A Pre-validation Study on In Vitro Skin Corrosivity Testing

The Report and Recommendations of ECVAM Workshop 6^1,2

Reprinted with minor amendments from ATLA 23, 219-255

Philip A. Botham³, Mark Chamberlain⁴, Martin D. Barratt⁴, Rodger D. Curren⁵, David J. Esdaile⁶, John R. Gardner⁷, Virginia C. Gordon⁸, Bernhard Hildebrand⁹, Richard W. Lewis³, Manfred Liebsch¹⁰, Pamela Logemann¹¹, Rosemarie Osborne¹², Maria Ponec¹³, Jean-François Régnier¹⁴, Winfried Steiling¹⁵, Arthur P. Walker¹⁶, and Michael Balls¹⁷
³ZENECA Central Toxicology Laboratory, Alderley Park, Macclesfield, Cheshire SK10 4TJ, UK; ⁴Environmental Safety Laboratory, Unilever Research, Colworth House, Sharnbrook, Bedford MK44 1LQ, UK; ⁵ Microbiological Associates Inc., 9900 Blackwell Road, Rockville, MD 20878, USA; ⁶Rhône-Poulenc Secteur Agro, 355 Rue Dostoievski, BP 153, 06903 Sophia Antipolis Cedex, France; ⁷Hazleton Europe, Otley Road, Harrogate, N. Yorkshire HG3 1PY, UK; ⁸In Vitro International, 16632 Milliken Avenue, Irvine, CA 92714, USA; ⁹BASF, ZHT-Z470, D-67056 Lüdwigshafen, Germany; ¹⁰ZEBET, Bundesinstitut für gesundheitlichen Verbraucherschutz und Veterinärmedizin (BgVV), Diedersdorfer Weg 1, D-12277 Berlin, Germany; ¹¹Advanced Tissue Sciences, 10933 North Torrey Pines Road, La Jolla, CA 92037-1005, USA; ¹²The Procter & Gamble Company, Miami Valley Laboratories, PO Box 538707, Cincinnati, OH 45253-8707, USA; ¹³Department of Dermatology, University Hospital Leiden, PO Box 9600, 2300 RC Leiden, The Netherlands; ¹⁴Elf Atochem, Department of Toxicology, 92091 Paris-la-Défense, France; ¹⁵Henkel KGaA, Henkelstrasse 67, D-40191 Dusseldorf, Germany; ¹⁶Apojay Consultancy, 6 Cragside, Whitley Bay, Tyne & Wear NE26 3DU, UK; ¹⁷ECVAM, JRC Environment Institute, 21020 Ispra (Va), Italy

¹ECVAM - The European Centre for the Validation of Alternative Methods. ²This document represents the agreed report of the participants as individual scientists.

Address for correspondence: Dr Mark Chamberlain, Environmental Safety Laboratory, Unilever Research, Colworth House, Sharnbrook, Bedford MK44 1LQ, UK

Address for reprints: ECVAM, TP 580, JRC Environment Institute, 21020 Ispra (VA), Italy

Preface

This is the report of the sixth of a series of workshops organised by the European Centre for the Validation of Alternative Methods (ECVAM). ECVAM's main goal, as defined in 1993 by its Scientific Advisory Committee, is to promote the scientific and regulatory acceptance of alternative methods which are of importance to the biosciences and which reduce, refine or replace the use of laboratory animals. One of the first priorities set by ECVAM was the implementation of procedures which would enable it to become well-informed about the state-of-the-art of non-animal test development and validation, and the potential for the possible incorporation of alternative tests into regulatory procedures. It was decided that this would be best achieved by the organisation of ECVAM workshops on specific topics, at which small groups of invited experts would review the current status of various types of in vitro tests and their potential uses, and make recommendations about the best ways forward (1).

The workshop on In Vitro Skin Corrosivity Testing was held in Angera, Italy on 12-14 January 1994, under the chairmanship of Philip Botham. The workshop was concerned primarily with the discussion of a pre-validation study on alternative methods for skin corrosivity testing. The results and conclusions of the pre-validation study are given in this report, in addition to some specific recommendations which relate to future requirements in skin corrosivity testing. A follow-up meeting was held at ECVAM in February 1995 to finalise the workshop report, to review further test optimisation conducted since January 1994, and to discuss the design of a validation study on alternative methods for corrosivity testing which is to be sponsored by ECVAM.

Introduction

The potential for chemical-induced skin corrosion is an important consideration in establishing procedures for the safe handling, packing, and transport of chemicals. The standard approach for assessing skin corrosion and irritation has been to apply the test chemical to the shaved skin of albino rabbits (2). The production of irreversible full-thickness necrosis of the skin (which is the endpoint of corrosion) is determined by visual inspection of the skin for up to 21 days following exposures to the test material of up to four hours duration. This test is included in international regulatory requirements for the testing of new chemicals, for example, the Code of Federal Regulations (3) and OECD testing guidelines (4).

Regulations originating with the United Nations (UN; 5) require the labelling of packaged chemicals for international transport purposes. The UN guidelines recommend that corrosives be classifed into potency categories, termed "packing groups". Packing groups I, II, and III are assigned on the basis of the capacity of a chemical, when tested on the intact skin of albino rabbits, to produce skin corrosion within 3 minutes, 1 hour or 4 hours, respectively. In October 1993, these UN guidelines were accepted by the U.S. Department of Transportation (DOT). European regulations (6) require classification according to risk phrases, determined according to whether a chemical causes corrosion within 3 minutes (R35; analogous to packing group I) or 4 hours (R34; analogous to packing groups II and III).

Testing for skin corrosion/irritation in laboratory animals has the potential to cause them discomfort or pain. For this reason, alternative methods for trying to identify corrosive substances have been developed. The severity of the skin lesion (i.e. tissue destruction) lends itself to using non-animal methods that can detect severe tissue damage. Thus, the feasibility of using non-animal tests for assessing skin corrosion is probably greater than for trying to detect toxic effects which are exerted by subtle multifactorial mechanisms.

As in several other areas of toxicity testing, the evolution of alternative tests for detecting chemicals which are corrosive to the skin has resulted in the production of model test systems which have been designed for a particular market and/or to fulfil a specific need. The in vitro test systems investigated in this pre-validation study cover a range of biological complexity: excised rat skin (the transcutaneous electrical resistance [TER] assay), a three-dimensional human skin model (Skin²), and a biochemical matrix (CORROSITEX^TM).

The most mature alternative method for corrosivity testing, the transcutaneous electrical resistance (TER) assay (7), has been used successfully as a routine in-house prescreen for several years. When used in screening mode, the TER test is employed to predict corrosivity potential rather than the degree of corrosive effect (i.e. potency), and it is used primarily to guide humane in vivo skin testing. The TER method has undergone intralaboratory and interlaboratory validation (7,8), although these studies would not meet current standards for the conduct of validation studies, such as those defined by Balls et al (9).

Recently, two commercially available test systems, CORROSITEX^TM (In Vitro International, Irvine, California; 10) and a protocol using Skin² (Advanced Tissue Sciences, La Jolla, California; 11), have been developed to predict the skin corrosivity potentials of chemicals and to rank them with respect to their degree of corrosive effect (typically to assign chemicals to UN packing groups). CORROSITEX^TM is an assay system based on the penetration of a test material through a non-living barrier constructed to have physicochemical properties similar to rat skin. It has been granted regulatory approval by the US Department of Transportation (DOT), in the form of an exemption (limited to defined chemical classes) for the detection of corrosives, such that a positive result in this non-animal test enables the chemical to be classified as a corrosive without this being confirmed in the standard animal procedure. However, a negative response does not obviate the need for an animal test for subsequent classification. The Skin² Model ZK1350 in vitro skin corrosion test, which involves the topical application of test materials to the stratum corneum of three-dimensional human skin cultures, has now also been granted similar DOT approval, despite the fact that neither CORROSITEX^TM nor the Skin² protocol have undergone formal interlaboratory validation.

The TER, CORROSITEX^TM, and Skin² tests are, therefore, at different stages of optimisation, evaluation and validation. The need for a pre-validation assessment of the tests, prior to the possible undertaking of a formal validation study, was recognised early in 1993 by scientists from the Central Toxicology Laboratory (CTL) at ZENECA and from the Environmental Safety Laboratory (ESL) at Unilever. Subsequently, representatives from laboratories which were familiar with the in vitro tests were canvassed regarding the possibility of them actively participating in such a pre-validation study.

The aims of the pre-validation study were:

to evaluate the relative performances of the TER, CORROSITEX^TM, and Skin² methods in correctly predicting defined corrosive and non-corrosive test chemicals;
to undertake an initial assessment of the interlaboratory variabilities in the methods, by conducting each test in at least two laboratories; and
to assess the relative states of optimisation, evaluation, and validation of the tests.

The test chemicals (Table I) were selected on the basis of their availability and the confidence with which they could be classified unambiguously as "corrosive" or "non-corrosive". In vivo data were obtained from in-house studies or from the scientific literature (12), or by reference to the 1991 Comité Européen des agents de Surface et leurs Intermédiaires Organiques (CESIO) classifications or to manufacturers' data. All animal tests were reported to have been conducted in accordance with OECD Test Guideline 404. Corrosivity classifications were from three sources: animal data, hazard data sheets, or Annex 1 of the Dangerous Substances Directive (13). The selection of the test chemicals was based primarily on the availability of the test material and associated data. Thus, the criteria for selection were not as stringent as those employed by other groups (for example, the European Centre for the Ecotoxicology and Toxicology of Chemicals [ECETOC]) when choosing the chemicals to be tested in formal validation studies. However, the test chemicals chosen were considered to be adequate for the primary objective of the study, i.e. for an assessment of the relative performances of the three tests.

Table I: Test Chemicals

Tradename	Chemicalname (if different)	Chemicalclass	Appearance
*CORROSIVES*
Acetic acid (glacial)^a		organic acid	clear liquid
Acrylic acid (99%)^a		organic acid	clear liquid
Armeen CD^b	Cocoamine	organic base	clear liquid
Armeen TD^b	Tallowamine	organic base	opaque gel
Arquad 16-50^b	Hexadecyltrimethyl-ammoniumchloride, 50% in isopropanol	cationic surfactant	clear liquid
Arquad DMMCB-50^c	Coco(C12)dimethylbenzyl-ammoniumchloride, 50% in aq. ethylene glycol	cationic surfactant	clear viscous liquid
Bromoacetic acid(8%)^a		organic acid	clear liquid
Bromoacetic acid(55.6%)^a		organic acid	clear liquid
Butylamine (40%)^a		organic base	clear liquid
Capric/caprylic(45:55) acid^b		organic acid	clear liquid
Caprylic acid^b		organic acid	clear liquid
Cyclohexylamine(11.9%)^a		organic base	clear liquid
1,4-Diaminobutane(30%)^a		organic base	clear liquid
Dichloroaceticacid (36.1%)^a		organic acid	clear liquid
Diethylamine (35%)^a		organic base	yellow liquid
Duoquad T-50^b	Pentamethyl-N-tallow-1, 3-propanediammonium chloride, 50% in isopropanol	cationic surfactant	yellow liquid
Formic acid (33.9%)^a		organic acid	clear liquid
Hexanoic acid^a		organic acid	clear yellow liquid
Mercaptoaceticacid (15.1%)^a		organic acid	clear liquid
Proxel BD^b (biocide A)	1,2-Benzisothiazolin-3-one(33%) in aq. propylene glycol	neutral organic	tan opaque liquid
Pyrrolidine (34.5%)^a		organic base	yellow liquid
Sodium hydroxide(4.88%)^a		inorganic	clear liquid
Sodium metasilicate^b		inorganic	granular powder^c
Sodium silicateA140^b		inorganic	clear gel
Synprolam 35X2^b	C13-15Alkyl-di(2-hydroxyethyl)amine	organic base	clear viscous liquid
*NON-CORROSIVES*
Armeen 2C^d	Dicocoamine	organic base	crystalline powder^c
Aromox DMMCD-W^b	Coco(C12)dimethylamineoxide (30%)	amine oxide	clear liquid
Arquad C-33-W^d	Coco(C12)trimethyl-ammoniumchloride, 33% in water	cationic surfactant	clear gel
Butylbenzene^a		neutral organic	clear liquid
Dequest 2000^e	Aminotris(methylphosphonicacid), 50% in water	organic acid	clear liquid
Dowanol PNB^f	Propylene glycoln-butyl ether	neutral organic	clear liquid
Elfan OS 46^d	C12-14α-Olefinsulphonate, Na salt	anionic surfactant	yellow viscousliquid
Empicol LZPV/C^d	Sodium dodecylsulphate	anionic surfactant	dry pellets^c
Empigen OB^d	Coco(C12)dimethylamineoxide (30%)	amine oxide	clear liquid
Empilan CME^d	Fatty acid monoethanolamidecoco	neutral organic	dry chips^c
Empilan KB2^d	Fatty alkylethoxylate2EO	neutral organic	white opaque cream
Ethomeen T/25^b	Polyoxyethylene(15)tallowamine	organic base	yellow viscousliquid
Genamin KDM-F^d	Behenyl(C20-22)trimethyl-ammoniumchloride, 80% in isopropanol	cationic surfactant	powdered flakes^c
Genapol LRO^d	Coco(C12)2EO sulphate,Na salt (70%)	anionic surfactant	clear gel
n-Hexanol^a		neutral organic	clear liquid
Hostaphat KLD^d	Alkyl(4EO)phosphoricacid	neutral organic	clear viscous liquid
Lauric acid^b		organic acid	fine powder^c
n-Nonanol^a		neutral organic	clear liquid
Oleic/caprylic(80:20) acid^b		organic acid	yellow liquid
Proxel AB^b(biocide B)	1,2-Benzisothiazolin-3-one(33%), aqueous	neutral organic	opaque tan liquid
Sodium perborate^e		inorganic inorganic	crystalline powder^c
Sodium percarbonate^e		inorganic	granular powder^c
Sodium silicateH100^b		inorganic	clear viscous liquid
Triethanolamine^a		organic base	clear viscous liquid
n-Undecanol^a		neutral organic	clear liquid

^aJacobs & Martens (12) classification from animal data.
^bOriginal animal data.
^cPrepared in distilled water at 1g/ml.
^dCESIO classification from animal data.
^eHarmonised Electronic Dataset (HEDSET) data.
^fManufacturers' data sheet and summary of test data.

In this pre-validation study, each in vitro test was conducted to the same agreed protocol in at least two different laboratories, in accordance with the principles of good laboratory practice (GLP). The laboratories involved encompassed industrial, contract, and government establishments located either in the USA or in Europe (Table II).

Table II: Tests and Laboratories Involved in the Pre-validation Study

TER

CORROSITEX^TM

Skin²

Rhône-Poulenc,France

ZENECA CTL,UK

MicrobiologicalAssociates, USA

In Vitro International,USA

Procter& Gamble, USA

ZEBET, Germany

HuntingdonResearch Centre, UK

TER assay = transcutaneous electrical resistance assay

Materials and Methods

Test Chemicals

Most of the test chemicals (25 corrosives, 25 non-corrosives; Table I) were commercially available, and were obtained from Aldrich (Gillingham, UK), Akzo (Hersham, UK), Monsanto (Basingstoke, UK), Unichema (Gouda, The Netherlands), Hoechst (Hounslow, UK), K & K Greeff (Croydon, UK), Albright and Wilson (Oldbury, UK), ZENECA (Macclesfield, UK) or ICI (Leatherhead, UK). The sources of the corrosive/non-corrosive classifications are indicated in Table I. All test samples were independently coded by Unilever ESL. Each chemical was assigned a single code number and complete sets of test materials were dispatched to the participating laboratories. The identities of the coded chemicals were unknown to the participants. To ensure the integrity of the raw data, test results were lodged with ECVAM prior to the code being broken. Subsequently, the identities of the test chemicals were revealed and the participating laboratories analysed their own data.

Transcutaneous Electrical Resistance Assay

In the in vitro skin corrosivity test developed by Oliver et al (7, 14, 15), substances are applied for up to 24 hours to the epidermal surfaces of skin discs obtained from humanely killed young rats. Corrosive substances produce an irreversible loss of normal stratum corneum integrity and function, which is measured as a reduction in the inherent TER below a predetermined (corrosive) threshold level; irritant substances do not reduce the TER below the threshold level. The original protocol (14) has been refined by the use of magnesium sulphate rather than sodium chloride as the electrolyte solution. This has been found to reduce the incidence of false positive results obtained with solvents and surfactants (16). The protocol used in this pre-validation study has been evaluated previously with 88 industrial substances (16), and with 20 test materials in a blind interlaboratory trial (8). The assay produces very few false negative results, but some false positive results are obtained with test materials containing surfactants and solvents.

Animals
Male Wistar albino rats, aged between 23 and 30 days, were used. These were supplied by Charles River (Manston, Kent, UK) or Iffa-Credo (69210 L'Arbresle, France).

Methodology
The TER assay was performed as described previously (8, 16). Rats (23-25 days old) were shaved to remove hair from the dorsal surface without abrading the skin, and were then washed in an antibiotic bath. Another antibiotic wash was performed three days later. At 28-30 days old, the rats were humanely killed. During this period, rats are in the telogen phase of hair growth. Thus, during the preparation stages there is no hair growth, and the stratum corneum recovers from the effects of shaving and is not damaged by bacterial growth. The quality of the stratum corneum is critical to the success of the assay, and so the stage of hair growth must be controlled by using animals of exactly the correct age.

The dorsal skin was removed from the rat as a single pelt. The excess fat was removed and the pelt was then mounted, epidermal side uppermost, onto polytetrafluoroethylene (PTFE) tubes (International Market Supply, Macclesfield, UK) and secured with a rubber 'O' ring. Excess tissue was trimmed away and the 'O' ring/PTFE tube interface sealed with soft paraffin wax. The tube was supported by a spring clip inside a plastic tube containing electrolyte solution (154 mM MgSO₄ in deionised or distilled water). Three discs were taken from each pelt and the TER was measured (as described below) as a quality control procedure. Only pelts showing a TER of greater than 10 K-ohms/skin disc were used in the assay. The quality control discs were discarded, and new discs from the acceptable pelts (up to nine discs per pelt) were mounted onto PTFE tubes, which were then randomised to avoid bias from individual pelts.

The test materials were applied to the epidermal surfaces of at least three skin discs per chemical, at room temperature for 24 hours. For liquids, 150 µl was applied; for solids, 100 mg (or sufficient to cover the skin disc) was applied, along with 150 µl water to ensure good contact with the skin. At the end of the exposure period, the chemicals were removed with a jet of tap water. The stratum corneum was rinsed with aqueous ethanol (70%), to reduce the surface tension, prior to the addition of electrolyte solution (3 ml). The TER was then measured using a resistance meter in alternating current mode (AIM Databridge, AIM Instruments, Huntingdon, UK).

Data Evaluation and Analysis
The mean TER for the skin discs was calculated for each substance. Test materials giving mean TER values below 5K-ohms/skin disc (the corrosive threshold; 8) are classified as skin corrosives.

CORROSITEX™

In the CORROSITEX™ system, a test material is applied directly to a biobarrier constructed to be similar to rat skin. If it alters the biobarrier sufficiently to be able to pass through into a second compartment, then the chemical is detected by a colour or physical change in a liquid (the "chemical detection system" [CDS]) which is located directly below the biobarrier. The time required for this change to occur (the "breakthrough time") is reported to be inversely proportional to the degree of corrosivity of the test material, i.e. the longer it takes to detect a change in the CDS, the less corrosive is the substance.

Chemicals
The coded chemicals were dispatched only to one of the laboratories conducting CORROSITEX™ (Microbiological Associates [MA], Rockville, Maryland). Following the completion of testing by MA, the remainder of the test materials were forwarded to In Vitro International (IVI; Irvine, California). IVI also received samples of the coded test materials from The Procter & Gamble Company (P&G; Cincinnati, Ohio), which had been excess to their requirements when undertaking the Skin² assay. However, there were two test materials (later identified as sodium silicate A140 and Synprolam 35X2, following breaking of the code) which could not be tested by IVI, because there were insufficient amounts remaining.

Reagents
CORROSITEX™ kits were supplied by IVI. The kits included liquid CDS, lyophilised biobarrier matrix, biobarrier diluent, membrane discs, and glass scintillation vials (used to hold all the components during the assay).

Methodology
Firstly, the compatibilities of the coded chemicals with the test kit were determined, i.e. whether they possessed the chemical or physical properties which enabled them to be detected by the CDS. Test material (50 µl) was mixed with CDS (500µl). If a noticeable colour or physical change occurred in the CDS within a 5 minute observation period, then the sample was considered to qualify for subsequent testing in the CORROSITEX™ system.

The biobarrier was prepared according to the instructions provided with the kits. In outline, CORROSITEX™ diluent (50 ml) was slowly mixed (with stirring for about 20 minutes) with biobarrier matrix (5 g) in a beaker kept in a water bath at 60-68°C. After solubilisation of the matrix, the hot solution was pipetted into discs placed in 24 well plates. Each biobarrier was inspected carefully and discarded if any air bubbles were present. The plates were sealed with plastic film and refrigerated (2-8°C) for at least two hours before use.

Triplicate vials were set up for each test material. CDS (22 ml) was pipetted into the vials, and a biobarrier disc was then placed into each vial. Test materials (liquids: 500 µl; solids: 500 mg) were placed onto the discs. All vials were left uncapped during the test. The time of the first physical or colour change of the CDS was recorded, either to the nearest minute (Laboratory A; MA) or to the nearest second (Laboratory B; IVI). A single pellet of sodium hydroxide, placed on the biobarrier, served as the positive control for all experiments. The assay acceptance criteria (Laboratory A) was that the positive control breakthrough time was within two standard deviations of the historical mean value (11.6 ± 1.1 min; n = 37).

Data Evaluation and Analysis
A chemical was considered to be corrosive if it penetrated the biobarrier and was detected by the CDS in less than four hours. Suggested cut-off times of 3 minutes and 1 hour can be used to assign chemicals to UN packing groups I and II, respectively. However, in this study only corrosive/non-corrosive classifications were determined.

The mean values from triplicate measurements were used to determine whether the test chemicals were corrosives or non-corrosives. Data are presented as means ± SD, to give some indication of intralaboratory and interlaboratory reproducibility. Interlaboratory reproducibility was also assessed by performing a regression analysis on the paired data from both laboratories.

Skin² Model ZK1350 In Vitro Skin Corrosion Test

The Skin² Model ZK1350 test is based on the topical application of test materials to the stratum corneum of three-dimensional human skin cultures (11, 17, 18). Following a 10 second exposure period, the extent of cell damage is determined using the MTT reduction assay, to assess the degree of corrosivity of the test chemical. The Skin² cultures are grown from neonatal human skin cells and contain dermal and epidermal components. Neonatal fibroblasts are seeded onto inert nylon mesh and are grown into a dermal tissue containing fibroblasts and naturally secreted extracellular matrix and growth factors. Keratinocytes are seeded on top of this dermal tissue, and they then differentiate into a functional epidermis. Basal, spinous, and granular layers of keratinocytes are present, as well as a multilayered stratum corneum (11). Biochemical and ultrastructural characterisation of the human skin cultures has demonstrated the presence of differentiation markers and metabolising enzyme activities comparable to those of intact human skin (11, 19, 20).

The Skin² Model ZK1350 test was performed in three independent laboratories, designated A, B, and C. The laboratories were familiar with the basic test procedure, but no specialised training on this test protocol was conducted. Test kits were purchased from Advanced Tissue Sciences, La Jolla, California. Each test kit contained 24 human skin cultures (9 x 9 mm), cell culture medium and various other items required to conduct the test. One kit was used to test five chemicals (each in a single experiment using quadruplicate cultures) using a 10 second exposure period. Four control cultures (treated with distilled water) per kit were evaluated.

Methodology
The test was performed as described in the directions for use included with each kit. On the day before the test was performed, each Skin² cell culture was removed from the surface of the agarose used for shipping purposes and was placed onto a Millicell culture insert above serum-free DMEM-based assay medium (1 ml). The cultures were incubated (5% CO₂, 90% humidity) overnight at 37°C. In those test kits which were to be used over a three-day period (Laboratory B), the Skin² cell cultures were placed onto a Millicell culture insert above maintenance medium (1 ml) and incubated for 24 hours. The medium was then changed to serum-free DMEM-based assay medium, and the cultures were incubated overnight prior to use on the following day.

Liquid and semi-solid test materials were evaluated undiluted. Test materials (15 µl - a volume sufficient to just cover the surface of each cell culture) were dispensed using positive displacement micropipettes onto small (18 mm diameter) glass coverslips. Powdered and granular materials were prepared by grinding them with a mortar and pestle. The ground material (1 g) was mixed with distilled water (1 ml) to obtain a "100% solution". Aliquots (15 µl) were then dispensed onto the glass coverslips. Ten of the 50 test materials were difficult to pipette accurately; with these, approximately 15 µl was spread on the coverslip over an area of about 9mm x 9mm.

The epidermal side of the skin cultures was placed onto the test material on the glass coverslip for exactly 10 seconds. The cultures were then washed with copious amounts of calcium-containing phosphate-buffered saline (PBS) to remove residual test material.

The effects of the test materials on cell viability were determined using the MTT reduction assay (21, 22), employing either MTT Topical Cytotoxicity Assay test kits (Advanced Tissue Sciences, CA) or reagents obtained from Sigma Chemical Co. (St. Louis, Missouri). Skin cultures were incubated on a shaker plate at 37°C for two hours with medium containing MTT (2 mg/ml; 2 ml per 35 mm well). The cultures were then washed with PBS and the insoluble blue formazan product extracted with isopropanol (4 ml/culture) for 60 min at room temperature. Aliquots (200 µl) of the isopropanol extracts were transferred to 96 well plates and the intensity of colour determined using a microplate reader set at a wavelength of 540 nm.

Data Evaluation and Analysis
For each test material, the average viability of the treated skin cultures (conducted in quadruplicate) was calculated as a percentage of the untreated control values using the equation: (mean A₅₄₀ of the chemical-treated cultures ÷ mean A₅₄₀ of the untreated control cultures) x 100. The percentage viability values were then used to classify the material as corrosive (<80% viability) or non-corrosive (>80%). Although data from the Skin² assay can be used to classify degrees of corrosivity (for example, to assign chemicals to UN packing groups), only corrosive/non-corrosive classifications were determined in this study.

Results and Discussion

Transcutaneous Electrical Resistance Assay

The data obtained with the TER assay are given in Tables III and IV. All the substances were compatible with the test system. The only technical difficulty encountered was in the rinsing of the skin discs with certain adherent materials (specifically, this was noted for the two formulations containing caprylic acid by Laboratory B).

Table III: In Vitro/In Vivo Comparisons

Laboratory	TER		CORROSITEX™		Skin²
Laboratory	A	B	A	B	A	B	C
Samples tested	50	50	50	48	50	50	50
Qualified samples	50^a	50^a	38	35	50^a	50^a	50^a
Corrosives identifiedcorrectly	24	22	17^b	16^b	16	24	21
Non-corrosivesidentified correctly	14	12	11^b	12^b	19	22	19
Concordance (%)	76	68	74^b	80^b	70	92	80
False positives	11	13	6^b	5^b	6	3	6
False negatives	1	3	4^b	2^b	9	1	4
Sensitivity (%)	96	88	81^b	89^b	64	96	84
Specificity (%)	56	48	65^b	71^b	76	88	76

^aall test substances were judged to "qualify" since they were all compatible with the test system.
^bvalues have been calculated on the basis of the samples which could be tested.

TER assay = transcutaneous electrical resistance assay.

Table IV: Corrosivity of the Test Chemicals as Determined in the Transcutaneous Electrical Resistance Assay

Chemical	LaboratoryA		LaboratoryB
	k-ohm/disc mean±SD^a	C/NC	k-ohm/disc mean±SD^b	C/NC
CORROSIVES
Aceticacid (glacial)	1.3± 0.4	C	1.5± 0.3	C
Acrylicacid	1.3± 0.3	C	1.3± 0.1	C
ArmeenCD	1.1± 0.1	C	1.5± 0.1	C
ArmeenTD	4.3± 1.1	C	2.5± 0.9	C
Arquad16-50	1.1± 0.4	C	0.9± 0.2	C
ArquadDMMCB-50	0.8± 0.1	C	0.5± 0.0	C
Bromoaceticacid (8%)	3.1± 0.5	C	3.6± 2.2	C
Bromoaceticacid (55.6%)	1.6± 0.3	C	2.2± 0.2	C
Butylamine	0.8± 0.1	C	1.8± 0.0	C
Capric/caprylicacid	1.6± 0.8	C	10.4± 6.8	NC
Caprylicacid	3.8± 1.8	C	13.4± 7.4	NC
Cyclohexylamine	1.1± 0.2	C	1.5± 0.0	C
1,4-Diaminobutane	0.8± 0.1	C	0.8± 0.1	C
Dichloroaceticacid	1.7± 0.4	C	3.0± 0.7	C
Diethylamine	0.9± 0.2	C	0.9± 0.1	C
DuoquadT50	2.6± 1.1	C	1.6± 0.1	C
Formicacid	1.8± 1.4	C	2.7± 0.2	C
Hexanoicacid	1.0± 0.4	C	1.4± 0.6	C
Mercaptoaceticacid	1.2± 0.2	C	1.6± 0.3	C
ProxelBD	10.5± 3.0	NC	9.4± 3.4	NC
Pyrrolidine	0.6± 0.1	C	0.8± 0.0	C
Sodiumhydroxide	1.3± 0.2	C	1.2± 0.2	C
Sodiummetasilicate	1.1± 0.2	C	0.9± 0.1	C
Sodiumsilicate A140	1.9± 1.2	C	2.9± 0.4	C
Synprolam35X2	2.0± 0.5	C	1.5± 0.3	C
NON-CORROSIVES
Armeen2C	13.3± 2.0	NC	13.2± 5.8	NC
AromoxDMMCD-W	1.2± 0.4	C	1.6± 0.0	C
ArquadC-33-W	2.3± 1.1	C	2.0± 0.4	C
Butylbenzene	3.9± 3.2	C	3.8± 2.2	C
Dequest2000	7.9± 4.0	NC	3.2± 0.5	C
DowanolPNB	6.0± 1.9	NC	7.1± 1.2	NC
ElfanOS 46	1.1± 0.3	C	0.8± 0.2	C
EmpicolLZPV/C	1.5± 0.3	C	1.4± 0.3	C
EmpigenOB	1.5± 0.6	C	1.4± 0.3	C
EmpilanCME	13.8± 4.9	NC	12.7± 2.2	NC
EmpilanKB2	7.6± 1.1	NC	2.7± 0.5	C
EthomeenT/25	12.3± 3.5	NC	13.6± 1.2	NC
GenaminKDM-F	6.7± 2.8	NC	6.9± 3.2	NC
GenapolLRO	1.8± 0.8	C	4.8± 2.0	C
n-Hexanol	3.6± 1.1	C	2.3± 0.3	C
HostaphatKLD	2.9± 0.9	C	3.3± 0.3	C
Lauricacid	6.9± 2.8	NC	14.6± 4.1	NC
n-Nonanol	2.5± 0.7	C	1.5± 1.0	C
Oleic/caprylicacid	10.6± 4.1	NC	18.5± 6.5	C
ProxelAB	11.0± 1.8	NC	9.4± 3.4	NC
Sodiumperborate	8.1± 5.3	NC	10.2± 2.1	NC
Sodiumpercarbonate	2.2± 2.0	C	1.6± 0.4	C
Sodiumsilicate H100	6.2± 2.9	NC	6.2± 0.9	NC
Triethanolamine	11.2± 1.9	NC	11.6± 5.3	NC
n-Undecanol	14.2± 3.0	NC	9.2± 3.0	NC

^an = 6;
^bn = 3

C - corrosive, NC - non-corrosive

The interlaboratory comparison of the TER assay results is shown in Figure 1. For the corrosive materials, 23 of the 25 were classified similarly by both laboratories; the two exceptions were capric/caprylic acid and caprylic acid (Table IV). Subsequently, it was shown that the false negative result in Laboratory B was caused by insufficient rinsing of the skin discs before measurement of the TER. The remaining layer of the test material acted as a barrier, thereby increasing the TER value. For the non-corrosive materials, again 23 of the 25 were classified similarly by both laboratories. The two substances which gave different results were Dequest 2000 and Empilan KB2 (Table IV).

Figure 1: Interlaboratory Comparison of Data from the Transcutaneous Electrical Resistance (TER) Assay

Of the 25 corrosive materials, Laboratory A classified 24 as corrosive and one (Proxel BD) as a non-corrosive; Laboratory B classified 22 as corrosive and three (Proxel BD, capric/caprylic acid, and caprylic acid) as non-corrosive (Table III). The false negative results from the two materials containing caprylic acid were the result of a technical problem, as described earlier. Following interlaboratory comparison of the results, these two materials were retested and gave positive results in Laboratory B. Of the 25 non-corrosives, Laboratory A classified 11 as corrosive and Laboratory B classified 13 as corrosive (Table III). The false positives found only in Laboratory B were Dequest 2000 and Empilan KB2.

The TER assay has been in regular use in several laboratories for over five years. The results do not allow prediction of the potential severity of corrosive effects (i.e. assignment into packing classes, etc.) but they provide a means to distinguish between potential corrosives and non-corrosives. The results of this pre-validation study confirm previous experiences with the assay and the published data. False negative results are rare, but false positive results are relatively common for materials containing certain solvents or surfactants (16). This is because they tend to solubilise the stratum corneum, thus allowing the passage of ions and reducing the electrical resistance. Users of the assay need to be aware of this problem and should either avoid testing certain products and/or take into account the chemical class of the test material when interpreting the results. Of the false positive results obtained in this study, sodium percarbonate was the only material which was not a solvent or a surfactant.

Thus, the TER assay was able to classify correctly 24 out of 25 corrosive materials. Although about half of the 25 non-corrosives were classified incorrectly, all but one of these would have been regarded as possible false positives on the basis of their chemical class. Therefore, with a knowledge of the chemical nature and physicochemical properties of the test materials, the TER assay is able to provide a good indication of the potential corrosivities of most test materials.

CORROSITEX™

The results obtained with the CORROSITEX™ test system for the 50 test chemicals are shown in Tables III and V. Laboratory B was not supplied with sufficient amounts of samples to be able to complete the testing of sodium silicate A140 and Synprolam 35X2.

In the initial compatibility determination, some of the test chemicals did not cause a visible change in the CDS, and thus did not qualify for testing (i.e. they were incompatible with the CORROSITEX™ test system). In Laboratory A, 38 out of the 50 chemicals qualified (76%); in Laboratory B, 35 out of the 48 chemicals tested were judged to qualify (73%). The two samples which could not be tested by Laboratory B were both qualified by Laboratory A. Only one chemical (caprylic acid) was qualified by one laboratory and not by the other. Thus, twelve (Arquad 16-50, Arquad DDMMCB, Duoquad T50, Proxel BD, butylbenzene, Dowanol PNB, Elfan OS 46, n-hexanol, Hostaphat KLD, lauric acid, n-nonanol, and n-undecanol) of the set of 50 chemicals were found to be incompatible with CORROSITEX™ in both laboratories. The 12 materials which were not qualified by both laboratories comprised seven neutral organics, three cationic surfactants, an organic acid, and an anionic surfactant. The one additional material not qualified by Laboratory B was an organic acid. It has been reported previously that many organic solvents are not detected by the CORROSITEX™ CDS (10).

The intralaboratory reproducibility (triplicate determinations) of CORROSITEX™ was very good. The coefficients of variation (CV) for Laboratory A, which recorded breakthrough times only to the nearest minute, ranged from 0-9.1% (mean: 3.2%; n = 23). The results obtained in Laboratory B, which actually recorded data to the nearest hundredth of a minute, were even more reproducible (CV: 0-0.5%; mean: 0.13%; n = 22).

Laboratory A classified 23 materials as corrosives and 15 as non-corrosives; Laboratory B identified 21 materials as corrosives and 14 as non-corrosives (Table V). Empicol LZPV/C was classified differently in the two laboratories, and caprylic acid was identified as a non-corrosive by Laboratory A but did not qualify ("NQ") according to Laboratory B. Similar breakthrough times were reported for the 21 chemicals which were classified as corrosives in both laboratories. Thirteen test materials were identified as non-corrosives by both laboratories.

Table V: Corrosivity of the Test Chemicals as Determined Using CORROSITEX™

Chemical	LaboratoryA		LaboratoryB
	Time(min)^a mean±SD	C/NC	Time(min)^a mean±SD	C/NC
CORROSIVES
Aceticacid (glacial)	21.7± 0.6	C	28.5± 0.0	C
Acrylicacid	28.0± 0.0	C	29.0± 0.0	C
ArmeenCD	159± 5.3	C	212± 0.6	C
ArmeenTD	>240	NC	>240	NC
Arquad16-50	NQ	?	NQ	?
ArquadDMMCB-50	NQ	?	NQ	?
Bromoaceticacid (8%)	28.7± 1.2	C	34.6± 0.2	C
Bromoaceticacid (55.6%)	6.3± 0.6	C	5.1± 0.0	C
Butylamine	28.3± 0.6	C	36.8± 0.1	C
Capric/caprylicacid	>240	NC	>240	NC
Caprylicacid	>240	NC	NQ	?
Cyclohexylamine	43.3± 1.2	C	48.7± 0.1	C
1,4-Diaminobutane	26.7± 0.6	C	30.9± 0.0	C
Dichloroaceticacid	12.7± 0.6	C	20.5± 0.0	C
Diethylamine	34.0± 0.0	C	33.0± 0.0	C
DuoquadT50	NQ	?	NQ	?
Formicacid	15.7± 0.6	C	18.5± 0.0	C
Hexanoicacid	95.3± 1.5	C	165± 0.4	C
Mercaptoaceticacid	37.7± 2.1	C	35.6± 0.1	C
ProxelBD	NQ	?	NQ	?
Pyrrolidine	27.7± 0.6	C	27.0± 0.0	C
Sodiumhydroxide	17.0± 1.0	C	22.0± 0.1	C
Sodiummetasilicate	18.0± 1.0	C	22.4± 0.0	C
Sodiumsilicate A140	20.0± 0.0	C	NT	?
Synprolam35X2	>240	NC	NT	?
NON-CORROSIVES
Armeen2C	>240	NC	>240	NC
AromoxDMMCD-W	>240	NC	>240	NC
ArquadC-33-W	>240	NC	>240	NC
Butylbenzene	NQ	?	NQ	?
Dequest2000	14.3± 0.6	C	15.1± 0.0	C
DowanolPNB	NQ	?	NQ	?
ElfanOS 46	NQ	?	NQ	?
EmpicolLZPV/C	140± 4.0	C	>240	NC
EmpigenOB	> 240	NC	>240	NC
EmpilanCME	>240	NC	>240	NC
EmpilanKB2	>240	NC	>240	NC
EthomeenT/25	>240	NC	>240	NC
GenaminKDM-F	>240	NC	>240	NC
GenapolLRO	>240	NC	>240	NC
n-Hexanol	NQ	?	NQ	?
HostaphatKLD	NQ	?	NQ	?
Lauricacid	NQ	?	NQ	?
n-Nonanol	NQ	?	NQ	?
Oleic/caprylicacid	>240	NC	>240	NC
ProxelAB	>240	NC	>240	NC
Sodiumperborate	63.3± 0.6	C	78.3± 0.1	C
Sodiumpercarbonate	69.7± 4.2	C	77.6± 0.0	C
Sodiumsilicate H100	45.0± 0.0	C	49.9± 0.0	C
Triethanolamine	42.7± 1.5	C	55.3± 0.1	C
n-Undecanol	NQ	?	NQ	?

^a Biobarrier breakthrough time (minutes), as described in Materials and Methods section

C - corrosive; NC - non-corrosive; NQ - not qualified (incompatible with the test system); NT - not tested (insufficient sample); ? - cannot be classified

The interlaboratory comparison of CORROSITEX™ breakthrough times is shown in Figure 2. For breakthrough times below about 40 minutes, the values obtained in both laboratories were very similar. However, for times greater than 40 minutes, faster breakthrough times were observed in Laboratory A than in Laboratory B. Thus, there could be some systematic difference in the determination of the endpoint between the two laboratories. Further investigation of the effects of environmental conditions on the data generated, such as possible differences due to the ambient temperature at which the assay is conducted, may help explain this discrepancy.

Figure 2: Interlaboratory Comparison of Data from the CORROSITEX™ Assay

Comparisons of the in vitro corrosivity classifications with those assigned to the test chemicals prior to undertaking the pre-validation study (Table I), which are mainly based on in vivo corrosivity data, are summarised in Table III. The classifications based on the in vitro data obtained by Laboratory A agreed with the assigned (in vivo) classifications for 28 of the 38 qualified test chemicals (74%). The corrosivities of six materials (Dequest 2000, Empicol LZPV/C, sodium perborate, sodium percarbonate, sodium silicate H100, and triethanolamine) were overestimated, and those of four materials (Armeen TD, capric/caprylic acid, caprylic acid, and Synprolam 35X2) were underestimated. The sensitivity and specificity were 81% and 65%, respectively. The classifications based on the in vitro data recorded by Laboratory B agreed with the assigned classifications for 28 of the 35 qualified test chemicals (80%). The corrosive effects of five materials (Dequest 2000, sodium perborate, sodium percarbonate, sodium silicate H100, and triethanolamine) were overestimated, while those of two materials (Armeen TD and capric/caprylic acid) were underestimated. The sensitivity and specificity were 89% and 71%, respectively. The materials whose corrosive effects were overestimated by both laboratories were three inorganics, an organic base, and an organic acid. Those which were underestimated were organic acids (two) and organic bases (two).

Following the completion of the blind part of this study, IVI determined that the corrosive effects of materials with low acid/alkaline reserve capacities were often overpredicted by CORROSITEX™. A screening test was, therefore, introduced to assign materials to one of four categories: A₁ - high acid content; B₁ - high base content; A₂ - low acid content; and B₂ - low base content. The cut-off breakthrough time for distinguishing between corrosives and non-corrosives was adjusted (from 240 minutes to 45 minutes) for materials designated A₂ or B₂. Thus, if these materials resulted in a change in the CDS within 45 minutes, they were classified as corrosives; if the breakthrough time was greater than 45 minutes, the material was considered to be non-corrosive. Post-hoc analysis of the data for the chemicals tested during this pre-validation study, to incorporate this new procedure for handling materials with low acid/alkaline reserve capacities, resulted in the re-classification of four (Laboratory A) or five (Laboratory B) chemicals (data not shown), and slightly increased the in vitro/in vivo concordance values to 79% (Laboratory A) and 83% (Laboratory B).

Skin² Model ZK1350 In Vitro Skin Corrosion Test

The results obtained using the Skin² test are summarised in Tables III and VI. All substances were compatible with the test system. Certain highly viscous test materials were difficult to pipette accurately, and the skin cultures were applied to these test materials with light pressure, to ensure even contact.

Comparison of the percentage viabilities from the three independent laboratories indicates reasonably good reproducibility, although there are several outliers (Figures 3A, 3B and 3C). An analysis of variance (ANOVA) showed no significant evidence of systematic interlaboratory differences based on the percentage viability data (P > 0.1). Comparison of the corrosive/non-corrosive classifications (Table VI) indicates agreement between all three laboratories for 30 of the 50 chemicals (i.e. 60%). Pairwise comparisons indicate better agreement between laboratories B and C (80%) than between Laboratory A and either of the other two laboratories (A/B, 70%; A/C, 60%). Although the laboratories were familiar with the basic assay procedures, this study was the first time that this specific Skin² protocol had been conducted in each of the three laboratories. It is probable that additional experience and training will improve the interlaboratory reproducibility of the results obtained.

Table VI: Corrosivity of the Test Chemicals as Determined Using the Skin² Test

Chemical	Laboratory A Viability(%)^a		Laboratory B Viability(%)^a		Laboratory C Viability(%)^a
Chemical	mean ± SD^b	C/NC	mean ± SDb	C/NC	mean ± SD^b	C/NC
CORROSIVES
Acetic acid (glacial)	4.1 ± 14.7	C	5.9 ± 7.8	C	24.3 ± 13.8	C
Acrylic acid	0 ± 1.7	C	1.9 ± 1.7	C	4.2 ± 2.4	C
Armeen CD	18.4 ± 8.8	C	7.4 ± 2.1	C	7.2 ± 1.5	C
Armeen TD	48.8 ± 13.0	C	15.5 ± 3.7	C	17.9 ± 2.9	C
Arquad 16-50	98.0 ± 7.7	NC	61.0 ± 14.8	C	50.3 ± 12.6	C
Arquad DMMCB-50	31.8 ± 20.4	C	44.3 ± 2.8	C	44.6 ± 20.7	C
Bromoacetic acid (8%)	47.6 ± 7.8	C	33.5 ± 12.2	C	40.2 ± 4.6	C
Bromoacetic acid (55.6%)	2.7 ± 8.2	C	7.3 ± 4.2	C	15.6 ± 7.9	C
Butylamine	0 ± 1.4	C	1.4 ± 0.9	C	2.3 ± 0.9	C
Capric/caprylic acid	72.8 ± 32.6	C	54.5 ± 4.4	C	69.9 ± 18.2	C
Caprylic acid	88.8 ± 22.3	NC	98.8 ± 43.2	NC	78.0 ± 25.4	C
Cyclohexylamine	1.3 ± 3.1	C	12.6 ± 3.5	C	15.2 ± 15.4	C
1,4-Diaminobutane	101 ± 24.3	NC	64.6 ± 31.7	C	61.6 ± 27.8	C
Dichloroacetic acid	22.5 ± 14.3	C	34.5 ± 29.4	C	14.6 ± 7.9	C
Diethylamine	0.7 ± 0.9	C	26.0 ± 13.7	C	10.0 ± 7.8	C
Duoquad T50	85.1 ± 9.1	NC	68.5 ± 22.4	C	79.9 ± 15.3	C
Formic acid	80.6 ± 27.9	NC	13.4 ± 20.6	C	24.2 ± 18.8	C
Hexanoic acid	85.8 ± 29.2	NC	63.0 ± 25.8	C	42.1 ± 7.7	C
Mercaptoacetic acid	95.0 ± 36.2	NC	59.0 ± 17.3	C	84.2 ± 12.0	NC
Proxel BD	95.4 ± 23.5	NC	66.1 ± 4.3	C	102 ± 11.7	NC
Pyrrolidine	8.9 ± 17.9	C	30.2 ± 29.6	C	9.1 ± 4.0	C
Sodium hydroxide	61.2 ± 40.0	C	55.7 ± 33.2	C	30.1 ± 10.9	C
Sodium metasilicate	65.7 ± 10.4	C	35.3 ± 28.6	C	106 ± 9.1	NC
Sodium silicate A140	82.4 ± 12.0	NC	73.7 ± 13.0	C	102 ± 6.5	NC
Synprolam 35X2	47.4 ± 22.3	C	29.2 ± 3.9	C	56.0 ± 12.6	C
NON-CORROSIVES
Armeen 2C	108 ± 26.9	NC	90.0 ± 26.0	NC	96.1 ± 5.7	NC
Aromox DMMCD-W	123 ± 7.4	NC	89.4 ± 10.5	NC	62.0 ± 10.1	C
Arquad C-33-W	83.0 ± 12.7	NC	68.7 ± 30.3	C	63.5 ± 22.9	C
Butylbenzene	134 ± 17.4	NC	112 ± 37.3	NC	79.9 ± 19.3	C
Dequest 2000	95.0 ± 28.3	NC	95.4 ± 17.0	NC	85.4 ± 14.8	NC
Dowanol PNB	75.0 ± 7.8	C	121 ± 24.4	NC	113 ± 32.4	NC
Elfan OS 46	93.5 ± 7.4	NC	92.3 ± 7.3	NC	103 ± 5.4	NC
Empicol LZPV/C	90.7 ± 25.5	NC	64.1 ± 6.8	C	108 ± 12.3	NC
Empigen OB	75.5 ± 5.1	C	88.8 ± 4.3	NC	77.8 ± 13.1	C
Empilan CME	73.3 ± 15.7	C	84.5 ± 20.0	NC	102 ± 7.9	NC
Empilan KB2	75.0 ± 32.1	C	66.3 ± 6.1	CM	73.3 ± 8.9	C
Ethomeen T/25	93.5 ± 21.0	NC	105 ± 25.3	NC	73.5 ± 7.9	C
Genamin KDM-F	77.3 ± 9.3	C	85.6 ± 17.3	NC	110 ± 3.4	NC
Genapol LRO	69.9 ± 24.8	C	90.5 ± 19.4	NC	95.0 ± 17.9	NC
n-Hexanol	103 ± 6.5	NC	85.2 ± 17.7	NC	87.4 ± 4.6	NC
Hostaphat KLD	83.7 ± 17.3	NC	84.5 ± 8.7	NC	94.2 ± 3.8	NC
Lauric acid	92.4 ± 7.8	NC	92.2 ± 21.5	NC	99.7 ± 14.9	NC
n-Nonanol	115 ± 15.0	NC	98.3 ± 15.7	NC	92.4 ± 12.7	NC
Oleic/caprylic acid	113 ± 5.7	NC	113 ± 25.6	NC	120 ± 14.3	NC
Proxel AB	98.6 ± 5.1	NC	82.9 ± 6.1	NC	104 ± 10.6	NC
Sodium perborate	95.3 ± 24.6	NC	130 ± 38.0	NC	108 ± 7.3	NC
Sodium percarbonate	112 ± 19.0	NC	106 ± 16.9	NC	93.7 ± 5.6	NC
Sodium silicate H100	91.7 ± 10.9	NC	106 ± 18.6	NC	88.4 ± 16.7	NC
Triethanolamine	117 ± 13.7	NC	85.6 ± 11.1	NC	102 ± 9.6	NC
n-Undecanol	97.5 ± 11.2	NC	93.5 ± 5.5	NC	103 ± 7.5	NC

^aCell viability, as a percentage of the control, was determined following topical treatment of human skin cultures with test chemical for 10 seconds.
^bn = 4; standard deviations (SD) for 4 skin samples per test material, conducted in parallel, are given. Since these values have been calculated with respect to four independent controls (mean = 100% viability), which themselves showed variation (SD ± ±10%), the SD values quoted should not be used for further statistical calculations.
^cC - corrosive (< 80% viability); NC - non-corrosive (> 80% viability)

Figure 3: Interlaboratory Comparison of Data from the Skin^2TM Assay

The ability of the Skin² test to predict the corrosive/non-corrosive classifications of the test chemicals is summarised in Table III. The concordance between the in vitro and in vivo classifications ranged from 70-92% for the three laboratories. There were four false negative results in either two or all three of the laboratories: caprylic acid, mercaptoacetic acid, Proxel BD, and sodium silicate. There were three false positive results in either two or all of the laboratories: Arquad C-33-W, Empigen OB, and Empilan KB2. Thus, the Skin² test is promising in terms of its ability to identify corrosives and non-corrosives.

The interlaboratory comparisons were acceptable, but formal training of the participants is recommended prior to a formal validation study, to ensure uniform conduct of the test across laboratories. To determine intralaboratory versus interlaboratory sources of variability, it would be useful to include more repeat experiments (for example, three independent experiments) with each chemical in each laboratory, rather than undertaking a single experiment with quadruplicate cell cultures.

General Discussion

As described in the Introduction, the purpose of this study was to evaluate three in vitro tests for their abilities to detect substances which are corrosive to skin. This was not a validation study in terms of any of the currently accepted criteria, such as those described by Balls et al (9). The principal objective was a relative comparison of the three assays, to determine which, if any, had been sufficiently well-developed to be considered for inclusion in a formal validation study.

No attempt was made to select the test materials from a wide range of chemical classes as, ideally, would be the case for a formal validation study. Test substances were selected on the basis of their skin corrosivity potentials and not their irritancy potentials, and so no conclusions can be drawn regarding the irritant effects of the test materials.

The relative performances of the tests have been assessed by comparing their interlaboratory reproducibilities and the values for their sensitivity, specificity, and concordance (Table III).

Interlaboratory Reproducibility

Transcutaneous Electrical Resistance Assay
For the corrosive substances, only two were identified differently in the two laboratories; similarly, for the non-corrosives, only two were classified differently. Thus, for the 50 substances tested, both laboratories agreed in their assessment for 46 materials (i.e. 92%).

CORROSITEX™
The assessment of CORROSITEX™ has to take into account those test substances which do not qualify (NQ) in the assay. In some cases, a substance may not qualify in one laboratory whereas another laboratory can obtain a result (for example, caprylic acid; Table V). Therefore, for this analysis NQ is taken to be an outcome of the assay, and is included in the interlaboratory comparison. Of the 25 corrosives tested, only one was classified differently by the two laboratories. However, two of the corrosive substances were not tested in one of the laboratories because there was insufficient sample available. For the non-corrosives, again only one substance was classified differently. Thus, for the 48 substances tested in both laboratories, the assessment agreed for 46 (i.e. 96%).

Skin² Model ZK1350 In Vitro Skin Corrosion Test

Three laboratories carried out the Skin² assay, whereas only two laboratories conducted both of the other tests. Agreement on the corrosive/non-corrosive classification across all three laboratories was obtained for 30 (i.e. 60%) of the test substances. Pairwise comparisons between the three laboratories indicated higher degrees of agreement of the results obtained (60%, 70% and 80%). Whilst this is better than the comparison across the three laboratories, the agreement is still markedly lower than that for the TER and CORROSITEX™ assays. The reasons for the relatively low reproducibility of the Skin² assay were not investigated as part of this study, but may relate to different degrees of training and experience across the participating laboratories. It is recommended that possible sources of interlaboratory variation be identified and resolved prior to the inclusion of the Skin² assay in a formal validation study.

Sensitivity/Specificity/Concordance

For assays which provide results that can be assigned to one of two categories, the analysis proposed by Cooper et al (23) may be used. The analysis provides several indices which can be used to assess the overall performance of assays in identifying correctly the hazard of test substances. Although this study was designed to assess the relative performances of the three in vitro tests, the indices derived are also useful in this context. The in vitro/in vivo comparisons for all the tests in each laboratory are given in Table III.

Concordance is a measure of the ability of an assay to assign a test substance into its true category. Sensitivity is a measure of the ability of an assay to identify correctly the "active" test substances. For these indices, there were no appreciable differences between the three in vitro tests. Specificity is a measure of the ability of an assay to discriminate between "active" and "inactive" test substances. In this particular case it is assessed by scoring the number of non-corrosives identified correctly. The highest specificity (76-88%) was found with the Skin² assay. CORROSITEX™ performed moderately well, giving specificities of 65% and 71%, whilst the TER assay showed relatively low specificity (values of 48% and 56%) because it gave too many false positives.

Conclusions and Recommendations

On the basis of the results of this study, it is impossible to conclude that any one test performed better than the other two. Each method has both strengths and weaknesses. The TER assay showed high sensitivity but was characterised by relatively low specificity (i.e. too many false positives). As with all ex vivo tests, the TER assay has both the advantage of close relevance to the in vivo animal model and the disadvantage of requiring animal tissue. CORROSITEX™ gave high concordance and specificity values, but its overall utility is reduced by the significant proportion of non-qualifying test materials. The overall performance of the Skin² assay was quite creditable, but one laboratory reported significantly different results from the other two, suggesting that there are technical issues which should be addressed.

There are well-established standard operating procedures for all of the tests, and it has been demonstrated that they can all be transferred from one laboratory to another. A clear, unambiguous endpoint is defined for each in vitro test, in order to distinguish between corrosive and non-corrosive substances. Although two of the methods, CORROSITEX™ and Skin², are marketed as having the capability to distinguish between degrees of corrosive effects (for example, to assign chemicals to the UN packing groups I-III), this was not addressed within this pre-validation study. To further clarify the state of optimisation and evaluation of the three tests, it is recommended that the following additional work is undertaken:

Investigations are undertaken to try to reduce the false positive rate of the TER assay.
The TER assay is assessed for its ability to distinguish between chemicals with different degrees of corrosivity (for example, the UN packing groups or EU classification groups R34 and R35).
Investigations are undertaken to try to modify the CORROSITEX™ CDS, specifically to reduce the number of non-qualifying test materials, and to reduce the subjectivity of the assay.
Investigations are conducted with the Skin² assay with the aim of reducing the interlaboratory variability (in terms of predicting corrosive/non-corrosive classifications).
The laboratories conducting the Skin² assay should consider revaluating the threshold value for distinguishing between corrosives and non-corrosives.

Depending on the outcome of this additional work, it is recommended that a formal validation study is then conducted involving the TER, CORROSITEX™ and Skin² assays. For such a study, a suitable set of test materials should be selected, probably from the reference chemical set which is currently being prepared by an ECETOC Task Force (A.P. Walker, personal communication). It is further recommended that such a study be undertaken in the light of recent recommendations concerning the conduct of validation studies (9), and taking into account the lessons learned during previous validation studies (24).

Furthermore, future validation studies should, wherever possible, incorporate knowledge from structure-activity relationships (qualitative and quantitative). Such information may be useful in guiding the selection of appropriate test chemicals, so that they cover different mechanistic classes and represent an even spread of in vivo potencies (25, 26). Appendix A provides an example of the application of quantitative structure-activity relationships (QSAR) to the prediction of skin corrosivity.

Acknowledgements

The authors would like to thank Julia Fentem (ECVAM, Italy) for compiling and editing this workshop report. They would also like to thank the following people for their contributions to the conduct and analysis of the pre-validation study: Lesley Earl, George Holland, and Penny Jones (Unilever ESL, UK), Agnés Arnoud and Sylvie Diot (Rhône-Poulenc, France), Heather J. Connolly and Nick, A.M. Hadfield (ZENECA CTL, UK), John Harbell, Brian Ignotz, Kathleen Phillips and Kathleen Wallace (Microbiological Associates, USA), Janis Demetrulias, Leonardo Epstein, Mary A. Perkins, Dierdre A. Roberts and Susan R. Wilkins (Procter & Gamble, USA), Beate Döring and Ferdinand Moldenhauer (ZEBET, Germany), and Patricia F. Uphill (Huntingdon Research Centre, UK).

References

Anon. (1994). ECVAM News & Views. ATLA 22: 7-11.
Draize, J.H., Woodand, G. & Calvery, H.O. (1944). Methods for the study of irritation and toxicity of substances applied topically to the skin and mucous membranes. Journal of Pharmacology and Experimental Therapeutics 82: 377-390.
Anon. (1991). Code of Federal Regulations. Method of Testing Corrosion to the Skin. Transportation Title 49, Part 173.136, Appendix A. Washington, D.C.: Office of the Federal Register, National Archives and Records Administration.
Anon. (1992). OECD Guideline for Testing of Chemicals, Number 404. (Adopted by the Council on 17 July 1992). 6 pp. Paris: Organisation for Economic Cooperation and Development.
Anon. (1977). United Nations Economic and Social Council. Joint meeting of the RID safety committee and the group of experts on the transportation of dangerous goods. Trans/GE 15/R 274: 2.
Anon. (1983). Guide to the classification and labelling of dangerous substances and preparations. Criteria for the choice of chemicals indicating special risks (R phrases) and safety advice (S phrases). Official Journal of the European Communities L257: 13-20.
Oliver, G.J.A., Pemberton, M.A. & Rhodes, C. (1988). An in vitro model for identifying skin corrosive chemicals. I. Initial validation. Toxicology in Vitro 2: 7-17.
Botham, P.A., Hall, T.J., Dennett, R., McCall, J.C., Basketter, D.A., Whittle, E., Cheesman, M., Esdaile, D.J. & Gardner, J. (1992). The skin corrosivity test in vitro. Results of an interlaboratory trial. Toxicology in Vitro 6: 191-194.
Balls, M., Blaauboer, B.J., Fentem, J.H., Bruner, L., Combes, R.D., Ekwall, B., Fielder, R.J., Guillouzo, A., Lewis, R.W., Lovell, D.P., Reinhardt, C.A., Repetto, G., Sladowski, D., Spielmann, H. & Zucco, F. (1995). Practical aspects of the validation of toxicity test procedures. The report and recommendations of ECVAM workshop 5. ATLA 23: in press.
Gordon, V.C., Harvell, J.D. & Maibach, H.I. (1994). Dermal corrosion, the CORROSITEX™ system: a DOT accepted method to predict corrosivity potential of test materials. In In Vitro Skin Toxicology - Irritation, Phototoxicity, Sensitization. Alternative Methods in Toxicology, Volume 10. Ed. A. Rougier, A.M. Goldberg & H.I. Maibach. pp. 37-45. New York: Mary Ann Liebert.
de Wever, B. & Rheins, L.A. (1994). Skin^2TM: An in vitro human skin analogue. In In Vitro Skin Toxicology - Irritation, Phototoxicity, Sensitization. Alternative Methods in Toxicology, Volume 10. Ed. A. Rougier, A.M. Goldberg & H.I. Maibach. pp. 121-131. New York: Mary Ann Liebert.
Jacobs, G. & Martens, M. (1987). Evaluation of the test method for skin irritation as prescribed by OECD and EEC. Journal of Toxicology - Cutaneous and Ocular Toxicology 6: 215-226.
Anon. (1984). Commission Directive 84/449/EEC of 25 April 1984 adapting to technical progress for the sixth time Council Directive 67/548/EEC on the approximation of laws, regulations and administrative provisions relating to the classification, packaging and labelling of dangerous substances: skin irritation. Official Journal of the European Communities L251: 106-108.
Oliver, G.J.A., Pemberton, M.A. & Rhodes, C. (1986). An in vitro skin corrosivity test - modifications and validation. Food and Chemical Toxicology 24: 507-512.
Oliver, G.J.A. (1990). The evaluation of cutaneous toxicity: past and future. In Skin Pharmacology and Toxicology: Recent Advances. Ed. C.L. Galli, C.N. Hensby & M. Marinovich. pp. 147-163. New York: Plenum Press.
Barlow, A., Hirst, R.A., Pemberton, M.A., Rigden, A., Hall, T.J., Botham, P.A. & Oliver, G.J.A. (1991). Refinement of an in vitro test for the identification of skin corrosive chemicals. Toxicology Methods 1: 106-115.
Naughton, G.K., Jacobs, L. & Naughton, B.A. (1989). A physiologic skin model for in vitro toxicity studies. In Alternative Methods in Toxicology, Volume 7. Ed. A.M. Goldberg. pp. 183-189. New York: Mary Ann Liebert.
Perkins, M.A. & Osborne, R. (1993). Development of an in vitro method for skin corrosion testing. Journal of Investigative Dermatology 100: 535.
Slivka, S.R., Landeen, L., Zimber, M.P. & Bartel, R.L. (1993). Biochemical characterization, barrier function, and drug metabolism of an in vitro skin model. Journal of Investigative Dermatology 100: 39-45.
Contard, P., Bartel, R.L., Jacobs II, L., Perlish, J.S., MacDonald II, E.D., Handler, L., Cone, D. & Fleischmajer, R. (1993). Culturing keratinocytes and fibroblasts in a three-dimensional mesh results in epidermal differentiation and formation of a basal lamina-anchoring zone. Journal of Investigative Dermatology 100: 34-38.
Mosmann, T. (1983). Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays. Journal of Immunological Methods 65: 55-63.
Triglia, D., Sherard Braa, S., Yonan, C. & Naughton, G.K. (1991). Cytotoxicity testing using neutral red and MTT assays on a three-dimensional human skin substrate. Toxicology in Vitro 5: 573-578.
Cooper, J.A., Saracci, R. & Cole, P. (1979). Describing the validity of carcinogen screening tests. British Journal of Cancer 39: 87-89.
Fentem, J.H., Prinsen, M.K., Spielmann, H., Walum, E. & Botham, P.A. Validation - lessons learned from practical experience. Submitted to Toxicology in Vitro.
Chamberlain, M. & Barratt, M.D. (1995). Practical applications of QSAR to in vitro toxicology illustrated by consideration of eye irritation. Toxicology in Vitro, in press.
Barratt, M.D., Castell, J.V., Chamberlain, M., Combes, R.D., Dearden, J.C., Fentem, J.H., Gerner, I., Giuliani, A., Gray, T.J.B., Livingstone, D.J., Provan, W.McL., Rutten, A.A.J.J.L., Verhaar, H.J.M. & Zbinden, P. (1995). The integrated use of alternative approaches for predicting toxic hazard. The report and recommendations of an ECVAM workshop. ATLA, in press.

A Pre-validation Study on In Vitro Skin Corrosivity Testing

The Report and Recommendations of ECVAM Workshop 61,2

Appendix 1

The Report and Recommendations of ECVAM Workshop 6^1,2