divided into the following layers (figure 6): - The epithelium layer, which is continuous with the conjunctival epithelium, and consists of a stratified tissue of fast-growing and easily-regenerated cells that multiply from the basal layer; - The Bowman s layer, which consists of irregularly-arranged collagen fibers; - The Corneal stroma, which represents 85-90% of the corneal thickness and consists of regularly-arranged collagen fibers and sparsely distributed interconnected keratocytes (specialized fibroblasts residing in the stroma); - The Descemet s membrane, which is a thin acellular layer that serves as the modified basement membrane of the corneal endothelium; - The Endothelium layer, which represents a monolayer of mitochondria-rich cells responsible for regulating fluid and solute transport between the aqueous humour and the corneal stroma. Ocular injury can be caused by different insults such as chemical (e.g., toxic, alkali or acid), mechanical (e.g., contunsion, compression, iris scratch, surgical trauma) or thermal (e.g., laser light) (Unger, 1990). In the case of a chemical insult, the chemical injury is determined by the chemical nature (e.g., category, class, and functional groups), its concentration, ph, the form and length of exposure and by the effectiveness of the repairing mechanisms (Berta, 1992; Pfister, 1983). If in most cases chemical injuries of the eye are relatively minor, alkaline and acidic substances may cause more severe ocular damage. Perhaps for this reason, more is known about alkali and acidic injuries than about other chemical injuries (McCulley, 1987). Other types of chemicals that may also injure the eye can be divided into inert chemicals, solvents and lachrymators (ECHA, 2008a appendix R.7.2-1). A brief description of the different mechanisms of action for these chemical types is given below. p 17 out of 84

19 1. Epithelium 2. Bowman s membrane (or anterior elastic lamina) 3. Corneal stroma (or substantia propria) a. Oblique fibers in the anterior layer of the substantia propria b. Lamellæ the fibers of which are cut across, producing a dotted appearance c. Corneal corpuscles appearing fusiform in section d. Lamellæ the fibers of which are cut longitudinally e. Transition to the sclera, with more distinct fibrillation, and surmounted by a thicker epithelium f. Small blood vessels cut across near the margin of the cornea 4. Descemet s membrane (or posterior elastic lamina) 5. Endothelium of the anterior chamber Figure 6: A vertical section of human cornea. 1) In a first stage (0 to 5 min), alkalies produce a high ph which saponifies the fatty components of the cell membrane leading to cell disruption. They also denaturate collagen and cause the swelling of collagen fibers and the loss of glycosaminoglycans in the corneal stroma. Depending on the depth of penetration, destruction to some or all of the following structures can occur: corneal and conjunctival epithelium, keratocytes, corneal nerves, endothelium of the cornea and blood vessels, cellular and vascular components of iris and ciliary body, and lens epithelium. In a second stage, the production of degradative and the release of intracellular substances with vasoactive and chemotactic action initiates the process of inflammation. This stage is primarily characterised by the formation of edema in the corneal stroma and by the invasion of leukocytes. In particular polymorphonuclear leukocytes (PMNs) are attracted to the region of tissue damage, infiltrating the stroma. By this time and as a third stage (1 week), the epithelium may start regenerating, and the circulation in the limbal blood vessels be renewed. These processes include the regeneration of epithelium gradually covering the denuded surface of the stroma, the synthesis of collagen and glycosaminoglycans, scar formation and the vascularization of the cornea. Corneal opacity begins to clear and, in mild to moderate cases, may completely resolve during this period. However, in some cases corneal ulceration can ensue, 2 or 3 weeks after the injury. During this fourth phase a progressive destruction of the cornea occurs and a stromal ulcer is formed, mediated by hydrolytic enzymes produced and released from the lysosomes of PMNs, coupled with inadequate collagen synthesis. Other sources of hydrolytic enzymes are the epithelial cells, fibroblasts, keratocytes and microorganisms. When the cornea has reepithelialized or when the corneal stroma becomes totally vascularized, corneal ulceration ceases. Epithelial healing blocks the penetration of PMNs into the corneal stroma from the tear film and also seems to inactivate the synthesis of collagenase by the white blood cells present in the stroma (Berta, 1992, Pfister, 1983; McCulley, 1987; Lemp, 1974). p 18 out of 84

20 Other effects ensuing from alkalies severe injuries may include the permanent loss of corneal innervation, with resultant neurotrophic keratitis which may occur in the late reparative phase. Injury to the lacrimal glands may also occur resulting in tear deficiency. Tear film abnormalities are one of the more common and important sequelae of all but mild alkali burns and lead to epithelial abnormalities manifested by epithelial cell damage and keratinization. Intense pain, lacrimation, and blepharospasm accompanying a chemical ocular insult may result from direct stimulation of free nerve endings located in the epithelium of the cornea and conjunctival lining. It probably acts as a defence mechanisms to minimize the potential damage to the eye by producing reflex mechanisms such as lacrimation, blinking, and head withdrawal. However, this stimulation and neurologic pain response, may also lead to neurogenic inflammation. Mediated by neuropeptides, the neurogenic inflammation could be responsible for inflammation, vasodilation and increased permeability, with increased number of cells of the immune system, in particular PMNs leukocytes that infiltrate the tear film over the front of the eye and conjunctiva (ECHA, 2008a; ILSI, 1996; McCulley, 1987; Pfister, 1983). 2) With regard to acid injuries, coagulation and precipitation of proteins is induced causing tissue destruction. In general, however, acids produce less severe injuries than do alkalies, due to the fact that the acid coagulates the epithelial surface, forming a relative barrier to penetration of the acid, and that the corneal stroma has a buffering effect on acid solutions with a ph less than 4. An acid insult will penetrate deep into the eye only in case of severe necrosis. Acids also react with collagen, resulting in a shortening of collagen fibers, which causes a rapid increase in intraocular pressure. The turbidity seen after acid burns is probably not caused by changes in collagen but rather by precipitation of the extracellular glycosaminoglycans (McCulley, 1987). 3) With regard to the other chemical types, solvents may dissolve lipids in plasma membranes of epithelial and underlying cells resulting in loss of the cells affected and, as a result tissue degradation, that might be, depending on the repair mechanisms (cell proliferation, tissue restoration), transient. Inert chemicals may cause effect due to large size, where protusions may cause direct puncture of the eye. Lachrymators may stimulate the sensory nerve endings in the corneal epithelium causing an increase in tearing (ECHA, 2008a). Besides the effects on cornea, inflammation of the conjunctiva can be induced. This includes dilation of the blood vessels causing redness, increased effusion of water causing swelling (oedema/chemosis) and an increase in the secretion of mucous leading to an increase in discharge. Iritis may also occur as a result from direct irritation or become a secondary reaction to the corneal injury. Once the iris is inflamed, infiltration of fluids can follow which affects the ability to adjust the size of the pupil and decreases the reaction to light leading to decreased visual acuity. Due to the richness of nerves in the iris, irritation also causes subjective symptoms such as itching, burning and stinging (ECHA, 2008a). Eye injury can be reversible or irreversible depending on the degree of damage and degree of repair. Damage to the corneal epithelium alone can repair quickly, often with no permanent eye damage. The cornea may still repair fairly well if the damage goes beyond the basement membrane into the superficial part of the stroma but the repair process may take days or even weeks to occur. Once the damage extends significantly into the stroma, corneal ulceration can occur due to the subsequent series of inflammatory processes. If damage extends to and beyond the endothelium, corneal perforation may occur which is irreversible and may cause permanent loss of vision (ECHA, 2008a). Maurer, Jester and co-workers (2002) have proposed that the level of ocular irritation is related to the extent of initial injury, and that regardless of the processes leading to tissue damage, the extent of initial injury is the principal factor determining the outcome of ocular irritation. The authors have examined the in vivo mechanistic basis for ocular irritation by using chemicals of different classes, including surfactants (anionic, cationic, and non-ionic), acids, alcohol, aldehyde, alkali and bleaches. They have shown that depth of injury to the cornea, in the early hours after exposure (usually 3 hours), can be predictive of the eventual degree and duration of the ocular lesions in the rabbit. In general, slight irritants were found to affect only the superficial corneal epithelium, mild and moderate irritants p 19 out of 84

21 principally affect the epithelium and superficial stroma, and severe irritants act through to deeper parts of the stroma, potentially as far as full stromal depth. The reactive chemistries (e.g. bleaches) showed a more-delayed onset of toxicity and emphasised the need for evaluation of depth of injury one day after exposure. These authors also concluded that the depth of injury measurements were more consistent for each test material, over time, than were the macroscopic tissue scores, and that reversibility of the lesions was correlated with the initial depth of the injury. The link between depth of injury (cellular lesions) and macroscopic observations could provide a mechanistic basis for the development of new in vitro/ex vivo assays for the prediction of depth of injury (depth of cytotoxicity) within the cornea (Maurer et al., 2002, Eskes et al., 2005) Method description according to OECD TG 405 The procedure is described in details in the OECD TG 405 (2002). The principal steps of the in vivo testing are described here below. a) Animals used Healthy young adult albino rabbit is recommended. A rationale for using other strains or species should be provided. b) Dose and application of the test substance Liquids: 0.1 ml, pump sprays should not be used. Solids: An amount that has a volume of 0.1 ml or a weight of no more than 100 mg, the test material should be ground to fine dust. The test substance is placed in the conjunctival sac of one eye of each animal after gently pulling the lower lid away from the eyeball. The lids are then gently held together for about one second in order to prevent loss of the material. The other eye, which remains untreated, serves as control. Aerosols: All pump sprays and aerosols should be collected prior to installation into the eye. The one exception is for pressurised aerosols which cannot be collected due to vaporisation. In such cases, the eye should be held open and the test substance administered to the eye in a simple burst of about one second, from a distance of 10 cm directly in front of the eye. This distance may vary depending on the pressure of the spray and its contents. An estimate of the dose may be made as outlined in the test guideline. c) Exposure time The eyes of the test animals should not be washed for at least 24 hours following instillation of the test substance, except for solids and in case of immediate corrosive or irritating effects. In the case of liquids and aerosols, a 24 hours washout may be used if considered appropriate. In the case of solids, if the test substance has not been removed from the eye by physiological mechanisms at the 1 hour post-treatment observation time, the eye may be rinsed with saline or distilled water. d) Sequential testing An initial test using one animal is recommended. If the results of the test indicate the substance to be corrosive or severe irritant to the eye, further testing for ocular irritancy should not be performed. If corrosive effect is not observed in the initial test, the irritant or negative response should be confirmed using up to two additional animals. It is suggested to conduct the confirmatory test in a sequential manner in one animal at a time, rather than exposing the two animals simultaneously. If the second animal reveals corrosive or severe irritant effects, the test is not continued. Additional animals may be needed to confirm weak or moderate irritant response. p 20 out of 84

22 e) Observation period Animals should be observed normally for 21 days post administration of the test substance to determine reversibility of effects. If reversibility is seen before 21 days, the experiment should be terminated at that time. However animals that do not develop ocular lesions may be terminated not earlier than 3 days post instillation. The eyes should be examined at 1, 24, 48, and 72 hours after the test substance application. In case of mild to moderate lesions, animals should be observed until the lesions clear or for 21 days, where observations should be performed at 7, 14, and 21 days. However, if animal shows continuing signs of severe pain or distress, the experiment should be terminated at any time. f) Grading of ocular reactions The grades of ocular reaction (conjunctivae, cornea and iris) should be recorded at each examination as described in table 2. Any other lesions in the eye (e.g., pannus, staining) or adverse systemic effects should also be reported. Table 2: Grading of Ocular Lesions Cornea Opacity: degree of density (readings should be taken from most dense area)* No ulceration or opacity 0 Scattered or diffuse areas of opacity (other than slight dulling of normal lustre); details of iris clearly visible 1 Easily discernible translucent area; details of iris slightly obscured 2 Nacrous area; no details of iris visible; size of pupil barely discernible 3 Opaque cornea; iris not discernible through the opacity 4 Maximum possible: 4 * The area of corneal opacity should be noted Iris Normal 0 Markedly deepened rugae, congestion, swelling, moderate circumcorneal hyperaemia; or injection; iris reactive to light (a sluggish reaction is considered to be an effect 1 Hemorrhage, gross destruction, or no reaction to light 2 Maximum possible: 2 Conjunctivae Redness (refers to palpebral and bulbar conjunctivae; excluding cornea and iris) Normal 0 Some blood vessels hyperaemic (injected) 1 Diffuse, crimson colour; individual vessels not easily discernible 2 Diffuse beefy red 3 Maximum possible: 3 Chemosis Swelling (refers to lids and/or nictating membranes) Normal 0 Some swelling above normal 1 Obvious swelling, with partial eversion of lids 2 Swelling, with lids about half closed 3 Swelling, with lids more than half closed 4 Maximum possible: 4 Examination of reactions can be facilitated by use of a binocular loupe, hand slit-lamp, biomicroscope, or other suitable device. After recording the observations at 24 hours, the eyes may be further examined with the aid of fluorescein. In case of the following other eye lesions, animals should be humanely killed: corneal perforation or significant corneal ulceration including staphyloma; blood in the anterior chamber of the eye; grade 4 p 21 out of 84

23 corneal opacity which persists for 48 hours; absence of a light reflex (iridial response grade 2) which persists for 72 hours; ulceration of the conjunctival membrane; necrosis of the conjunctivae or nictitating membrane; or sloughing. This is because such lesions generally are not reversible Limitation of the Draize eye rabbit test Extensive literature on the limitations of the in vivo Draize eye irritation test can be found. While the current test method is widely used, the test has often been reason for criticism. Besides the fact that the Draize test can be very painful to the rabbits, some of the drawbacks referred in literature and detailed below are: - the type and duration of exposure of the test material, - the limited reproducibility within and between the laboratories that may be due amongst other factors to the subjectivity in the allocation of the respective scores and to the rabbits biological variability, - the differences in physiology and sensitivity to tested substances between rabbit and human eyes. - ethical issues Test material exposure The Draize rabbit eye test was initially developed to evaluate products that come into contact with the eye and ocular adnexa such as ophtamological preparations and cosmetics. Only later it has been incorporated into testing guidelines for industrial chemicals, household products or pesticides to estimate accidental exposure to the human eye (Wilhelmus, 2001). However, such exposure conditions, i.e., in the conjunctival sac and manual eyelid closure that prolong the duration and extend of exposure, are not consistent with accidental exposures of the human eye which generally occur by direct contact with the corneal surface (Griffith et al., 1980). It is believed that the response experienced by humans would be more readily duplicated in animals by corneal application than by conjunctival instillation (ILSI, 1996). Another criticism is that in the Draize rabbit eye test the precise exposure times and/or delivery of dosage remain actually unknown. It might depend on the time the animals take to flush the test material from the eyes, and on the test material properties. For example, for aqueous and non-viscous formulations the standard instillation results in a rapid removal of the material within seconds/minutes through blinking with the nictitating membrane (third eye-lid) and grooming by the rabbit. This contrasts with the situation for sticky pastes for example, which cannot be removed that easily. In these cases, the contact time may vary from a couple of minutes to 24 h, because rinsing the eye is not allowed before the 24 hours reading. The most dramatic variation in contact time and dosage occurs with solids. Even if applied as a 0.1 ml equivalent (the content of the cul-de-sac), the actual amount of a powder/solid that stays in contact with the eye is unpredictable. The contact time may also vary from a couple of minutes to 24 h depending on the guidelines and their application. Indeed, if generally rinsing the eye is not allowed before the 24 hours reading, in the updated OECD TG 405 and EU B.5 test methods an exception is made for solids, which may be washed 1 hour after exposure. Still enclosure of test materials in the conjunctival cul-de-sac in combination with mechanical damage, could have devastating effects. For example in the case of poorly water-soluble solids with distinct cytotoxic properties, swelling of the conjunctivae may occur making it even more difficult for the animal to remove the test material. Such forced continuous exposure for up to 24 hours may result in a complete closure of the eye lids by the abundant production of colloidal discharge which often forms a sealing crust. Upon opening these sealed eye-lids, purulent discharge, and other inflammatory debris are released. The degree of swelling of the conjunctivae can be sufficiently severe such that removal of any remains of the test substance is hardly possible anymore. Also in the case of less severe effects, the eye can become vulnerable to microbiological infection, or secondary inflammation process, causing initial mild to moderate effects during the first days after exposure developing into more severe and prolonged effects during the 21 day observation period. p 22 out of 84

24 Such uncertainty in the true amount, duration and delivery of the test material, may lead to variations on the way animals lacrimate and flush out the test material resulting in possible variation in the reactions from the animals even before the scoring of effects takes place (Prinsen, 2006) Variability of the Draize eye irritation test Several studies report on a limited within- and between-laboratory reproducibility of the Draize rabbit eye test, found especially in the middle range of mild to moderate irritating compounds (Wilhelmus, 2001; ILSI, 1996; Balls et al., 1999). In addition to the variations linked to the exposure conditions as described above (e.g., rate of release of the test product from the delivery vehicle, the amount of reflex tearing, the exposure duration and timing of post-exposure irrigation) other described sources of variation include the subjective scoring systems and the differences in the individual animal responses of the same species and strain (Wilhelmus, 2001; Eskes et al., 2005; Balls et al., 1999). The largest study on variability is perhaps the one from Weil and Scala (1971) who studied 9 test substances in up to 24 laboratories. The authors show that the Draize eye test can produce quite variable results among laboratories as well as within certain laboratories. Certain materials were rated as the most irritating tested by some laboratories and, contrariwise, as the least irritating by others. Figure 7 represents an extract of Weil and Scala s findings, where the maximum and minimum eye irritation scores found for the different tested substances and laboratories are given (the minimum and maximum attributable rates being 0 and 110 respectively). As it can be seen, the test material with the smallest variability had scores varying from 0 to 63. The authors suggest that the primary reason for the observed extreme variation between laboratories is in the reading of reactions. Unconscious bias or definite tendencies to over- or under-read reactions or misinterpret the meaning of descriptive terms may have accounted for that. In addition, variation in interpretation of and performance of the procedures was also reported as a component for the observed interlaboratory variability (Weil and Scala, 1971). Marzulli and Ruggles (1973) have studied 7 test materials in 10 laboratories and have confirmed the findings from Weil and Scala. Although the authors suggest that laboratories were able in most cases to distinguish eye irritants from non irritants, statistically significant differences were found between collaborators with regard to the tissue readings. Figure 7: Extract from Weil and Scala (1971). Minimum and maximum scores for eye irritation of individual rabbits. p 23 out of 84

25 A quantification of the Draize rabbit eye test variation has been done by Cormier et al. (1996). The authors have estimated the coefficient of variation (CV) of the Draize Maximum Average Scores (MAS, see chapter 3.5 for details) obtained for 1 material tested in several laboratories in 13 studies, and found it to be in the order of 38%. The authors also calculated the CVs from the MAS scores obtained at the 24h observation time in the work of Weil and Scala (1971), and have found it to range from 42 to 59% for the 9 test materials tested in 24 laboratories. Gettings and co-workers (1996) have studied 25 surfactant-based formulations using 6 replicate animals tested in different randomized blocks to reflect within-laboratory between-test variability. They have found a higher standard error for the middle range of mild to moderate irritants. Earl and coworkers (1997) made a review of existing studies on the Draize rabbit eye test variability and show that variability, as measured by the standard deviation (SD) of the MAS (see chapter 3.5 for details), was consistently found to be greatest in the middle ranges of irritancy, and the lowest at the extremes of the scales. Ohno and co-workers (1999) also confirmed these findings and showed that variation in Draize scores was larger for mild and moderate irritants (MAS scores from 15 to 50) with standard deviations going up to 50 over the MAS scores. The authors also showed that variability could be due to differences in the grading techniques depending on the institutes and scientist, but also due to the sensitivity of the rabbit eyes from the individual animals and/or strains Inter-species differences and Predictive Capacity of the Draize eye irritation test Rabbits are often preferred over other animals for their large eyes with well-described anatomy and physiology, ease of handling, and availability. However, the eyes of rabbits appear to present several anatomical and physiological differences with regard to the human eye, they are not able to measure ocular pain and sting, and are in general more sensitive to irritating substances than the eyes of humans (Roggeband et al., 2000; Gershbein and McDonald, 1977; Wilhelmus, 2001; ILSI, 1996). Physiologically, the rabbit eyes show lower tear production, less developed blink reflex, a thinner cornea (0.4 for rabbits versus 0.53 to 0.54 for humans) and a larger corneal surface area (Beckeley, 1965; Wilhelmus, 2001; ILSI, 1996). Furthermore, the Bowman s layer is not present in the rabbit eyes, and rabbits have larger conjunctival sac which allows for larger test volumes to be instilled than what could be accounted for on human accidental exposure (Wilhelmus, 2001; Curren and Harbell, 1998; ILSI, 1996). On the other hand, rabbits have a nictating membrane that functions as a third eyelid and may help removing irritating substances from the corneal surface although in a different way than in humans. Rabbit eyes were also reported to have different constituents of the tear film, and different ocular pigmentation (Wilhelmus, 2001; ILSI, 1996). These elements could all contribute for differences in the responses of the rabbit eyes to irritants with respect to the human eyes. Several studies have shown that rabbits seem to be amongst the most sensitive species that react to ocular irritation insults. Gershbein and McDonald (1977) have studied the corneal irritancy developed by various species by testing four shampoos and two cationic detergents. They found that corneal sensitivity was highest in the rabbit, hamster and mouse; intermediate in the rat and guinea pig, and possibly lowest in the dog, cat, rhesus monkey and chicken. Beckeley and co-workers (1969) compared the ocular irritation effects of a soap formulation in humans, monkeys and rabbits. The authors also showed that rabbits over-predicted the human responses, and that the monkey was more accurately predicting the nature of the human hazard. Bito (1984) also described the rabbit eye as the most sensitive of the species studied, whereas the primate eyes were the least sensitive, and proposed an evolutionary reasoning for such differences. Griffith and co-workers (1980) have compared the effects of 21 test materials in rabbits and man. The authors showed that the rabbit eye develops more intense responses to many chemicals than does the human eye, and that the period of recovery extends beyond that seen in typical chemical exposures in man. Freeberg and co-workers (1984) showed that rabbits produced more severe eye responses than those reported from human eye accidents with ten household consumer products. The same group of authors have further compared the reactions of rabbits and humans to four household p 24 out of 84

26 products (Freeberg et al., 1986), and showed again that the Draize eye test was poorly predictive of the human recovery time, i.e., the rabbit and human mean time to clear presented a correlation of 0.35 to Roggeband and co-workers (2000) confirmed such findings, by testing two undiluted liquid detergents in 29 human volunteers and 12 rabbits. The authors found that effects in the rabbit were greater than the effects observed in man Ethical issues Perhaps because of the fact that the Draize rabbit eye test produces can be very painful and result in readily visible suffering, trauma and reactions in the rabbit eyes, animal activists have often used this assay as a symbol for cruelty. In the 1980s, the animal right activist Henry Spira specifically targeted the Draize eye test by publishing a full-page advertisement in the New York Times asking, "How many rabbits does Revlon blind for beauty's sake?" (figure 8). In addition, the existence of extensive literature on the limitations of the assay and on its poor scientific quality, as well as the major developments which took place since the 80s to advance alternatives to reduce, replace and refine the Draize rabbit eye (Balls et al., 1999; Eskes et al., 2005), make some authors consider that the ethical balance weights against the Draize eye test (Wilhelmus, 2001; Rowan, 1980). Figure 8: Full-page advertisement seeking to influence public opinion about the Draize eye test in the New York Times, appeared on 15 April 1980 and 7 October 1980 (from Wilhelmus, 2001). p 25 out of 84

27 3 Classification and Labelling The in vivo single tissue and reversibility observations used for classifying substances for eye irritation or severe irritation effects according to the UN GHS, EU DSD and EU CLP are given below. Note that such classification systems are to be considered within the framework of the sequential testing strategies recommended by the UN, OECD and EU (for details see chapter 1.5.2). In these strategies a new in vivo rabbit eye test is performed only as a last step, i.e. when the assessments in all the other tiers have produced negative results, and whereby only one animal is required for the testing of severe eye irritants. Furthermore, details for the EU DPD are not shown since it is not based on classifiable in vivo observed effects (for details refer to EC, 1999) The UN Globally Harmonised System (GHS) for classification & labelling (UN, 2003, 2009) The United Nations Globally Harmonized System for the classification and labelling of hazardous chemicals is summarized in Table 3. It proposes the two following categories: irreversible effects on the eye / serious damage to eyes classified as Category 1, and reversible eye effects / irritating to eyes classified as Category 2. Within category 2 an optional sub-categorisation is allowed, where 2A is given to substances irritating to eyes as described in table 3, and 2B is given for mildly irritating to eyes when effects are fully reversible within 7 days of observation. Table 3: GHS criteria for classification (extracted from UN, 2003, 2009). Scores are calculated as the mean scores following grading at 24, 48 and 72 hours after installation of the test material. Category 2A Reversible eye effects / Irritating to eyes (reversible within 21 days) Category 1 Irreversible effects on the eye / serious damage to eyes Corneal opacity at least 2 animals (out of 3) 1.0 at least 2 animals (out of 3) 3.0 Iritis at least 2 animals (out of 3) 1.0 at least 2 animals (out of 3) 1.5 Conjuctival redness at least 2 animals (out of 3) 2.0 n.a. Conjunctival oedema (chemosis) Reversibility Other effects n.a. not applicable at least 2 animals (out of 3) 2.0 Optional Cat. 2B (mildly irritating to eyes): all effects fully reversible within 7 days n.a. n.a. At least one animal effects not expected to reverse, or not fully reversed within 21 days Animals with grade 4 cornea lesions and other severe reactions (e.g., destruction of cornea) observed at any time during the test, as well as discoloration of the cornea by a dye substance, adhesion, pannus, and interference with the function of the iris or other effects that impair sight. However, the GHS classification system does not yet provide with detailed decision criteria for classification in case of studies where more than 3 animals are used. This may cause uncertainties as how the different regulatory systems may apply the GHS classification system. In particular, it remains unclear which proportion and/or how many animals are required for applying the Cat. 1 or Cat. 2 classification based on the tissue effects as described in table 3. In addition, it does not specify what is meant by fully reversibility. Personal communication have indicated that full reversibility may be considered when scores of 0 were observed for all endpoints of relevance, i.e., corneal opacity, iris lesions, conjunctival redness and conjunctival chemosis at a certain day up to and including day 21 of the study. p 26 out of 84

28 3.2. The new EU CLP classification system (EC, 2008b) The new EU CLP classification system is similar to the UN GHS classification system as described above. It defines the two following categories: irreversible effects on the eye / serious eye damage classified as Category 1, and reversible effects on the eye / irritating to eyes classified as Category 2. In this later category, only one category is applied for irritating substances without the optional GHS sub-categorisation (Cat. 2A and 2B). Moreover, the REACH Guidance on the application of the CLP criteria (ECHA, 2009) has recently established rules on how to apply the EU CLP in case that more than 3 animals are used. The following detailed decision criteria are given: In case of 6 rabbits the following applies: - Classification as serious eye damage Category 1 if at least in one animal effects on the cornea, iris or conjunctiva that are not expected to reverse or have not fully reversed within an observation period of normally 21 days; and/or at least 4 out of 6 rabbits show a mean score of 3 for the cornea and/or 1.5 for the iris - Classification as eye irritation Category 2 if at least 4 out of 6 rabbits show a mean score of 1 for the cornea and/or 1 for the iris and/or 2 conjunctival erythema and/or 2 conjunctival swelling In case of 5 rabbits the following applies: - Classification as serious eye damage Category 1 if at least in one animal effects on the cornea, iris or conjunctiva that are not expected to reverse or have not fully reversed within an observation period of normally 21 days; and/or at least 3 out of 5 rabbits show a mean score of 3 for the cornea and/or 1.5 for the iris - Classification as eye irritation Category 2 if at least 3 out of 5 rabbits show a mean score of 1 for the cornea and/or 1 for the iris and/or 2 conjunctival erythema and/or 2 conjunctival swelling In case of 4 rabbits the following applies: - Classification as serious eye damage Category 1 if at least in one animal effects on the cornea, iris or conjunctiva that are not expected to reverse or have not fully reversed within an observation period of normally 21 days; and/or at least 3 out of 4 rabbits show a mean score of 3 for the cornea and/or 1.5 for the iris - Classification as eye irritation Category 2 if at least 3 out of 4 rabbits show a mean score of 1 for the cornea and/or 1 for the iris and/or 2 conjunctival erythema and/or 2 conjunctival swelling In this case the categories 1 and 2 are used if 4 of 6 rabbits show a mean score as outlined in the criteria. Likewise, if the test was performed with 4 or 5 animals, for at least 3 individuals the mean score must exceed the values laid down in the classification criteria. A single animal showing irreversible or otherwise serious effects consistent with corrosion will necessitate classification as serious eye damage Category 1 irrespective of the number of animals used in the test. p 27 out of 84

29 However, no clear mentions are made regarding what is meant by fully reversible effects; or for the effects of corneal score of 4 observed at any time where 1 animal out of 3 is needed to assign category The EU DSD classification system (EC, 2001) The EU outphasing DSD classification and labelling system has been applied in accordance to the Commission Directive 2001/59/EC (EU, 2001) and is based on the eye irritation testing method B.5 of Annex V as reported in the Commission Directive 2004/73/EC (EU, 2004), which is equivalent to the OECD TG 405 (2002). The EU classification criteria for ocular lesions (eye irritation / severe eye irritation) are summarized in Table 4 and reported below as quoted in the Commission Directive 2001/59/EC (EU, 2001): The following risk phrases shall be assigned in accordance with the criteria given: R36 Irritating to eyes - Substances and preparations which, when applied to the eye of the animal, cause significant ocular lesions which occur within 72 hours after exposure and which persist for at least 24 hours. Ocular lesions are significant if the mean scores of the eye irritation test cited in Annex V have any of the following values: - cornea opacity equal to or greater than 2 but less than 3, - iris lesions equal to or greater than 1 but not greater than 1.5, - redness of the conjunctivae equal to or greater than 2.5, - oedema of the conjunctivae (chemosis) equal to or greater than 2, or, in the case where the Annex V test has been completed using three animals if the lesions on two or more animals, are equivalent to any of the above values except that for iris lesion the value should be equal to or greater than 1 but less than 2 and for redness of the conjunctivae the value should be equal to or greater than 2.5. In both cases all scores at each of the reading times (24, 48 and 72 hours) for an effect should be used in calculating the respective mean values. - Substances or preparations which cause significant ocular lesions, based on practical experience in humans. - Organic peroxides except where evidence to the contrary is available. R41 Risk of serious damage to eyes - Substances and preparations which, when applied to the eye of the animal cause severe ocular lesions which occur within 72 hours after exposure and which persist for at least 24 hours. Ocular lesions are severe if the means of the scores of the eye irritation test in Annex V have any of the values: - cornea opacity equal to or greater than 3, - iris lesion greater than 1.5. The same shall be the case where the test has been completed using three animals if these lesions, on two or more animals, have any of the values: - cornea opacity equal to or greater than 3, - iris lesion equal to 2. In both cases all scores at each of the reading times (24, 48 and 72 hours) for an effect should be used in calculating the respective mean values. Ocular lesions are also severe when they are still present at the end of the observation time. Ocular lesions are also severe if the substance or preparation causes irreversible colouration of the eyes. - Substances and preparations which cause severe ocular lesions, based on practical experience in humans. Table 4: EU Criteria for Classification (EU, 2004). All scores at each of the reading times (24, 48 and 72 hours) for an effect should be used in calculating the respective mean values. p 28 out of 84

30 Corneal opacity Iris lesion Redness of conjunctivae Oedema of conjunctivae (chemosis) Reversibility Other effects n.a. not applicable R36 Irritant to the eyes 2.0 mean score 3.0 or animals (of 3) mean score 1.5 or animals (of 3) 2.0 mean score 2.5 or 2 animals (of 3) 2.5 mean score 2.0 or 2 animals (of 3) 2.0 n.a. - Organic peroxides (except if evidence to the contrary) - Substances causing significant ocular lesions in humans R41 Risk of serious damage to eyes mean score 3.0 or 2 animals (of 3) 3.0 mean score 1.5 or 2 animals (of 3) = 2.0 n.a. n.a. Ocular lesions still present at the end of the observation time (a maximum of 21 days) - Substances causing irreversible colouration to eyes - Substances causing severe ocular lesions in humans In the EU DSD classification system, the decision criteria to be applied in cases where more than three animals are used were given in the text of the Dangerous Substance Directive (EC, 2001). However, no detailed criterion was given on what was meant by clearance of lesions (or reversibility of effects). Personal communication from the European Chemicals Bureau informed that clearance of lesions (or reversibility of effects) was usually considered when scores of 0 were observed for all 4 endpoints of relevance (i.e., corneal opacity, iris lesions, conjunctival redness and conjunctival chemosis) at a certain day up to (and including) day 21 of the study Comparison of classification systems The comparison between the classification criteria according to the EU DSD, the EU CLP and the UN GHS is shown in figure 9 and table 5. For substances to be classified as eye irritants (R36 or Category 2) the new EU CLP and the UN GHS appear to be more sensitive than the outphasing EU DSD classification system. Indeed, the corneal and conjunctival redness scores required are lower (1 and 2 respectively for the EU CLP instead of 2 and 2.5 in the EU DSD). Furthermore, for substances to be classified as severe eye irritants (Category 1) additional criteria are given in comparison to the EU DSD (R41). For example, a Cat. 1 classification should be assigned if animals present grade 4 cornea lesions at any time during the test. Figure 9. Comparison of tissue scores leading to classification according to the EU DSD (EC, 2001), the EU CLP (EC, 2008b) and the UN GHS (UN, 2003, 2009). p 29 out of 84

31 Table 5. Comparison of reversibility and other criteria used for the classification according to the EU DSD (EC, 2001), the EU CLP (EC, 2008b) and the UN GHS (UN, 2003, 2009). Reversibility Other effects Irritants to eye EU DSD R36 GHS & EU CLP Cat.2 n.a. Optional Cat. 2B (mildly irritating to eyes): all effects fully reversible within 7 days - Organic peroxides (except if evidence to the contrary) - Substances causing significant ocular lesions in humans - (Organic) peroxides, unless evidence suggest otherwise Severe irritants to eye EU DSD R41 GHS & EU CLP Cat.1 Ocular lesions still present at the end of the observation time (a maximum of 21 days) At least one animal effects not expected to reverse, or not fully reversed within 21 days - Substances causing irreversible colouration to eyes - Substances causing severe ocular lesions in humans - Animals with grade 4 cornea lesions and other severe reactions (e.g., destruction of cornea) observed at any time during the test, - discoloration of the cornea by a dye substance, adhesion, pannus, and interference with the function of the iris or other effects that impair sight - Hydropheroxides 3.5. Other classification systems Two additional classification systems for eye irritation scoring may be relevant for critically evaluating the present report. In particular the understanding of the (Modified) Maximum Average Scores, (M)MAS, may be useful to interpret the evaluation of the in vivo Draize data as presented in chapter 2.3. Furthermore the description of the US EPA classification system may be helpful to interpret the results of the validation studies of in vitro alternatives to partially replace the Draize rabbit eye test as shown in chapter The (Modified) Maximum Average Scores If the current regulatory classification and labelling systems are based on single tissue scores as well as reversibility and other observations (see above), weighed summed scores, such as the Maximum Average Score (MAS) and the Modified MAS (MMAS) have been extensively used e.g., for the safety assessment of cosmetics or the validation of alternative test methods. The MAS weighed summed scoring system was initially described in the original Draize protocol (Draize et al., 1944). It provides with a graded continuous numerical scale where the individual tissue scores are weighted with a heavy bias for corneal injury, since injury to the cornea has the greatest probability of producing irreparable eye damage. In addition to corneal opacity, iritis, conjunctival redness and conjunctival oedema, the MAS considers the area covered by the corneal injury and the discharge induced in the conjunctivae (table 6). The MAS is obtained by averaging the individual animal weighted scores at each time of observation, and then selecting the highest (maximum) of theses averages. The corneal opacity and area scores are then multiplied together and then multiplied again by a weighting factor of 5; resulting in a maximum corneal score of 80. The iris score is multiplied by a weighting factor of 5; resulting to a maximum p 30 out of 84

32 score of 10. The scores for the three conjunctival parameters are added together and then the total is multiplied by a weighting factor of 2; resulting in a maximum score of 20. The overall score for each rabbit is calculated by adding the values for each parameter; the maximum total score is 110 (see table 6). The MAS value was later modified to the MMAS, where the scores represents the maximum value occurring at 24 hours or later after application, i.e., not including the results obtained 1 hour after treatment (ECETOC 1998). One of the major criticisms of the (M)MAS scoring system is that it does not take into account reversibility and/or irreversibility of effects. Moreover, Prinsen showed that MMAS values did not necessarily correlate with the EU or GHS classification of the substances tested (Balls et al., 1999). Table 6: Scale of Weighted Scores for Grading the Severity of Ocular Lesions (Draize et al., 1944) Lesion Score 1 Cornea A. Opacity Degree of density (area which is most dense is taken for reading) Scattered or diffuse area details of iris clearly visible 1 Easily discernible translucent areas, details of iris slightly obscured 2 Opalescent areas, no details of iris visible, size of pupil barely discernible 3 Opaque, iris invisible 4 B. Area of cornea involved One quarter (or less), but not zero 1 Greater than one quarter, but less than one-half 2 Greater than one-half, but less than three quarters 3 Greater than three quarters up to whole area 4 Score equals A x B x 5 Total maximum = 80 Iris A. Values Folds above normal, congestion, swelling, circumcorneal injection (any one or all of these or combination of any thereof), iris still reacting to light (sluggish reaction is positive) 1 No reaction to light, hemorrhage; gross destruction (any one or all of these) 2 Score equals A x 5 Total possible maximum = 10 Conjunctivae A. Redness (refers to palpebral conjunctiva only) Vessels definitely injected above normal 1 More diffuse, deeper crimson red, individual vessels not easily discernible 2 Diffuse beefy red 3 B. Chemosis Any swelling above normal (includes nictitating membrane) 1 Obvious swelling with partial eversion of the lids 2 Swelling with lids about half closed 3 Swelling with lids about half closed to completely closed 4 C. Discharge Any amount different from normal (does not include small amount observed in inner canthus of normal rabbits 1 Discharge with moistening of the lids and hairs just adjacent to the lids 2 Discharge with moistening of the lids and considerable area around the eye 3 Score equals (A + B + C) x 2 Total maximum = 20 1 Scores of 0 are assigned for each parameter if the cornea, iris, or conjunctiva are normal. p 31 out of 84

33 The EPA Classification System The U.S. Environmental Protection Agency classification and labelling system is based on the guidelines given in the Label Review Manual (EPA, 2003), and on the test method described in the Acute Eye Irritation Health Effect Test Guideline (EPA, 1998). The EPA criteria for eye irritation classification are summarized in Table 7. Table 7. The EPA Criteria for Classification (EPA, 2003) EPA Category Category I Category II Category III Category IV Classification criteria Corrosive (irreversible destruction of ocular tissue) or corneal involvement or irritation persisting for more than 21 days Corneal involvement or other eye irritation clearing 1 in 8-21 days Corneal involvement or other eye irritation clearing in 7 days or less Minimal effects clearing in less than 24 hours 1 Clearing is defined as corneal opacity or iritis < 1 and conjunctival redness or chemosis < 2 (EPA, 1998). According to EPA (1998), the following ocular scores are considered as positive: Corneal opacity 1 Iritis 1 Conjunctival redness 2 Conjunctival chemosis 2 An animal is considered positive if any of the above-mentioned grades occurs at any of the grading periods, and 1 positive animal out of generally a maximum of 6 is required for the assignment to an irritant category (OECD, 1999). When below the scores above, ocular effects can be considered as having reversed, in contrast to the EU and GHS classification systems where scores of 0 are required. The EPA classification system presents significant differences in comparison to the GHS/EU CLP and EU DSD classification systems. According to the GHS/EU CLP and EU DSD classification systems, the mean scores following grading at 24, 48 and 72 hours are used as a basis for the classification, whereas, according to the EPA, an animal is considered positive if any of the positive grades occurs at any of the grading periods (OECD, 1999). Furthermore, the GHS/EU CLP and EU DSD require that tissue effects are observed in at least 2 animals out of 3, whereas the EPA considers that 1 positive animal out of generally a maximum of 6 is considered sufficient for the assignment to an irritant category (OECD, 1999). As a consequence, the EPA classification system results in general in a higher number of classified substances (EPA Cat. III) with respect to the GHS and EU classification systems. On the other hand, persistence and clearing of effects are also defined differently for EPA as compared to the GHS/EU CLP and EU DSD classification systems. As an example, a substance which is classified EPA Cat. IV needs to have effects clearing within 24 hours, where clearing is defined as corneal opacity or iritis < 1 and conjunctival redness or chemosis < 2 (EPA, 2003, 1998). As a consequence, such substance might still have scores higher than 0 for these endpoints after 24 hours. However, according to the GHS/ EU CLP and EU DSD classification systems, if there is persistence of any tissue effects with scores higher than 0 still present at day 21 of observation, such substance would be considered a severe irritant labelled as a Category 1 or R41 (EC, 2004; UN, 2003, 2009). As such, if scores for corneal opacity or iritis become lower than 1 and/or conjunctival redness or chemosis lower than 2 in 24 hours, but are still higher than 0 at day 21, an EPA Cat. IV substance could have a GHS Cat. 1 or EU R41 classification. p 32 out of 84

34 4 In vitro Alternative Methods for Eye Irritation Testing 4.1. Introduction Perhaps because of the cruelty and the high public concern of its testing procedures, eye irritation has been a pioneer field in where major efforts were undertaken as early as in the 1980 s and 1990 s to develop, evaluate and validate in vitro methods to replace the Draize Eye Irritation test. Six major multi-laboratories studies were undertaken: - the EC/HO study in where 9 assays were evaluated, 60 chemicals tested and a total of 37 laboratories participated (Balls et al., 1995); - the COLIPA study in where 10 different assays were evaluated, 55 materials tested and a total of 32 laboratories participated (Brantom et al., 1997); - the BGA/BMBF study in where 2 assays were evaluated, 166 substances tested and up to 13 laboratories participated (Spielmann et al., 1996); - the CTFA study in where 24 test methods were evaluated and 53 materials tested (e.g., Gettings et al., 1996); - the IRAG study, in where data on 29 test methods were evaluated received from 41 laboratories (Bradlaw et al., 1997); - and the MHW/JCIA study, in where 16 test methods were evaluated, 38 substances tested and a total of 27 laboratories participated (Ohno et al., 1999). Altogether these efforts resulted in the evaluation of around 30 in vitro alternative test methods, the testing of hundreds of test materials and the participation of an extensive number of laboratories. At that time some assays showed good reproducibility and reliability, but no single method was found able to replace the Draize rabbit eye test. The main reasons identified for such outcome were: a) the variability of the in vivo eye irritation responses linked to the subjectivity of scoring, uncontrolled exposure conditions and variability of animal responses; b) the fact that in vitro tests only partially model the complex in vivo eye irritation response; c) the protocols and prediction models which might have been insufficiently developed at that time; and d) the choice of statistical approaches for analyzing the data which might not have been appropriate (for review see Balls et al., 1999). Though not formally validated, the usefulness of the in vitro methods that have undergone extensive evaluation has been established within regulatory agencies, industry and contract research organizations for specific and limited purposes (Worth and Balls, 2002). To advance the validation of in vitro alternatives that might ultimately replace the Draize rabbit eye test a thorough review was carried out in 2004 on the status of the most promising alternatives for eye irritation testing (Eskes et al., 2005). In particular, the most promising alternatives to replace the animal test were identified, which can be divided in three major groups: a) Organotypic Methods - Bovine Corneal Opacity and Permeability Test (BCOP) - Isolated Chicken Eye test (ICE) - Isolated Rabbit Eye (IRE) - Hen`s egg test on the Chorio-Allantoic Membrane (HET-CAM Assay) b) Cytotoxicity- / cell function- based methods - Neutral Red Release assay (NRR) - Red blood cell (RBC) haemolysis test - Fluorescein leakage (FL) - Cytosensor Microphysiometer (CM) c) Reconstructed human tissue models - EpiOcular TM - SkinEthic TM reconstituted Human Corneal Epithelium (HCE) model p 33 out of 84

35 It was also acknowledged that the range of criteria for injury and inflammation covered by the Draize rabbit eye test is unlikely to be covered by a single in vitro test. As such, several recommendations to progress validation efforts in view to replace the Draize rabbit eye test were made. Among those an important one was to make use of testing strategies that combines and utilises the strengths of individual in vitro test methods to address required ranges of irritation potential and/or chemical classes (Eskes et al., 2005). For that purpose, ECVAM has organised an expert meeting in February 2005 to identify potential test strategies based on the current uses and applicability of in vitro methods to eye irritation. Two testing schemes were identified proposing the use of a Bottom-Up (begin with using test methods that can accurately identify non-irritants) or Top-Down (begin with using test methods that can accurately identify severe irritants) progression of in vitro tests, based on expected irritancy of substances (for details see chapter and Scott et al., 2010). In addition to the assays identified as most promising, other assays have been submitted and were considered for evaluation, including the Ocular Irritection assay, a membrane based assay; the Slug Mucosal Irritation (SMI) assay, based on the biological reactions of slugs; and the Low Volume Eye Test (LVET) which is a refinement to the Draize rabbit eye test in which only a tenth of the sample volume used in the standard Draize test is applied directly to the cornea instead of the conjunctival sac. The performances and applicability domains of individual alternative methods which are considered as most promising to populate the proposed testing strategies were, or are currently being, determined through validation studies in a joint effort between ECVAM, it s US counterpart ICCVAM, COLIPA and industry. The validation studies on the four organotypic methods and on the four cell cytotoxicity- and cell function- based assays have been finalized and are described in more details below. These efforts led to the formal validation of two organotypic assays, the BCOP and the ICE, and of two cytotoxicity- / cell function based assays, the CM and the FL, for defined applicability domains as described below. The validation study on the LVET has also been finalized and will be discussed in chapter 5 on other alternatives to eye irritation testing. Finally, the validation efforts on the reconstructed human tissue models, on the Ocular Irritection and on the SMI assays are still ongoing and will be addressed in more details in chapter Validated assays The background on the validation studies that led to the formal endorsement of the scientific validity of the BCOP, ICE, CM and FL will be described hereafter Organotypic assays Further to an ECVAM survey with regulatory authorities, in July 2004 (Worth and Balls, 2002), the European Commission stated in its Manual of Decisions for Implementation of the 6th and 7th Amendments to Directive 67/548/EEC on Dangerous Substances, that positive outcomes of four organotypic assays, BCOP, ICE, IRE, and HET-CAM, were accepted for the classification and labelling of severe eye irritants (R41), but that a negative result required confirmation by an in vivo test (EC, 2004). Subsequently, between 2003 and 2006, ICCVAM-NICEATM conducted, together with ECVAM, a retrospective validation study on these four assays in which the ability of these assays for detecting severe eye irritants and ocular corrosives was evaluated. After peer review, two assays, the BCOP and ICE tests, were endorsed as scientifically valid to identify ocular corrosives and severe irritants in the US and in the EU (ICCVAM, 2007; ESAC, 2007). OECD Test Guidelines on the two test methods were recently adopted in September 2009 as TG 437 for the BCOP and TG 438 for the ICE and will be described here-after (OECD, 2009a,b). For the two other organotypic assays evaluated, the HET- CAM assay and the IRE test, ESAC recommended that further work was performed before a statement on their scientific validity to identify ocular corrosives and severe irritants could be made. p 34 out of 84

36 With regard to the evaluation of the four organotypic assays for identifying substances not classified as ocular irritants, a retrospective analysis of the collected data has been recently carried out by ICCVAM. The BCOP assay was the only organotypic assay that appeared suitable for the identification of non-classified materials (ICCVAM 2009a,b). Furthermore, none of the test methods was recommended for full replacement, because none of the methods are able to identify the mild/moderate ranges of ocular irritancy. However, the final report on this validation study is not yet available. As a follow-up work, ECVAM is also working on further improvements of the prediction models of the organotypic assays and on the analyses of their usefulness within the framework of testing strategies by using data mining techniques Cytotoxicity- and cell function-based assays The retrospective validation of the four cytotoxicy- and cell function- based assays, i.e., NRR, RBC, FL and the CM was carried out by ECVAM between May 2006 and October The study was based on the retrospective collection of existing data compiled according to the ECVAM Modular Approach to Validation and Weight-of-Evidence Principles (Hartung et al., 2004; Balls et al., 2006). Based on the final results, recommendations were made by the Validation Management Group (VMG) on the validity of the NRR and FL to be used in a bottom-up approach, discriminating non-irritants (GHS and EU nonclassified) from all other classes, and the CM and FL in a top-down approach, discriminating severe irritants (GHS Cat 1, EU R41) from all irritant classes, for defined applicability domains. In July 2009, ESAC endorsed the CM (Invittox protocol 102 modified) and the FL (Invittox protocol 71) as scientifically valid for being considered for regulatory purposes as an initial step within a Top-Down approach to identify ocular corrosives and severe irritants (GHS Cat 1, EU R41) from all other classes for water-soluble chemicals (substances and mixtures). Furthermore, the CM was considered to have been scientifically validated and to be ready for consideration for regulatory use as an initial step within a bottom-up approach to identify non-irritants (GHS NC, EU NC) from all other classes, only for water-soluble surfactants, and water-soluble surfactant-containing mixtures (ESAC, 2009a). With regard to the remaining tests, ESAC considered that the available evidence was insufficient to support a recommendation that they are ready for consideration for regulatory use. Notably, for those assays considered useful to initiate a Bottom-Up approach, different VMG and ESAC recommendations were made. Possible reasons for such differences could be: 1) The acceptance criteria applied for the identification of substances not classified as irritants. An ECVAM internal analysis of a total of 2039 new and existing chemicals showed that, by only considering the within-test variability of in vivo responses, the Draize test may underpredict up to 10% Cat. 2 (GHS) or 22% R36 (EU) substances as non classified. If the in vivo between laboratory variability was to be added to the within-test variability, similar to what was considered in the validation study of cell-based assays, the overall in vivo variability might even be larger. Based on that, the VMG accepted up to 5-10% false negatives in the non severe irritancy range according to the GHS (Cat. 2) and EU (R36) classification and labelling systems and a not too high percent of false positives. On the other hand the ESAC accepted only 0% false negatives but were more flexible towards false positives. In addition ESAC considered not only the GHS and EU but also the EPA classification system. As explained in chapter 3.5.2, the EPA classification system differs considerably from the GHS and EU classification system as it considers positive effects and reversibility in a very different manner. This can result in particular in several GHS Non Classified substances that would have an EPA classification and therefore a higher false negative rate p 35 out of 84

37 observed for the EPA classification system in comparison to the GHS and EU classification systems. 2) Differences in the subjective weighting of the evaluated evidence. A parallel evaluation made by ICCVAM on the CM assay led to the same recommendations as those made by ESAC, although a final report is not yet available (ICCVAM, 2009a). Despite their limited applicability domains, these validated assays, and in particular the CM recommended for the identification of non-classified substances, might contribute to further decreasing animal testing for eye irritation. Indeed, 79% of newly registered substances (out of 2497) were found to be non-irritants, and 16% severe irritants (Scott et al., 2010) The Bovine Corneal Opacity and Permeability assay (BCOP): Severe Eye Irritation & Non- Classification As described above, the BCOP has been validated and endorsed as scientifically valid to (ICCVAM, 2007; ESAC, 2007): - identify ocular corrosives / severe irritants (GHS Cat. 1 / R41) An OECD Test Guidelines on this test method has been adopted in 2009 as TG 437 (OECD, 2009a). Furthermore, the BCOP has also been recently recommended as suitable for the identification of nonclassified materials but not as full replacement (ICCVAM 2009a,b). The same protocol was used in this validation study with the only difference being the Prediction Model used in the Decision Criteria (see chapter j and k). However, the final report on this validation study is not yet available Principles of the test The BCOP test method is an organotypic model that uses isolated corneas from the eyes of freshly slaughtered cattle. It provides short-term maintenance of normal physiological and biochemical function of the bovine cornea in vitro with the help of a corneal holder (figure 10). Test substances are applied to the epithelial surface of the cornea by addition to the anterior chamber of the corneal holder. Toxic effects of a test substance to the cornea are measured by its ability to induce opacity and increased permeability in the isolated bovine cornea. Corneal opacity is measured quantitatively as the amount of light transmission through the cornea with the help of an opacitometer. Permeability is measured quantitatively as the amount of sodium fluorescein dye that passes across the full thickness of the cornea, as detected in the medium in the posterior chamber with the help of a visible light spectrophotometer. Both measurements are used to calculate an In Vitro Irritancy Score (IVIS), which is used to assign an in vitro irritancy hazard classification category for prediction of the in vivo ocular irritation potential of a test substance. The protocol is based on the methodology reported by Gautheron and co-workers, and from Invittox protocol 124 (Gautheron et al., 1992; Invittox protocol 124, 1997). p 36 out of 84

38 Figure 10. The BCOP corneal holders (from OECD TG 437, 2009a). The corneal holders are made of an inert material (e.g., polypropylene), and comprise two halves (an anterior and posterior chamber). Each chamber terminates in a glass window, through which opacity measurements are recorded. The corneas are placed endothelial side down on the posterior chambers and the anterior chambers are placed on the epithelial side of the corneas. The chambers are maintained in place by three stainless screws located on the outer edges of the chamber. The glass window at the end of each chamber can be removed to easy access to the cornea. O-rings are located between the glass window and the chamber to prevent leaks. Two holes on the top of each chamber permit introduction and removal of medium and test compounds. They are closed with rubber caps during the treatment and incubation periods. The corneal holders can be obtained commercially from different sources or can be constructed. It is important that holders are adapted to the size of the corneas used (see below) Preparation of the corneas according to OECD TG 437 a) Source and age of bovine eyes Bovine eyes are obtained from cattle sent to slaughterhouses killed either for human consumption or for other commercial uses. Only healthy animals considered suitable for entry into the human food chain are used as a source of corneas for use in the BCOP. Variations in corneal dimensions can result when using eyes from animals of different ages. Generally cattle between 12 and 60 months old are used. Because of their larger size, eyes from cattle older than 60 months are not typically used. On the other hand, eyes from cattle less than 12 months of age present corneal thickness and corneal diameter considerably smaller than that reported for adult cattle. However, the use of corneas from 6 to 12 months old animals is possible as it present some advantages such as increased availability, a narrow age range that may lead to decreased variability, and decreased hazards related to potential worker exposure to Bovine Spongiform Encephalopathy. It is important to ensure that the appropriate corneal holders are used depending on the size of the corneas and age of the animals. The traditional holders were designed for adult cattle from 12 to 60 months old. In the case young cattle from 6 to 12 months old are used, the corneal holders used should have the appropriate dimensions to hold the smaller cornea. In particular, the ratio of exposed corneal surface area to posterior chamber volume should be the same as the ratio in the traditional corneal holder. This is necessary to assure that permeability values are correctly determined for the calculation of the IVIS by the proposed formula. b) Collection and transport of eyes to the laboratory Eyes are collected by slaughterhouse employees. To minimize mechanical and other types of damage to the eyes, the eyes should be enucleated as soon as possible after death. To prevent exposure of the eyes to potentially irritant substances, the slaughterhouse employees should not use detergent when rinsing the head of the animal. Eyes are immersed completely in Hanks Balanced Salt Solution (HBSS) in a suitably sized container, and transported to the laboratory in such a manner as to minimize deterioration and/or bacterial contamination. This is ensured by e.g., keeping the container containing the eyes on wet ice, or by adding antibiotics to the HBSS used to store the eyes during transport (e.g., penicillin at 100 IU/mL and streptomycin at 100 μg/ml). In addition, the time p 37 out of 84

39 interval between collection of the eyes and use of corneas in the BCOP should be minimized (typically collected and used on the same day). All eyes used in the assay should be from the same group of eyes collected on a specific day. c) Selection of the eyes used in the BCOP The eyes, once they arrive at the laboratory, are carefully examined for defects including increased opacity, scratches, and neovascularization. Only corneas from eyes free of such defects are to be used. The quality of each cornea is also evaluated following the preparation of the corneas (see below). d) Preparation of the corneas Corneas free of defects are dissected with a 2 to 3 mm rim of sclera remaining to assist in subsequent handling, with care taken to avoid damage to the corneal epithelium and endothelium. Isolated corneas are mounted in the appropriate designed corneal holders (figure 10). Both chambers are filled to excess with pre-warmed Eagle's Minimum Essential Medium (EMEM) (posterior chamber first), ensuring that no bubbles are formed. The device is then equilibrated at 32 ± 1 C for at least one hour to allow the corneas to equilibrate with the medium and to achieve normal metabolic activity, to the extent possible (the approximate temperature of the corneal surface in vivo is 32 C). Because the heat capacity of water is higher than that of air, water provides more stable temperature conditions for incubation. Therefore, the use of a water bath for maintaining the corneal holder and its contents at 32 ± 1ºC is recommended. However, air incubators might also be used, assuming precaution to maintain temperature stability (e.g., by pre-warming of holders and media) Method description The methods described below are based on the protocol described in the OECD TG 437 (2009a), with the only exception being the Decision Criteria used to identify non-classified test materials for eye irritation (point k), which is based on the recent evaluation from ICCVAM (2009b). a) Quality control of the isolated corneas and selection of negative (or solvent) control corneas Following the equilibration period, fresh pre-warmed EMEM is added to both chambers and baseline opacity readings are taken for each cornea. Any corneas that show macroscopic tissue damage (e.g., scratches, pigmentation, neovascularization) or an opacity >7 opacity units are discarded. It is important to note that the opacitometer should be calibrated with opacity standards as described in the OECD TG 437. The mean opacity of all equilibrated corneas is calculated. A minimum of three corneas with opacity values close to the median value for all corneas are selected as negative (or solvent) control corneas. Since all corneas are excised from the whole globe, and mounted in the corneal chambers, there is the potential for artefacts from handling upon individual corneal opacity and permeability values (including negative control). Furthermore, the opacity and permeability values from the negative control corneas are used to correct the test article and positive control-treated corneal opacity and permeability values in the IVIS calculations. The remaining corneas are distributed into treatment and positive control groups. b) Number of replicates Each treatment group (test substance, concurrent negative and positive controls) consists of a minimum of three corneas. c) Negative (solvent), positive and benchmark controls p 38 out of 84

40 Concurrent negative or solvent/vehicle controls and positive controls should be included in each experiment. Negative or solvent control When testing a liquid substance at 100%, a concurrent negative control (e.g., 0.9% sodium chloride solution or distilled water) is included in the BCOP test method so that nonspecific changes in the test system can be detected and to provide a baseline for the assay endpoints. It also ensures that the assay conditions do not inappropriately result in an irritant response. When testing a diluted liquid, surfactant, or solid, a concurrent solvent/vehicle control group is included in the BCOP test method so that nonspecific changes in the test system can be detected and to provide a baseline for the assay endpoints. Only a solvent/vehicle that has been demonstrated to have no adverse effects on the test system can be used. Positive control A known ocular irritant is included as a concurrent positive control in each experiment to verify that an appropriate response is induced. In case the BCOP assay is used to identify corrosive or severe irritants, ideally the positive control should be a reference substance that induces a severe response in this test method. However, to ensure that variability in the positive control response across time can be assessed, the magnitude of irritant response should not be excessive. Examples of positive controls for severe eye irritancy of liquid test substances are 10% sodium hydroxide or dimethylformamide. An example of a positive control for severe eye irritancy of solid test substances is 20% (weight to volume) imidazole in 0.9% sodium chloride solution. Positive controls to be used for the identification of non-classified test materials have not yet been defined. Benchmark controls Benchmark substances are useful for evaluating the ocular irritancy potential of unknown chemicals of a specific chemical or product class, or for evaluating the relative irritancy potential of an ocular irritant within a specific range of irritant responses. d) Dose and application of the test substance Two different treatment protocols are used, one for liquids and surfactants (solids or liquids), and one for non-surfactant solids: - Liquids are tested undiluted, while surfactants are tested at a concentration of 10% w/v in a 0.9% sodium chloride solution, distilled water, or other solvent that has been demonstrated to have no adverse effects on the test system. Semi-solids, creams, and waxes are typically tested as liquids. Appropriate justification should be provided for alternative dilution concentrations. Corneas are exposed to liquids and surfactants for 10 minutes. Use of other exposure times should be accompanied by adequate scientific rationale. - Non-surfactant solids are usually tested as solutions or suspensions at 20% concentration in a 0.9% sodium chloride solution, distilled water, or other solvent that has been demonstrated to have no adverse effects on the test system. In certain circumstances and with proper scientific justification, solids may also be tested neat by direct application onto the corneal surface using the open chamber method. Corneas are exposed to solids for four hours, but as with liquids and surfactants, alternative exposure times may be used with appropriate scientific rationale. Different treatment methods can be used, depending on the physical nature and chemical characteristics (e.g., solids, liquids, viscous vs. non-viscous liquids) of the test substance. The critical factor is ensuring that the test substance adequately covers the epithelial surface and that it is adequately removed during the rinsing steps. A closed-chamber method is typically used for nonp 39 out of 84

41 viscous to slightly viscous liquid test substances, while an open-chamber method can be used for semi-viscous and viscous liquid test substances and for neat solids. The application of the test substance can therefore be carried out as follows: - In the closed-chamber method, sufficient test substance (750 μl) to cover the epithelial side of the cornea is introduced into the anterior chamber through the dosing holes on the top surface of the chamber, and the holes are subsequently sealed with the chamber plugs during the exposure. It is important to ensure that each cornea is exposed to a test substance for the appropriate time interval. - In the open-chamber method, the window-locking ring and glass window from the anterior chamber are removed prior to treatment. The control or test substance (750 μl, or enough test substance to completely cover the cornea) is applied directly to the epithelial surface of the cornea using a micropipette. If a test substance is difficult to pipette, the test substance can be pressure-loaded into a positive displacement pipette to aid in dosing. The pipette tip of the positive displacement pipette is inserted into the dispensing tip of the syringe so that the material can be loaded into the displacement tip under pressure. Simultaneously, the syringe plunger is depressed as the pipette piston is drawn upwards. If air bubbles appear in the pipette tip, the test article is removed (expelled) and the process repeated until the tip is filled without air bubbles. If necessary, a normal syringe (without a needle) can be used since it permits measuring an accurate volume of test substance and an easier application to the epithelial surface of the cornea. After dosing, the glass window is replaced on the anterior chamber to recreate a closed system. e) Washing After the exposure period, the test substance, the negative control, or the positive control substance is removed from the anterior chamber and the epithelium washed at least three times (or until no visual evidence of test substance can be observed) with EMEM (containing phenol red). Phenol red containing medium is used for rinsing since a colour change in the phenol red may be monitored to determine the effectiveness of rinsing acidic or alkaline materials. The corneas are washed more than three times if the phenol red is still discoloured (yellow or purple), or the test substance is still visible. Once the medium is free of test substance, the corneas are given a final rinse with EMEM (without phenol red), to ensure removal of the phenol red from the anterior chamber prior to the opacity measurement. The anterior chamber is then refilled with fresh EMEM without phenol red. f) Post-exposure incubation period For liquids and surfactants, after rinsing, the corneas are incubated for an additional 2 hours at 32 ± 1ºC. Longer post-exposure time may be useful in certain circumstances and could be considered on a case-by-case basis. Corneas treated with solids are rinsed thoroughly at the end of the four-hour exposure period, but do not require further incubation. g) Endpoints measured At the end of the post-exposure incubation period for liquids and surfactants and at the end of the fourhour exposure period for non-surfactant solids, the opacity and permeability of each cornea are recorded. Also, each cornea is observed visually and pertinent observations recorded (e.g., tissue peeling, residual test substance, non-uniform opacity patterns). These observations could be important as they may be reflected by variations in the opacitometer readings. - Corneal opacity is determined by the amount of light transmission through the cornea. It is measured quantitatively with the aid of an opacitometer, resulting in opacity values measured on a continuous scale. - Corneal permeability is determined by the amount of sodium fluorescein dye that penetrates all corneal cell layers (i.e., the epithelium on the outer cornea surface through the endothelium on the inner cornea surface). 1 ml sodium fluorescein solution (4 or 5 mg/ml when testing liquids and surfactants or non-surfactant solids, respectively) is added to the anterior chamber of the corneal p 40 out of 84

42 holder, which interfaces with the epithelial side of the cornea, while the posterior chamber, which interfaces with the endothelial side of the cornea, is filled with fresh EMEM. The holder is then incubated in a horizontal position for 90 ± 5 min at 32 ± 1 ºC. The amount of sodium fluorescein that crosses into the posterior chamber is quantitatively measured with the aid of UV/VIS spectrophotometry. Spectrophotometric measurements evaluated at 490 nm are recorded as optical density (OD 490 ) or absorbance values, which are measured on a continuous scale. The fluorescein permeability values are determined using OD 490 values based upon a visible light spectrophotometer using a standard 1 cm path length. Alternatively, a 96-well microtiter plate reader may be used provided that: (i) the linear range of the plate reader for determining fluorescein OD 490 values can be established; and (ii), the correct volume of fluorescein samples are used in the 96-well plate to result in OD 490 values equivalent to the standard 1 cm path length (this could require a full well [usually 360μL]). h) Data evaluation Once the opacity and mean permeability (OD 490 ) values have been corrected for background opacity and the negative control permeability OD 490 values, the mean opacity and permeability OD 490 values for each treatment group are combined in an empirically-derived formula to calculate the In Vitro Irritancy Score for each treatment group as follows: IVIS = mean opacity value + (15 x mean permeability OD 490 value) The opacity and permeability values could also be evaluated independently to determine whether a test substance induced corrosivity or severe irritation through only one of the two endpoints (ICCVAM, 2007). i) Study acceptance criteria A test is considered acceptable if: - The positive control gives an IVIS that falls within two standard deviations of the current historical mean. Such historical mean is to be updated at least every three months, or each time an acceptable test is conducted in laboratories where tests are conducted infrequently (i.e., less than once a month). - The negative or solvent/vehicle control responses result in opacity and permeability values that are less than the established upper limits for background opacity and permeability values for bovine corneas treated with the respective negative or solvent/vehicle control. j) Decision criteria used for the identification of severe eye irritants (GHS Cat.1 / R41) According to the OECD TG 437 (2009a): A substance that induces an IVIS 55.1 is defined as a corrosive or severe irritant. If the test substance is not identified as an ocular corrosive or severe irritant, additional testing should be conducted for classification and labelling purposes. k) Decision criteria used for the identification of non-classified substances for eye irritancy According to ICCVAM (2009b): A substance that induces an IVIS 3.0 is defined as a non-classified test material. If the test substance is not identified as a non-classified material, additional testing should be conducted for classification and labelling purposes Proficiency testing according to OECD TG 437 (severe eye irritants) p 41 out of 84

43 For any laboratory initially establishing the BCOP assay for the identification of severe eye irritants, testing of proficiency chemicals as recommended in the OECD TG 437 should be carried out (Table 8). A laboratory may use these chemicals to demonstrate their technical proficiency in performing the BCOP test method prior to submitting BCOP assay data for regulatory hazard classification purposes. Ten proficiency chemicals were selected to represent the range of responses for local eye irritation/corrosion, which is based on results from the in vivo rabbit eye test carried out in accordance to the OECD TG 405 (i.e., Categories 1, 2A, 2B, or Not Classified according to the UN GHS). However, in the case the assay is used to identify ocular corrosives/severe irritants, there are only two test outcomes for classification purposes to demonstrate proficiency ( corrosive/severe irritant = cat. 1 or non-corrosive/non-severe irritant=not-classified or cat. 2 ) to demonstrate proficiency. Other selection criteria were that substances are commercially available, there are high quality in vivo reference data available, and there are high quality data from the two in vitro methods for which Test Guidelines were developed for the identification of severe eye irritants (BCOP and ICE). Proficiency testing chemicals for the BCOP assay to be used for the identification of non-classified test materials have not yet been defined. Table 8: Recommended substances for demonstrating technical proficiency with BCOP (extracted from OECD TG 437, 2009a) Chemical CASRN Chemical Class 1 Physical Form GHS In Vivo Classification 2 In Vitro Classification 3 Benzalkonium chloride (5%) Onium compound Liquid Category 1 Corrosive / Severe Irritant Chlorhexidine Amine, Amidine Solid Category 1 Corrosive / Severe Irritant Dibenzoyl-Ltartaric acid Carboxylic acid, Ester Solid Category 1 Corrosive / Severe Irritant Imidazole Heterocyclic Solid Category 1 Corrosive / Severe Irritant Trichloroacetic acid (30%) Carboxylic Acid Liquid Category 1 Corrosive / Severe Irritant 2,6-Dichlorobenzoyl chloride Acyl halide Liquid Category 2A Ethyl-2-methylacetoacetate Ketone, Ester Liquid Category 2B Ammonium nitrate Inorganic salt Solid Category 2A Glycerol Alcohol Liquid Not Labeled Non corrosive / Non severe irritant Non corrosive / Non severe irritant Non corrosive / Non severe irritant Non corrosive / Non severe irritant Hydrocarbon Noncorrosive / n-hexane Liquid Not Labeled (acyclic) Non severe irritant Abbreviations: CASRN = Chemical Abstracts Service Registry Number 1 Chemical classes were assigned to each test substance using a standard classification scheme, based on the National Library of Medicine Medical Subject Headings (MeSH) classification system (available at http// 2 Based on results from the in vivo rabbit eye test (OECD TG 405) and using the UN GHS classification system (UN, 2003, 2009). 3 Based on results in BCOP and ICE Known applicability and limitations p 42 out of 84

44 Identification of Severe Irritants The BCOP test method can be used, under certain circumstances and with specific limitations, to classify substances as ocular corrosives and severe irritants as defined by the UN GHS Category 1, the EU DSD R41 and the US EPA Category 1. It is not considered valid as a complete replacement for the in vivo rabbit eye test, as the BCOP is recommended for use as part of a tiered-testing strategy where test substances (including single components and multi-component formulations) can be classified as ocular corrosives or severe irritants without further testing in rabbits. The Predictive Capacity of the BCOP assay to identify ocular corrosives and severe eye irritants according to the GHS, EU or EPA classification systems was found to be during the validation study (ICCVAM, 2007): Accuracy 79% (113/143) to 81% (119/147) False positive rate 19% (20/103) to 21% (22/103) False negative rate 16% (7/43) to 25% (10/40) (numbers between brackets indicate the number of substances from which the rates were calculated). High false positive rates were found for alcohols and ketones and high false negative rate was found for solids in the validation database (ICCVAM, 2007). When substances within these chemical and physical classes were excluded from the database, the predictive capacity of BCOP across the GHS, EU and EPA classification systems was found to be (ICCVAM, 2007): Accuracy 87% (72/83) to 92% (78/85) False positive rate 12% (7/58) to 16% (9/56) False negative rate 0% (0/27) to 12% (3/26) In case the assay is used to identify ocular corrosives/severe irritants, false negative rates are not critical since such substances would be subsequently tested in rabbits or with other adequately validated in vitro tests, depending on regulatory requirements, using a sequential testing strategy in a weight of evidence approach. As such the only limitation of the BCOP assay for identifying severe eye irritants is that positive results obtained with alcohols or ketones should be interpreted cautiously due to risk of over-prediction. Furthermore, the ICCVAM validation database did not allow for an adequate evaluation of some chemical or product classes (e.g., formulations). However, investigators could consider using this test method for all types of test substances (including formulations), whereby a positive result could be accepted as indicative of an ocular corrosive or severe irritant response. Identification of Non-Classified test materials The Predictive Capacity of the BCOP assay to identify non-classified test materials for eye irritancy according to the GHS, EU or EPA classification systems was found to be during the validation study (ICCVAM, 2009b): Accuracy 64% (76/118) to 83% (154/186) False negative rate 0% (0/97) to 6% (8/141) False positive rate 53% (24/45) to 70% (63/90) In case the assay is used to identify non-classified test materials, false positive rates are not critical since such substances would be subsequently tested in rabbits or with other adequately validated in vitro tests, depending on regulatory requirements, using a sequential testing strategy in a weight of evidence approach. The BCOP test method was recommended to be used as a screening test to identify substances which require no classification (i.e., not classified as EU R41 or R36; GHS Category 1, 2A, or 2B). Because 568 of the significant lesions associated with 50% (4/8) of the EPA Category III substances that were false negative in BCOP (i.e., identified as Category IV), the BCOP cannot be recommended as a screening test to identify EPA Category IV substances. Other applicability and limitations p 43 out of 84

45 Based on an evaluation of available data and corresponding performance (sensitivity and specificity), the BCOP test method appeared as not recommendable to identify substances from all hazard categories as defined by the GHS, EPA, and EU classification systems (ICCVAM, 2009a). From a mechanistic point of view, the BCOP assay models some of the ocular effects evaluated in the Draize rabbit eye test (such as corneal effects) and to some degree their severity. However, it does not model conjunctival and iridal injuries. Also, although the reversibility of corneal lesions cannot be evaluated per se in the BCOP assay, it has been proposed, based on rabbit eye studies, that an assessment of the initial depth of corneal injury can be used to distinguish between irreversible and reversible effects (Maurer et al., 2002) The Isolated Chicken Eye test (ICE): Severe Eye Irritation As described in chapter 4.2.1, the ICE has been validated and endorsed as scientifically valid to (ICCVAM, 2007; ESAC, 2007): - identify ocular corrosives / severe irritants (GHS Cat. 1 / R41) An OECD Test Guidelines on this test method has been adopted in 2009 as TG 438 (OECD, 2009a) Principles of the test The ICE test method is an organotypic model that provides short-term maintenance of the chicken eye in vitro. Toxic effects by the test substance to the eyes are measured by (i) a qualitative assessment of corneal opacity, (ii) a qualitative assessment of damage to corneal epithelium based on application of fluorescein to the eye (fluorescein retention), (iii) a quantitative measurement of increased corneal thickness (swelling), and (iv) a qualitative evaluation of macroscopic morphological damage to the corneal surface. The corneal opacity, swelling, and damage assessments following exposure to a test substance are assessed individually and are then combined to derive an Eye Irritancy Classification. The protocol is based on the work from Prinsen and co-workers (Prinsen and Koeter, 1993; Invittox protocol 80, 1994) Preparation of the eyes according to OECD TG 438 a) Source and age of chicken eyes Historically, eyes collected from chickens obtained from a slaughterhouse where they are killed for human consumption have been used for this assay, eliminating the need for laboratory animals. Only the eyes of healthy animals considered suitable for entry into the human food chain are used. Although a controlled study to evaluate the optimum chicken age has not been conducted, the age and weight of the chickens used historically in this test method are that of spring chickens traditionally processed by a poultry slaughterhouse (i.e., approximately 7 weeks old, kg). b) Collection and transport to the laboratory Heads should be removed immediately after sedation of the chickens, usually by electric shock, and incision of the neck for bleeding. A local source of chickens close to the laboratory is recommended to minimize deterioration and/or bacterial contamination. Indeed, the time interval between collection of the chicken heads and use of eyes in the ICE test method should be minimized (typically within two hours) and should be demonstrated to not compromise the assay results (based on the selection criteria for the eyes, as well as the positive and negative control responses). Because eyes are dissected in the laboratory, the intact heads are transported from the slaughterhouse at ambient temperature in plastic boxes humidified with towels moistened with isotonic saline. All eyes used in the assay should be from the same group of eyes collected on a specific day. p 44 out of 84

46 c) Preparation of the eyes The eyelids are carefully excised, taking care not to damage the cornea. Corneal integrity is quickly assessed with a drop of 2% (w/v) sodium fluorescein applied to the corneal surface for a few seconds, and then rinsed with isotonic saline. Fluorescein-treated eyes are then examined with a slit-lamp microscope to ensure that the cornea is undamaged (i.e., fluorescein retention and corneal opacity scores 0.5). If undamaged, the eye is further dissected from the skull, taking care not to damage the cornea. The eyeball is pulled from the orbit by holding the nictitating membrane firmly with surgical forceps, and the eye muscles are cut with a bent, blunt-tipped scissor. It is important to avoid causing corneal damage due to excessive pressure (i.e., compression artefacts). When the eye is removed from the orbit, a visible portion of the optic nerve should be left attached. Once removed from the orbit, the eye is placed on an absorbent pad and the nictitating membrane and other connective tissue are cut away. The enucleated eye is mounted in a stainless steel clamp with the cornea positioned vertically. The clamp is then transferred to a chamber of the superfusion apparatus (figure 11). The clamps should be positioned in the superfusion apparatus such that the entire cornea is supplied with the regular stream of isotonic saline drops. The chambers of the superfusion apparatus should be temperature controlled at 32 ± 1.5 C. d) Selection of the eyes used for testing After being placed in the superfusion apparatus, the eyes are again examined with a slit-lamp microscope to ensure that they have not been damaged during the dissection procedure. Corneal thickness should also be measured at this time at the corneal apex using the depth measuring device on the slit-lamp microscope. Eyes with; (i), a fluorescein retention score of > 0.5; (ii) corneal opacity > 0.5; or, (iii), any additional signs of damage should be replaced. For eyes that are not rejected based on any of these criteria, individual eyes with a corneal thickness deviating more than 10% from the mean value for all eyes are to be rejected. Users should be aware that slit-lamp microscopes could yield different corneal thickness measurements if the slit-width setting is different. The slit-width should be set at mm. p 45 out of 84

47 Figure 11. The ICE superfusion apparatus and eye clamps (from OECD TG 438, 2009b). The apparatus can be obtained commercially or constructed. It may be modified to meet the needs of an individual laboratory (e.g., to accommodate a different number of eyes). p 46 out of 84

48 Method description according to OECD TG 438 a) Equilibration and baseline recordings (time = 0) Once all eyes have been examined and approved, the eyes are incubated for approximately 45 to 60 minutes to equilibrate them to the test system prior to dosing. Following the equilibration period, a zero reference measurement is recorded for corneal thickness and opacity to serve as a baseline (i.e., time = 0). The fluorescein score determined at dissection is used as the baseline measurement for that endpoint. b) Number of replicates Each treatment group and concurrent positive control consists of at least three eyes. The negative control group or the solvent control (if using a solvent other than saline) consists of at least one eye. c) Negative, solvent/vehicle, positive and benchmark controls Concurrent negative or solvent/vehicle controls and positive controls should be included in each experiment. Negative control: Physiological saline is used when testing liquids at 100% or solids, to detect nonspecific changes in the test system, and to ensure that the assay conditions do not inappropriately result in an irritant response. Solvent/vehicle control: the respective solvents/vehicles used when testing diluted liquids should be tested concurrently as controls for the same reasons as the negative controls. Only a solvent/vehicle that has been demonstrated to have no adverse effects on the test system can be used. Positive control: As the ICE assay is recommended to identify corrosive or severe irritants, the positive control should be a reference substance that induces a severe response in this test method. However, to ensure that variability in the positive control response across time can be assessed, the magnitude of the severe response should not be excessive. Sufficient in vitro data for the positive control should be generated such that a statistically defined acceptable range for the positive control can be calculated. If adequate historical ICE test method data are not available for a particular positive control, studies may need to be conducted to provide this information. Examples of positive controls for liquid test substances are 10% acetic acid or 5% benzalkonium chloride, while examples of positive controls for solid test substances are sodium hydroxide or imidazole. Benchmark controls: Benchmark substances are useful for evaluating the ocular irritancy potential of unknown chemicals of a specific chemical or product class, or for evaluating the relative irritancy potential of an ocular irritant within a specific range of irritant responses. d) Dose and application of the test substance Immediately following the zero reference measurements, the eye (in its holder) is removed from the superfusion apparatus, placed in a horizontal position, and the test substance is applied to the cornea. Liquid test substances are typically tested undiluted, but may be diluted if deemed necessary (e.g., as part of the study design). The preferred solvent for diluted substances is physiological saline. Other solvents may also be used if shown appropriate. Liquid test substances are applied to the cornea such that the entire surface of the cornea is evenly covered with the test substance; the standard volume is 0.03 ml. Solid substances should be ground, if possible, as finely as possible in a mortar and pestle, or comparable grinding tool. The powder is applied to the cornea such that the surface is uniformly covered with the test substance; the standard amount is 0.03 g. p 47 out of 84

49 e) Exposure time The test substance (liquid or solid) is applied for 10 seconds and then rinsed from the eye with isotonic saline (approximately 20 ml) at ambient temperature. The eye (in its holder) is subsequently returned to the superfusion apparatus in the original upright position. f) Observation period Treated corneas are evaluated before treatment (pre-treatment) and then starting at 30, 75, 120, 180, and 240 minutes (± 5 minutes) after the post-treatment rinse. g) Endpoints measured The endpoints evaluated are corneal opacity, swelling, fluorescein retention, and morphological effects (e.g., pitting or loosening of the epithelium). After the final examination at four hours, users are encouraged to preserve eyes in an appropriate fixative (e.g., neutral buffered formalin) for possible histopathological examination. All of the endpoints, with the exception of fluorescein retention (which is determined only at pretreatment and 30 minutes after test substance exposure) are determined at each of the above time points. Photographs are advisable to document corneal opacity, fluorescein retention, morphological effects and, if conducted, histopathology Corneal swelling Corneal swelling is determined from corneal thickness measurements made with an optical pachymeter on a slit-lamp microscope. It is expressed as a percentage and is calculated from corneal thickness measurements according to the following formula: corneal thickness at time t corneal thickness at time = 0 x 100 corneal thickness at time = 0 The mean percentage of corneal swelling for all test eyes is calculated for all observation time points. Based on the highest mean score for corneal swelling, as observed at any time point, an overall category score is then given for each test substance (see point h). Corneal opacity Corneal opacity is calculated by using the area of the cornea that is most densely opacified for scoring (Table 9). The mean corneal opacity value for all test eyes is calculated for all observation time points. Based on the highest mean score for corneal opacity, as observed at any time point, an overall category score is then given for each test substance (see point h). Table 9. Corneal opacity scores. Score Observation 0 No opacity 0.5 Very faint opacity 1 Scattered or diffuse areas; details of the iris are clearly visible 2 Easily discernible translucent area; details of the iris are slightly obscured Severe corneal opacity; no specific details of the iris are visible; size of the 3 pupil is barely discernible 4 Complete corneal opacity; iris invisible p 48 out of 84

50 Fluorescein retention The mean fluorescein retention value for all test eyes is calculated for the 30-minutes observation time point only (Table 10), which is used for the overall category score given for each test substance (see point h). Table 10. Fluorescein retention scores. Score Observation 0 No fluorescein retention 0.5 Very minor single cell staining Single cell staining scattered throughout the treated area of the 1 cornea 2 Focal or confluent dense single cell staining 3 Confluent large areas of the cornea retaining fluorescein Morphological effects Morphological effects include pitting of corneal epithelial cells, loosening of epithelium, roughening of the corneal surface and sticking of the test substance to the cornea. These findings can vary in severity and may occur simultaneously. The classification of these findings is subjective according to the interpretation of the investigator. h) Data evaluation and decision criteria used for the identification of severe eye irritation Once each endpoint has been evaluated, ICE classes are assigned based on predetermined scales. Interpretation of corneal thickness, opacity and fluorescein retention using four ICE classes is done as shown in Tables 11, 12 and 13 respectively. The ICE classes for each endpoint and the morphological effects observed are then combined to generate an Irritancy Classification for each test substance as shown in Table 14. Table 11. ICE classification criteria for corneal thickness. Mean Corneal Swelling (%)* ICE Class 0 to 5 I >5 to 12 II >12 to 18 (>75 min after treatment) II >12 to 18 ( 75 min after treatment) III >18 to 26 III >26 to 32 (>75 min after treatment) III >26 to 32 ( 75 min after treatment) IV >32 IV *Corneal swelling scores based on thickness measures made with the help of a Haag-Streit BP900 slit-lamp microscope with depth-measuring device no. I and slit-width setting at 9½, equalling mm. Slit-lamp microscopes could yield different corneal thickness measurements if the slit-width setting is different. Table 12. ICE classification criteria for opacity. Mean Maximum Opacity Score* *See Table 9. ICE Class I II III IV p 49 out of 84

51 Table 13. ICE classification criteria for mean fluorescein retention. Mean Fluorescein Retention Score at 30 minutes post treatment* ICE Class I II III IV *See Table 10. The overall In Vitro Irritancy classification for a test substance is then determined by applying the irritancy classification that corresponds to the combination of categories obtained for corneal swelling, corneal opacity, and fluorescein retention and applying the scheme as shown in table 14. Table 14. Overall in vitro irritancy classifications. Classification Corrosive/Severe Irritant *Combinations less likely to occur. Combinations of the 3 Endpoints 3 x IV 2 x IV, 1 x III 2 x IV, 1 x II* 2 x IV, 1 x I* Corneal opacity 3 at 30 min (in at least 2 eyes) Corneal opacity = 4 at any time point (in at least 2 eyes) Severe loosening of the epithelium (in at least 1 eye) If the test substance is not identified as an ocular corrosive or severe irritant, additional testing should be conducted for classification and labelling purposes. j) Study acceptance criteria A test is considered acceptable if the concurrent negative or vehicle/solvent controls and the concurrent positive controls give an Irritancy Classification that falls within non-irritant and severe irritant/corrosive classes, respectively Proficiency testing according to OECD TG 438 (severe eye irritation) For any laboratory initially establishing the ICE assay, testing of proficiency chemicals as recommended in the OECD TG 438 should be carried out (Table 8). A laboratory may use these chemicals to demonstrate their technical competence in performing the ICE test method prior to submitting ICE data for regulatory hazard classification purposes. Ten proficiency chemicals were selected to represent the range of responses for local eye irritation/corrosion, which is based on results in the in vivo rabbit eye test TG 405 (i.e., Categories 1, 2A, 2B, or Not Classified according to the UN GHS). However, considering the validated usefulness of these assays (i.e., to identify ocular corrosives/severe irritants only), there are only two test outcomes for classification purposes to demonstrate proficiency ( corrosive/severe irritant = cat. 1 or non-corrosive / non-severe irritant = non-classified or Cat. 2 ) to demonstrate proficiency. Other selection criteria were that substances are commercially available, there are high quality in vivo reference data available, and there are high quality data from the two in vitro methods for which Test Guidelines were developed for the identification of severe eye irritants (ICE and BCOP). p 50 out of 84

52 Known applicability and limitations Identification of Severe Irritants The Isolated Chicken Eye test method can be used, under certain circumstances and with specific limitations, to classify substances as ocular corrosives and severe irritants, as defined by the UN GHS Category 1, EU DSD R41 and the US EPA Category I. It is not considered valid as a complete replacement for the in vivo rabbit eye test. The ICE is recommended for use as part of a tiered testing strategy for regulatory classification and labelling where test substances (including substances with a single component and multi-component formulations) that are positive in this assay can be classified as ocular corrosives or severe irritants without further testing in rabbits. The Predictive Capacity of the ICE test method to identify ocular corrosives and severe irritants according to the GHS, EU or EPA classification systems was found to be during the validation study (ICCVAM, 2007): Accuracy 83% (120/144) to 87% (134/154) False positive rate 6% (7/122) to 8% (9/116) False negative rate 41% (13/32) to 50% (15/30). (number between brackets indicate the number of substances from which the rates were calculated). A high false positive rate was found for alcohols and high false negative rates were found for solids and surfactants in the validation database (ICCVAM, 2007). When substances within these chemical and physical classes were excluded from the database, the predictive capacity of the ICE across the GHS, EU and EPA classification systems was found to be (ICCVAM, 2007): Accuracy 91% (75/82) to 92% (69/75) False positive rates 5% (4/73) to 6% (4/70) False negative rates 29% (2/7) to 33% (3/9). As the purpose of this assay is to identify ocular corrosives/severe irritants only, false negative rates are not critical since such substances would be subsequently tested in rabbits or with other adequately validated in vitro tests, depending on regulatory requirements, using a sequential testing strategy in a weight of evidence approach. As such, the only limitation of the ICE assay for identifying severe eye irritants is that positive results obtained with alcohols should be interpreted cautiously due to risk of over-prediction. Furthermore, the current validation database did not allow for an adequate evaluation of some chemical or product classes (e.g., formulations). However, investigators could consider using this test method for testing all types of substances (including formulations), whereby a positive result could be accepted as indicative of an ocular corrosive or severe irritant response. Finally, from a mechanistic point of view, the ICE assay models some of the ocular effects evaluated in the rabbit ocular irritancy test method and to some degree their severity. However, it does not model conjunctival and iridal injuries. Also, although the reversibility of corneal lesions cannot be evaluated per se in the ICE test method, it has been proposed, based on rabbit eye studies, that an assessment of the initial depth of corneal injury can be used to distinguish between irreversible and reversible effects (Maurer et al., 2002). Identification of other ranges of irritancy It has to be noted that the recent ICCVAM evaluation of the ICE test method to identify other ranges of irritancy, suggested that the ICE was not recommended to identify substances from all hazard categories as defined by GHS, EPA and EU classification systems. Furthermore, it also suggested that the ICE test method was not recommended to be used as a screening test to identify substances which require no classification (i.e., non-classified as GHS Cat. 1, 2A or 2B, EU R41 or R36, and EPA Cat. III, II or I). This was due to the fact that, although the false negative rate for the GHS system was 6% (4/62), among these false negatives there was a Category 1 substance (ICCAM, 2009a). p 51 out of 84

53 4.5. The Cytosensor Microphysiometer assay (CM): Severe Eye Irritation & Non-Classification As described in chapter 4.2, the CM has been recently endorsed as scientifically valid to (ESAC, 2009a; ICCVAM, 2009a): - identify ocular corrosives and severe irritants (GHS Cat 1, EU R41) from all other classes for water-soluble chemicals (substances and mixtures); - and to identify substances not classified as irritants (GHS NC, EU NC) from all other classes, only for water-soluble surfactants, and water-soluble surfactant-containing mixtures. A test guideline is currently under preparation for submission to EU and OCED. The validated assay is based on a standard Invittox protocol 102 (modified), which will be described hereafter. The same protocol is used for the two regulatory uses of the assay with the only difference being the Prediction Model used in the Decision Criteria (see chapter j and k) Principles of the test The Cytosensor Microphysiometer is a cell function-based assay that estimates the potential ocular irritancy of a test substance by measuring its induced modification in the metabolic rate of treated cultures of mouse L929 fibroblast cells. Cells are maintained in flow chambers and are exposed to increasing concentrations of the test material (figure 12). Toxic concentrations induce a decreased metabolic rate of the cultured cells. Changes in metabolic rate are measured indirectly as a function of changes in the extracellular acidification. The dose which induces a 50% decrease in metabolic rate, the MRD 50 value (in units of mg/ml) is used to estimate the potential ocular irritation of the test material. A B Figure 12. A. Diagram of the operating components of the Cytosensor; B. The Cytosensor chamber with the transwell in place (Curren et al., 2008). The Cytosensor Microphysiometer Equipment measures the extracellular acidification rate of cell cultures. It includes: 1) the cytosensor microphysiometer units which include eight built-in peristaltic pumps for each channel, 2) sensor chambers, 3) a computer, 4) a printer. Adherent cells are seeded in the capsule cup. Each cell culture-containing cell capsule (capsule cup and spacer assembly) is loaded into the sensor chamber. The bottom of the sensor chamber is made of silicon sensor chip. This chip is capable of detecting very small changes in ph. Low-buffered medium is perfused across the cells in a stop/flow manner. When the flow is stopped, the change in ph due to acidic metabolites (e.g., lactate and CO 2 ) build-up is detected by the silicon sensor. The acidification of the medium occurs at a reproducible rate in the presence of a normal, undamaged cell population. Cells which have received a toxic insult will produce an altered acidification rate Method description according to Invittox Protocol 102 The validated assay corresponds to a slightly modified version of the Invittox protocol 102 (1996) that will be described hereafter, as available from the documents of the validation study that shall be soon available at the ECVAM website (Curren et al., 2008). Due to the recent validation of the assay (July 2009), an official test guideline is not yet available. p 52 out of 84