The second ECVAM workshop on phototoxicity testing.

The Report and Recommendations of ECVAM Workshop 42^1,2

Reprinted with minor amendments from ATLA 28: 777-814.

Horst Spielmann,³ Lutz Müller,⁴ Dietrich Averbeck,⁵ Michael Balls,⁶ Susanne Brendler-Schwaab,⁷ José V. Castell,⁸ Rodger Curren,⁹ Odile de Silva,¹⁰ Neil K. Gibbs,¹¹ Manfred Liebsch,³ Will W. Lovell,¹² Hans F. Merk,¹³ J. Frank Nash,¹⁴ Norbert J. Neumann,¹⁵ Wolfgang J.W. Pape,¹⁶ Peter Ulrich¹⁷ and Hans-Werner Vohr⁷
³ZEBET BgVV, Diedersdorfer Weg 1, 12277 Berlin, Germany; ⁴Novartis Pharma AG, WSH 2881.228, 4002 Basel, Switzerland; ⁵Institut Curie--Section de Recherche, UMR218 CNRS 26 Rue d'Ulm, 75248 Paris Cedex 05, France; ⁶ECVAM, JRC, Institute for Health and Consumer Protection, TP 580, 21020 Ispra (VA), Italy; ⁷Bayer AG, FB Toxikologie, Abt. Cancerogenität & Genotoxizität, P.O. Box 10 17 18, 42096 Wuppertal, Germany; ⁸Hospital Universitario La Fe, Centro de Investigacion, Avda. de Campanar 21, 46009 Valencia, Spain; ⁹Institute for In Vitro Sciences Inc., 21 Firstfield Road, Suite 220, Gaithersburg, MD 20878, USA; ¹⁰L'Oréal, Direction des Sciences du Vivant, Relations Extérieures, 1 Avenue Eugéne Schueller, 93600 Aulnay-sous-Bois, France; ¹¹University of Manchester Dermatology Centre, Photobiology Unit, Hope Hospital, Manchester M6 8HD, UK; ¹²Unilever Research, SEAC Toxicology, Colworth House, Sharnbrook, Bedford MK44 1LQ, UK; ¹³Universitätsklinikum der Rheinisch-Westfälischen Technischen Hochschule Aachen, Hautklinik, Pauwelsstrasse 30, 52074 Aachen, Germany; ¹⁴The Procter & Gamble Company, Human Safely--Skin Care, Sharon Woods Technical Centre, 11511 Reed Hartman Highway, Cincinnati, OH 45241, USA; ¹⁵Heinrich-Heine-Universität, Hautklinik, Moorenstrasse 5, 40225 Düsseldorf, Germany; ¹⁶Beiersdorf AG, Safety Assessment Centre Cosmed, K.St. 4284--Safety Assessment/Scientific Services, Unnastrasse 48, 20245 Hamburg, Germany; ¹⁷Novartis Pharma AG, PCS-Tox/Path, WSH-2881.3.29, 4002 Basel, Switzerland

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: Professor Horst Spielmann, ZEBET, BgVV Diedersdorfer Weg 1, 12277 Berlin, Germany. e-mail: zebet@bgvv.de

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

Preface:

This is the report of the forty-second of a series of workshops organised by the European Centre for the Validation of Alternative Methods (ECVAM). ECVAM was established in 1992 by the Commission of the European Communities (EC) to carry into effect article 23 of Council Directive 86/609/EEC on the protection of animals used for experimental and other scientific purposes (1). ECVAM is part of the Institute for Health & Consumer Protection (IHCP) of the Joint Research Centre of the European Commission, located at Ispra (Italy). The main goal of ECVAM is to promote the regulatory acceptance of alternative methods which are of importance to the biosciences and which reduce, refine or replace the use of laboratory animals. The activities of the centre are focused on facilitating, coordinating, supporting and/or organising international studies, preferably in collaboration with other bodies.

One of the first priorities set by ECVAM was the implementation of procedures that 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 in vitro tests and their potential uses, and would make recommendations about the best ways forward.

The first ECVAM workshop on in vitro phototoxicity testing (2) set the stage for the first successful validation study of in vitro phototoxicity tests (3, 4). Meanwhile, the in vitro 3T3 neutral red uptake (3T3 NRU) phototoxicity test has been accepted for regulatory purposes within the European Union. Since this in vitro photocytotoxicity test does not cover all areas of phototoxicity, the European Cosmetic, Perfumery and Toiletry Association (COLIPA) and the ECVAM Scientific Advisory Committee (ESAC) suggested that additional in vitro tests should be developed and validated to cover the entire field of phototoxicology.

To set priorities for collaborative studies, the second ECVAM workshop on in vitro phototoxicity testing was held on 22-27 June 1999, in Berlin at ZEBET (the National German Centre for Documentation and Evaluation of Alternatives to Testing in Animals), at the BgVV (Federal Institute for Health Protection of Consumers and Veterinary Medicine). The workshop was chaired by Horst Spielmann and Lutz Müller (then at BfArM, Berlin, Germany, now at Novartis, Basel, Switzerland), and was attended by photobiologists, toxicologists and photodermatologists from industry, academia and government. The current status of advanced testing methods was discussed, focusing on both the safety of new chemicals in humans and the reduction of testing in animals. A welcome contribution to the workshop was the provision by dermatologists from the Austrian, German and Swiss photopatch test group of unpublished clinical data collected over the past 12 years.

Consensus was reached at the workshop that validation efforts for in vitro photogenotoxicity tests should be given a high priority at ECVAM, since international regulatory agencies--for example, the US Food and Drug Administration (FDA; 5)--have been focusing on chemical photocarcinogenicity in rodents. In addition, the development and validation of in vitro methods for assessing the photoallergy potential of chemicals, and the application of 3-dimensional human skin models for the in vitro phototoxicity testing of finished products should be funded. The workshop also agreed on a general testing strategy covering all aspects of in vitro phototoxicity that will ensure the highest level of consumer protection and avoid testing in animals.

Introduction

In vitro phototoxicity testing was the topic of the second ECVAM workshop, in 1993 (2). The task given to participants of the first phototoxicity workshop was to plan a validation study on the most promising in vitro phototoxicity tests and to identify an optimum set of test chemicals, based on high-quality in vivo data in humans. The participants were chosen in collaboration with the COLIPA task force on phototoxicity testing, and represented an excellent selection of European experts from industry and academia. For everyone who had the pleasure of participating in this workshop, the stimulating discussions with the late Brian E. Johnson from Aberdeen, remain one of the most memorable experiences. Whenever our discussion reached an impasse, Brian, who was wheelchair-bound due to multiple sclerosis, would go back to his room and come up with the publication that we had been missing. Thanks to him, our discussion was then immediately back on course, and we were able to solve almost all of our scientific problems during our time in Angera, near Ispra.

Taking into account the subsequent progress in photobiology, the participants of the second ECVAM workshop on phototoxicity testing were asked to describe, discuss and evaluate the current status of phototoxicity testing both in vitro and in vivo. To cover the most important areas in phototoxicology, firstly, small break-out groups and, at a later stage, all of the participants discussed the following main topics: terminology, mechanisms of photosensitisation reactions, acute phototoxicity, photopatch testing in humans, photoallergy and photogenotoxicity/photocarcinogenicity. For each of these topics, the workshop agreed on recommendations to ECVAM and other funding agencies, as well as on testing strategies for the safety assessment of chemicals. The workshop report will cover each of the major aspects of phototoxicity separately.

Terminology

Chemical photosensitisation as an adverse reaction can be induced by a broad spectrum of industrial or therapeutic agents, which can enter the body by being swallowed, injected or topically applied. It is essential to clarify the technical terms related to light-induced and, in particular, ultraviolet (UV) radiation-induced toxicology. Although this document will address genotoxic effects induced by exposure to light/UV radiation alone, in general, the terminology is restricted to defining toxicological and technical terms of acute toxicity induced by a chemical plus light (Figure 1).

Figure 1: Photobiological effects of chemicals

Photosensitisation is defined as a process in which reactions to normally ineffective radiation doses are induced in a system by the introduction of a specific, radiation-absorbing substance (the photosensitiser) that causes another substance (the substrate) to be changed by the same dose of radiation. When used to describe the reaction of skin to an exogenous chemical and UV or visible radiation, the term includes both phototoxic and photoallergic reactions.

Phototoxicity is an acute toxic response elicited after the first exposure of skin to certain chemicals and subsequent exposure to light/UV radiation, or that is similarly induced by skin irradiation after the systemic administration of a chemical (the photosensitiser).

Photoirritation is a particular type of phototoxicity: the term is used to describe only those phototoxic skin reactions that are induced by chemicals 0-72 hours after exposure to light/UV radiation ("acute reaction").

Photoallergy is an acquired immunological reactivity, which does not occur on first treatment with a photosensitiser and light/UV radiation, and needs an induction period of one or two weeks before skin reactivity can be demonstrated by administration of photosensitiser and irradiation with light/UV radiation.

Photogenotoxicity/photomutagenicity is a genotoxic response, which is observed after exposure to a chemical (a photosensitiser) and a (non)-genotoxic dose of light/UV radiation.

Photocarcinogenicity is defined as carcinogenicity induced by a chemical and repeated application of light/UV radiation. The term "photo co-carcinogenesis", is used if UV-induced tumorigenesis is enhanced by a chemical.

UV and visible light wavebands. The ultraviolet spectral band designations recommended by the CIE (Commission Internationale de L'Eclairage) are: UVA = 315-400 nm; UVB = 280-315 nm; and UVC = 100-280 nm Other designations are also used; the division between UVB and UVA is often placed at 320 nm Furthermore, UVA may be divided into UV-A1 and UV-A2, with a division made at about 340 nm The visible-light spectral band (400-750 nm) covers the range between ultraviolet and infrared radiation.

Dose of radiation is defined as the quantity of UV radiation or light incident on a surface, measured in Joules per square metre, J/m² (irradiance: j/s/m² = W/m²)

Two general types of light sources are used in phototoxicity testing: light/UV radiation sources with a limited emission spectrum, such as fluorescent tubes for UVA or UVB or mercury arc lamps; and sources with an emission spectrum simulating solar light. The approximate solar simulation given by filtered mercury-metal halide lamps, as used in the 3T3 NRU phototoxicity test, is useful in the routine testing laboratory. More-specialist sources can be used in the research laboratory, for example, the European Centre for the Ecotoxicology and Toxicology of Animals (ECETOC) source for phototransformation (photodegradation) studies.

It is important when considering any photobiological effect, to take into account the emission spectrum of the lamp and the action spectrum of the phenomenon in question. Thus, although accurate radiometry over a broad spectral range is essential for interlaboratory comparisons, a simple summation of radiation within a waveband is often misleading. However, information on irradiance over time is essential, for example, for interlaboratory comparisons.

Mechanisms of Photosensitised Reactions

The chemistry and biology of photosensitised reactions (Figure 2) have been extensively reviewed (2, 6-8). Absorption of light in the 300-750 nm range indicates the ability of a molecule to be activated to an excited singlet state, and is an essential feature of photosensitisers. Following excitation, some absorbing molecules may dissipate energy in several ways, including photolysis/photodegradation, which result in photoproducts that may be toxic. The determination of the UV/visible absorption spectrum should therefore precede any testing for photosensitising properties in biological systems in vitro or in vivo.

Figure 2: Mechanisms involved in phototoxicity

After absorption of a photon of UV/visible radiation, photosensitisers have two systems of electronically excited states, the singlet and triplet states. The singlet state is usually short-lived, but may cross to form a triplet state with a longer lifetime. With very few exceptions, photosensitised oxidations proceed by way of the triplet photo sensitiser. Therefore, effective photo sensitisers are usually those providing a high yield of a long-lived triplet (see energy transfer in Figure 2). The triplet state of the photosensitiser can subsequently react via two major pathways (6; Figure 2): a) by electron or hydrogen transfer (free radical) processes (Type I reaction), which may or may not require oxygen; or b) by energy transfer (typically) to oxygen (Type II reaction), to form excited-state singlet oxygen.

The relative contributions of the Type I and Type II processes depend on the chemical nature of the photosensitiser and the substrate, the reaction conditions (solvent, pH, concentrations of photosensitiser, substrate and oxygen) and, in some cases, on whether the photosensitiser absorbs light into its first or second absorption band.

Even though it is longer-lived than the singlet, the triplet state of a photosensitiser still typically only has a submillisecond lifetime. Therefore, for efficient photosensitisation, the photosensitiser needs to be excited near to its biological target. Cellular targets for photosensitised oxidations include the plasma membrane, cytoplasmic organelles and the nucleus, depending on the uptake and localisation of the photosensitiser.

Stable photodegradation products of some photosensitisers can act as toxins or as photosensitisers. However, other reaction types may proceed efficiently with only short-lived, low-yield excited states, especially when the photosensitiser is dark-bound to a macromolecule (for example, 8-MOP to DNA, or tetrachlorosalicylanilide to serum albumin).

Photoallergy is considered to be delayed-type hypersensitivity mediated by the formation of a photosensitiser-protein conjugate (9). Therefore, protein-binding to skin proteins is an essential, but not exclusive, feature of photoallergens.

Photogenotoxicity and photocarcinogenicity (Figure 1), as endpoints of chemical UV sensitisation, have recently come to the attention of regulators and scientists in the drug and cosmetic industries. Standard methods for assessing these endpoints have yet to be established.

Human Photopatch Testing

The photopatch test

The German, Austrian and Swiss photopatch test group was founded in 1984, to investigate photoallergic reactions, and the epidemiology in central Europe of photoallergy (10). During the first period of the project (1985-1990), 32 substances were tested in patients suspected to be photosensitive. The high frequency of clinically non-relevant reactions in photopatch testing (mostly phototoxic) was identified as a major problem (11). To reduce such clinically non-relevant reactions, in the second period of the multicentre study (1991-1997), the set of test substances was modified in the following manner: a few substances were taken from the test set, because they did not induce any responses when applied topically to the skin, for example, furosemide, or because topical application in the photopatch test was associated with a high risk of photo sensitisation, for example, tiaprofenic acid. In addition, the test concentration of promethazine was reduced from 1% to 0.1% to diminish the high numbers of phototoxic reactions that were observed in the first test period with this compound. A few new substances were added to the test set, most of them sunscreens (12). Thus, a modified set of 26 test substances was used in the second test period (Table I).

Table I: Chemicals used in photopatch testing period II, 1991-1997

Substance	Concentration
Tetrachlorosalicylanilide Bromosalicylchloranilide Tribromosalicylanilide Buclosamide Fenticlor	0.1 1.0 1.0 5.0 1.0
Hexachlorophene Bithionol Triclosan Sulphanilamide Chlorpromazine	1.0 1.0 2.0 5.0 0.1
Promethazine Carprofen Quinidine Musk ambrette Chlorothiazide	0.1 5.0 1.0 5.0 1.0
Balsam of Peru Compositae mix p-Aminobenzoic acid 2-Hydroxyl-4-methoxybenzophenone 3-(4-Methylbenzylidene)camphor	25.0 6.5 10.0 10.0 10.0
4-Isopropyldibenzoylmethane 4-tert-Butyl-4'-methoxydibenzoylmethane^a 2-Ethylhexyl p-dimethylaminobenzoate^a 2-Ethylhexyl p-methoxycinnamate^a Phenylbenzimidazolesulphonic acid^a Isoamyl p-methoxycinnamate^a	10.0 10.0 10.0 10.0 10.0 10.0

^aAdditionally integrated in the photopatch set of test chemicals of study part II

According to the standard procedure of photopatch testing used in the first test period (Table II), it was the aim of this multicentre study to apply substances from the modified second test set to a large group of photosensitive patients (10, 11). After evaluation of the data from the second test period (1991-1997), the results of the first and second periods of testing were compared (11). In this process, all the data were subjected to computer-assisted analysis, in order to classify all positive test reactions either as contact (non-specific, toxic) or photoinduced reactions. The photoinduced reactions were further defined as phototoxic or photoallergic, and a substance-specific reaction pattern was determined for each chemical tested (13).

Table II: Photopatch testing procedure

Test materials applied via Finn Chambers to the back for 24 hours

Irradiation with 10 J/cm² UVA (320-400 nm)

Reading immediately, and 24, 48 and 72 hours later

Controls: non-irradiated patch test, UVA irradiation of normal back skin

In the first test period, data from 1129 patients were evaluated. Of 2859 positive test reactions, 28.6% were excluded as plain contact reactions, 71.3% were found to be photoinduced, 3.8% of which were classified as photoallergic. In the second test period, data from 1261 patients were evaluated. 1415 positive test reactions were recorded, 28.7% of these were excluded from further analysis, since they were due to plain contact reactions. The remaining 71.3% of the skin reactions were classified as photoinduced, 8.1% of which were photoallergic reactions. From the results of the two periods of the central European photopatch testing project, the major photoallergens identified belonged to the following groups of chemicals: non-steroidal anti-inflammatory drugs (NSAIDs), disinfectants, phenothiazines and sunscreens (Table III).

Table III: Rank order of phototoxic reactions in test period II, 1991-1997

Test chemical	Frequency (%)
Chlorpromazine Promethazine Carprofen Fenticlor Hexachlorophene	5.63 5.15 5.15 2.93 2.54
Balsam of Peru Triclosan 2-Hydroxy-4-methoxybenzophenone 4-Isopropyldibenzoylmethane Buclosamide	2.06 1.74 1.35 1.35 1.11
Musk ambrette 2-Ethylhexyl p-methoxycinnamate Bithionol 3-(4-Methylbenzylidene) camphor 2-Ethylhexyl p-dimethylaminobenzoate	1.11 1.03 0.95 0.95 0.87
Tribromosalicylanilide Sulphanilamide Quinidine p-Aminobenzoic acid Tetrachlorosalicylanilide	0.79 0.79 0.71 0.71 0.63
Chlorothiazide Compositae mix Isoamyl p-methoxycinnamate 4-tert-Butyl-4'-methoxydibenzoylmethane Phenylbenzimidazolesulphonic acid Monobromosalicylchloranilide	0.63 0.56 0.56 0.40 0.40 0.24

Photoinduced reactions

All of the substances tested in the two periods of the project induced photoreactions. However, their frequency varied, i.e. some provoked phototoxic reactions, others induced photoallergic reactions, and the reactions of a considerable number could not clearly be assigned to either of these two groups.

Photoirritation
All of the test substances induced phototoxic reactions in both periods of the study at different levels of intensity. In test period I, tiaprofenic acid induced phototoxic reactions most frequently, followed by promethazine, carprofen, chlorpromazine and hexachlorophene. Since tiaprofenic acid was excluded from the second period of the study, chlorpromazine induced phototoxic reactions most frequently in period II, followed by promethazine, carprofen, fentichlor and hexachlorophene (Table III).

Photoallergic reactions
Photoallergic reactions were induced significantly less frequently in comparison to phototoxic reactions, in both periods of the study. Photoallergic reactions were observed in 3.8% of the patch tests in period I, and in 8.1% in period II. In period I, tiaprofenic acid headed the list of compounds inducing photoallergic reactions, followed by fentichlor, carprofen, 4-isopropyldibenzoylmethane, and 2-hydroxy-4-methoxybenzophenone. After excluding tiaprofenic acid from period II testing, fentichlor, carprofen, chlorpromazine, 2-hydroxy-4-methoxybenzophenone and 4-isopropyldibenzoylmethane were identified as the leading photoallergens (Table IV).

Unclassified reactions
In period I, reactions that could not clearly be classified as either phototoxic or photoallergic were induced by tiaprofenic acid, promethazine, fentichlor, carprofen and chlorpromazine. In test period II, the reactions were produced by fentichlor, chlorpromazine, carprofen, promethazine and 2-hydroxy-4-methoxybenzophenone

Table IV: Rank order of chemicals by photoallergic reaction in test period II, 1991-1997

Test chemicals	Frequency
Fenticlor Carprofen Chlorpromazine 2-Hydroxy-4-methoxybenzophenone 4-Isopropyldibenzoylmethane	1.82 1.19 0.95 0.63 0.56
Tetrachlorosalicylanilide Promethazine Isoamyl p-methoxycinnamate Tribromosalicylanilide Bithionol	0.40 0.40 0.40 0.32 0.32
Balsam of Peru Monobromosalicylchloranilide Buclosamide Compositae mix p-Aminobenzoic acid	0.32 0.24 0.24 0.24 0.24
4-tert-Butyl-4'-methoxydibenzoylmethane 2-Ethylhexyl p-methoxycinnamate Sulphanilamide Quinidine Musk ambrette	0.16 0.16 0.16 0.08 0.08
3-(4-Methylbenzylidene)camphor Phenylbenzimidazolesulphonic acid Hexachlorophene Triclosan Chlorothiazide 2-Ethylhexyl p-dimethylaminobenzoate	0.08 0.08 0.00 0.00 0.00 0.00

Analysis of photopatch test reactions according to specificity of reaction pattern

The cumulative data of all the photoinduced skin reactions were analysed according to the changes in reaction pattern over time, from graphs depicting intensity of the skin reaction for all of the test substance of periods I and II. The graphic presentation of these data revealed four major patterns of reaction (13).

The decrescendo pattern, peaking immediately after irradiation and decreasing in intensity thereafter. This pattern is characterised by erythema or by a combination of erythema and infiltration, sometimes accompanied by a few papulovesicles.

A combination of the decrescendo and crescendo patterns. The immediate response, which decreased in intensity (decrescendo), is characterised by erythema with minor infiltration. In addition to these signs of inflammation, the late reaction is characterised by a few papulovesicles.

The plateau reaction pattern is characterised by erythema and infiltration, which do not change in intensity over time.

The delayed or crescendo pattern peaks 48 hours after irradiation, and is characterised by papulovesicles in addition to erythema and infiltration.

The decrescendo pattern
In period I of the study, a classic decrescendo pattern was elicited by quinidine. This photoreaction is typically limited to erythema in a distinctive decrescendo shape. The reaction induced by quinidine is a typical example of a purely phototoxic reaction of immediate onset (Figure 3). Interestingly, most of the compounds that are phototoxic after systemic administration, and also plant extracts and derivatives, induced a photoreaction of the decrescendo pattern.

Figure 3

The combined reaction pattern
The time-course of the photoreaction induced by tetrachlorosalicylanilide (Figure 4) is a typical example of the reaction pattern of the second category, i.e. an immediate reaction of the decrescendo type overtaken by a delayed crescendo reaction. Most of the halogenated salicylanilides, disinfectants, p-aminobenzoic acid, musk ambrette, and also fragrance mixtures, provoked the combined reaction pattern.

Figure 4: Photoreactions induced by tetrachlorosalicylanilide in test period I: a typical example of a combined patter

The plateau pattern
As reported for period I, tiaprofenic acid induced photoreactions typical of the plateau pattern, which is characterised by erythema and infiltration and a few papulovesicles between day 2 and day 4 ( Figure 5). Carprofen, promethazine, chlorpromazine and fentichlor induce reactions of this type. The majority of reactions could not be classified properly, although they seem to be induced by prolonged phototoxic reactions.

Figure 5: Photoreactions induced by tiaprofenic acid in test period I: A typical example of a plateau pattern

The crescendo pattern
This category includes the classic photoallergens, for example fentichlor and sunscreens. The photoreactions of isoamyl p-methoxycinnamate shown in Figure 6 are typical for the delayed crescendo pattern. This pattern is compatible with a classic type IV reaction.

Figure 6: Photoreactions induced by isoamyl p-methoxycinnamate in test period I: A typical example of a crescendo (photoallergic) pattern:

Evaluation of the results from human photopatch testing

The high frequency of test reactions that could not be properly assigned to a specific photoreaction pattern in period I, was markedly reduced after the introduction of specific modifications in period II. In particular, the exclusion of tiaprofenic acid and the reduction of the test concentration of promethazine decreased the number of nonspecific positive test reactions, thereby increasing the specificity. For example, in test period I, of a total of 2041 photoreactions, 3.8% were classified as photoallergic reactions, in comparison to 8.1% out of 1009 in period II. Although tiaprofenic acid, the leading photoallergen of period, had been omitted from the set of test substances, the percentage of photoallergic reactions was almost twice as high in period II as in period I, i.e. 8.1% versus 3.8%. The increase in photoallergic reaction in period II resulted predominantly from a significant decrease in non-photoallergic test reactions, and from an increased induction of photoallergic reactions by the UV-filter chemicals (sunscreens; Table I), which were added to the set of test substances in period II. In period I, 17.5% of the photoallergic reactions were induced by the sunscreens, 2-hydroxy-4-methoxybenzophenone, 4-isopropyldibenzoylmethane and p-aminobenzoic acid. The addition of UV-filters in test period II led to a significant increase in photoallergic reactions (25.5%). Thus, in period II, specific modifications in the evaluation of the photoreactions resulted in a higher specificity of the photopatch testing procedure.

In period I, tiaprofenic acid was the leading photoallergen, followed by fentichlor, carprofen, 2-hydroxy-4-methoxybenzophenone and 4-isopropyldibenzoylmethane. In period II, fentichlor, carprofen, chlorpromazine, and the sunscreens 2-hydroxy-4-methoxybenzophenone and 4-isopropyldibenzoylmethane, were the leading photoallergens (Table IV). In summary, virtually the same chemicals revealed a high photoallergic potential in both periods of the study. This result confirms the high reproducibility of the photopatch testing procedure.

The reproducibility of the two photoreaction patterns in the two test periods, and the persistence of the plateau pattern of promethazine, even after a reduction in the test concentration, demonstrates the specificity of the reaction pattern for each test chemical (5). In addition to the well-known phototoxic (category 1) and photoallergic (category 4) pattern, computer-assisted analysis of the specificity of the reaction pattern over time for each test chemical revealed a combined pattern (category 2) and the plateau pattern (category 3). This additional analysis facilitated the classification of individual photoreactions in one of the four photoreaction patterns.

The plateau pattern and the combined pattern give more insight into clinical photoreactions, which previously could not be classified. Although the mechanisms underlying plateau reactions remain to be determined, the analysis of every single individual (plateau) test reaction suggests that they most probably represent a delayed phototoxic reaction.

Evaluation of the results from human photopatch testing

It should be noted that photopatch tests are merely indicative of possible reaction mechanisms, i.e. phototoxicity and photoallergy. The underlying mechanism must be confirmed by experimental laboratory studies.

Since many chemicals are capable of inducing both photoallergic and phototoxic effects, a complex photoreaction pattern resulting from an immediate (phototoxic) and a delayed (photoallergic) reaction may be observed. Such a combined pattern was, in fact, elicited by tetrachlorosalicylanilide. Consequently, a few clinical photoreactions, which hitherto could not be classified, could now be interpreted as photoallergic reactions superimposed by immediate phototoxic reactions. The two newly added photoreaction patterns (plateau and combined) seem to hold promise for improving the specificity of the photopatch test procedure, and, in particular, for a better definition of photoallergic reactions.

An updated set of test chemicals for the photopatch test (Table V) should contain photosensitisers that were of significance in the past and which will remain relevant today and in the future (14). Thus, a standardised set of test chemicals for photopatch testing should be continually updated, according to the results of testing and to clinical and/or experimental experience.

Table V. The photopatch test chemicals: test period III

Test chemical	Concentration
Standard set
Tetrachlorosalicylanilide Monobromosalicylchloranilide Hexachlorophene Bithionol Sulphanilamide	0.1 1.0 1.0 1.0 5.0
Promethazine Quinidine Musk ambrette Perfume mix p-Aminobenzoic acid	0.1 1.0 5.0 8.0 10.0
Ethylhexyl p-dimethylaminobenzoate 1-(4-Isopropylphenyl)-3-phenyl-1,3-propanedione 4-tert-Butyl-4-methoxydibenzoylmethane Isoamyl p-methoxycinnamate 2-Ethylhexyl p-methoxycinnamate	10.0 10.0 10.0 10.0 10.0
3-(4-Methylbenzylidene)camphor Phenylbenzimidazolesulphonic acid Oxybenzone Sulisobenzone	10.0 10.0 10.0 10.0
Special set
Tribosalan Chlorpromazine Thiourea Olaquindox	1.0 1.0 0.1 1.0

Acute Phototoxicity Testing

The 3T3 NRU phototoxicity (= photocytotoxicity) test

Background and results of the validation study
To validate the 3T3 NRU phototoxicity test, three studies were conducted, all of which were aimed at different goals and started with different levels of information. This must be kept in mind, in order to appreciate how the studies were conducted. A major problem in the field of phototoxicity arises from the fact that the number of test chemicals backed by high-quality in vivo data in either animals or humans is very limited. Moreover, when the prevalidation study was started in 1992, the 3T3 NRU phototoxicity test had just been developed without a prediction model, and the concept of a biostatistically based prediction model had not yet been put forward. It was suggested for the first time as an essential element of experimental validation at the second Amden (ECVAM/European Research Group for Alternatives in Toxicity Testing [ERGATTI] validation workshop in 1994 (15). Later, the Interagency Coordinating Committee on Validation of Alternative Methods (ICCVAM) and the Organisation for Economic Cooperation and Development (OECD) accepted that a validation study must include an evaluation of the reliability and relevance of the prediction model to be used in a new test for a specific purpose (16).

In contrast to all the previous OECD Test Guidelines, the toxicity test described in this guideline is aimed at correctly predicting the phototoxic potential in relation to clinical human data. Thus, in addition to describing how to conduct the method appropriately, a prediction model is included as an essential part of the method. Taking into account the results of several experimental validation trials, the prediction model allows the phototoxic potential of new test chemicals in humans to be predicted with a very high accuracy. None of the toxicity tests that have been accepted into the OECD guidelines so far have contained a biostatistically based prediction model that allows toxic potential in humans to be predicted.

Prevalidation study 1992-1994
Before this study was started, a COLIPA task force had identified a list of 20 test chemicals, 11 of which were phototoxic (+pt) and 9 non-phototoxic (-pt) in animal studies and/or in humans. It was the aim of the study to identify in vitro phototoxicity tests that would correctly predict the phototoxic potential of test chemicals. In order to reduce technical problems at this early stage, the majority of the chemicals chosen for test development and prevalidation were soluble in water/culture medium. The prevalidation study was not conducted as a blind study with coded chemicals (17).

The 3T3 NRU phototoxicity test study was conducted in eight laboratories and, from the 158 data points obtained, and using discriminant analysis, the photoirritation factor (PIF) prediction model was developed to discriminate between +pt and -pt chemicals. A cutoff value of PIF 5 resulted from the statistical analysis of the set of test chemicals used in the prevalidation study. Applying this prediction model, all of the test chemicals were correctly classified in the 3T3 NRU phototoxicity test.

Identical results were obtained in a study conducted independently of the EU/COLIPA study at the Hatano Research Institute, in Japan, in 1994, with the 3T3 NRU phototoxicity test according to the same test protocol, including the PIF prediction model and with the same set of chemicals (18).

Formal validation study 1994-1996
It became quite obvious during the prevalidation study that the quality of the available animal data was poor. To overcome this problem, in 1993, clinical experts were invited to the first ECVAM workshop on in vitro phototoxicity testing, in order to identify a set of test chemicals backed by high-quality in vivo data. At this ECVAM workshop, a list of high-quality data from standardised human photopatch testing was made available, for both acute phototoxicity and photoallergy. The list of these two groups of chemicals, mostly drugs, including the incidence of photochemical side-effects, was published in the ECVAM workshop report (2). A major result of the ECVAM workshop was the compilation of a "list of test chemicals with sufficient human data", which covered all classes of currently known phototoxic drugs and chemicals. The list of test chemicals was the same as that shown in Table VI, which relates to the red blood-cell phototoxicity (Photo-RBC) test.

Table VI: Results of the combined Photo-RBC test in phase II of the EU/COLIPA validation project on photoirritation in vitro

No.	Chemical	*In vivo*	Assessment
No.	Chemical	*In vivo*	PHF	(delta)OD_max	Final
1.	2-Hydroxy-4-methoxybenzophenone	-	-	-	-
10.	Chlorhexidine dihydrochloride	-	-	-	-
15.	Hexachlorophene	-	-	+	+
17.	Sodium lauryl sulphate	-	-	-	-
24.	p-Aminobenzoic acid (PABA)	-	-	+	+
25.	Penicillin G	-	-	-	-
14.	Furosemide	+?	-	-	-
23.	Ofloxacin	+	-	-	-
2.	5-Methoxypsoralene (5-MOP)	+	-	-	-
3.	6-Methylcoumarin	+	+/-	+	+
4.	Acridine hydrochloride	+	+/-	+	+
5.	Acridine free base	+	+	+	+
6.	Amiodarone	+	+/-	-	+/-
7.	Anthracene	+	+/-	+	+
8.	Bergamot oil	+	-	+	+
9.	Bithionol	+	+/-	+	+
11.	Chlorpromazine	+	+/-	+	+
12.	Demeclocycline	+	-	+	+
13.	Fenofibrate	+	+	+	+
16.	Ketoprofen	+	+	+	+
18.	Musk ambrette	+	+	+	+
19.	Nalidixic acid sodium salt	+	+/-	+	+
20.	Nalidixic acid free acid	+	-	+	+
21.	Neutral red	+	+	+	+
22.	Norfloxacin	+	-	+	+
26.	Promethazine	+	-	+	+
27.	Protoporphyrin IX free acid	+	+	+	+
28.	Protoporphyrin IX disodium	+	+	-	+
29.	Rose bengal	+	+	+	+
31.	Tiaprofenic acid	+	+	+	+

Data from Pape et al. (30).
PHF = photohaemolysis factor, (delta)OD_max = optical density.
+ = positive, - = negative, +/- = equivocal results.

Taking this information into account, a formal validation study on the 3T3 NRU phototoxicity test was conducted under the sponsorship of the European Commission and COLIPA, as recommended by ECVAM. It was the goal of the study to assess whether the 3T3 NRU phototoxicity test could correctly predict the phototoxic potentials of test chemicals in humans. Therefore, an unbalanced set of test chemicals was chosen, with 5 -pt and 25 +pt chemicals. In order to assess problems of bio availability, three chemicals were used as both water-soluble salts and insoluble free acids or free bases (acridine-hydrochloride and acridine-free base; nalidixic acid-free acid and nalidixic acid-sodium salt; protoporphyrine IX-free acid and protoporphyrin IX-disodium salt). It was, of course, an essential goal of this part of the study to confirm and/or improve the PIF prediction model with a new set of chemicals tested under blind conditions.

An interesting result of the study was that the 3T3 NRU phototoxicity test was able to predict correctly the phototoxic potential of test chemicals in humans, irrespective of their solubility (3). This proves that solubility is not, but bioavailability is, a relevant criterion in identifying the phototoxic potential of a chemical correctly, and this aspect was carefully covered by the process laid down in the test protocol. However, since this study was conducted under blind conditions and since insufficient guidance was given on appropriate solvents and limit concentrations of solubility, in many instances the chemicals were not tested under optimal conditions.

UV-filter chemicals study 1997-1998
At the request of the Scientific Committee of Cosmetology and Non-Food Products (SCCNFP), the expert advisory committee on cosmetics, which reports to the Health and Consumer Protection Directorate General of the European Commission, a set of the most commonly used UV-filter chemicals, which are not phototoxic in vivo and, in most cases, are poorly soluble in water, was tested in a blind trial with the 3T3 NRU phototoxicity test. To obtain information on the optimum test concentration and on the predictivity of the 3T3 NRU phototoxicity test, an equivalent set of proven phototoxic chemicals was tested. To avoid the inappropriate use of solvents, information was provided with each of the coded test chemicals on both the most appropriate solvent and the solubility limits in that solvent.

The results of the study confirm that the 3T3 NRU phototoxicity test is able to handle test chemicals irrespective of their solubility (4). A thorough biostatistical analysis showed that no false-positive results were obtained up to 100 µg/ml, whereas some false-positive results were obtained at higher concentrations. The report also shows that all of the -pt UV-filter chemicals were tested at least up to the highest test concentrations recommended after the independent assessment of solubility.

Taking into account that some test concentrations reported for the formal validation study were nominal rather than actual concentrations, they must not be compared to concentrations at which the same chemical was used in the UV-filter study. Thus, the only data from which a reliable recommendation for maximum test concentrations can be derived were those produced in the UV-filter study, which shows that false-positive results may be obtained at concentrations of more than 100 µg/ml.

The results from the UV-filter study clearly demonstrate that the 3T3 NRU phototoxicity test does not provide a yes or no answer, but that the result is clearly dependent on the test concentration applied, with an increasing risk of false-positive results at high test concentrations. Thus, expert judgment is required when data from the 3T3 NRU phototoxicity test are used for regulatory purposes.

It must be stressed that, among the toxicity tests currently available, the 3T3 NRU phototoxicity test has a unique position, since it is the only toxicity test that predicts the situation in humans, due to the availability of high-quality human in vivo photopatch test data, against which it has been experimentally validated. We are not aware of any other toxicity test that has been accepted for regulatory purposes, which meets this important criterion.

Current use, technical improvements and regulatory acceptance
Early in the year 2000, the 3T3 NRU phototoxicity test was officially accepted by the European Commission and the EU Member States into Annex V of EU Council Directive 67/548/EEC for the classification and labeling of hazardous chemicals (19). In accordance with EU Council Directive 86/609/EEC (1), which regulates the use of experimental animals, the 3T3 NRU phototoxicity test must now be used to determine the phototoxic potential of chemicals, and animal tests are prohibited for this purpose in all EU Member States.

The protocol can be used successfully with human keratinocytes, as demonstrated in a study conducted under blind conditions with the chemicals of phase 11 of the EU/COLIPA validation study and of the UV-filter study (20).

Other tests

Human 3-D skin models in phototoxicity testing
Reconstituted human skin models, available commercially or in a few experienced laboratories, are of three different types: dermal models (containing skin fibroblasts), epidermal models (containing skin keratinocytes and a stratum corneum), and full skin models (containing fibroblasts, keratinocytes and a stratum corneum). Since the last two types contain viable, metabolising primary skin cells and a skin barrier, both are frequently referred to as "3-D skin models". Human skin models have been used successfully in routine laboratory investigations, since they are relevant to the organ of interest. For in vitro toxicity testing, standardisation and control of 3-D skin models need to be defined clearly, in order to ensure that reliable and reproducible data are obtained.

In contrast to commonly used cell cultures, such as 3T3 mouse fibroblasts, human skin models permit the topical application of various types of chemicals and preparations, and seem to have fewer limitations relating to solubility problems. In 3-D skin models, test materials can be applied undiluted, even at extreme pH values or as "insoluble" substances, as shown, for instance, by the successful use of such models for testing corrosives.

The first promising data obtained with skin model phototoxicity tests were reported in 1994-1995, with a full skin model (21, 22) and an epidermal model (23). The results obtained with the full skin model were confirmed in 1997 (24). Since the commercial production of the full skin model, Skin^2TM ended in 1996, the test protocol was successfully adapted to the use of the epidermal model, EpiDerm™ (25), and later was evaluated in an ECVAM prevalidation study, which revealed promising results from three laboratories (26). The test is currently established in several laboratories of the European cosmetics industry (27), and has been successfully adapted for the epidermal model, SKINETHIC™ (28). Efforts undertaken to optimise the phototoxicity test protocol and the prediction model, when transferring them from the full skin model to the epidermal model (25), revealed that the basic test protocol and prediction model did not need to be changed. Several studies (22, 24, 25) reported that in vivo photoallergens that are not simultaneously acute photoirritants (for example, coumarin, 6-methylcoumarin, musk ambrette), are classified as negative by the skin model phototoxicity tests. Finally, dermal models, which do not contain a skin barrier, show a sensitivity to phototoxic chemicals that is similar to that of photocytotoxicity tests, such as the 3T3 NRU phototoxicity test (29). Therefore, they do not provide any advantage in a phototoxicity testing strategy.

Assuming that the 3-D skin models comply with the best available laboratory and scientific standards, they will offer the following advantages in comparison to the 3T3 NRU phototoxicity test.

Pure chemicals or complex mixtures can be applied in a way that simulates the application of preparations topically to the skin.
The test concentrations applied are closer to real exposure conditions, including those used in dermatological patch testing.
Histology can be performed on exposed and control samples.
Exposure to light can be better adapted to real-life situations; for example, exposure time and spectrum of simulated sunlight (a higher dose of short-wave light in the U-VB range).
Depending on the barrier function of the stratum corneum, adsorption and penetration of the original chemicals or molecules created during the exposure of skin models will provide more-relevant results than tests performed on simpler models (fewer false positives).

The following disadvantages also have to be taken into account.

The number of commercial suppliers of such models is limited (for example, MatTek™, SKINETHIC™, EPISKIN™, CellSystems™).
Lack of skin appendices, such as hair follicles or sebum and sweat glands, which may be sensitive areas in vivo, could be of relevance, although there is no experimental evidence so far that this is actually a drawback.
Until now, there has been no convincing evidence that the photoallergic effects of chemicals can be predicted in 3-D skin models.
Since 3-D skin models are still very expensive--currently about 25-90 Euro per sample--they are not yet suitable for routine large-scale testing.

In conclusion, during the EU/COLIPA validation project, it was shown that the 3T3 NRU phototoxicity test can correctly predict the phototoxic potentials of chemicals by using a monolayer cell culture. Since this may not be relevant when chemicals are applied topically to the skin at lower concentrations in finished products, there was a need for test method development to permit the application of complex formulations directly to reconstructed skin models.

In the ECVAM prevalidation study on the EpiDerm phototoxicity test, an appropriate test protocol was developed, successfully leading to reliable test results in three laboratories for ten test chemicals (26). From the results obtained in this project, there is some evidence that 3-D skin models can be used for the assessment of the potency of a phototoxin applied topically to the skin. Therefore, it can be concluded that 3-D skin models can play an essential role as adjuncts in a testing strategy for the sequential evaluation of phototoxicity.

The combined Photo-RBC test
In the EU/COLIPA validation programme on "photoirritation in vitro", two core tests and a number of mechanistically based tests were performed, to examine their suitability as regulatory tests for phototoxicity testing. In addition to the 3T3 NRU phototoxicity test, as second core test, the Photo-RBC test was evaluated in a prevalidation study within the validation programme.

In the protocol for the Photo-RBC test, two endpoints are determined in erythrocytes, namely, photohaemolysis and met-haemoglobin (met-Hb) formation. The endpoints are assessed by measuring changes in optical density at 525 nm for photohaemolysis and at 630 nm for met-Hb formation. Furthermore, a prediction model was used in the Standard Operating Procedure (SOP), taking into account two cutoff values: the photohaemolysis-factor (PHF) > 3.0 for photo haemolysis, and optical density (OD) (delta)OD_max > 0.05 for met-Hb formation. Three laboratories agreed on the SOP, and the 30 chemicals (25 phototoxic and five non-phototoxic chemicals) that had been selected for the validation study (3) were tested under blind conditions.

The results of the prevalidation study on the Photo-RBC suggested that the transfer of the SOP between laboratories had not been very efficient, and that additional training of the laboratories would be helpful. Most importantly, the results obtained in the lead laboratory (Table VI) clearly indicate that an experienced laboratory will obtain correct classifications for both phototoxic and non-phototoxic test chemicals, when the SOP is applied correctly. The biostatistical analysis shows a good overall in vitro/in vivo correlation of the results, including an acceptable accuracy (83.3%), sensitivity (87.5%), and positive predictivity (91.3%; 30). The low specificity (66.7%) and low negative predictivity (57.1%) most probably result from the unbalanced number of positive and negative test chemicals, in particular, from the low number of non-phototoxic substances in the set of test chemicals.

A test protocol was developed to identify photosensitisers and to study phototoxic potential of light-activated chemicals from their ability to lyse freshly isolated erythrocytes and/or to oxidise oxyhaemoglobin (oxy-Hb) after exposure to simulated sunlight comprising UVB, UVA and visible light (31, INVITTOX protocol number 81¹). In contrast to the established photohaemolysis test (32, 33), the new protocol included photochemically induced met-Hb formation (31, 34, 35) as the second endpoint, since met-Hb may be produced during extensive exposure to sunlight, especially as a result of type I photodynamic reactions (36). Furthermore, in contrast to keratinocytes or fibroblasts, that are used in the 3T3 NRU phototoxicity test protocol, mammalian erythrocytes are readily available. Moreover, due to their specific cellular defence mechanisms, erythrocytes can be exposed to more-intensive UV irradiation, even to UVB, than can cells from other tissue. Finally, since erythrocytes do not contain a nucleus, they are not susceptible to photogenotoxic effects.

Thus, results from the combined Photo-RBC test can be considered as interesting additional information within an overall photo-safety testing strategy, when in vitro methods are the initial step of testing. In addition, the Photo-RBC test provides mechanistic information on two different types of photodynamic reactions (met-Hb formation for type I reactions, and photohaemolytic effects as primary type II reactions). As summarised in Table VI, when the results obtained with both endpoints in the Photo-RBC test in the final assessment are combined, a good overall fit with the in vivo evaluations is obtained.

The three false-negative results can be explained in the following manner (31). For furosemide, a relevant phototoxic potency in vivo is not proven, since the clinical data in humans are insufficient (3). Ofloxacin was tested as a clinical preparation at a very low concentration of the active ingredient, which may not reveal its phototoxic potential. 5-MOP may have provided a negative result due to the lack of sensitive target molecules in erythrocytes.

False-positive results were obtained with PABA and hexachlorophene in the PhotoRBC test. PABA is easily oxidised by oxy-Hb in the dark in a pseudo-enzymic reaction, which may be enhanced under light exposure (or increased temperature).

In summary, as a general result of the prevalidation study according to the SOP, the combined Photo-RBC test can be performed reproducibly, and it provides relevant mechanistic information on photodynamic reactions, which add important information for the evaluation of the photo-safety of test chemicals, within a wider testing strategy starting with the 3T3 NRU phototoxicity test, which does not provide information on photodynamic reactions. An additional advantage of RBC cells is their resistance to the short-wave UVB part of sunlight, which enables RBC cells in the Photo-RBC test to be exposed to the entire solar spectrum for prolonged periods.

The yeast assay
The use of mammalian cells and bacteria in phototoxicity and photogenotoxicity testing may be limited by the high sensitivity of these organisms to UV light and, in particular, to UVB. The facultative anaerobic yeast Saccharomyces cerevisiae, a eukaryotic organism, has the advantage of being relatively insensitive to exposure to sunlight or to prolonged exposure to drugs in aqueous solution, in final galenic forms and ointments, even when oxygen tension is low (as in the presence of an oil or cream). Thus, when compared to mammalian cells and other testing organisms, S. cerevisiae offers some advantages for testing the phototoxic potentials of chemicals.

In this organism, phototoxicity can easily and cheaply be tested qualitatively and quantitatively, directly on plates with complete growth medium or after plating of exposed cell suspensions, by measuring the density of the outgrowing cells or colony forming ability (clonogenic survival; 37).

When using yeast, DNA damage can be demonstrated by comparing the responses of DNA repair-deficient strains with those of DNA repair-competent strains (37). In this way, even the specific mechanism of photo-induced lethality can be assessed, since mutant strains are available in all known DNA repair pathways (38). The use of such mutants can significantly improve the sensitivity of the test. Furthermore, photomutagenicity and mitotic recombination can readily be assessed (37, 39).

With regard to testing for the photosensitising or photoprotective potentials of medical, cosmetic or pharmaceutical compounds, S. cerevisiae offers the advantage that it can withstand quite long exposures to UVB, UVA, solar simulated or solar radiation (40-42). It is thus useful for testing reciprocity between light exposure (fluence) and drug concentration, and dose-rate effects, and for evaluating the photocytotoxic and photomutagenic potentials of photosensitising or photoprotecting agents under conditions close to clinical use in patients.

In general, the AMES test with Salmonella typhimurium is more sensitive than a yeast assay for genotoxicity testing even though the latter assay may also provide useful information (43-45). Using the yeast assay, Averbeck (46) and Morlié et al. (40) were among the first to clearly demonstrate the specific properties of some bifunctional psoralens, for example, 8-methoxypsoralen (8-MOP) or 5-methoxypsoralen (5-MOP), to induce genetic damage (mutations and mitotic recombination) in the presence of UVA or simulated solar radiation (40, 41, 46). These observations were confirmed in cultured mammalian and human cells (47). Since the exposure to doses of UV and UVA radiation could be increased in comparison to those used for bacterial and mammalian cells, it has been possible to test combinations of the psoralens with UV-filters by using conditions which mimic human exposure (48). For example, the photoprotective effects of solar filters against the induction of genotoxic effects by 5-MOP could be demonstrated (41, 49). Among other assays, yeast was recently chosen for use within a general strategy for photogenotoxicity testing (50-52). Interestingly, the photomutagenic effects of the fluoroquinolones were detected in this system (51).

In order to examine the effects of metabolism of neat compounds, mixed or in a near-final galenic form, the photomutagenic and photogenotoxic capacities of suction blister fluids derived from human skin after application of the test compound to the skin of volunteers were determined in yeast. It was possible to detect the active genotoxic principle bergamot oil in the interstitial liquid of human skin (42). This is of interest, since the immunosuppressive effects of exposure to sunlight, and photocarcinogenic responses, imply the activation of genes (53, 54).

In summary, in using the yeast assay, it has to be noted that:

In this assay, not only can the neat compound be tested, but also its final galenic form combined with other compounds. Ethanolic or oily solutions of test compounds can be spread onto yeast plated on solid growth medium, before exposure to UV radiation, in order to test for phototoxicity and/or photogenotoxicity; and
The yeast assays can be performed under conditions (for example, prolonged UV exposures) that are relatively close to actual or expected human exposures, but would be too phototoxic or too photogenotoxic for bacteria and human cells in vitro.

Photoallergy

In photodermatology and immunotoxicological risk assessment, the precise predictive differentiation of photoallergenic from acute phototoxic/photoirritation reactions induced by low-molecular weight compounds, is very important. Although clinical criteria in the examination of skin lesions have been, and continue to be, of proven value, they are not without their limitations. The animal models used so far for the screening of photoallergic properties of compounds are very heterogenous and time-consuming. To achieve proper risk assessment, the animal models currently in use have to be replaced, with the aim of improving hazard identification by predicting photoreactions in humans more precisely.

Human and experimental evidence for photoallergy

The major obstacle in clinical and experimental practice is the difficulty of determining the nature of photoreactivity, i.e. of photoallergic or phototoxic reactions. There are significant human data in the literature about photoreactive compounds. However, the classification of the photoallergic and phototoxic proper-ties of chemicals is mostly based on clinical parameters, which are not always sufficient. The major difference between photoirritancy and photoallergy is the activation of (photo)antigen-specific immunocompetent cells. A prerequisite for the induction of photoallergic reactions is the presentation of the modified (photo)antigen to specific T-cells. Taking this mechanism of photoallergy into consideration, the involvement of specifically reacting T-lymphocytes or antibodies could be one tool for use in discriminating between both photoreactions. One in vitro assay that could be modified for such an approach is the so-called lymphocyte transformation test (LTT; 55). However, the LTT is so far a retrospective assay, which needs lymphocytes from sensitised patients.

For pre-clinical screening purposes, the most promising test is the modified local lymph node assay (UV-LLNA or UV-IMDS [IMDS = integrated model for the differentiation of skin reactions]) which takes into account the immunological mechanism of the reaction (56-58).

The following list results from studies providing clear evidence for the mechanism of the hazardous potential of photoallergy in humans, guinea-pigs and mice. A summary is given in Table VII.

Table VII: Proposed standards for the validation of the integrated model for the differentiation of skin reactions test

Class of compounds	Patch test	T-cell	Animal data			Chemical mechanism
Class of compounds	Patch test	T-cell	Guinea-pig	PMT		Chemical mechanism
Halogenated salicylanilides (for example, TCSA, TBSA)	+	+	+	+	+	+ PA
Halogenated thiolic phenols (for example, bithionol, fentichlor)	+	-	+	+	+	? PT, PA
Aromatic nitro (for example, musk ambrette)	+	-	+	+	neg.	+ PA
Coumarins (for example, 6-methylcoumarin)	+	?	(+) -	+ ?	+	- (PT), PA
Benzophenones (for example, ketoprofen, thiaprofen, PT PA carprofen	+	-	+ -	+	-	+ PT, PA
Aromatic amines (for example, sulphonamides, PABA)	+	+	+	?	+	+ (PT), PA
Aromatic nitro-oxides (for example, omadine, quindoxin)	+	-	+	+	+	? PA
Tetracyclines (for example, doxycycline, oxytetracycline, chlortetracycline, tetracycline)	+	+	?	?	-	+ ?
Phenothiazines (for example, chlorpromazine, promethazine)	+	?	+	+	+ -	+ PT, PA
Fluoroquinolones (for example, lomefloxacin, sparfloxacin)	+^a	-	+^a	+^a	(+)^a PT, (PA)	-

PT = phototoxic reaction (photoirritation), PA = photoallergic reaction.
^aOnly after oral/intravenous application; no reaction after topical application. + = Positive, - = negative, ? = equivocal results.
TCSA = 3,5,3',4'-tetrachlorosalicylanilide, TBSA = 3,5,4'-tribromosalicylanilide, PMT = photomaximisation test.

Halogenated salicylanilides
3,5,3',4'-Tetrachlorosalicylanilide (TCSA): experimental photoallergy has been reported in humans (59), guinea-pigs (for example, 60-64) and mice (56, 65-68).

3,5,4'-Tribromosalicylanilide (TBS): photoallergic reactions to TBS were complicated by the purity of the sample tested. The report of Kaidbey & Kligman (59) accords with marketing experience. They found no reactions to a pure sample of TBS, but readily produced reactions to dibromosalicylanilides, and concluded that the photosensitising potential of TBS was due to the presence of dibrominated contaminants.

Thiobisphenols
Bithionol: in experimental studies, phototoxic reactions were reported in 8 of 18 persons after using 5% bithionol with 28.5 J/cm² UVA (69). Photoallergy was elicited in 3 of 25 subjects after using 1% bithionol with 4 J/cm² UVA in the photomaximisation procedure (59). Photoallergic reactions have been reported in guinea-pigs (61, 63, 70) and mice (66, 67).

Fentichlor: photoallergy in mice was reported by Gerberick & Ryan (67).

Aromatic nitro-compounds
Musk ambrette: in the photomaximisation test in volunteers, musk ambrette was negative (59), which led to a recommendation that the guinea-pig could be the animal of choice for photoallergy testing (71). Musk ambrette has been reported to be photoallergic, in the absence of significant phototoxicity, after topical application to guinea-pigs (62, 70, 72-75), and mice (66-68).

Coumarins
6-Methylcoumarin (6-MC): in experimental studies in volunteers, 6-MC was not phototoxic, in contrast with 7-MC. However, 6-MC was photoallergenic in the human photomaximisation test (76, 77). The US regulatory position has been discussed (78). In guinea pigs, 6-MC has not been reported to be photoirritant after topical application. In photoallergy tests, it has been reported to be negative (63, 70) or weakly active (73, 79, 80). A photoallergic potential for 6-MC was also obtained in mice (65, 67, 68), although a further report in mice was rather equivocal (66).

Benzophenones
Ketoprofen: ketoprofen has been reported not to be photoallergenic in a photomaximisation procedure (81) or in a guinea-pig test (82). However, it showed clear photoallergenic potential in a sensitive guinea pig test (T. Maurer, personal communication), as well as in a mouse model (H-W. Vohr, personal communication).

Aromatic amines
Sulphanilamide: phototoxic and photoallergic reactions were first distinguished after applying sulphanilamide.
PABA: photoallergy to PABA in guinea-pigs was reported by Gerberick & Ryan (62).

Aromatic nitro-oxides
Omadine: photoallergy to sodium omadine has been reported after using the photomaximisation procedure on humans (71) and in mice (65, 67).

Quindoxin: photodermatitis to quindoxin was reported in the UK in the early 1970s, and to olaquindox in Germany in the 1990s (see 83). Quindoxin was photoallergenic in guinea-pigs (W. Lovell, unpublished observations), and olaquindox has been reported to be photoallergenic in mice (58, 84).

Tetracyclines
Topical tetracycline preparations are rarely contact sensitisers, but may yellow the skin. In addition, the skin may fluoresce under black fluorescent light. Together with the photoreactions found in in vitro investigations with tetracyclines (see Table I), this effect led to concern about photoreactions in vivo. However, no definite report has been published regarding photoallergic properties.

Phenothiazines
Chlorpromazine (CPZ): Phototoxic reactions were produced by topical application of 0.5%, aqueous solutions plus UVA irradiation, whereas photoallergic reactions were elicited by 0.025% CPZ plus UVA (85). Experimental phototoxicity was reported after the intradermal injection of 0.1 mg with about 18 J/cm² UVA (76). Photoallergy was elicited by topical application of 0.2% CPZ with 4 J/cm² UVA via the photomaximisation technique (59). In 27 volunteers, phototoxic reactions could be induced with a minimal concentration of 0.5% and 4 J/cm² of UVA, whereas, in patients photoallergic to chlorpromazine, 0.1% CPZ and 2 J/cm² of UVA were sufficient (86). Photoallergic reactions have been elicited topically in guinea-pigs (63, 87), and both topically (58, 65-68, 88) and systemically (58, 89) in mice.

Promethazine: in a volunteer study, promethazine did not induce phototoxic reactions when one tablet of 25 mg was given and light doses of 5-72 J/cm² were used (90). Promethazine was photoallergic, but was much weaker than chlorpromazine in a guinea-pig test (63). Photoallergic reactions were also reported in guinea-pigs by Guillot et al. (70) and in mice by Vohr et al. (88).

Fluoroquinolones
The fluoroquinolones are a family of broad-spectrum antibiotics, primarily intended for systemic use. There is much epidemiological evidence from clinical safety trials that many of the family of fluoroquinolones are phototoxic (photoirritant) in human skin. Carefully conducted trials involving the photo-testing of volunteers in defined wavebands of UVR, have supported the epidemiological evidence and quantified the relative potencies of various fluoroquinolones (for example, 91-93). The compounds primarily absorb in the UVA (315-400 nm) range of the solar spectrum and phototoxic potency varies from the ability of compounds such as BAY y3118 to photosensitise the skin by a factor of over 30, to compounds such as moxafloxacin, which have no phototoxic potential (94). Several fluoroquinolones have also been reported to be phototoxic in guinea-pig skin (95, H-W. Vohr, personal communication) and mice (84, 96). Although most of the quinolones seem to cause phototoxic reactions exclusively, there are reports of photoallergic properties of some nuoroquinolones, such as enoxacin (84, 97-99).

Testing strategy

In vivo
Until now, photoallergic hazard identification has largely been performed in guineapig models. Since these animal procedures are expensive, time-consuming, and harmful to the animals, and objective endpoints are still lacking, a new testing strategy is required that takes account of objectivity of hazard identification and animal welfare. Recent reports showed that a combination of the LLNA and the primary mouse-ear swelling test (MEST), the IMDS (58), could probably provide a reasonable alternative in terms of animal welfare and objectivity There has only been intralaboratory prevalidation of the photo-LLNA until now, which showed that this model also provides the possibility of discriminating between photoallergic and phototoxic reactions. To verify the validity and robustness, an interlaboratory validation of the respective test protocol, monitored by an international panel of experts, should be conducted. A testing strategy incorporating the LLNA and the use of structure-activity relationships is shown in Figures 7 and 8.

Figure 7: Testing for contact (photo) allergenic potential

Figure 8: Photoallergy testing scheme

IMDS = integrated model for the differentiation of skin reactions.
LLNA = local lymph node assay
LNC = lymph node cells
LN = lymph node
MHC = mature Langerhans cells

In vitro
Photobinding to protein is a common property of photoallergens, and a test based on photobinding of photoallergens to human serum albumin has been proposed (100). The model has been demonstrated to detect all the photoallergens tested so far. While photobinding to protein is considered a necessary condition for photoallergy, it is not sufficient for discrimination between photoallergens and photoirritants. Additional mechanistic tests, including tests for photooxidation, are required to enable the discrimination of photoallergens (101). The photobinding model, in conjunction with a test of photooxidation of histidine, was used to test the 30 chemicals selected for the EU/COLIPA phototoxicity trial. Although the test chemicals were heavily biased toward phototoxicity, the photobinding model showed excellent detection of photoallergens, and in conjunction with a test for photooxidising potential, differentiation between photoallergens and phototoxins was largely achieved (102).

A hierarchical scheme for the in vitro hazard assessment of substances for photoallergic potential should start with a review of the chemistry, structure-activity relationships and phototoxicology of the test item (Figure 9). If experimental evidence is required, UV/visible absorbance can be used as a pre-screen; substances that absorb significantly have the potential for adverse photochemistry and should be tested further (101). It has been proposed that the 3T3 NRU phototoxicity test could be used as a general screen for the photosensitising potential of absorbing substances (Figure 9). If these screening tests are positive, in vitro screening tests for photobinding to protein and for photooxidation should be performed (100-102). If the substance is an efficient photo-oxidiser, photoirritancy may be the problem, rather than photoallergy. If the photobinding test is negative, a potential for photoallergy is not expected, and further testing for photoallergenicity should not be required. If the test material binds to protein in the absence of significant photo-oxidation, a photoallergy potential is predicted. The safety evaluation could be discontinued, but if there is further interest in the test substance, additional studies can be considered. For example, skin-penetration studies can be carried out, as no penetration may mean there will be no problem. A further option may be to consider the use of radiolabelled test substances, in order to investigate further and to quantify the in vitro photobinding. In vivo testing may be considered.

Figure 9: Phototoxicology: hazard identification

A biologically based model of photoallergic potential could be useful in hazard assessment in practice (Figure 8). Early upregulation of the inflammatory mediator, IL-1b, by dendritic cells, and their endocytic activation by immunogenic haptens, have been suggested as potential in vitro indicators of immunogenicity (103, 104). Modification of such systems for photoallergens could form an in vitro biological test of the present photochemical model. Animal testing, however, would be required to address the problem of potency in vivo.

Potency

No experimental approach has been established to date for assessing potency for photoallergy, Although one could design a test system for the photoallergy potency of test chemicals in humans, this is impossible, for ethical reasons. Therefore, risk assessment in humans has to rely on clinical case reports and photopatch testing.

A promising preclinical screening test for photoreactions is the IMDS mentioned above. At least after oral application, the determination of the potency of a test substance to induce photoirritancy or photoallergy seems feasible.

Metabolism

The generation of photoreactive metabolites must not be underestimated and should, therefore, be considered in the general toxicological evaluation. A modern testing strategy for photoallergic potential should integrate relevant photo sensitisers of the past and of the present, as well as photosensitisers that may prove to be relevant in the future. Thus, the spectrum of test substances will have to be modified over time, according to test results and to clinical or experimental findings. It must be taken into account, from a general perspective, that the metabolic capacity of in vitro systems may be limited in comparison to the situation in vivo, irrespective of the organ or tissue, and thus may provide a spectrum of metabolites which are different from the situation in vivo.

Photochemical Genotoxicity/Carcinogenicity (photogenotoxicity/photocarcinogenicity)

Human data

UVB is a known human skin carcinogen, and its direct genotoxic effects are considered to be an important factor in this process (105). UVB is also known to be a skin carcinogen in rodents (106). Conversely, there is a lack of documented human photochemical carcinogens that can be used to evaluate the potential relevance of photochemical genotoxicity and rodent co-carcinogenicity testing. To date, the combination of 8-methoxypsoralen (8-MOP) and UVA, known as PUVA therapy, as used in the treatment of psoriasis, is the only well-documented human photochemical genotoxic carcinogen (107, 108). Although this effect in human skin could be predicted by results in mouse phototumorigenesis studies, apart from the psoralens and fluoroquinolones (109, 110), no other groups of photo-tumorigens have been studied in any detail. Given the scarcity of data, both in animal models and in humans, the exact relevance of this type of modelling is not clear (111). At this time, the limited data sets of human and rodent photochemical carcinogens and non-carcinogens make the evaluation of the human relevance of short-term photochemical genotoxicity tests problematic. Despite these limitations, genotoxicity can be addressed by using tests that hold the greatest promise in assessing potential photochemical mutagenic/clastogenic effects, akin to standard genotoxicity/carcinogenicity testing (112).

Current status of photochemical genotoxicity testing

Many different genotoxicity test systems have been employed for photochemical genotoxicity testing. In principle, most standard in vitro genotoxicity tests can, and have been, adapted for testing the consequences of photoactivation of chemicals in terms of damage to the genetic material. The potential of a few classes of chemicals, such as psoralens, phenothiazines or compounds used in photodynamic therapy (for example, porphyrin derivatives), to induce genotoxic effects after irradiation, has been known for several years (113-115). More recently, the photochemical genotoxicity of fluoroquinolone antibiotics has been reported (116-120). Additional important contributions regarding test method development and standardisation have been made by several groups in recent years (45, 121-123). Recommendations with regard to the conduct of tests for photochemical genotoxicity have recently been elaborated by an international expert working group. Their report constitutes a valuable manual for laboratories involved in the conduct of these assays (124).

Regarding product safety testing strategies, the guideline of the SCCNFP stipulates that both a bacterial test for gene mutation and an in vitro test for chromosomal aberrations in mammalian cells should be performed in the presence of UV radiation (125).

Rationale for testing for photochemical genotoxicity

In considering the carcinogenic potentials of new chemical entities, an evaluation of genotoxic potential has become an integral part of the toxicological assessment. A multitude of in vitro and in vivo mutagenicity test systems have been devised, and regulatory guidelines for this application have been issued. From these data and the parallel experience in standard genotoxicity testing, testing for photochemical genotoxicity should be considered in the phototoxicological assessment of chemicals. Importantly, the main purpose of such testing is to make an assessment of the likelihood of a compound to turn into a photochemical carcinogen after activation with UV or visible radiation.

Test systems and strategies to test for photochemical genotoxicity (Figure 10)

Figure 10: A generalised approach to photocarcinogenicity testing

^aTests for photomutagenicity are usually not needed if comprehensive testing for phototoxicity did not indicate a phototoxic potential

Tests for photochemical genotoxicity may be considered for compounds that are intended for use on the skin or are widely systemically distributed with significant skin exposure, if a phototoxic potential was proven or, for nonphototoxic compounds, if there is evidence of absorbance of UV and/or visible light and the compound belongs to a chemical class containing known phototoxic or photogenotoxic compounds.

An established absence of a phototoxic action in relevant investigations is considered to be a strong indicator that the respective compound would be unlikely to possess photogenotoxic properties, although it may not be a sufficient criterion for forgoing testing.

For product safety testing, it seems reasonable to focus evaluation and validation on those in vitro systems that are recommended as standard genotoxicity systems in regulatory testing strategies, such as OECD or International Conference on Harmonisation (ICH) guidelines. Several arguments can be put forward for a preference for systems employing mammalian cells in vitro over those involving lower eukaryotes or prokaryotes. Such a preference is based, in part, on the higher complexity of the cellular model and on the fact that phototoxicity testing in vitro is done in mammalian cells. There is no general consensus on which test system would constitute a minimal test battery for a sufficient assessment of a test compound for photochemical genotoxicity (124). As no photochemical genotoxin is known which is exclusively positive for gene mutations, and since the recognised photochemical mechanisms are strongly clastogenic, it may be suggested that a test for photochemical clastogenicity would suffice for screening purposes.

Thus, the photochemical genotoxicity tests could follow a general sequence that includes:

a test for induction of chromosome damage (for example, aberrations or micronuclei) in vitro;
a test for induction of gene mutations in mammalian cells or (less preferable) in bacteria; and
the Comet assay, assessing photochemically induced DNA damage (116, 117). However, since this test is currently not accepted in standard batteries for genotoxicity testing of chemicals, it should be regarded as an ancillary tool.

It can also be noted that the Comet assay is the only genotoxicity test that has been successfully applied to testing for photochemical genotoxicity in vivo. Positive results have been reported for the fluoroquinolone clinafloxacin (119).

Metabolic activation

Genotoxicity testing in vitro is usually done with the inclusion of an external metabolic activation system such as rat liver S9 fraction. There is no evidence that enzymic metabolic activation plays a crucial role in the generation of a photochemical genotoxin. Furthermore, there is evidence that the addition of proteinaceous material such as that present in rat liver S9 can abolish the photogenotoxic effects of some compounds. Thus, the use of an external metabolic activation system such as liver S9 is not advocated for photogenotoxicity testing in vitro. However, there are a limited number of examples where metabolism of the parent compound shifts the absorption spectrum. Therefore, researchers are encouraged to look for and to investigate potential candidate compounds for which metabolism is suggested to result in increased photobinding or phototoxicity.

Light source

Light sources, in general, have been discussed in section 3 (Terminology). A solar simulator has been recommended for use in photochemical genotoxicity testing. The characteristics of the irradiation source need to be described in detail. Specifically, it is important to give information on the spectral output for determining the UVB:UVA ratio of the light source. In general, solar irradiation possesses a ratio of around 1:20, although this value depends very much on location, time of day and atmospheric conditions (the longer the light path through the atmosphere, the lower the ratio; clouds and dust further attenuate the UVB component). It should be noted that the UVB (290-320 nm) content of the light source may need to be attenuated, in order to increase the sensitivity of the test method used, notably when monolayer mammalian cells and, most especially, repair-deficient organisms, are used.

In essence, it is important that the characteristics of the light source, i.e. the spectral output, are known and are provided together with any results from photogenotoxicity studies.

Test systems for photochemical co-carcinogenesis

A test for photochemical co-carcinogenesis in animals has been advocated by the FDA, for example, for drugs that are applied to skin or which reach the skin systemically (126). The only model that is currently available for assessing such effects in a Good Laboratory Practice (GLP)-like format is the SKH1(hr/hr) albino hairless mouse model. The endpoints of judgement are the number and size of papilloma formation on skin and the latency, after a general duration of treatment of 40 weeks. Various UV sources and intensities have been used for the assessment of photochemical carcinogenesis. As currently designed, the mechanistic understanding that is provided by the model is very limited, since highly tumorigenic doses of UVB are most often used and histopathology is not an integrated part of tumour investigation (111). In a modified test approach using a non-tumorigenic protocol of doses from a broad-band fluorescent UVA source, 8-MOP, as well as some fluoroquinolones, led to a high incidence of skin papilloma (110). As stated above, the extremely limited data on photosensitised human skin tumorigenesis, together with only a marginally larger amount of data from mouse studies, makes the relevance of the rodent photocarcinogenesis models difficult to assess. Overall, the predictivity of this rodent photocarcinogenicity model for the human situation is not known, and thus the relevance of the rodent studies for human health risk cannot be reliably assessed.

Guidance on testing for photochemical genotoxicity (Figure 10)

The direct genotoxic effects of UVB can lead to high mutation frequencies or chromosomal damage at irradiation doses comparable to only seconds or minutes of solar irradiation at environmental levels. Such a high responsiveness to UVB will limit the applicable irradiation doses to levels that might, theoretically, be too low for the activation of a comparatively weak photochemical genotoxin. Therefore, it may be necessary to attenuate the UVB content of the light source used for compounds that do not exclusively absorb in the UVB region. Irradiation through a plastic lid will generally meet these requirements, since it predominantly reduces the UVB part of the spectrum. Note, however, that lids from different manufacturers have different UV transmission characteristics. It is also possible to use a quartz glass cover or more simply, food-wrapping plastic film, neither of which affect the spectral composition. It is recommended that a full dose-response curve of the test system to the irradiation source in the absence of test compound is included in the report.
A specific, generally applicable irradiation dose (for example, equivalent to a certain, environmentally relevant solar irradiation period) cannot be defined for the various test systems in use, mainly because of their different sensitivities to the UVB effects. The irradiation per se should yield a small, reproducible genotoxic or toxic effect. When testing sunscreens, it might be of interest to describe the protective effect of the test compound. In such a case, it is advisable to use an irradiation dose which elicits a more sizeable effect, so that attenuation of the UV genotoxicity is measured more easily.
There is no clear-cut precedent for an exclusively UVB-absorbing compound which is established as a photochemical genotoxin. Indeed, for a compound exclusively absorbing in the UVB region, the protective (shielding) effect is most likely to predominate over any potential photochemical genotoxic action. Similar considerations also hold true for phototoxicity and photosensitisation. The EU test on in vitro phototoxicity (19) states that "UVA and visible regions are usually associated with photosensitisation, whereas UVB is of less importance and directly highly cytotoxic, increasing its cytotoxicity through 1000 fold from 313 nm to 280 nm."
Doses of UVB and UVA should be determined and reported separately. The simplest way that might be done is by using separate detectors for UVA and UVB that have been calibrated for the specific spectral outputs of the source employed or by determination of the spectral output of the irradiation source with the aid of a spectroradiometer and then integrating the spectral irradiance of the UVB and UVA ranges in parallel with the experiment. It has to be noted that, unless the meter is calibrated to the particular spectral output of the source (including any changes that occur due to filtering), measurements with UVA detectors, and even more so with UVB detectors, are notoriously unreliable. This is presumably due to slight variations in the filtration properties of the filters around 315-320 nm, as used to discriminate the spectral regions of UVA and UVB, where there is a strong increase of irradiation energy of the solar simulators and slight differences could have a large impact.
As indicated in Figure 10, photostable chemicals need to be tested, since they may act as photocatalysts for singlet oxygen production, as outlined in the section on mechanisms.

Hazard Identification in Phototoxicity Testing

At the first ECVAM workshop on phototoxicity testing (1), a tiered testing strategy was recommended for assessing the phototoxic potential of test chemicals (Figure 9). According to the tiered-testing approach, in the first step, validated in vitro phototoxicity tests were recommended, although they did not exist when the first workshop was held. It was assumed that, when no phototoxic potential could be detected in a selected set of validated in vitro phototoxicity tests, one may proceed to clinical testing in humans without any preclinical testing for phototoxicity in animals.

However, the early tiered-testing approach for phototoxicity did not take into account additional endpoints, such as photoallergy and photogenotoxicity/photocarcinogenicity. Due to a lack of experimental and clinical experience, the two new endpoints had not been considered at the first ECVAM workshop on phototoxicity. The participants in the current workshop have therefore updated the tiered flow chart for phototoxicity testing, which, today, is more complex. It is derived from the phototoxicity testing strategy of test guideline B.41 of Annex V of EU Council Directive 67/548/EEC for the classification and labelling of hazardous chemicals (19).

Figure 9 indicates that the 3T3 NRU phototoxicity test has the potential to detect most photoallergens and photogenotoxins, as well as photoirritants. This has been amply shown by the prevalidation and validation studies (2, 3). Hence, a negative result in the 3T3 tests for a compound tested at concentrations of up to 100 µg/ml is good evidence of the absence of adverse photobiological effects (1, 3, 4). However, additional information on photoallergic and photogenotoxic/photocarcinogenic potential should be obtained before clinical testing of a new chemical in humans is recommended. Unfortunately, there are no validated in vitro or in vivo tests for the evaluation of photoallergy and photogenotoxic/photocarcinogenic potential.

Recommendations

Human photopatch testing

An updated set of test chemicals for the photopatch test (Table V) must contain photosensitisers that were of significance in the past and which will remain relevant today and into the future (13). Thus, a standardised set of test chemicals for photopatch testing should be updated continually, according to results of experimental testing and clinical experience.

Acute phototoxicity testing

3T3 NRU phototoxicity test

The classification results of the validation of the 3T3 NRU phototoxicity test should be evaluated against the most recent human data set from the German, Austrian and Swiss photopatch test group.
The need for the inclusion of a metabolical activation step, however, could not be proven with the wide spectrum of test chemicals used during the prevalidation and validation of the 3T3 NRU phototoxicity test. The use of human keratinocytes should be considered as an alternative to 3T3 mouse fibroblasts, if inclusion of a metabolising system seems appropriate.

Human 3-D skin models

An exchange of experience and coordination of the activities within different evaluation projects using 3-D skin models is recommended. The prevalidation and/or validation of additional assays based on 3-D human skin models is imperative.
A prevalidation study of the EpiDerm™ phototoxicity test model, which has been successfully finished (26), should be published.
The results obtained with the EpiDerm™ and the SKINETHIK™ models in other laboratories (27, 28) should be published.
Data on the testing of finished products are urgently needed, to support the need for including them in a validation study.
Comparative data on the metabolising capacities of skin models (127) is needed, in order to facilitate their use in routine safety testing and the regulatory acceptance of data produced with them.
The use of skin models for photopotency testing should be evaluated. The need for prevalidation and/or validation of additional assays based on 3-D human skin models was stressed by the participants of the workshop.

Photoallergy

Recent reports showed that a combination of the LLNA and the primary MEST (IMDS) might provide a reasonable alternative in terms of animal welfare and objectivity. So far, there has only been an intralaboratory prevalidation of this photo-LLNA, which showed that the model offers the possibility of discriminating between photoallergic and phototoxic reactions. To verify the validity and robustness of the method, an interlaboratory validation of the test protocol, monitored by an international panel of experts, should be conducted.
Interlaboratory validation of the in vitro photobinding model, in conjunction with a test for photo-oxidising potential, is recommended, since this model has been shown to be able to detect photoallergenic potential, as distinct from photoirritation.
Research into the use of dendritic cells, antigen processing and cytokine production is recommended, to provide additional models for the detection of photoallergens as opposed to photoirritants.

Photogenotoxicity studies

Thus far, testing for photochemical genotoxicity has not been sufficiently standardised, and only a few laboratories are conducting such assays on a regular basis. Hence, further studies are recommended on photogenotoxicity testing in vitro, which should mainly address the following topics.

Further evaluation of the links between phototoxicity and photogenotoxicity, for example, through the testing of compounds that have been used to validate the 3T3/NRU assay.
Evaluation of pairs of phototoxic/nonphototoxic, yet UV light-absorbing, compounds.
Clarification of whether metabolism is important for the generation of photochemical genotoxins.
An exploration of the potential of photochemical genotoxins to induce cell transformation in vitro, for example, in the SHE cell-transformation assays. Potential candidates for such studies are established photochemical genotoxins that have been shown to induce skin tumours in animal studies, such as some psoralens, polycyclic hydrocarbons and fluoroquinolones.
Evaluation of the photogenotoxic potential of compounds in short-term in vivo skin assays, i.e. the use of the Comet assay with epidermal suspensions from phototoxin + UV/visible radiation-treated mouse or human skin.
Following up patients who are taking known phototoxins chronically (for example, fluoroquinolones in cystic fibrosis, or phenothiazines in chronic mental illness) for increased skin cancer risk.
Evaluating whether the use of sources (for example, typical solar simulators) with a high UVB output (known to be genotoxic and carcinogenic in its own right) limit the ability of current models to detect photogenotoxins and photocarcinogens.

Acknowledgements

We are indebted to Dieter Traue and Susanne Boy (both at ZEBET) for their skilful assistance in the preparation of the manuscript.

References

Anon. (1986). Council Directive 86/609/EEC of 24 November 1986 on the approximation of laws, regulations and administrative provisions of the Member States regarding the protection of animals used for experimental and other scientific purposes. Official Journal of the European Communities L358: 1-29.
Spielmann, H., Lovell, W.W., Hölzle, E., Johnson, B.E., Maurer, T., Miranda, M.A., Pape, W.J.W, Sapora, O. & Sladowski, D. (1994). In vitro phototoxicity testing. The report and recommendations of ECVAM workshop 2. ATLA 22: 314-348.
Spielmann, H., Balls, M., Dupuis, J., Pape, W.J., Pechovitch, G., de Silva, O., Holzhütter, H-G., Clothier, R., Desolle, P., Gerberick, F., Liebsch, M., Lovell, W.W., Maurer, T., Pfannenbecker, U., Potthast, J.M., Csato, M., Sladowski, D., Steiling, W. & Brantom, P. (1998). The international EU/COLIPA in vitro phototoxicity validation study: results of phase II (blind trial). Part 1: the 3T3 NRU phototoxicity test. Toxicology in Vitro 12: 305-327.
Spielmann, H., Balls, M., Dupuis, J., Pape, W.J.W, de Silva, O., Holzhütter, H-G., Gerberick, F., Liebsch, M., Lovell, W.W. & Pfannenbecker, U. (1998). A study on UV filter chemicals from Annex VII of European Union Directive 76/768/EEC, in the in vitro 3T3 NRU phototoxicity test. ATLA 26: 679-708.
Anon (2000). Guidance for Industry: Photosafc~v Testing. U.S. Department of Health and Human Services--Food and Drug Administration, Center for Drug Evaluation and Research (CDER), January 2000, Pharmacology and Toxicology; Rockville, MD (USA). Web site http://www.fda.gov/cder/guidance/index.htm.
Johnson, B.E. (1987). Light sensitivity associated with drugs and chemicals. In The Physiology and Pathology of the Skin (ed. A. Jarret), pp. 2541-2606. San Diego, CA, USA: Academic Press.
Johnson, B.E. (1992). Drug and chemical photosensitization. In The Environmental Threat to the Skin (ed. R. Marks & G. Plewig), pp. 57-66. London, UK: Martin Dunitz.
Spikes, J.D. (1989). Photosensitization. In The Science of Photobiology (ed. K.C. Smith), 2nd edn, pp. 79-110. New York, USA: Plenum Press.
Pendlington, R.U. & Barrat, K.R. (1990). Molecular basis of photocontact allergy. International Journal of Cosmetic Sciences 12: 91-103.
Lehmann, P. (1991). Die deutschsprachige Arbeitsgemeinschaft Photopatch-Test (DAPT), Hautarzt 41: 295-297.
Hözle, E., Neumann, N., Hausen, B., Przybilla, B., Schauder, S. & Honigsmann, H. (1991). Photopatch testing: the 5-year experience of the German, Austrian, and Swiss Photopatch Test Group. Journal of the American Academy of Dermatology 25: 59-68.
Neumann, N.J., Hö1zle, E., Plewig, G., Schwarz, T., Panizzon, R.G., Breit, R., Ruzicka, T. & Lehmann, P. (2000). Photopatch-testing: the 12-years experience of the German, Austrian, and Swiss photopatch test group. Journal of the American Academy of Dermatology, in press.
Neumann, N.J., Hölzle, E., Lehmann, P., Benedikter, S., Tapernoux, B. & Plewig, G. (1994). Pattern analysis of photopatch test reactions. Photodermatology Photoimmunology and Photomedicine 10: 65-73.
Runger, T.M., Lehmann, P, Neumann, N.J., Matthies, C., Schauder, S. & Ortel, B. (1995). Empfehlungen einer Photopatch-Test Standardreihe durch die deutschsprachige Arbeitsgruppe "Photopatch-Test". Hautarzt 46: 240-243.
Balls, M., Blaauboer, B.A., Fentem, J., Bruner, L., Combes, R.D., Ekwall, B., Fielder, R.A., 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 23:129-147.
OECD (1996). Final Report of the OECD Workshop on Harmonization of Validation and Acceptance Criteria for Alternatiue Toxicological Test Methods, 60pp. Paris, France: OECD.
Spielmann, H., Balls, M., Döring, B., Holzhütter, H.G., Kalweit, S., Klecak, G., L'Eplattenier, H., Liebsch, M., Lovell, W.W., Maurer, T., Moldenhauer, F., Moore, L., Pape, W., Pfannbecker, U., Potthast, J., de Silva, O., Steiling, W. & Willshaw, A. (1994). EEC/COLIPA project on in vitro phototoxicity testing: first results obtained with a Balb/c 3T3 cell phototoxicity assay. Toxicology in Vitro 8: 793-796.
Wakuri, S., Tanaka, N. & Ono, H. (1995). In vitro phototoxicity assay using culture cells. Proceedings of the 8th Annual Meeting of the JSAAE in Tokyo 1994; abstract p.19. AATE (Alternatives to Animal Testing and Experimentation) 3: 67.
Anon (2000). Commission Directive 2000/33/EC of 25 April 2000, adapting to technical progress for the 27th 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. ANNEX II: B.41. In vitro 3T3 NRU Phototoxicity Test. Official Journal of the European Communities L136: 90-107.
Clothier, R., Willshaw, A., Cox, H., Garle, M., Bowler, H. & Combes, R. (1999). The use of human keratinocytes in the EU/COLIPA international in vitro phototoxicity test validation study and the ECVAM/COLIPA study on UV filter chemicals. ATLA 27: 247-259.
Edwards, S.M., Donnelly, T.A., Sayre, R.M., Rheins, L.A., Spielmann, H. & Liebsch, M. (1994). Quantitative in vitro assessment of phototoxicity using a human skin model: Skin². Photodermatology Photoimmunology and Photomedicine 10: 111-117.
Liebsch, M., Döring, B., Donnelly, T.A., Logemann, P, Rheins, L.A. & Spielmann, H. (1995). Application of the human dermal model Skin² ZK 1350 to phototoxicity and skin corrosivity testing. Toxicology in Vitro 9: 557-562.
Roguet, R., Cohen, C. & Rougier, A. (1994). A reconstituted human epidermis to assess cutaneous irritation, photoirritation and photoprotection in vitro. In Alternative Methods in Toxicology, Vol. 10, In Vitro Skin Toxicology--Irritation, Phototoxicity, Sensitization (ed. A. Rougier, A. Goldberg & H. Maibach), pp. 141-149. New York, USA: Mary Ann Liebert.
Api, A.M. (1997). In vitro assessment of phototoxicity. In Vitro Toxicology 10: 339-350.
Liebsch, M., Barrabas, C., Traue, D. & Spielmann, H. (1997). Entwicklung eines neuen in vitro Tests auf dermale Phototoxizität mit einem Modell menschlicher Epidermis, EpiDerm^TM. Alternativen zu Tierexperimenten (ALTEX) 14: 165-174.
Liebsch, M., Traue, D., Barrabas, C., Spielmann, H., Gerberick, G.F., Cruse, L., Diembeck, W., Pfannenbecker, U., Spieker, J., Holzhütter, H-G., Brantom, P., Aspin, P. & Southee, J. (1999). Prevalidation of the EpiDerm Phototoxicity Test. In Alternatives to Animal Testing II: Proceedings of the Second International Scientific Conference Organised by the European Cosmetic Industry Brussels, Belgium (ed. D. Clark, S. Lisansky & R. Macmillan), pp. 160-t66. Newbury UK: CPL Press.
Jones, P., King, A., Lovell, W. & Earl, L. (1999). Phototoxicity testing using 3-D reconstructed human skin models. In Alternatives to Animal Testing II: Proceedings of the Second International Scientific Conference Organised by the European Cosmetic Industry, Brussels, Belgium (ed. D. Clark, S. Lisansky & R. Macmillan), pp. 138-141. Newbury, UK: CPL Press.
Bernard, F.X., Barrault, C., Deguery, A., de Wever, B. & Rosdy, M. (1999). Development of a highly sensitive phototoxicity assay using the reconstructed human epidermis SkinEthik. In Alternatives to Animal Testing II: Proceedings of the Second International Scientific Conference Organised by the European Cosmetic Industry, Brussels, Belgium (ed. D. Clark, S. Lisansky & R. Macmillan), pp. 167-174. Newbury, UK: CPL Press.
Augustin, C., Collombel, C. & Damour, O. (1997). Use of dermal equivalent and skin equivalent models for identifying phototoxic compounds in vitro. Photodermatology Photoimmunology and Photomedicine 13: 27-36.
Pape, W.J.W., Maurer, T., Pfannenbecker, U. & Steiling, W. (2000). The red blood cell phototoxicity test (photohaemolysis and haemoglobin oxidation): EU/COLIPA validation programme on phototoxicity (Phase II). ATLA, in press.
Pape, W.J.W., Brandt, M. & Pfannenbecker, U. (1994). Combined in vitro assay for photohaemolysis and haemoglobin oxidation as part of a phototoxicity test system assessed with various phototoxic substances. Toxicology in Vitro 8: 755-757.
Kahn, G. & Fleischaker, B.I. (1971). Red blood cell haemolysis by photosensitizing compounds. Journal of Investigative Dermatology 56: 85-90.
Johnson, B.E,, Walker, E.M. & Hetherington, A.M. (1986). In vitro models for cutaneous phototoxicity. In Skin Models: Models to Study Function and Disease of Skin (ed. R. Marks & G. Plewig), pp. 264-281. Berlin, Germany: Springer-Verlag.
Traynor, N.J., Johnson, B.E. & Gibbs, N.K. (1966). Photohaemolysis assay for drug phototoxicity complicated by "bleaching of released haemoglobin". Toxicology in Vitro 10: 619-624.
Okamoto, Y., Ryu, A. & Ohkoshi, K. (1999). In vitro alternatives and phototoxicity testing. I. Evaluation of in vitro phototoxicity assays. ATLA 27: 639-644.
Winterbourn, C.C. (1986). Free-radical production and oxidative reactions of haemoglobin. Environmental Health Perspectives 64: 321-330.
Averbeek, D. & Moustacchi, E. (1991). Response to ultraviolet radiation in a simple eukaryote (yeast): genetic control and biological consequences. In Laboratory Manual, NATO ASI Series: Photobiological techniques (ed. R.H. Pottier, R.H. Douglas, P. Mathis & D.P Valenzeno), pp. 165-186. New York, USA: Plenum Press.
Friedberg, E.C., Walker, G.C. & Siede, W. (1995). DNA Repair and Mutagenesis, pp. 1-698. Washington, DC, USA: ASM Press.
Chételat A., Albertim, S., Dresp, J.H., Strobel, R. & Gocke, E. (1993). Photomutagenesis test development. I. 8-Methoxypsoralen, chlorpromazine and sunscreen compounds in bacterial and yeast assays. Mutation Research 292: 241-250.
Morliére, P, Hönigsmann, H., Averbeck, D., Dardalhon, M., Hüppe, G., Ortel, B., Santus, R. & Dubertret, L. (1988). Phototherapeutic, photobiologic and photosensitizing properties of khellin. Journal of Investigative Dermatology 90: 720-724.
Averbeck, D., Averbeck, S., Dubertret, L., Young, A.R. & Morliére, P. (1990). Genotoxicity of bergapten and bergamot oil in Saccharomyces cerevisiae. Journal of Photochernistrv and Photobiology B: Biology 7: 209-230.
Moysan, A., Morliére, P, Averbeck, D. & Dubertret, L. (1993). Evaluation of phototoxic and photogenotoxic risk associated with the use of photosensitizers in suntan preparations: application to tanning preparations containing bergamot oil. Skin Pharmacology 6: 282-291.
Carls, N. & Schiestl, R.H. (1994). Evaluation of the yeast DEL assay with 10 compounds selected by the International Program on Chemical Safety for the evaluation of short-term tests for carcinogens. Mutation Research--DNA Repair 320: 293-303.
Galli, A. & Schiestl, R.H. (1998). Effect of Salmonella assay negative and positive carcinogens on intrachromosomal recombination in S-phase arrested yeast cells. Mutation Research 419: 53-68.
Dean, S.W., Lane, M., Dunmore, R.H., Ruddock, S.P., Martin, C.N., Kirkland, D.A. & Loprieno, N. (1991). Development of assays for the detection of photomutagenicity of chemicals during exposure to UV light. I. Assay development. Mutagenesis 6: 335-341.
Averbeck, D. (1985). Relationship between lesions photoinduced by mono- and bifunctional furocoumarins in DNA and genotoxic effects in diploid yeast. Mutation Research 148: 217-233.
Averbeck, D. (1989). Yearly review: recent advances in psoralen phototoxicity mechanism. Photochemistry and Photobiology 50: 859-882.
Knowland, J., McKenzie, E.A., McHugh, PJ. & Cridland, N.A. (1993). Sunlight-induced mutagenicity of a common sunscreen ingredient. FEBS Letters 324: 309-313.
Marrot, L., Belaidi, J.P., Chaubo, C., Meunier, J.R., Perez, P. & Agapakis-Causse, C. (1998). An in vitro strategy to evaluate the phototoxicity of solar UV at the molecular and cellular level: application to photoprotection assessment. European Journal of Dermatology 8: 403-412.
Marrot, L. & Agapakis-Causse, C. (2000). Differences in the photogenotoxic potential of two fluoroquinolones as shown in diploid yeast strain (Saccharomyces cerevisae) and supercoiled plasmid DNA. Mutation Research 468: 1-9.
Agapakis-Causse, C., Bosca, F., Castell, J.V., Hernandez, D., Marin, M.L., Marrot, L. & Miranda, M.A. (2000). Tiaprofenic acid-photosensitized damage to nucleic acids: a mechanistic study using complementary in vitro approaches. Photochemistry and Photobiology 71: 499-505.
Averbeck, D. & Averbeck, S. (1994). Induction of the genes RAD54 and RNR2 by various DNA damaging agents in Saccharomyces cerevisiae. Mutation Research--DNA Repair 315: 123-138.
Dubertret, L., Santus, R. & Morliére, P., eds. (1995). Ozone, Sun, Cancer, Molecular and Cellular Mechanisms. Prevention, pp. 1-223. Paris, France: INSERM.
Krutman, J., Ahrends, C., Roza, L. & Arlett, C.F (1996). The role of DNA damage and repair in ultraviolet B radiation-induced immunomodulation: relevance for human photocarcinogenesis. Photochemistry and Photobiology 63: 394-396.
Firket, H., Leclercq, J. & Geubelle, E (1971). A simplified procedure for the use of the "blast-like transformation" of lymphocytes in cultures as a routine clinical test: results in a group of allergic children. Acta Allergologica 26: 191-199.
Homey, B., von Schilling, C., Bluemel, J., Schuppe, H.C., Ruzicka, T., Ahr, H.J., Lehmann, P. & Vohr, H.W. (1998). An integrated model for the differentiation of chemical -induced allergic and irritant skin reactions. Toxicology and Applied Pharmacology 153: 83-94.
Ulrich, P, Homey, B. & Vohr, H.W (1998). A modified local lymph node assay for the differentiation of contact photoallergy from phototoxicity by analysis of cytokine expression in skin-draining lymph node cells. Toxicology 125: 149-168.
Vohr, H.-W., Blümel, J., Blotz, A., Homey, B. & Ahr, H.J. (2000). An intra-laboratory validation of the integrated model for the differentiation of skin reactions (IMDS): discrimination between (photo)allergic and (photo)irritant skin reactions in mice. Archives of Toxicology 73: 501-509.
Kaidbey, K.H. & Kilgman, A.M. (1980). Photomaximisation test for identifying photoallergic contact sensitisers. Contact Dermatitis 6: 161-169.
Harber, L.C., Targovnik, S.E. & Baer, R.L. (1967). Contact photosensitivity patterns to halogenated salicylanilides in man and guineapigs. Archives of Dermatology 96: 646-656.
Morikawa, F., Nakayama, Y., Fukuda, M., Hamano, M., Yokoyama, Y., Nagura, T., Ishihara, M. & Toda, K. (1974). Techniques for evaluation of phototoxicity and photoallergy in laboratory animals and man. In Sunlight and Man (ed. T.B. Fitzpatrick, M.A. Pathak, L.C. Harber, M. Seiji & A. Kukita), pp. 529-557. Tokyo, Japan: University of Tokyo Press.
Gerberick, G.F. & Ryan, C.A. (1989). Contact photoallergy testing of sunscreens in guineapigs. Contact Dermatitis 20: 251-259.
Maurer, T. (1984). Experimental contact photoallergenicity: guinea-pig models. Photodermatology 1: 221-231.
Lovell, W.W. & Sanders. D.J. (1990). Dose-response study of ultraviolet-radiation for induction of photoallergy to tetrachlorosalicylanilide in guinea-pigs. Photodermatology 7: 192-197.
Maguire, H.C. & Kaidbey, K. (1982). Experimental photoallergic contact dermatitis A mouse model. Journal of Investigative Dermatology 79: 147-152.
Miyachi, Y. & Takigawa, M. (1983). Mechanisms of contact photosensitivity in mice III predictive testing of chemicals with photoallergenic potential in mice. Archives of Dermatology 119: 736-739.
Gerberick, G.F. & Ryan, C.A. (1990). A predictive mouse ear swelling model for investigating topical photoallergy. Food and Cemical Toxicology 28: 361-368.
Gerberick, G.F. & Ryan, C.A. (1990). Use of UVB and UVA to induce and elicit contact photoallergy in the mouse. Photodermatology 7: 13-19.
Kaidbey, K.H. & Kligman, A.M. (1978). Identification of topical photo sensitizing agents in humans. Journal of Investigative Dermatology 70:149151.
Guillot, J.P. & Martini, M.C. (1985). A new method for the assessment of phototoxic and photoallergic potentials by topical applications in the albino guinea-pig. Journal of Toxicology--Cutaneous and Ocular Toxicology 4: 117-133.
Kligman, A.M. & Kaidbey K.H. (1982). Human models for identification of photosensitizing chemicals. Journal of the National Cancer Institute 69: 269-272.
Kochevar, I.E., Zalar, G.L., Einbinder, J. & Harber, L.C. (1979). Assay of contact photosensitivity to musk ambrette in guinea-pigs. Journal of Investigative Dermatology 73: 144-146.
Ichikawa, H., Armstrong, R.B. & Harber, L.C. (1981). Photoallergic contact dermatitis in guinea-pigs: improved induction technique using Freund's complete adjuvant. Journal of Investigative Dermatology 76: 498-501.
Parker, R.D., Buehler, E.V. & Newmann, E.A. (1986). Phototoxicity, photoallergy and contact sensitization of nitro musk perfume raw materials. Contact Dermatitis 14: 103-109.
Lovell, W.W. & Sanders, D.J. (1988). Photoallergic potential in the guinea-pig of the nitromusk perfume ingredients musk ambrette, musk moskene, musk xylene, musk ketone and musk tibetene. International Journal of Cosmetic Science 10: 271-279.
Kaidbey, K.H. & Kligman, A.M. (1978). Photocontact allergy to 6-methylcoumarin. Contact Dermatology 4: 277-282.
Kaidbey, K.H. & Kligman, A.M. (1981). Photosensitization by coumarin derivatives: structure-activity relationships. Archives of Dermatology 117:259-263.
Eiermann, H.J. (1979). Regulatory considerations: AETT and 6-methylcoumarin. Drug and Cosmetics Industry 124: 37-39.
Jordan, W.P. (1982). The guinea-pig as a model for predicting photoallergic contact dermatitis. Contact Dermatitis 8: 109-116.
Buehler, E.V, Newmann, E.A. & Parker. R.D. (1985). Use of the occlusive patch to evaluate the photosensitive properties of chemicals in guinea-pigs. Food and Chemical Toxicology 23: 689-694.
Mozzanica, N. & Pigatto, P.D. (1990). Contact and photocontact allergy to ketoprofen: clinical and experimental study. Contact Dermatitis 23: 336-340.
Endo, T., Katsuta, T. & Inoue, H. (1985). Toxicological studies of ketoprofen gel-type ointment phototoxicity and photocontact sensitivity tests in guinea-pigs. Pharmacometrics 40: 159-164.
Schauder, S., Schroder, W. & Geier, J. (1996,). Olaquindox-induced airborne photoallergic contact dermatitis followed by transient or persistent light reactions in 15 pig breeders. Contact Dermatitis 35: 344-354.
Blotz, A., Blilmel, J., Ahr, H.J. & Vohr, H-W. (1998). Analyses of dermal photoreactivity after oral application of fluoroquinolones with the IMDS. Immunobiology 199: 3-5.
Epstein, S. (1968). Chlorpromazine photosensitivity: phototoxic and photoallergic reactions. Archives ofDermatology 98: 354-363.
Schauder, S. (1985). How to avoid phototoxic reactions in photopatch testing with chlorpromazine. Photodermatology 2: 95-100.
Hoshino, T., Ishida, K., Irie, T., Uekama, K. & Ono, T. (1989). An attempt to reduce the photosensitizing potential of chlorpromazine with the simultaneous use of beta-cyclodextrins and dimethyl-beta cyclodextrins in guinea-pigs. Archives of Dermatological Research 281: 60-65.
Vohr, H-W., Homey, B., Schuppe, H. & Kind, P. (1994). Photoreactions detected in a modified local lymph node assay in the mouse. Photodermatology Photoimmunology and Photomedicine 10:57-64.
Giudici, P.A. & Maguire, H.C. (1985). Experimental photoallergy to systemic drugs. Journal of Investigative Dermatology 85: 207-211.
Rosen, K. & Swanbeck, G. (1982). Phototoxic reactions from some common drugs provoked by a high-intensity UV-A lamp. Acta DermatoVenereologica 62: 246-248.
Vousden, M., Ferguson, J., Richards, J., Bird, N. & Allen, A. (1999). Evaluation of phototoxic potential of gemifloxacin in healthy volunteers compared with ciprofloxacin. Chemotherapy 45: 512-20.
Man, I., Traynor, N.J. & Ferguson, J. (1999). Recent developments in fluoroquinolone phototoxicity. Photodermatology. Photoimmunology and Photomedicine 15: 32-33.
Man, I., Murphy, J. & Ferguson, J. (1999). Fluoroquinolone phototoxicity: a comparison of moxifloxacin and lomefloxacin in normal volunteers. Journal of Antimicrobial Chemotherapy 43: Suppl. B, 77-82.
Traynor, N.J., Barratt, M.D., Lovell, W.W., Ferguson, J. & Gibbs, N.K. (2000). Comparison of an in vitro, cellular phototoxicity model against controlled clinical trials of fluoroquinolone skin phototoxicity Toxicology in Vitro 14: 275-283.
Akter, U., Niwa, M., Nose, T., Kaida, T., Matsuno, H., Kozawa, O., & Uematsu, T. (1998). Effects of several agents on UVB and UVA plus systemic fluoroquinolone-induced erythema of guinea-pig skin evaluated by reflectance colorimetry. Free Radical Biology and Medicine 24: 1113-1119.
Marutani, K., Matsumoto, M., Otabe, Y, Nagamuta, M., Tanaka, K., Miyoshi, A., Hasegawa, T., Nagano, H., Matsubara, S. & Kamide, R. ( 1993). Reduced phototoxicity of a fluoroquinolone antibacterial agent with a methoxy group at the 8 position in mice irradiated with long-wavelength UV light. Antimicrobial Agents in Chemotherapy 37:2217-2223.
Kawabe, Y., Mizuno, N. & Sakakibara, S. (1989). Photoallergic reaction caused by enoxacin. Photo-Dermatology 6: 57-59.
Schauder, S. (1990). Photosensitivität nach Enoxacin und Xipamid: Kombinierte phototoxische und photoallergische Reaktion auf Enoxacin, photoallergische Reaktion auf Xipamid mit nachfolgender transienter Lichtreaktion. (Photosensitivity following enoxacin, photo-allergic reaction to xipamide with subsequent transient light reaction). Zeitschrift für Hauthrankheiten 65: 253-262.
Izu, R., Gardeazabal, J., Gonzalez, M., Landa, N., Raton, J.A. & Diaz-Perez, J.L. (1992). Enoxacin-induced photosensitivity: study of two cases. Photodermatology Photoimmunology and Photomedicine 9: 86-88.
Barratt, M.D. & Brown, K.R. (1985). Photochemical binding of photoallergens to human serum albumin: a simple in vitro method for screening potential photoallergens. Toxicology Letters 24: 1-6.
Lovell, W.W. (1993). A scheme for in vitro screening of substances for photoallergenic potential. Toxicology in Vitro 7: 95-102.
Lovell, W.W. & Jones, P.A. (2000). An evaluation of mechanistic in vitro tests for the discrimination of photoallergic and photoirritant potential. ATLA 28: 707-724.
De Silva, O., Basketter, D., Barratt, M.D., Corsini, E., Cronin, M.T.D., Das, P.K., Degwert, J., Enk, A., Garrigue, J.L., Hauser, C., Kimber, I., Lepoittevin, J-P., Peguet, J. & Ponec, M. (1996). Alternative methods for skin sensitisation testing. The report and recommendations of ECVAM workshop 19. ATLA 24: 683-705.
Becker, D., Kuhn, U., Lempertz, U., Enk, A., Saloga, J. & Knop, J. (1997). Flow-cytometric screening for the modulation of receptor-mediated endocytosis in human dendritic cells: implications for the development of an in vitro technique for predictive testing of contact sensitisers. Journal of Immunological Methods 203: 171-180.
IARC (1992). Solar and Ultraviolet Radiation: Evaluation of Carcinogenic Risks to Humans. International Agency for Research on Cancer UARC) Monographs 55: 148 pp. Lyons, France: IARC.
Nataraj, A.A., Black, H.S., Kripke, M.L. & Ananthaswamy, H.N. (1995). Mechanism of induction of p53 mutation in 8-methoxypsoralen + UV-A (PUVA)induced mouse skin cancers. Proceedings of the American Association for Cancer Research Annual Meeting 36: 192.
Stern, R.S. & Laird, N. (1994). The carcinogenic risk of treatment for severe psoriasis. Cancer 73: 2759-2764.
Stern, R.S., Nichols, K.T. & Väkevä, L.H. (1997). Malignant melanoma in patients treated for psoriasis with methoxsalen (psoralen) and ultraviolet A radiation (PUVA). New England Journal of Medicine 336: 1041-1045.
Johnson, B.E., Gibbs, N.K. & Ferguson, J. (1997). Quinolone antibiotic with potential to photosensitize skin tumorigenesis. Journal of Photochemistry and Photobiology B: Biology 37: 171-173.
Klecak, G., Urbach, F. & Urwyler, H. (1997). Fluoroquinolone antibacterials enhance UVA-induced skin tumors. Journal of Photochemistry and Photobiology B: Biology 37: 174-181.
Forbes, P.D. (1996). Relevance of animal models of photocarcinogenesis to humans. Photochemistry and Photobiology 63: 357-362.
Müller, L., Kasper, A., Kersten, B. & Zhang, J. (1998). Photochemical genotoxicity and photochemical carcinogenesis: two sides of a coins Toxicology Letters 102-103: 383-387.
Averbeck, D. & Moustacchi, E. (1979). Genetic effects of 3-carboxymethylpsoralen, angelicin, psoralen and 8-methoxypsoralen plus 365 nm irradiation in S. cerevisiae. Mutation Research 68:133-148.
Jose, J.G. (1979). Photornutagenesis by chlorinated phenothiazine tranquilizers. Proceedings of the National Academy of Sciences USA 76: 74457449.
Gomer, C.J. (1980). DNA damage and repair in CHO cells following hematoporphyrin photoradiation. Cancer Letters 11: 161-167.
Chételat, A.A., Albertini, S. & Gocke, E. (1996). The photomutagenicity of fluoroquinolones in tests for gene mutation, chromosomal aberration, gene conversion and DNA breakage (Comet assay). Mutagenesis 11: 497-504.
Reavy, H.J., Traynor, N.J. & Gibbs, N.K. (1997). Photogenotoxicity of skin phototumorigenic fluoroquinolone antibiotics detected using the single cell gel electrophoresis (Comet) assay. Photochemistry and Photobiology 66: 368-373.
Gocke, E., Albertim, S., Chetelat, A.A., Kirchner, S. & Muster, W. (1998). The photomutagenicity of fluoroquinolones and other drugs. Toxicology Letters 102-103: 375-381.
Bulera, S.J., Theiss, J.C., Festerling, T.A. & de la Iglesia, F.A. (1999). in vitro photogenotoxic activity of clinafloxacin: a paradigm predicting photocarcinogenicity Toxicology and Applied Pharmacology 156: 222-230.
Snyder, R.D. & Cooper, C.S. (1999). Photogenotoxicity of fluoroquinolones in Chinese hamster V79 cells: dependency on active topoisomerase II. Photochemistry, and Photobiology 69: 288-293.
Chételat, A.A., Albertim, S., Dresp, J.H., Strobel, R. & Gocke, E. (1993). Photomutagenesis test development. I. 8-Methoxypsoralen, chlorpromazine and sunscreen compounds in bacterial and yeast assays. Mutation Research 292: 241-250.
Chételat, A.A., Dresp, J.H. & Gocke, E. (1993). Photomutagenesis test development. Il. 8-Methoxypsoralen, chlorpromazine and sunscreen compounds in chromosomal aberration assays using CHO cells. Mutation Research 292: 251-258.
Kersten, B., Zhang, J., Brendler-Schwaab, S.Y., Kasper, P. & Müller, L. (1999). The application of the in vitro micronucleus test in Chinese hamster V79 cells to detect drug-induced photogenotoxicity. Mutation Research 445: 55-71.
Gocke, E., Maller, L., Guzzie, P.J., Brendler-Schwaab, S., Bulera, S., Chignell, C.F., Henderson, L.M., Jacobs, A., Murli, H., Snyder, R.D. & Tanaka, N. (2000). Considerations on photochemical genotoxicity: report of the International Workshop on Genotoxicity Test Procedures Working Group. Environmental and Molecular Mutagenesis 35: 173-184.
Anon (1998). European Commission, Directorate General XXIV. The Scientific Committee On Cosmetic Products And Non-Food Products Intended For Consumers--SCCNFP/0069/98 Final. Opinion on "in vitro methods to assess phototoxicity in the safety evaluation of cosmetics ingredients or mixtures of ingredients"; adopted by the plenary session of the SCCNFP of 25 November 1998.
Jacobs, A., Avalos, J., Brown, P. & Wilkin, J. (1999). Does photosensitivity predict photocarcinogenicity? International Journal of Toxicology 18: 191-198.
Roguet, R., Beek, H., Boelsma, E., Bracher, M., Faller, C., Harris, L, Lotte, C., Dreher, F. & Ponec, M. (1999). Testing and improvement of reconstructed skin kits in order to elaborate European standards: first results. ATLA 27: Special Issue, 333.

The second ECVAM workshop on phototoxicity testing.

The Report and Recommendations of ECVAM Workshop 421,2

The Report and Recommendations of ECVAM Workshop 42^1,2