Tiered application of the neutral red release and EpiOcular™ assays for evaluating the eye irritation potential of agrochemical formulations

Regulatory Toxicology and Pharmacology 81 (2016) 407e420

Contents lists available at ScienceDirect

Regulatory Toxicology and Pharmacology journal homepage: www.elsevier.com/locate/yrtph

Tiered application of the neutral red release and EpiOcular™ assays for evaluating the eye irritation potential of agrochemical formulations Raja S. Settivari a, *, Ricardo Acosta Amado b, Marco Corvaro c, Nicolo R. Visconti a, Lynn Kan a, Edward W. Carney a, Darrell R. Boverhof a, Sean C. Gehen b a b c

The Dow Chemical Company, Midland, MI, USA Dow AgroSciences LLC, Indianapolis, IN, USA Dow Chemical Services UK Ltd, Milton Park, Abingdon, Oxon, UK

a r t i c l e i n f o

a b s t r a c t

Article history: Received 20 July 2016 Received in revised form 20 September 2016 Accepted 27 September 2016 Available online 28 September 2016

Agrochemical formulations have been underrepresented in validation efforts for implementing alternative eye irritation approaches but represent a significant opportunity to reduce animal testing. This study assesses the utility of the neutral red release assay (NRR) and EpiOcular™ assay (EO) for predicting the eye irritation potential of 64 agrochemical formulations relative to Draize data. In the NRR, formulations with an NRR50 value
50 mg/mL were categorized as UN GHS Cat 1 and those >250 mg/mL were classified as UN GHS Non Classified (NC). The accuracy, sensitivity, and specificity were 78, 85 and 76% and 73, 85 and 61% for identifying UN GHS 1 and NC formulations, respectively. Specificity was poor for formulations with NRR50 > 50 to
250 mg/mL. The EO (ET-40 method) was explored to differentiate formulations that were UN GHS 1/2 and UN GHS NC. The EO resulted in accuracy, sensitivity, and specificity of 65%, 58% and 75% for identifying UN GHS NC formulations. To improve the overall performance, the assays were implemented using a tiered-approach where the NRR was run as a first-tier followed by the EO. The tiered-approach resulted in improved accuracy (75%) and balanced sensitivity (73%) and specificity (77%) for distinguishing between irritating and non-irritating agrochemical formulations. © 2016 Elsevier Inc. All rights reserved.

Keywords: Ocular irritation Neutral red release assay EpiOcular assay In vitro Eye irritation Agrochemical formulations

1. Introduction Assessment of acute eye irritation potential is required for registration of agrochemical active substances and formulations prior to commercialization (EPA, 2007; Sanco, 2010). Eye irritation data can be obtained with accepted test methods such as those adopted by the Organization for Economic Cooperation and Development (OECD). The acute hazard characterization data derived from these tests is used for hazard communication purposes including classification and labeling and for defining the appropriate personal protection equipment (PPE) required for safe use of the product (EPA, 2014). Historically, the rabbit Draize ocular irritation test has been considered the global standard for assessing eye irritation potential

* Corresponding author. E-mail address: [email protected] (R.S. Settivari). http://dx.doi.org/10.1016/j.yrtph.2016.09.028 0273-2300/© 2016 Elsevier Inc. All rights reserved.

of chemicals. Per the most recent OECD test guideline (OECD TG 405), chemical-mediated ocular effects are evaluated in up to three albino rabbits. The effects are graded according to (a) the severity of lesions produced in the cornea, iris, and conjunctiva and (b) the duration over which the lesion(s) persist. While it has served as the standard approach for many years, the Draize test has several limitations including subjectivity related to qualitative scoring of effects, biological variability in animal response as well as anatomical differences between rabbit and human eyes like nictitating membrane, cornea thickness and lack of sufficient tearing effects (Williams et al., 1982; Scott et al., 2010; Adriaens et al., 2014; Barroso et al., 2016). Considering the above concerns as well as the ethical considerations related to animal testing, increased emphasis has been placed on developing mechanism-based non-animal alternatives to the Draize test. While there are several modes of action by which chemicals can cause ocular injury upon direct contact, most often the degree of ocular irritation is driven primarily by direct cytotoxicity (Scott

408

R.S. Settivari et al. / Regulatory Toxicology and Pharmacology 81 (2016) 407e420

et al., 2010; Hackett and McDonald, 1991; Fox and Boyes, 2008). A number of alternative model systems have been developed to assess the cytotoxic response of chemicals in different culture systems. These can be divided into (i) organotypic models: Bovine Corneal Opacity and Permeability (BCOP) assay; Isolated Chicken Eye (ICE) test and Hen's Egg Test-Chorioallantoic Membrane (HETCAM) assay, etc. (ii) 2-dimensional cell culture models: Neutral Red Release (NRR); Neutral Red Uptake (NRU); Short Time Exposure (STE) test; Fluorescence Leakage (FL) test; Cytosensor Microphysiometer (CM) test, etc. (iii) 3-dimensional cell models: EpiOcular™ (EO) assay; Human Corneal Epithelial Model (HCE), etc. While some of these methods have undergone formal validation and have internationally adopted guidelines, others are still under development. In each of these test systems, the test material is applied directly to the cells or tissue to mimic human topical exposure and then cell viability or membrane integrity is determined as an indicator of potential ocular effects. These alternative methods have many potential advantages including reduction in animal use, permitting robust concentrations-response evaluation, shorter turnaround time etc. With all of these alternative methods, accuracy in comparison to in vivo data-sets needs to be determined. While the ultimate goal is to maximize both sensitivity and specificity, it is of particular importance to reduce occurrence of false negative predictions especially for severe eye irritants (UN GHS Cat 1; which may cause extensive ocular damage) so as to ensure that resulting hazard communication materials are adequately health protective (for appropriate PPE usage). Ultimately, if products are misclassified, the appropriate PPE and precaution may not be indicated and therefore could represent an issue for safe handling and use by down-stream users. The NRR assay is a cytotoxicity-based alternative test method that uses mouse fibroblasts or human keratinocytes to identify potential ocular irritants. In this assay, the cells are incubated with a water-soluble weak cationic dye, 3-amino-7-dimethylamino-2methylphenazine hydrochloride (neutral red; NR), which selectively accumulates within lysosomes (due to the pH difference in lysosomes and cytoplasm) of healthy cells. Subsequent exposure of the NR dye-loaded cells to potential ocular irritants results in cell membrane damage and release of the NR dye which is quantified to correlate to the eye irritation potential of the chemical (Reader et al., 1990). The NRR assay has relatively short exposure period (i.e. 1 min) and permits evaluating test chemicals at concentrations up to 100%. The NRR assay was included in several inter-laboratory studies and external validation studies, including an ECVAM retrospective validation study (Zuang, 2001). It was one of six alternate test methods that had better correlation with the in vivo Draize ocular test for surfactants and hydro-alcohol formulations belonging to Global Harmonization System (UN GHS) category 1, however, it tended to over-predict UN GHS category 2 and category Non Classified (NC) substances (Gettings et al., 1994, 1996). The EpiOcular assay is another in vitro cytotoxicity-based assay for determining eye irritation potential of chemicals. The tissues used in this assay are reconstructed from primary human keratinocytes, which are cultured for several weeks to form a highly differentiated, multi-layered stratified squamous epithelium that is morphologically similar to that found in the human cornea. In the EpiOcular assay, the test chemical is applied topically and the exposure time required to reduce tissue viability by 40% of controls is measured (ET-40 method) to discriminate chemicals belonging to UN GHS NC from ocular irritants. The performance of the EpiOcular Eye Irritation Test (EIT) was evaluated by the European Union Reference Laboratory for Alternatives to Animal Testing (EURL ECVAM) and Cosmetic Europe between 2008 and 2013. Recently an OECD test guideline was published for the EIT for the identification of UN GHS Cat NC compounds from ocular irritants (OECD TG 492).

Both the NRR and the EpiOcular assays have been evaluated for their performance primarily using single compounds or a few simple mixtures in their respective validation studies. Agrochemical formulations are diverse and often complex mixtures containing one or more active ingredients and various co-formulants (e.g. solvents, surfactants, etc.). The combination and concentrations of co-formulants are specific to agrochemical formulations and serve a function in delivering the active ingredient effectively in field applications (adjuvants, antifoam, biocides etc). Often the agrochemical formulations consist of high concentrations of active ingredients and co-formulants as they are intended to be diluted several-fold before their use. These characteristics may impact osmolarity and surface tension thereby presenting challenges to successfully apply these alternative methods. Successful application of alternative methods for formulations would provide a major advancement for the agrochemical industry in reducing animal use since one active ingredient can be formulated to make multiple formulated products, each requiring an evaluation of eye irritation potential prior to registration. The purpose of the current study was to evaluate the performance of the NRR and EpiOcular assays alone and in combination for a variety of agrochemical formulations. In this study, a total of 64 and 51 agrochemical formulations were evaluated in the NRR and EpiOcular assays, respectively (Table 1). In each case, in vivo ocular irritation data was already available for these formulations. As the NRR and EpiOcular assays cover different regions of the eye irritation spectrum, these assays when used in a tiered–approach are expected to complement each other and provide better predictions compared to the existing in vivo Draize eye irritation data. Therefore, the data from both assays were also applied in tiered–approach and the results were compared to existing in vivo Draize data. Based on the performance of the selected test methods on the agrochemical formulations included in this study, the NRR and EpiOcular assays were concluded to provide encouraging performance for assessing ocular irritation potential when applied in a tiered-manner. 2. Materials and methods 2.1. Test materials All prototypical compounds (sodium dodecyl sulfate, benzalkonium chloride, Tween™ 80, Triton™ X-100, polyethylene glycol 400, ammonium lauryl sulfate, benzethonium chloride, and Tween™ 20) used to demonstrate technical proficiency of the conducting laboratory (i.e., The Dow Chemical Company) were purchased from Sigma-Aldrich, St. Louis, MO. In the main study, 64 agrochemical formulations corresponding to eleven formulations types (Tables 1A and 1B) were assessed. All agrochemical formulations (liquids and solids) were of commercial quality, obtained from Dow AgroSciences, Indianapolis, IN and represented all categories (Cat) of eye irritation potential according to the UN GHS classification and labeling (ST/SG/AC.10/30/Rev.5 e UN e New York and Geneva, 2013) (Table 1B). Results of in vivo rabbit studies were used in order to establish the classification according to the UN GHS criteria. For the scope of this study, the following definitions were used: ‘Severe Irritant’ for UN GHS Cat 1, ‘Moderate Irritant’ for UN GHS Cat 2A, ‘Mild irritant’ for UN GHS Cat 2B, ‘non classified’ for those formulations which do not meet UN GHS criteria for classification (UN GHS Cat NC) and a general descriptor of ‘Irritants’ for UN GHS Cat 2 (Cat 2A and 2B). All the in vivo studies for the selected agrochemical formulations have been previously performed as part of the regulatory requirements and no additional animal experiments were conducted for the purpose of this study. Cell culture reagents including Dulbecco's Modified Eagle's Medium (DMEM), Dulbecco's phosphate buffered saline (DPBS)

R.S. Settivari et al. / Regulatory Toxicology and Pharmacology 81 (2016) 407e420

409

Table 1A Agrochemical formulation types tested in the current study using the Neutral Red Release (NRR) & EpiOcular™ (ET-40 method) assays. Formulation typea

CropLife CropLife definition code

Soluble concentrate

SL

Emulsifiable EC Concentrate Suspension concentrate SC (¼ flowable concentrate) Wettable Powder WP Micro-emulsion

ME

Water Dispersible Granules Oil Dispersion

WG OD

Emulsion oil in water

EW

Capsule suspension

CS

Suspo-emulsion

SE

Emulsion, water in oil

EO

a

AI presentation

Continuum phase

A clear to opalescent liquid to be applied as a solution of the active ingredient after Salt dilution in water. The liquid may contain water-insoluble formulants. A liquid, homogeneous formulation to be applied as an emulsion after dilution in water. Organic solvent soluble (ester or oil soluble salt) A stable suspension of active ingredient(s) with water as the fluid, intended for dilution Water insoluble solid with water before use. A powder formulation to be applied as a suspension after dispersion in water.

Water insoluble solid

A clear to opalescent, oil and water containing liquid, to be applied directly or after dilution in water, when it may form a diluted microemulsion or a conventional emulsion. A formulation consisting of granules to be applied after disintegration and dispersion in water. A stable suspension of active ingredient(s) in a water- immiscible fluid, which may contain other dissolved active ingredient(s), intended for dilution with water before use. A fluid, heterogeneous formulation consisting of a solution of pesticide in an organic liquid dispersed as fine globules in a continuous water phase. A stable suspension of capsules in a fluid, normally intended for dilution with water before use. A fluid, heterogeneous formulation consisting of a stable dispersion of active ingredient(s) in the form of solid particles and of water-non-miscible fine globules in a continuous water phase. A fluid, heterogeneous formulation consisting of a solution of pesticide in water dispersed as fine globules in a continuous organic liquid phase.

Salt and either ester or oil soluble salt Water insoluble solid Organic-solvent insoluble solid Salt and either ester or oil soluble salt Organic solvent soluble (ester or oil soluble salt) Organic solvent soluble (ester or oil soluble salt) and water insoluble solid Organic solvent soluble (ester or oil soluble salt) and water insoluble solid

Water Organic Water

Solid carrier Water

Solid carrier Organic Water Water Water

Organic

As defined in the CropLife's Catalog of pesticide formulation types and international coding system, Technical Monograph no. 2, 6th Edition, 2008.

Table 1B Summary of all the agrochemical formulations tested in the current study. UN GHS classification

Physical statea

Totals

Cat 1

L S L S L S L S

12 1 10 1 8 1 26 5

Cat 2A Cat 2B Non classified

Formulation typeb

13

SL

EC

4

3

SC

WP

ME

WG

OD

EW

CS

SE

4

EO 1

1 11

2

3

1

3

1

1 9

1

3

1

2

2

6

1

1 31

2

1

4

1

6

4

5

a

L ¼ liquid; S ¼ solid. SL ¼ Soluble concentrate; EC ¼ Emulsifiable Concentrate; SC ¼ Suspension concentrate; WP ¼ Wettable powder; ME ¼ Micro-emulsion; WG ¼ Water Dispersible Granules; OD ¼ Oil Dispersion; EW ¼ Emulsion oil in water; CS ¼ Capsule suspension; SE ¼ Suspo-emulsion; EO ¼ Emulsion, water in oil. b

with calcium and magnesium chloride, Hepes buffer, penicillin (10,000 U/ml)-streptomycin (10,000 mg/mL) mixture and fetal bovine serum (FBS) as well as neutral red (NR) dye were purchased from Sigma-Aldrich Chemical Co (St. Louis, MO, USA) or Gibco BRL (Grand Island, NY, USA). Culture reagents for EpiOcular assay including DMEM-based medium, positive control (Triton™ X-100), extraction medium (isopropyl alcohol), MTT concentrate and diluents were obtained from MatTek Corporation (Ashland, MA, USA). 2.2. Cell cultures The NRR assay was originally developed using Balb/c mouse embryonic fibroblast cell line, 3T3 (Reader et al., 1989, 1990; Zuang, 2001; Balls et al., 1991). Subsequently, other cell types, including normal human epidermal keratinocytes or the spontaneously immortalized human keratinocyte cell line, HaCaT were used for conducting the NRR assay (Zuang, 2001). Current literature on the cell types suggests either comparable NRR results or one cell line being more sensitive over the other depending on the tested chemicals as well as the conducting laboratory (Benavides et al.,

2004; Zanatta et al., 2008; Maier et al., 1991; Sanchez et al., 2004). As no specific guidance was available on appropriate cell type for testing agrochemical formulations, in the present study NRR assay was initially conducted using both 3T3 and HaCaT cell lines. In this analysis, a set of 11 surfactants and agrochemical formulations were evaluated and the results were compared to published NRR and/or prior in vivo data to determine the appropriate cell line to be used in this study (Table 2). The 3T3-L1 cells (ATCC CCL92.1) were cultured in DMEM medium (American Type Culture Collection; ATCC, Manassas, VA) supplemented with 10% fetal calf serum, 0.4% fungizone and 1% penicillin (10,000 U/ml) e streptomycin (10,000 mg/mL) mixture at 37  C, 5% CO2. The spontaneously immortalized human keratinocyte cell line HaCaT was cultured in DMEM medium with GlutaMAX™-l (Life technologies, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS) and 0.55 mg/mL Geneticin® (G-418; Life technologies, Carlsbad, CA) in humidified atmosphere at 37  C and 5% CO2. Both cell lines were routinely sub-cultured into 75 cm2 culture flasks at approximately 80% confluency, for a maximum of 25 continuous passages.

410

R.S. Settivari et al. / Regulatory Toxicology and Pharmacology 81 (2016) 407e420

Table 2 List of prototypical compounds and agrochemical formulations tested in the NRR assay using 3T3 and HaCaT cells. Test item

Sodium dodecyl sulfate Benzalkonium chloride Tween™ 80 Triton™ X-100 Polyethylene glycol 400 Ammonium lauryl sulfate Benzethonium Chloride Tween™ 20 DAS-003 DAS-004 DAS-058 a

In vivo interpretation

Severe Severe Mild Severe Mild Severe Severe Mild Severe Severe Non irritant

Published NRR50 (mg/mL)

1.1 ± 0.2 (Balls et al., 1991) 0.4 ± 0.1 (Balls et al., 1991) >830 (Balls et al., 1991) <10 (Clothier, 1992) 860 ± 43 (Reader et al., 1989) 8.3 ± 2.1 (Hubbard et al., 1994) 14 ± 5 (Balls et al., 1991) 433 ± 44 (Balls et al., 1991) NA NA NA

In-house NRR50 values (mg/mL) 3T3

HaCaT

Interpretation (3T3/HaCaT)

0.7 2.0 >1000 12.5 846.0 0.1 16.9 >1000 37.1 40.2 a 25.0

2.1 2.0 >1000 8.1 846.0 0.2 7.5 >1000 37.7 37.6 >1000

Severe/Severe Severe/Severe Mild/Mild Severe/Severe Mild/Mild Severe/Severe Severe/Severe Mild/Mild Severe/Severe Severe/Severe a Severe/Non irritant

Detachment of cells; No cytotoxicity was noted.

2.3. The neutral red release assay The NRR assay was performed as described in INVITTOX protocol No. 54, with minor modifications (Reader et al., 1989). For chemical exposure, the cells were seeded at a density of 15,000 cells/well (1  105 cells/ml; 150 ml/well), in the central 60 wells of a clear, flat bottom 96-well plate and cultured at 37  C and 5% CO2. At confluency (approximately 48-h of culture), the cells were incubated with NR dye (5 mg/mL; Sigma-Aldrich, St. Louis, MO) for three hours under standard cell culture conditions. After incubation, the NR dye-containing media was replaced with fresh pre-warmed medium to terminate NR dye uptake by the cells. The cells were then directly exposed to each test material at eight concentrations. The test items that are either liquids or gels were tested at the highest concentration of 100% and seven additional concentrations (dissolved in DPBS). In the case of solids and highly viscous gels (pastes), the highest possible pipetable uniform suspension (dissolved in DPBS followed by sonication) was used as the highest concentrations. Each 96-well testing plate consisted of up to 2 test chemicals tested in triplicate wells at eight concentrations, the positive control (5 mg/mL sodium hydroxide; 6 wells), negative control (DPBS; 6 wells) and no-cell blank (2 wells). The cells were exposed to each test material for one minute at room temperature, followed by up to 3 washes (Vs. 1-time described in INVITTOX protocol No. 54) to remove the NR dye released from lysed cells as a result of treatment. The cells were then incubated on a microtitreplate shaker at room temperature for three minutes with extraction solution (containing 50% ethanol absolute-1% acetic acid in distilled water) to release NR dye from cells that were intact following treatment. The optical density of the resulting solution was measured at a wavelength of 540 nm in a BMG LABTECH FLUOstar Omega spectrophotometer (Ortenberg, Germany). The concentration of test item producing a 50% release of pre-loaded NR dye (NRR50) was obtained by non-linear interpolation of the concentrations-response curve. The NRR50 value was used to determine eye irritation potential of the test items. All agrochemical formulations were tested on at least two separate days (independent experiments) and an average of the independent studies is presented in this manuscript. Each replicate was considered acceptable when the following criteria were met: (a) no cytotoxicity in the negative control and (b) greater than 90% cytotoxicity in the positive control. The NRR assay results were interpreted based on the criteria described in Table 3. 2.4. The EpiOcular™ assay The EpiOcular OCL-200 kit was purchased from MatTek Corporation (Ashland, MA, USA). In this, each test material was applied to

Table 3 The NRR assay interpretation criteria. NRR50 (mg/mL)

UN GHS classification

ECVAM-proposed criteria 0 to 250 UN GHS Cat >250 to 600 UN GHS Cat >600 UN GHS Cat Modified classification criteria <50 UN GHS Cat >50 and
250 UN GHS Cat >250 UN GHS Cat a

Label

1 2a 2B or aNC

Severe irritant Moderate irritant Mild irritant or Non Classified

1 2 a NC

Severe irritant Moderate or Mild irritant Non Classified

NC ¼ Non Classified.

the tissue surface at either 50 ml (liquids and suspensions) or 100 mg (solids) in duplicates. Highly viscous test materials were applied to cover the flat end of a dosing device (a sterile push pin with the point cut off) and the pin was then placed onto the tissue, sample side down. Following application of test item, EpiOcular tissues were incubated for at least three-time points (3, 15, 30/ 60 min) at 37  C in 5% CO2/95% air. Tissues treated with the positive control, Triton™ X-100 were incubated for 15 and 45 min and those incubated with negative control, deionized water were incubated at the longest exposure time period (i.e. 30 or 60 min). Following treatment, tissues were thoroughly rinsed with DPBS, and incubated with 5 mL of assay medium for about 10 min, rinsed with DPBS and evaluated for treatment-related cytotoxicity using a tetrazolium-based assay. The EpiOcular tissues were incubated with 3-(4,5-dimethylthiazo-2-yl)-2,5-diphenyl-tetrazolium bromide (MTT) dye (5 mg/mL) for 3 h at 37 C in 5% CO2/95% air. The viable cells of the EpiOcular tissue reduce the MTT dye into a blue MTT formazan precipitate. Following incubation, the cultures were washed with DPBS and incubated with dimethyl sulphoxide overnight (in sealed plastic bags) at room temperature to extract the purple formazan product from the cells. Following extraction, 200 ml of formazan solution for each sample was transferred to a 96-well plate and optical densities were quantified at 600 and 630 nm wavelengths in a BMG LABTECH FLUOstar Omega spectrophotometer (Ortenberg, Germany). The ratio of absorbance (%) of each test item to that of negative control was represented as relative viability. The mean viability values for replicate tissues were obtained. The irritation category classification was determined based on the relative cytotoxicity as determined by ET40 value (Hayashi et al., 2012). The ET40 represents the time for manifestation of 40% cytotoxicity as determined by linear interpolation between time points above and below 40% cytotoxicity. A test item that had ET40
30 min was categorized as an irritant (UN GHS Cat 1 or 2), while items with ET40 > 30 min as non classified (UN GHS NC).

R.S. Settivari et al. / Regulatory Toxicology and Pharmacology 81 (2016) 407e420

2.5. Tiered-testing approach In this retrospective top-down tiered-approach, the NRR assay was considered as the 1st tier for the identification of severe irritants (NRR50
50 mg/mL) and formulations not requiring classification (NRR50 > 250 mg/mL). The EpiOcular assay (ET-40 method) was considered as the 2nd tier test for the evaluation of formulations that had NRR50 between 50 and 250 mg/mL. For these formulations, the NRR results were considered inconclusive and only EpiOcular data was considered for final interpretation. The overall accuracy (the proportion of correct positive and negative outcomes of the test methods) of the tiered-testing approach was evaluated in relation to existing in vivo Draize eye test data. 3. Results 3.1. Physico-chemical properties and eye irritation potential The physical and chemical properties of a test substance influence corneal penetration and the extent of the ocular damage. As formulations represent complex mixtures, evaluation of physicochemical properties is an important first step in the eye irritation evaluation process. The physical and chemical properties to consider include physical form, as well as chemical properties including pH and buffering capacity, lipid/water partition coefficient and protein reactivity. The OECD guideline 405 states that the strongly acidic or alkaline substances (pH
2 or  11.5) do not require testing owing to their potential corrosive properties. Likewise, Worth and Cronin (2001)proposed that chemicals with pH <3.2 or >8.6 could be classified as irritants. In the present study, the pH values as well as the acid/alkali reserve of the SL-type formulations (proof of concept, as SL-type formulations were present in all UN GHS categories; Table 1B) were evaluated as described earlier (Young et al., 1988) (OECD 122). In general, the pH values (1% aqueous solution) of tested agrochemical formulations were near the neutral zone. No clear correlation was observed between the buffering capacity and in vivo eye irritation categorization for the SL-type formulations with pH > 4.5 and < 8.3 tested in this study (Supplementary Table 1). 3.2. Technical proficiency with the NRR assay To demonstrate in-house laboratory technical proficiency and reproducibility, eight test chemicals listed in earlier publications were evaluated in the NRR assay (Table 2). In this analysis five severe and three mild irritants were tested using the spontaneously immortalized human keratinocytes, HaCaT cells and the mouse embryonic fibroblast cell line, 3T3 cell lines. The treatment-related cell survival rates were measured at eight concentrations, in triplicate wells per concentration. The concentration resulting in 50% release of the neutral red dye (NRR50, expressed in mg/mL) was calculated and the results were interpreted following ECVAMproposed criteria (Table 3). The severe irritants increased NR dye release (cell death) in a concentrations-dependent manner; whereas, milder irritants did not exhibit substantial cytotoxicity. As shown in Table 2, the in-house results for all the proficiency compounds were comparable with the NRR results reported in published literature as well as prior generated in vivo Draize eye irritation data. Overall, the NRR response between the 3T3 and HaCaT cells were similar with the two cell lines giving the same classification and rank order for the proficiency compounds. These results demonstrated technical proficiency of the laboratory to effectively conduct the NRR assay. In general, the ECVAM-suggested thresholds provided good predictions for single chemicals representing severe and mild irritant categories.

411

3.3. NRR test results for agrochemical formulations (ECVAM prediction model) As comparable results were observed for the above-listed proficiency compounds in 3T3 and HaCaT cell lines, next a subset of agrochemical formulations (DAS-003, DAS-004, and DAS-058) were tested in both cell lines (Table 2). The NRR results for agrochemical formulations were comparable between the cell types; however, the 3T3 cells exhibited challenges with regard to adherence to the 96- well micro-plate surface following treatments, resulting in lifting of the cells in the absence of apparent cytotoxicity (Fig. 1; Table 2). As a result, DAS-058 was over-predicted as severe irritant using 3T3 cells, while correctly predicted as nonirritant using HaCaT cells. The 3T3 cells did not exhibit such limitation when tested with the selected proficiency compounds. To limit such false positive predictions (due to adherence issues) for agrochemical formulations, subsequent NRR experiments were conducted using only the HaCaT cells. Next, 64 agrochemical formulations were tested in the NRR assay, of which, 13 belonged to UN GHS Cat 1, 11 to UN GHS Cat 2A, 9 to UN GHS Cat 2B and 31 were UN GHS Cat NC, (according to the existing in vivo Draize test data). The selected formulations (solids and liquids) represented 11 types of agrochemical formulations (Table 1B). Each formulation was tested at 8 concentrations and NRR50 values were determined. The NRR data for formulations were interpreted according to the ECVAM retrospective validation model (Table 3). When following this approach, the NRR assay correctly predicted all 13 severe irritants (UN GHS Cat 1). Amongst the UN GHS Cat 2A formulations, NRR assay correctly predicted only 2 of 11 formulations, while 8 were over-predicted as UN GHS Cat 1 and one was under-predicted as UN GHS Cat 2B/NC (Fig. 2a). Likewise, out of 40 formulations belonging to UN GHS Cat 2B/NC, 14 were correctly predicted, 21 were over-predicted as UN GHS Cat 1, and 5 formulations were over-predicted as UN GHS Cat 2A. Overall, the ECVAM retrospective validation model provided accurate predictions for severe irritants (UN GHS Cat 1), however, exhibited a higher tendency to over-predict moderate and mild irritants and non classified formulations, when compared to the in vivo data (Fig. 2A; Tables 4 and 5). These results are similar to earlier publications where the NRR assay was shown to over-predict moderate and mild irritants (Harbell et al., 1997). As the ECVAM retrospective classification criteria for NRR assay was empirically derived predominantly based on data for milder, water soluble, and relatively less complex compounds, there was a need to refine the NRR classification criteria that would suit classification of complex multi-component agrochemical formulations. In this study, two modifications were made to the ECVAM retrospective validation model: i) categorizing formulations into UN GHS Cat 1, 2, and NC instead of UN GHS Cat 1, 2A, and 2B/NC and ii) changing the categorization thresholds. Per the modified approach, formulations with NRR50
50 mg/mL were considered severe irritants (UN GHS Cat 1), >50 and
250 mg/mL as irritants (UN GHS Cat 2), and >250 mg/mL as non-classified (UN GHS Cat NC) (Table 3). Thresholds were empirically determined to maximize the sensitivity and specificity of UN GHS Cat 1 and UN GHS Cat NC formulations. 3.4. Classification based on modified NRR prediction model When interpreted using the modified approach, the NRR assay correctly predicted 11 of 13 severe irritants with 2 formulations (liquid, ME-type formulations) under-predicted as UN GHS Cat 2 (DAS-005 (NRR50: 97 mg/mL) and DAS-014 (NRR50: 72 mg/mL)) (Fig. 2B). Of the 20 irritants tested in this analysis, the NRR assay predicted 8 formulations as UN GHS Cat 2, while over-predicted 7 as UN GHS Cat 1 and under-predicted 5 as UN GHS Cat NC. Amongst

412

R.S. Settivari et al. / Regulatory Toxicology and Pharmacology 81 (2016) 407e420

Fig. 1. Comparison of NRR test results using 3T3 and HaCaT cells. The NRR test results were comparable between 3T3 and HaCaT cells for Triton™ X-100 (a and b), DAS-003 (c and d). However, DAS-058 (e and f) exhibited false positive results in the 3T3 cells due to lifting of the cells in the absence of apparent cytotoxicity. This formulation was correctly identified as non-classified in HaCaT cells.

R.S. Settivari et al. / Regulatory Toxicology and Pharmacology 81 (2016) 407e420

413

Fig. 2. The performance of the in vitro NRR assay when interpreted based on (a) ECVAM thresholds (NRR50 ¼ 250 and 600 mg/mL) and (b) proposed modified thresholds (NRR50 ¼ 50 and 250 mg/mL. Data distribution is presented using the scatter dot-plots, the first and third quartiles of the data distribution are at the ends of the error bars, the median value is the line within the bar.

the 31 UN GHS NC formulations, the NRR assay correctly identified 19, however, over-predicted 5 as UN GHS Cat 1 and 7 as UN GHS Cat 2 (Table 5). Overall, the NRR assay exhibited better predictions for the identification of formulations belonging to UN GHS Cat 1, followed by UN GHS Cat NC; however the assay exhibited lack of predictivity compared to in vivo Draize ocular data for UN GHS Cat 2, i.e. when NRR50 was between 50 and 250 mg/mL (Fig. 2). A similar trend was observed for the NRR results between the solid and liquid type formulations (Suppl Table 2). 3.5. EpiOcular™ test results for agrochemical formulations As it became clear that the NRR did not provide adequate predictive performance in the mid-range of NRR50 values (those with NRR50 values between 50 and 250 mg/L), the EpiOcular assay was selected as a potential second-tier assay to improve the overall predictive power. The EpiOcular was conducted on 51 formulations, with an emphasis on ensuring coverage of all formulation types and those with NRR50 values between 50 and 250 mg/L (Table 1A). The formulations selected represented UN GHS Cat 1 (n ¼ 12), UN GHS Cat 2A (n ¼ 10), UN GHS Cat 2B (n ¼ 9) and UN GHS Cat NC (n ¼ 20). Each formulation was tested at three exposure time points and an ET40 value was calculated as described above. The formulations were categorized as potential ocular irritants (UN GHS Cat 1/Cat 2) or non classified (UN GHS Cat NC) when ET40 was <30 or 30 min, respectively (Table 6). Of the 31 formulations belonging to UN GHS Cat 1/Cat 2, the EpiOcular assay classified 18 as irritants (ET40 < 30 min), while 13 were under predicted as non classified (ET40 > 30 min). The EpiOcular assay correctly predicted 15 of 20 UN GHS NC formulations, and 5 were over predicted as irritants (Fig. 3; Table 7). Overall, with in vivo Draize as a reference, the EpiOcular assay exhibited better predictions for the identification of formulations belonging to UN GHS Cat NC and tended to underpredict true irritants (Table 7). A similar trend was noted for the

EpiOcular results between the solid and liquid type formulations (Suppl Table 3).

3.6. Combined testing approach As a stand-alone assessment, the NRR assay with the modified criteria correctly predicted 11 of 13 UN GHS Cat 1 and 19 of 31 UN GHS Cat NC formulations. Of the tested 20 UN GHS Cat 2 formulations that were tested, the NRR correctly predicted only 8 formulations. The EpiOcular assay as standalone correctly predicted 18 of 31 UN GHS Cat 1/2 and 15 of 20 UN GHS Cat NC formulations. As the NRR assay provided encouraging performance to identify formulations representing UN GHS Cat 1 and NC categories and the EpiOcular assay for identifying UN GHS Cat NC from irritants, the performance of both assays were evaluated when applied in a combined approach, consisting of NRR as tier-1 and EpiOcular as tier-2 test (Fig. 4). In this approach, test materials were assigned to UN GHS Cat 1 or NC when the NRR50 value was less than 50 or greater than 250 mg/mL, respectively. For the chemicals that had NRR50 between 50 and 250 mg/mL (n ¼ 17; 10 formulations belonging to UN GHS Cat 1/2 and 7 belonging to UN GHS Cat NC), the EpiOcular results were considered for final interpretation (in bold in Table 8), as in this range the NRR test exhibited higher misclassification compared to in vivo Draize rabbit eye test (Fig. 2). As the second tier assay, the EpiOcular assay correctly predicted 6 out of 10 formulations belonging to UN GHS Cat 1/2 and 5 out of 7 formulations belonging to UN GHS Cat NC. Overall, the tieredapproach correctly identified 12 of 13 UN GHS Cat 1, 5 of 20 UN GHS Cat 2, and 24 of 31 UN GHS Cat NC, compared to the in vivo rabbit Draize data (Tables 8e10). Therefore, the tiered-approach provided encouraging results to identify UN GHS Cat 1 and NC formulations; however, this combined approach misclassified majority of formulations representing UN GHS Cat 2.

414

R.S. Settivari et al. / Regulatory Toxicology and Pharmacology 81 (2016) 407e420

Table 4 Assessment of agrochemical formulations for eye irritation potential based on the NRR assay. The formulations were evaluated following ECVAM suggested validation model and modified approach. Test sample

Actual (in vivo) UN GHS

Physical state

Agrochemical formulation type

Results NRR50 (mg/ml)

NRR50 classification ECVAM model

Modified model

Cat 1 (0
250); Cat 2A (>250
600); Cat 2B Cat 1 (0
50); Cat 2 (>50
250); Cat or NCa (>600) NCa (>250) DAS-001 DAS-002 DAS-003 DAS-004 DAS-005 DAS-006 DAS-007 DAS-008 DAS-009 DAS-010 DAS-011 DAS-012

1 1 1 1 1 1 1 1 1 1 1 1

Solid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid

WP ME SL SL ME EC EC EO EC SL SL ME

6.8 7.7 37.0 38.0 97.0 34.1 45.3 37.2 16.0 13.0 19.0 31.0

Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat

1 1 1 1 1 1 1 1 1 1 1 1

Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat

1 1 1 1 2 1 1 1 1 1 1 1

DAS-014 DAS-013 DAS-015 DAS-016 DAS-017 DAS-018 DAS-019 DAS-020 DAS-021 DAS-022 DAS-023 DAS-032

1 2A 2A 2A 2A 2A 2A 2A 2A 2A 2A 2A

Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Solid

ME EC OD ME SL SL OD OD EC SE EC WG

72.0 7.4 88.0 26.0 30.4 65.7 41.0 80.0 369.0 >1000 340.0 11.7

Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat

1 1 1 1 1 1 1 1 2A 2B or NC 2A 1

Cat Cat Cat Cat Cat Cat Cat Cat NC NC NC Cat

2 1 2 1 1 2 1 2

DAS-024 DAS-025 DAS-026 DAS-027 DAS-028 DAS-029 DAS-030 DAS-031 DAS-042 DAS-033 DAS-034 DAS-035 DAS-036 DAS-037 DAS-038 DAS-039 DAS-040 DAS-041 DAS-045 DAS-046 DAS-047 DAS-048 DAS-049 DAS-050 DAS-051 DAS-052 DAS-053 DAS-054 DAS-055 DAS-056 DAS-057 DAS-058 DAS-059 DAS-060 DAS-061 DAS-062 DAS-063 DAS-064 DAS-065 DAS-066

2B 2B 2B 2B 2B 2B 2B 2B 2B NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC

Liquid Liquid Liquid Solid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Solid Solid Liquid Liquid Liquid Liquid Solid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Solid Solid Liquid

EC EW SL WG EC SE EW EC OD ME OD WG WG EC OD EW EW WG SE SE CS SC SE SC CS CS SE EW SC SC SL SL EW EW CS CS CS WG WG EW

10.0 56.2 22.0 67.0 57.0 >1000 214.0 521.0 94.0 25.1 16.1 15.0 79.0 51.3 38.0 54.0 90.0 57.0 146.0 321.0 231.0 350.2 476.0 >500 >1000 >1000 >1000 34.0 436.0 846.3 645.7 >1000 >1000 >1000 >1000 >1000 >1000 >1000 257.0 431.0

Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat

1 1 1 1 1 2B or 1 2A 1 1 1 1 1 1 1 1 1 1 1 2A 1 2A 2A 2B or 2B or 2B or 2B or 1 2A 2B or 2B or 2B or 2B or 2B or 2B or 2B or 2B or 2B or 2A 2A

Cat Cat Cat Cat Cat NC Cat NC Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat Cat NC Cat NC NC NC NC NC NC Cat NC NC NC NC NC NC NC NC NC NC NC NC

1 2 1 2 2

a

NC ¼ Non Classified.

NC

NC NC NC NC

NC NC NC NC NC NC NC NC NC

1

2 2 1 1 1 2 2 1 2 2 2 2 2

1

R.S. Settivari et al. / Regulatory Toxicology and Pharmacology 81 (2016) 407e420

415

Table 5 Performance of the NRR assay for predicting eye irritation potential of agrochemical formulations using two NRR prediction models. ECVAM NRR prediction model

In vivo

UN GHS Cat 1 UN GHS Cat 2A UN GHS Cat 2B/NC

UN GHS Cat 1 (NRR50
250)

UN GHS Cat 2A (NRR50 > 250 and
600)

UN GHS Cat 2B/NC (NRR50 > 600)

13 8 19

0 2 7

0 1 14

Modified NRR prediction model

In vivo

UN GHS Cat 1 UN GHS Cat 2A/2B UN GHS Cat NC

UN GHS Cat 1 (NRR50
50)

UN GHS Cat 2A/2B (NRR50 > 50 and
250)

UN GHS Cat NC (NRR50 > 250)

11 7 5

2 5 7

0 5 19

Table 6 Assessment of agrochemical formulations for eye irritation potential based on the EpiOcular™ (ET-40 method) assay. Test sample

Actual (in vivo) UN GHS

Physical state

Agrochemical formulation typeb

Results EpiOcular™ ET40 (min)

Classification ET40 EpiOcular™ Cat 1/Cat 2 (<30 min); Cat NCa (30 min)

DAS-001 DAS-002 DAS-003 DAS-004 DAS-005 DAS-006 DAS-007 DAS-008 DAS-009 DAS-010 DAS-011 DAS-014

1 1 1 1 1 1 1 1 1 1 1 1

Solid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid

WP ME SL SL ME EC EC EO EC SL SL ME

8.1 25.6 <3 <3 3.9 41.5 14.8 41.6 7.2 <3 4.6 51.7

Cat Cat Cat Cat Cat NC Cat NC Cat Cat Cat NC

DAS-013 DAS-015 DAS-017 DAS-018 DAS-019 DAS-020 DAS-021 DAS-022 DAS-023 DAS-032

2A 2A 2A 2A 2A 2A 2A 2A 2A 2A

Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Solid

EC OD SL SL OD OD EC SE EC WG

<3 15.7 5.6 6.1 >30 45.5 >60 >60 36.6 >30

Cat Cat Cat Cat NC NC NC NC NC NC

1/Cat 1/Cat 1/Cat 1/Cat

DAS-024 DAS-025 DAS-026 DAS-027 DAS-028 DAS-029 DAS-030 DAS-031 DAS-042 DAS-033 DAS-034 DAS-035 DAS-036 DAS-037 DAS-038 DAS-039 DAS-040 DAS-041 DAS-045 DAS-046 DAS-047 DAS-048 DAS-049 DAS-054 DAS-055 DAS-056 DAS-064 DAS-065 DAS-066

2B 2B 2B 2B 2B 2B 2B 2B 2B NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC

Liquid Liquid Liquid Solid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Solid Solid Liquid Liquid Liquid Liquid Solid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Solid Solid Liquid

EC EW SL WG EC SE EW EC OD ME OD WG WG EC OD EW EW WG SE SE CS SC SE EW SC SC WG WG EW

7.4 >60 10.2 19.9 22.1 >60 >60 39.2 23.2 >30 21.1 19.1 >30 31.8 >30 28.2 >60 >30 >30 >30 18.5 >60 >30 11.0 >60 >60 >60 >60 46.8

Cat NC Cat Cat Cat NC NC NC Cat NC Cat Cat NC NC NC Cat NC NC NC NC Cat NC NC Cat NC NC NC NC NC

1/Cat 2

a

1/Cat 1/Cat 1/Cat 1/Cat 1/Cat

2 2 2 2 2

1/Cat 2 1/Cat 2 1/Cat 2 1/Cat 2 2 2 2 2

1/Cat 2 1/Cat 2 1/Cat 2

1/Cat 2 1/Cat 2 1/Cat 2

1/Cat 2

1/Cat 2

1/Cat 2

NC ¼ Non Classified. SL ¼ Soluble concentrate; EC ¼ Emulsifiable Concentrate; SC ¼ Suspension concentrate; WP ¼ Wettable powder; ME ¼ Micro-emulsion; WG ¼ Water Dispersible Granules; OD ¼ Oil Dispersion; EW ¼ Emulsion oil in water; CS ¼ Capsule suspension; SE ¼ Suspo-emulsion; EO ¼ Emulsion, water in oil. b

416

R.S. Settivari et al. / Regulatory Toxicology and Pharmacology 81 (2016) 407e420

Fig. 3. The performance of the EpiOcular™ (ET-40 method) assay for identifying agrochemical formulations with non-irritant (UN GHS Cat NC) and irritant (UN GHS Cat 1 þ 2) potential. Data distribution is described using the scatter dot-plots, the first and third quartiles of the data distribution are at the ends of the error bars, the median value is the line within the bar.

Table 7 Performance of the EpiOcular (ET-40 method) assay for predicting eye irritation potential of agrochemical formulations. Epiocular prediction Model

In vivo

UN GHS Cat 1/2 UN GHS Cat NC

UN GHS Cat 1/2

UN GHS Cat NC

18 5

13 15

4. Discussion After direct ocular contact, agrochemical active substances and their formulations may cause local injury to the frontal tissues and substructures of the eye (corneal opacity, iritis, conjunctival redness, and/or conjunctival chemosis). Therefore, testing for eye irritation potential is a critical component of the safety evaluation process for the agrochemical industry. Over the past decade, a number of in vitro and ex vivo methods have been developed, providing the opportunity to move away from in vivo testing for determination of this endpoint in near future. To accomplish this there is a need for research focused on evaluating the performance of these testing methods for various classes and types of chemicals and in particular for materials such as agrochemical formulations that are often complex chemical mixtures. Evaluation of formulations with in vitro approaches represents an important and significant opportunity to reduce animal testing as each active ingredient is used in multiple formulated products that also require an evaluation of eye irritation potential. The NRR assay has been in use for several years and has been found to be particularly useful for assessment of chemicals such as

surfactants or surfactant-based mixtures that are capable of causing immediate adverse reactions on contact to the eye (Zuang, 2001; Hubbard et al., 1994). The assay has the advantages of being relatively inexpensive, rapid throughput, and also provides concentrations-response information. Recently, the EpiOcular Eye Irritation Test has been extensively characterized and currently has an OECD testing guideline (OECD 492; 2015) for distinguishing mono-constituent substances and mixtures (including agrochemical formulations) that do not require classification for eye irritation (UN GHS Cat NC) from irritating substances. Unlike the NRR single cell layer culture, the EpiOcular tissue consists of multi-layered, highly differentiated stratified squamous epithelia, which closely resembles the human corneal epithelium. This study was undertaken with the aim of identifying an in vitro eye irritation testing strategy covering the full range of ocular irritation potential for formulated agrochemical products. To accomplish this, the applicability domain and performance of the NRR and EpiOcular assays were assessed by comparing results to existing in vivo Draize eye irritation data-sets. The prediction performance of the assays was evaluated when applied as stand-alone and also when used in accordance with a tiered-approach. The combined approach was considered because of recent recognition that no single in vitro test procedure is likely capable of predicting the full range of eye irritation responses (categories) observable in the in vivo rabbit Draize eye test (Scott et al., 2010). In the present study, the NRR assay was conducted as a first step for 64 agrochemical formulations based on its recognized high sensitivity for identifying severe irritants. The NRR results were initially interpreted following the ECVAM-suggested classification criteria and compared with existing in vivo Draize ocular test data. While the NRR assay correctly predicted all tested UN GHS Cat 1 formulations (n ¼ 13), there were a high number of false positives in the UN GHS Cat 2A and UN GHS Cat 2B/NC categories, i.e. the NRR assay demonstrated high sensitivity but poor specificity. This finding is consistent with prior published results and with the observations of an ECVAM Peer Review Panel (PRP) (Zuang, 2001). Subsequently, to better balance the sensitivity and specificity of the assay for agrochemical formulations, the thresholds of the ECVAM retrospective validation model were modified (Table 3), which resulted in an overall accuracy, sensitivity, and specificity of 78, 85, and 76% respectively, for identifying UN GHS Cat 1 vs. UN GHS Cat 2/NC formulations, compared to rabbit Draize eye irritation data (Table 10). In this approach, the NRR assay correctly predicted 11 of 13 severe irritants, while under-predicted DAS-005 and DAS014 (Table 4). The analysis of the in vivo Draize study results for the 13 UN GHS Cat 1 formulations indicated that the selected formulations had in 12 cases persistency as the only driver for classification while in 1 case, severity of injury in one animal led to the classification. In case of DAS-005, all three treated rabbits exhibited ocular tissue irritation, although the severity of injury did not meet UN GHS Cat 1 criteria at 24-48-72 h (i.e. the average scores for corneal opacity was <3 and for Iris <1.5 in all animals). Therefore, just based on severity of ocular irritation UN GHS Cat 2 classifications would be triggered for this formulation. However, the treated animals exhibited variability in persistence of tissue damage (recovered in 7, 14, and 28 day post-instillation in the three animals respectively). Thus, UN GHS Cat 1 classification was triggered based on persistence of irritation beyond 21 days in a single animal (conjunctival redness, score of 1, which recovered at day 28; demonstrating a non-permanent damage). However, in a recent publication from the Cosmetic EU, a UN GHS Cat 1 classification based only on persistence of lower conjunctival effects (score 1) in the absence of any other Cat 1-triggering effects were concluded to be highly questionable (Barroso et al., 2016).

R.S. Settivari et al. / Regulatory Toxicology and Pharmacology 81 (2016) 407e420

417

Fig. 4. Framework for the assessment of agrochemical formulations for eye irritation potential using NRR and EpiOcular™ (ET-40 method) in vitro methods in a tiered-approach.

The Draize test for the second under-predicted formulation (DAS-014) in NRR was performed in a single rabbit. The ocular damage observed in this animal would potentially trigger a UN GHS Cat 2 classification, however as the animal did not exhibit recovery by 21 days post-instillation, this formulation was assigned UN GHS Cat 1. Interestingly, recent publications from Cosmetic EU (Adriaens et al., 2014; Barroso et al., 2016) have advocated for a critical revision of the UN GHS/EU CLP decision criteria for the Cat 1 classification of chemicals. One of the major suggestions was that all classifiable ocular effects should be present in 60% of the animals to drive a classification from a regulatory point of view. Others have raised concerns regarding inter- and intra-laboratory variations in Draize assay responses and called for modifying the interpretation and UN GHS classification criteria. Some of the factors resulting in Draize assay variability are: indeterminate exposure of test material volume over time, inconsistent restraint procedures to prevent pawing of the irritated eye, subjective scoring, biological variability etc. (Scott et al., 2010; Adriaens et al., 2014; Barroso et al., 2016; Wahlberg, 1983). Consideration of these factors in appropriate interpretation and classification of the rabbit Draize results may further improve performance of the NRR assay. The NRR assay exhibited greater discordance for identifying UN GHS Cat 2 formulations. In the in vivo rabbit Draize eye test, reversibility of ocular tissue damage (within 21 days posttreatment) is one of the primary drivers for classifying UN GHS Cat 2 compounds. However, the NRR and majority of the existing alternative test methods are shorter in experimental duration (minutes to hours), and lack anatomical complexity of in vivo eye. As a result, currently there are no validated in vitro test methods available to accurately detect UN GHS Cat 2 ocular irritating

substances. Important to note that even the in vivo rabbit Draize eye irritation test has been reported to show higher biological variability for UN GHS Cat 2 compounds (Prinsen, 2006; Lotz et al., 2016). The NRR assay, demonstrated encouraging performance to identify UN GHS Cat NC formulations. Amongst the 31 UN GHS NC formulations, the NRR assay correctly identified 19, however, overpredicted 5 as UN GHS Cat 1 and 7 as UN GHS Cat 2. The overprediction noted in NRR assay for UN GHS Cat NC could again be partially due to its lack of sufficient anatomical complexity. In the EpiOcular assay, 51 agrochemical formulations were tested and categorized as potential ocular irritants (UN GHS Cat 1/ Cat 2) or non-classified (UN GHS Cat NC) using the criteria described in the methods. The EpiOcular assay exhibited overall accuracy, sensitivity, and specificity of 65, 58, and 75%, respectively, compared to in vivo rabbit Draize results (Table 10). The performance of the EpiOcular assay in the present study was slightly lower compared to prior published EpiOcular-EIT results for agrochemical formulations (Kolle et al., 2015), which could partially be due to the differences in chemical set size, distribution of compounds with irritation potential (disproportionate dataset), as well as the diversity and complexity of tested formulations. The tiered approach followed in this manuscript is conceptually consistent with EPA's recently proposed (2015) top-down integrated testing strategy for predicting eye irritation potential of antimicrobial products intended for cleaning purposes. In EPA's approach, BCOP is used as the first step for predicting materials having oxidizing chemistry or those expected to be severe or moderate ocular irritants. The BCOP assay is used to identify EPA Cat I, II, and III/IV (as one group) for ocular irritation. The EpiOcular

418

R.S. Settivari et al. / Regulatory Toxicology and Pharmacology 81 (2016) 407e420

Table 8 Assessment of agrochemical formulations for eye irritation potential using NRR and EpiOcular™ in vitro methods in a tiered-approach. Test sample

Actual (in vivo) GHS

Physical state

Agrochemical formulation typea

Results NRR50 (mg/ml)

Results EpiOcular™ ET40 (min)

Tier classificationb

DAS-001 DAS-002 DAS-003 DAS-004 DAS-005 DAS-006 DAS-007 DAS-008 DAS-009 DAS-010 DAS-011 DAS-012 DAS-014

1 1 1 1 1 1 1 1 1 1 1 1 1

Solid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid

WP ME SL SL ME EC EC EO EC SL SL ME ME

6.8 7.7 37.0 38.0 97.0 34.1 45.3 37.2 16.0 13.0 19.0 31.0 72.0

8.1 25.6 <3 <3 3.9 41.5 14.8 41.6 7.2 <3 4.6 NT 51.7

Cat 1 Cat 1 Cat 1 Cat 1 Cat 1/Cat 2 Cat 1 Cat 1 Cat 1 Cat 1 Cat 1 Cat 1 Cat 1 NC

DAS-013 DAS-015 DAS-016 DAS-017 DAS-018 DAS-019 DAS-020 DAS-021 DAS-022 DAS-023 DAS-032

2A 2A 2A 2A 2A 2A 2A 2A 2A 2A 2A

Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Solid

EC OD ME SL SL OD OD EC SE EC WG

7.4 88.0 26.0 30.4 65.7 41.0 80.0 369.0 >1000 340.0 11.7

<3 15.7 NT 5.6 6.1 >30 45.5 >60 >60 36.6 >30

Cat 1 Cat 1/Cat 2 Cat 1 Cat 1 Cat 1/Cat 2 Cat 1 NC NC NC NC Cat 1

DAS-024 DAS-025 DAS-026 DAS-027 DAS-028 DAS-029 DAS-030 DAS-031 DAS-042 DAS-033 DAS-034 DAS-035 DAS-036 DAS-037 DAS-038 DAS-039 DAS-040 DAS-041 DAS-045 DAS-046 DAS-047 DAS-048 DAS-049 DAS-050 DAS-051 DAS-052 DAS-053 DAS-054 DAS-055 DAS-056 DAS-057 DAS-058 DAS-059 DAS-060 DAS-061 DAS-062 DAS-063 DAS-064 DAS-065 DAS-066

2B 2B 2B 2B 2B 2B 2B 2B 2B NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC NC

Liquid Liquid Liquid Solid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Solid Solid Liquid Liquid Liquid Liquid Solid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Liquid Solid Solid Liquid

EC EW SL WG EC SE EW EC OD ME OD WG WG EC OD EW EW WG SE SE CS SC SE SC CS CS SE EW SC SC SL SL EW EW CS CS CS WG WG EW

10.0 56.2 22.0 67.0 57.0 >1000 214.0 521.0 94.0 25.1 16.1 15.0 79.0 51.3 38.0 54.0 90.0 57.0 146.0 321.0 231.0 350.2 476.0 >500 >1000 >1000 >1000 34.0 436.0 846.3 645.7 >1000 >1000 >1000 >1000 >1000 >1000 >1000 257.0 431.0

7.4 >60 10.2 19.9 22.1 >60 >60 39.2 23.2 >30 21.1 19.1 >30 31.8 >30 28.2 >60 >30 >30 >30 18.5 >60 >30 NT NT NT NT 11.0 >60 >60 NT NT NT NT NT NT NT >60 >60 46.8

Cat 1 NC Cat 1 Cat 1/Cat Cat 1/Cat NC NC NC Cat 1/Cat Cat 1 Cat 1 Cat 1 NC NC Cat 1 Cat 1/Cat NC NC NC NC Cat 1/Cat NC NC NC NC NC NC Cat 1 NC NC NC NC NC NC NC NC NC NC NC NC

2 2

2

2

2

NT ¼ Non Tested. NC ¼ Non Classified. a SL ¼ Soluble concentrate; EC ¼ Emulsifiable Concentrate; SC ¼ Suspension concentrate; WP ¼ Wettable powder; ME ¼ Micro-emulsion; WG ¼ Water Dispersible Granules; OD ¼ Oil Dispersion; EW ¼ Emulsion oil in water; CS ¼ Capsule suspension; SE ¼ Suspo-emulsion; EO ¼ Emulsion, water in oil. b For chemicals that had NRR50 between 50 and 250 mg/mL, the EpiOcular results were considered for final interpretation (bold font).

R.S. Settivari et al. / Regulatory Toxicology and Pharmacology 81 (2016) 407e420

419

Table 9 Performance of the tiered-approach for predicting eye irritation potential of agrochemical formulations. Tiered-approach

In vivo

UN GHS Cat 1 UN GHS Cat 2A/2B UN GHS Cat NC

UN GHS Cat 1

UN GHS Cat 1 or 2

UN GHS Cat NC

11 6 5

1 5 2

1 8 24

Table 10 Comparative performance of single NRR and EpiOcular™ assays versus the tiered-approach. Assay

Accuracy %

Sensitivity %

A) Non classified (negatives) versus irritants plus severe irritants (positives) NRR 73 85 EpiOcular™ (ET-40 method) 65 58 Tiered-approach 75 73 B) Severe Irritants (positives) versus irritants plus non classified (negatives) NRR 78 85 EpiOcular™ N/A N/A Tiered-approach 78 85

Specificity %

Sample size n

TP/FN n

TN/FP n

61 75 77

64 51 64

28/5 18/13 24/9

19/12 15/5 24/7

76 N/A 76

64 N/A 64

11/2 N/A 11/2

39/12 N/A 39/12

N/A: Not Applicable. Non classified: (UN GHS NC) TP/FN: True Positives/False Negatives. Irritants: (UN GHS Cat 2) TN/FP: True Negatives/False Positives. Severe Irritants: (UN GHS Cat 1).

or Cytosensor microphysiometer (CM) assays are proposed as second tier test for distinguishing EPA Cat III and IV for ocular irritation. On the contrary, for materials that are not supposed to be oxidizing or not expected to be severe or moderately irritating to the eye, the EpiOcular or CM assays are proposed as the Tier-1 study, for predicting EPA Cat I & II (as one group), Cat III, and Cat IV. In this approach, BCOP assay is used as the second tier test to distinguish between EPA Cat I and II. In the present study, the NRR assay was implemented instead of BCOP as it provides the advantage of better throughput, concentrations-response analysis, utilization of a human cell line, and avoids the necessity of access to a slaughter house for an ocular tissue. In addition, it was recently reported that BCOP has poor sensitivity for prediction of eye irritation potential of agrochemical formulations (Kolle et al., 2015). In the EPA-proposed criteria for the EpiOcular assay, 50% reduction in cell viability is considered as the threshold for classification, while in the present study, 40% cytotoxicity was considered as the appropriate threshold for agrochemical formulations (consistent with OECD 492). Overall, the tiered-testing approach used in this manuscript provided encouraging results; however, limitations were also noted. For instance, UN GHS Cat 2 formulations were mostly misclassified in the tiered-testing approach owing to the simplistic tissue nature of the NRR and EpiOcular assays and shorter exposure conditions. It is possible that inclusion of additional alternative assays that allow the measurement of reversibility of cornea damage (up to 21 days as in the in vivo Draize test) may further improve the performance of the tiered-testing system for identifying UN GHS Cat 2 formulations (Piehl et al., 2011). Furthermore, recent mechanistic understanding of the potential modes of action and the importance of the stroma for identifying reversible effects may potentially drive the development of alternative tests to UN GHS Cat 2 formulations in near future (Scott et al., 2010; Bhhatarai et al., 2016). In light of the recent recognition of the need to revise in vivo rabbit Draize test interpretation criteria, it is likely that the emerging in vitro assays, including the ones utilized in this manuscript, may provide more promising results for identifying potential ocular irritants.

5. Conclusions In this manuscript, a retrospectively developed tiered-testing approach relying on the NRR and EpiOcular assays was implemented to assess the spectrum of eye irritation potential for 11 types of agrochemical formulations. Results suggested that the NRR assay as a standalone served as an encouraging tool to evaluate UN GHS Cat 1 and NC formulations, while the EpiOcular assay demonstrated encouraging performance for identifying UN GHS Cat NC formulations. Implementation of these assays in a tiered manner further improved performance for identifying UN GHS Cat 1 and NC formulations; however, the model exhibited limitations to identify UN GHS Cat 2 formulations. Future efforts will focus on expanding the chemical set, including methods to assess tissue repair and refining the tiered-testing approach to improve performance for UN GHS Cat 2 formulations and extending applicability to additional types of chemistries and complex mixtures.

Acknowledgments The authors are grateful to Drs. Pamela Spencer, Daniel Wilson, Matthew LeBaron, Reza Rasoulpour, and Michael Woolhiser of The Dow Chemical Company for critical review of the manuscript. The authors are thankful to Lindsay Sosinski, Mathew Koehler, Coreena Cheney, and Elizabeth Srebinski of The Dow Chemical Company for their contributions.

Transparency document Transparency document related to this article can be found online at http://dx.doi.org/10.1016/j.yrtph.2016.09.028

Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.yrtph.2016.09.028.

420

R.S. Settivari et al. / Regulatory Toxicology and Pharmacology 81 (2016) 407e420

References Adriaens, E., Barroso, J., Eskes, C., Hoffmann, S., McNamee, P., Alepee, N., et al., 2014. Retrospective analysis of the Draize test for serious eye damage/eye irritation: importance of understanding the in vivo endpoints under UN GHS/EU CLP for the development and evaluation of in vitro test methods. Arch. Toxicol. 88 (3), 701e723. Balls, M., Reader, S., Atkinson, K., Tarrant, J., Clothier, R., 1991. Non-animal alternative toxicity tests for detergents: genuine replacements or mere prescreens? J. Chem. Technol. Biotechnol. 50 (3), 423e433. Barroso, J., Pfannenbecker, U., Adriaens, E., Alepee, N., Cluzel, M., De Smedt, A., et al., 2016 Mar 21. Cosmetics Europe compilation of historical serious eye damage/ eye irritation in vivo data analysed by drivers of classification to support the selection of chemicals for development and evaluation of alternative methods/ strategies: the Draize eye test Reference Database (DRD). Arch. Toxicol. http:// dx.doi.org/10.1007/s00204-016-1679-x [Epub ahead of print]. Benavides, T., Mitjans, M., Martinez, V., Clapes, P., Infante, M.R., Clothier, R.H., et al., 2004. Assessment of primary eye and skin irritants by in vitro cytotoxicity and phototoxicity models: an in vitro approach of new arginine-based surfactantinduced irritation. Toxicology 197 (3), 229e237. Bhhatarai, B., Wilson, D.M., Parks, A.K., Carney, E.W., Spencer, P.J., 2016. Evaluation of TOPKAT, Toxtree, and derek nexus in silico models for ocular irritation and development of a knowledge-based framework to improve the prediction of severe irritation. Chem. Res. Toxicol. 29 (5), 810e822. Clothier RH. Invittox Protocol No. 541992. EPA, 2007. 40 CFR Parts 9 and 158. Pesticides; Data Requirements for Conventional Chemicals. Environmental Protection Agency. No 207/Friday, October 26,. Federal Register, 72. EPA, 2014. Label Review Manual, Chapter 7; Precautionary Statements. Environmental Protection Agency. Fox, D.A., Boyes, W.K., 2008. Toxic responses of the ocular and visual system. In: CD, Klasen (Ed.), Casarett and Doull's Toxicology: the Science of Poisons, 7th ed. McGraw-Hill, New York, pp. 665e697. Gettings, S.D., Dipasquale, L.C., Bagley, D.M., Casterton, P.L., Chudkowski, M., Curren, R.D., et al., 1994. The CTFA Evaluation of Alternatives Program: an evaluation of in vitro alternatives to the Draize primary eye irritation test. (Phase II) oil/water emulsions. Food Chem. Toxicol. Int. J. Publ. Br. Ind. Biol. Res. Assoc. 32 (10), 943e976. Gettings, S.D., Lordo, R.A., Demetrulias, J., Feder, P.I., Hintze, K.L., 1996. Comparison of low-volume, Draize and in vitro eye irritation test data. I. Hydroalcoholic formulations. Food Chem. Toxicol. Int. J. Publ. Br. Ind. Biol. Res. Assoc. 34 (8), 737e749. Hackett, R.B., McDonald, T.O., 1991. Eye irritation. In: Marzulli, F., Maibach, H. (Eds.), Advances in Modern Toxicology: Dermatoxicology, 4th ed. Hemisphere Publishing Corporation, Washington, DC, pp. F749eF815. Harbell, J.W., Koontz, S.W., Lewis, R.W., Lovell, D., Acosta, D., 1997. IRAG working group 4. Cell cytotoxicity assays. Interagency Regulatory Alternatives Group. Food Chem. Toxicol. Int. J. Publ. Br. Ind. Biol. Res. Assoc. 35 (1), 79e126. Hayashi, K., Mori, T., Abo, T., Ooshima, K., Hayashi, T., Komano, T., et al., 2012. Twostage bottom-up tiered approach combining several alternatives for identification of eye irritation potential of chemicals including insoluble or volatile substances. Toxicol. In Vitro Int. J. Publ. Assoc. BIBRA 26 (7), 1199e1208. Hubbard, A.W., Moore, L.J., Clothier, R.H., Sulley, H., Rollin, K.A., 1994. Use of in vitro methodology to predict the irritancy potential of surfactants. Toxicol. In Vitro 8, 689e691.

Kolle, S.N., Rey Moreno, M.C., Mayer, W., van Cott, A., van Ravenzwaay, B., Landsiedel, R., 2015. The epiocular eye irritation test is the method of choice for the in vitro eye irritation testing of agrochemical formulations: correlation analysis of epiocular eye irritation test and BCOP test data according to the UN GHS, US EPA and Brazil ANVISA classification schemes. Altern. Lab. Anim. 43 (3), 181e198. Lotz, C., Schmid, F.F., Rossi, A., Kurdyn, S., Kampik, D., De Wever, B., et al., 2016. Alternative methods for the replacement of eye irritation testing. ALTEX 33 (1), 55e67. Maier, K., Schmitt-Landgraf, R., Siegemund, B., 1991. Development of an in vitro test system with human skin cells for evaluation of phototoxicity. Toxicol. In Vitro Int. J. Publ. Assoc. BIBRA 5 (5e6), 457e461. OECD, 2015. Test No. 492: Reconstructed human Cornea-like Epithelium (RhCE) test method for identifying chemicals not requiring classification and labelling for eye irritation or serious eye damage. OECD Publishing, Paris. http://dx.doi.org/ 10.1787/9789264242548-en. Piehl, M., Carathers, M., Soda, R., Cerven, D., DeGeorge, G., 2011. Porcine corneal ocular reversibility assay (PorCORA) predicts ocular damage and recovery for global regulatory agency hazard categories. Toxicol. In Vitro Int. J. Publ. Assoc. BIBRA 25 (8), 1912e1918. Prinsen, M.K., 2006. The Draize Eye Test and in vitro alternatives; a left-handed marriage? Toxicol. In Vitro Int. J. Publ. Assoc. BIBRA 20 (1), 78e81. Reader, S.J., Blackwell, V., O'Hara, R., 1989. A vital dye release method for assessing the short-term cytotoxic effects of chemicals and formulations. ATLA 17, 28e33. Reader, S.J., Blackwell, V., O'Hara, R., Clothier, R.H., Griffin, G., Balls, M., 1990. Neutral red release from pre-loaded cells as an in vitro approach to testing for eye irritancy potential. Toxicol. In Vitro Int. J. Publ. Assoc. BIBRA 4 (4e5), 264e266. Sanchez, L., Mitjans, M., Infante, M.R., Vinardell, M.P., 2004. Assessment of the potential skin irritation of lysine-derivative anionic surfactants using mouse fibroblasts and human keratinocytes as an alternative to animal testing. Pharm. Res. 21 (9), 1637e1641. Sanco, 2010. Annex to SANCO 11802/2010/rev.7; Part B 02.02. European Commission. Scott, L., Eskes, C., Hoffmann, S., Adriaens, E., Alepee, N., Bufo, M., et al., 2010. A proposed eye irritation testing strategy to reduce and replace in vivo studies using Bottom-Up and Top-Down approaches. Toxicol. In Vitro Int. J. Publ. Assoc. BIBRA 24 (1), 1e9. Wahlberg, J.E., 1983. Assessment of skin irritancy: measurement of skin fold thickness. Contact Dermat. 9 (1), 21e26. Williams, S.J., Graepel, G.J., Kennedy, G.L., 1982. Evaluation of ocular irritancy potential: interlaboratory variability and effect of dosage volume. Toxicol. Lett. 12 (4), 235e241. Worth, A.P., Cronin, M.T., 2001. The use of pH measurements to predict the potential of chemicals to cause acute dermal and ocular toxicity. Toxicology 169 (2), 119e131. Young, J.R., How, M.J., Walker, A.P., Worth, W.M., 1988. Classification as corrosive or irritant to skin of preparations containing acidic or alkaline substances, without testing on animals. Toxicol. In Vitro Int. J. Publ. Assoc. BIBRA 2 (1), 19e26. Zanatta, C.F., Ugartondo, V., Mitjans, M., Rocha-Filho, P.A., Vinardell, M.P., 2008. Low cytotoxicity of creams and lotions formulated with Buriti oil (Mauritia flexuosa) assessed by the neutral red release test. Food Chem. Toxicol. Int. J. Publ. Br. Ind. Biol. Res. Assoc. 46 (8), 2776e2781. Zuang, V., 2001. The neutral red release assay: a review. Altern. Lab. Anim. 29 (5), 575e599.