Non-animal Methods for Toxicity Testing

Eye Irritation/Corrosion

Last Updated: December 6, 2007

Print |

Related Forum Topics

One component in the safety assessment of many types of products is the evaluation of their potential to cause eye injury. Eye irritation is defined as "the production of changes in the eye following the application of test substance to the anterior surface of the eye, which are fully reversible within 21 days of application" (UNECE, 2004). Eye corrosion (serious eye damage) is defined 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" (UNECE, 2004).

The anterior surface of the eye is covered by the cornea and the conjunctiva. The cornea is the most exposed area and, therefore, the most likely part of the eye to be involved in a chemical exposure to the eye. Chemical injury to the cornea can result in the loss of vision. Accordingly, corneal injury in the animal test for eye irritation/corrosion accounts for 73% of the total ocular toxicity score. For these reasons, the cornea has been the primary tissue modeled in in vitro alternative models for accessing eye injury.

The cornea is composed of three cellular layers. The outermost layer, the corneal epithelium, is composed of 5-7 epithelial cell layers in the human cornea. The surface layers of cells are connected by tight junctions that modulate the permeation of molecules into the tissue. The stroma, beneath the corneal epithelium, is composed of keratocytes (fibroblast-like cells) interspersed in the stromal collagen matrix. A single cell layer of corneal endothelial cells forms the innermost cellular layer of the cornea. Nerve cells penetrate through the cornea into the corneal epithelium, making it one of the most highly innervated tissues in the body; these cells are readily perturbed when the corneal epithelial tight junctions are disrupted.

Many mild eye irritants act by disrupting or damaging only the surface cells of the corneal epithelium, and the corneal can repair this type of damage within a short time. The stronger the eye irritant, the deeper it penetrates into the next layer of the cornea, the stroma. Very damaging materials might penetrate deep enough to cause irreversible injury, including damage to the corneal endothelial cell layer, where tight junctions modulate the penetration of water and other substances from the cornea to the aqueous humor.

The conjunctiva covers the remaining surface of the eye and is also important in protecting the eye from environmental insults. Injury to the conjunctiva has been assigned about 18% of the in vivo eye injury score in the Draize test but has largely been ignored in in vitro assessments of chemical eye injury. Conjunctival injury may be irrelevant in moderate to severe eye injury, where the corneal effects are largely predictive of reversibility and outcome, but could be useful in assessing milder effects, especially for products used in and around the eye (Ward, et al., 2000). For example, the conjunctiva can exhibit different mechanisms of injury than the cornea due to its different physiology. The conjunctiva contains goblet cells, which secrete the mucin layer that protects the surface of the eye, as well as immune and vascular components important in the eye irritation response.

The iris is the third ocular tissue assessed for response to an irritant in the in vivo eye test. It is generally agreed that an in vitro iris assessment for most substances is not necessary, as iris responses occur only upon significant disruption to the ocular surface barrier of the cornea (Bagley, et al., 2006). Therefore, the degree of corneal injury should be predictive of potential iris effects.

The Animal Test(s)

Developed in 1944, the Draize rabbit eye irritation test remains the standard method for evaluating the ocular irritation/corrosion potential of a substance for regulatory purposes (Draize, et al., 1944; Friedenwald, et al., 1944; ILSI TCAAT, 1996). In this test, a material is instilled into one eye of albino rabbits (the other eye serving as the negative control), and the response of the animals is monitored using a standardized scoring system for injury to the cornea, conjunctiva, and iris. Ocular responses are scored at 1, 24, 48, and 72 hours. The animals are observed until the full magnitude and reversibility of the ocular injury can be evaluated—for up to 21 days. Reversibility of the ocular injury is an important component in the classification of a substance as an eye irritant versus an eye corrosive. Different modifications of the test require different numbers of animals, although no more than three animals is the current standard.

The Organisation for Economic Co-operation and Development (OECD) Test Guideline (TG) 405 Acute Eye Irritation/Corrosion, the OECD Guidance Document No. 14, and Chapter 3.3 of the Globally Harmonized System (GHS) for Classification and Labeling of Chemicals (UNECE, 2004) describe internationally accepted guidelines for eye irritation/corrosion studies. Many national agencies will accept the GHS beginning in 2008 (EPA, 2006). Most test guidance documents, including the OECD and GHS, indicate that a chemical found to be strongly irritating or corrosive in skin studies does not need to be tested in eye studies because the response can be assumed to be at least that severe in the eye.

Regulatory authorities in most countries require ocular safety assessments and commonly have some version of the Draize rabbit eye test as part of their testing guidelines. The Introduction and Rationale... section of the Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM) ocular background review documents provides an overview of the guidance and regulations for ocular testing and explains the classification systems that different authories use for eye irritation.

Draize rabbit eye data has proved to be highly variable, generally overpredictive of human eye injury, and sometimes incorrect due to species differences in the ocular response to specific substances. However, there has been renewed interest in a variant of the traditional Draize test, the low-volume eye test (LVET). In this test, one-tenth the dosing volume of the traditional test is placed directly on the cornea, as opposed to the conjunctival sac. In addition to giving responses closer to those observed in humans, it has the potential utility of providing mechanistic data (Jester, et al., 1998; Maurer, et al., 2002; Jester, 2006). The LVET, used with confocal microscopy, has been shown to correlate with the degree of corneal injury for some classes of chemicals. A significant amount of new animal data (only a few of the publications are cited above) are being generated to demonstrate that the method is an improved in vivo model for use in the validation of in vitro eye irritation methods. In 2003, one of the new activities that the European Centre for the Validation of Alternative Methods (ECVAM) endorsed was "the evaluation of alternative methods aiming to refine the existing animal test, such as the Low-Volume-Eye-Test."

Non-animal Alternative Methods

A number of non-animal test methods have been developed in the search for a replacement for the Draize rabbit eye test. Protocols for many of the in vitro methods are available at the AnimAlt-Zebet Database and the ECVAM database service on alternative methods to animal experimentation (DB-ALM). Protocols for the four methods reviewed by ICCVAM for assessing severe eye irritants and corrosives are available in Background Review Documents (BRD). Many additional methods and ocular models used for ophthalmic and toxicological research and testing have been reported in the literature.

Alternative methods for assessing the eye irritation/corrosion potential of substances have been reviewed in recent symposia and publications (Curren & Harbell, 1998; ECVAM, 2002; Eskes, et al., 2005; ICCVAM, 2006; Salem & Katz, 2003). The ICCVAM/National Toxicology Program Interagency Center for Evaluation of Alternative Toxicological Methods (NICEATM) Ocular Toxicity Scientific Symposium I: Mechanisms of Chemically-Induced Ocular Injury and Recovery of May 2005 also reviewed current research and in vitro models, but the presentations and meeting summary are not yet available on ICCVAM's website. A summary of at least some of the ocular models and assay endpoints that have been developed are summarized in Tables 1 and 2 [works in progress].

Table 1. Non-animal alternative methods for evaluating eye irritation and corrosion.*

Test Type Assay Reference(s)
Human volunteer clinical eye testing Makeup applied to outer eyelid for comparison to in vitro results Debbasch, et al., 2005
Vapor exposures Cometto-Muñiz, et al., 2007
Isolated eye assays Bovine Corneal Opacity and Permeability (BCOP) Gautheron, et al., 1992; Sina, et al., 1998; ICCVAM, 2006; Ubels, et al., 2004
Porcine Corneal Opacity and Permeability (PCOP) Van den Berghe, et al., 2005; Xu, et al., 2000
Isolated Chicken Eye (ICE) assay Prinsen & Koëter, 1993; Prinsen, 1996; ICCVAM, 2006
Isolated Rabbit Eye (IRE) Cooper, et al., 2001; ICCVAM, 2006
Isolated Mouse Eye  
Human Cornea (discarded from eyebanks)  
Chicken egg membrane assays Chorioallantoic Membrane Vascularization Assay (CAMVA) Bagley, et al., 1994; 1999
Hen's Egg Test - Chorioallantoic Membrane (HET-CAM) assay) Spielmann, et al., 1997; ICCVAM, 2006; Mehling, et al., 2007
Reconstituted human cornea models Human corneal equivalent Griffith, et al., 1999
Reconstituted rabbit cornea models 3D corneal tissue construct Zieske, et al., 1994
Reconstituted bovine cornea models Epithelium and stroma Minami, et al., 1993; Parnigotto, et al., 1998
3D human corneal epithelial cell models HCE-T human corneal epithelial cell model Clothier, et al., 2000; Kahn & Walker, 1993; Kahn, et al., 1993; Kruszewski, et al., 1995; 1997; Ward, et al., 1997; 2003
SkinEthic HCE model; CEPI Debbasch, et al., 2005; Van Goethem, et al., 2006; Mehling, et al., 2007
Coty corneal epithelial model Doucet, et al., 2006
3D epithelial cell models EpiOcular Jones, et al., 2001
MDCK fluorescein leakage Tchao, 1988; Shaw, et al., 1990; Jones, et al., 2001
Tissue equivalent assay Osborne, et al., 1995; Curren, et al., 1997
3D human conjunctival epithelial cell models Human conjunctival model Ward, et al., 2000
Monolayer epithelial cell cultures Human corneal epithelial cells Ward, et al., 1997
Rabbit corneal cells; SIRC cell line North-Root, et al., 1982; Grant, et al., 1992; Yang & Acosta, 1994
Various cultured cells Harbell, et al., 1997
Epithelial and fibroblast cell lines Pasternak & Miller, 1995
Red blood cells Red cell hemolysis Pape, et al., 1987; Lagarto, et al., 2006; Martinez, et al., 2006; Mehling, et al., 2007
Neural cell models (for detecting neurogenic ocular pain) TRPV1-expressing neuroblastoma SH-SY5Y cells Lilja & Forsby, 2004; Lilja, et al., 2007
Acellular models EYTEX/Irritection Curren, et al., 1997
Hemoglobin denaturation Liao, et al., 2004
Computational models SAR/(Q)SAR Abraham, et al., 2003; Gerner, et al., 2005; Kulkarni & Hopfinger, 1999; Kulkarni, et al., 2001; Li, et al., 2005; Pospisil & Holzhütter, 2001; Tsakovska, et al., 2007

*This table is a work in progress

A few of the non-animal methods for evaluating eye irritation and/or eye corrosion are described below.

BCOP Assay: The BCOP assay uses enucleated cow eyes that would otherwise be discarded at slaughterhouses. The cornea is isolated from the rest of the eye and maintained in a holder. A test substance is applied to this isolated cornea for a specified time then removed, and the effect of the substance on the permeability of the cornea to fluorescein (a colored dye) and the increase in corneal opacity (transmission of light through the cornea) are determined. As stated previously, corneal opacity is the primary component of the Draize rabbit eye test—hence, the BCOP assay has a direct mechanistic link to the rabbit test.

The BCOP assay has undergone a number of interlaboratory studies, and some companieshave used it for internal product decisions for many years. It has been found to be most predictive of the rabbit test for moderate to severe/corrosive substances. Additional endpoints are sometimes evaluated, and there are variations in the protocols used by different labs. The assay has been used to test a wide range of product types including liquids, powders, and creams. Standardized protocols for the BCOP assay are available, and contract labs conduct the assay for companies needing data for regulatory submissions.

Table 2. Endpoints used for assessing eye irritation and corrosion in ocular models.*

Endpoint Assay Reference(s)
Cytotoxicity/viability/proliferation MTT (tetrazolium salt; mitochondrial dehydrogenase dye reduction) Clothier, et al., 2000; Ward, et al., 1997
Lactate dehydrogenase (LDH)  
Neutral red release (NRR)  
Neutral red uptake (NRU) Pasternak & Miller, 1995
Lactate release Clothier, et al., 2000; Ward, et al., 2000
Akamar blue™ Clothier, et al., 2000
Methylene blue  
WST-1 (tetrazolium salt that yields water soluble cleavage products) Huhtala, et al., 2003
Histology/ultrastructure Histology in BCOP model Cater & Harbell, 2006; Cooper, et al., 2001; Ubels, et al., 2004
Transmission electron microscopy (TEM) in cell models Ward, et al., 1997
Barrier function Fluorescein permeability in isolated eye assays (BCOP, etc.) Gautheron, et al.,1992; Sina, et al., 1998; Ubels, et al. 2004; Van den Berghe, et al., 2005; Xu, et al., 2000
Fluorescein leakage/MDCK cells (FL) Tchao, 1988; Shaw, et al., 1990; Jones, et al., 2001
Transepithelial fluorescein permeability (TEP) Kahn & Walker, 1993; Kahn, et al., 1993; Kruszewski, et al., 1995; 1997; Ward, et al., 1997; 2000; 2003
Transepithelial electrical resistance (TER or TEER) and other measures of tight junctions Wang, et al., 2004; Ward, et al., 1997; 2000
Inflammatory mediator release or expression Cytokines Burbach, et al., 2001; Smit, et al., 2003; Debbasch, et al., 2005
Arachadonic acid metabolites  
Adhesion and other receptor molecule expression ICAM-1 Yannariello-Brown, et al., 1998
CD14 and toll-like receptor 4 Song, et al., 2001
Stress gene/protein expression HO-1  
Toxicogenomics Boulton & Wride, 2006
Transcription factor modulation NF?-b Xu, et al., 2000
Cellular metabolism Lactate release/glucose uptake  
pH/cytosensor microphysiometer  
Other Total glutathione content Pasternak & Miller, 1995
ATP content Pasternak & Miller, 1995
Methionine incorporation Pasternak & Miller, 1995
Cell migration  
Cell differentiation  
Cytoskeletal changes  

*This table is a work in progress

Isolated chicken eye: The ICE assay uses enucleated chicken eyes obtained from slaughterhouses. The eyes are placed in an apparatus, kept moist, and treated with the test substance. Three responses of the cornea are evaluated: corneal swelling, corneal opacity, and fluorescein retention. The irritation potential of a substance is calculated from the mean values of these measurements.

EpiOcular assay: "MatTek's EpiOcular™ corneal model consists of normal, human-derived epidermal keratinocytes that have been cultured to form a stratified, squamous epithelium similar to that found in the cornea. The epidermal cells, which are cultured on specially prepared cell culture inserts using serum free medium, differentiate to form a multi-layered structure that closely parallels the corneal epithelium.... EpiOcular has been utilized with several common tests of cytotoxicity and irritancy, including MTT, IL-1a, PGE2, LDH, and sodium fluorescein permeability."

HCE-T TEP assay: The HCE-T TEP assay measures the dose-dependent effect of a test material on fluorescein transepithelial permeability (TEP) across stratified cultures of human corneal epithelial cells (HCE-T cell line). HCE-T cells are grown at the air-liquid interface on a collagen membrane where the cells stratify and differentiate to form an epithelial barrier similar to the corneal surface. The concentration of a test material that causes fluorescein retention by the HCE-T cultures to decrease to 85% relative to the control cultures (FR85) is the assay endpoint. TEP and transepithelial electrical resistance (TER) are not equally altered by a treatment, indicating that fluorescein permeability in the HCE-T model is regulated by tissue properties in addition to tight junction integrity, including the multiple cell layers, cell viability, and desmosomal junction integrity. Other endpoints that have been evaluated using the HCE-T model include lactate release, PGE2 release, various cytokines, and MTT dye uptake.

MDCK FL assay: The Fluorescein Leakage (FL) assay measures the dose-dependent effect of a test material on the amount of fluorescein that penetrates across a monolayer of MDCK cells cultured on permeable membrane culture inserts. MDCK cells form tight junctions that prevent passage of the fluorescein unless damaged by an applied chemical.

Validation and Acceptance of Non-animal Alternative Methods

In vitro ocular methods have a long and disappointing history regarding test method validation (Balls, et al., 1999; Spielmann & Liebsch, 2001). A recent approach to validation of ocular alternative test methods involves the separate assessment of methods for eye irritation and eye corrosion. ICCVAM/NICEATM has already evaluated four methods for their ability to assess substances that are severe/corrosive to the eye, and ECVAM is conducting a retrospective assessment of the validity of additional in vitro methods for their ability to assess substances that are mild to moderately irritating or nonirritating to the eye.

In 2003, four alternative test methods were nominated to ICCVAM/NICEATM for review as screening methods for severe eye irritation/corrosion: the Isolated Chicken Eye (ICE), Isolated Rabbit Eye (IRE), Hen's Egg Test-Chorioallantoic Membrane (HET-CAM), and Bovine Corneal Opacity and Permeability (BCOP) assays. Several EU authorities had already accepted these methods for the classification of severe/corrosive eye irritants. The ICCVAM review of these methods concluded with an expert panel report in March 2005 and with ICCVAM's endorsement of the BCOP and ICE methods as valid screens for corrosive and severe eye irritants. However, ICCVAM has determined that negative results still require in vivo testing before a conclusion of nonsevere/corrosive can be made.

The ICCVAM endorsement was stated as follows: "In 2007, BCOP and ICE recommended as screening tests for identifying corrosives and severe irritants, with certain limitations; HET-CAM and IRE not recommended for regulatory hazard classification purposes until further developed and evaluated." The specific limitations on the uses of these in vitro methods are described in the documents on the ICCVAM website. The US Food and Drug Administration (FDA) and US Environmental Protection Agendy (EPA) reported previous acceptance of BCOP data in specific circumstances and are expected to continue to do so considering ICCVAM's recommendations.

The ECVAM Scientific Advisory Committee (ESAC) statement on the ICCVAM retrospective study of the four in vitro screening assays for ocular corrosives/severe irritants endorsed the validity of the BCOP and ICE methods for use in a tiered strategy as part of the weight of evidence approach. ESAC indicated further work is needed for the IRE and HET-CAM methods (ESAC Statement, April 27, 2007).

Current non-animal methods considered valid for regulatory testing purposes (for limited applications) are listed in Table 3.

Table 3. In vitro ocular test methods considered valid for limited regulatory testing applications.


Test Purpose

Validation Authority


Bovine Corneal Opacity and Permeability (BCOP) assay Eye corrosion/severe irritation



Isolated Chicken Eye (ICE) assay Eye corrosion/severe irritation



Isolated Rabbit Eye (IRE) assay Eye corrosion


Hen's Egg Test—Chorio-Allantoic Membrane assay (HET-CAM) Eye corrosion



(a) Although not formally endorsed as valid, positive outcomes can be used for classifying and labeling substances as severe eye irritants (R41) in the EU

Validation studies for two cell-based in vitro test methods for assessing ocular irritants, the Gillette HCE-T TEP assay and the MatTek EpiOcular assay, were completed in 2001. The prevalidation study results for the HCE-T TEP assay were presented at the 2000 Alternative Toxicological Methods for the New Millennium meeting in Bethesda, MD, US (Ward, et al., 2003) and were submitted to the ICCVAM Ocular Toxicity Working Group (OTWG) in January 2001. The validation study was completed in January 2001, and the results were prepared for submission to the OTWG in 2001. Validation study results from the first phase of the EpiOcular study were also submitted to ICCVAM and reviewed by the OTWG. Additional studies were conducted and submitted to ECVAM for review. ICCVAM's nonsevere ocular irritant page now indicates only that ICCVAM is reviewing the validation status of in vitro methods for identifying nonsevere ocular irritants and nonirritants. NICEATM is still requesting the submission of in vivo and/or in vitro ocular data that can be added to its database of alternative methods for assessing nonsevere irritants.

ECVAM is also reviewing the validation status of several additional in vitro methods (Fluorescein leakage, Red blood cell lysis, Neutral red release, and Cytosensor microphysiometer). A prevalidation study has been completed for the SkinEthic HCE model (Van Goethem, et al., 2006), which may be included in the ECVAM assessment.