In Vitro Phototoxicity Testing

The Report and Recommendations of ECVAM Workshop 21,2

Reprinted with minor amendments from ATLA 22, 314-348.

Horst Spielmann3, Will W. Lovell4, Erhard Hölzle5, Brian E. Johnson6,Thomas Maurer7, Miguel A. Miranda8, Wolfgang J.W. Pape9, Orazio Sapora10 and Dariusz Sladowski11
3ZEBET, Bundesinstitut für gesundheitlichen Verbraucherschutz und Veterinärmedizin (BgVV), Diedersdorfer Weg 1, D-12277 Berlin, Germany; 4Unilever Research, Colworth House, Sharnbrook, Bedford MK44 1LQ, United Kingdom; 5Department of Dermatology, Hautklinik, Universitäts-Krankenhaus Eppendorf, D-2000 Hamburg, Germany; 6Photobiology Unit, Department of Dermatology, Ninewells Hospital and Medical School, Dundee DD1 9SY, United Kingdom; 7Preclinical Safety K135.284, Ciba-Geigy Ltd, CH-4002 Basel, Switzerland; 8Departimento Quimica, Universidad Polytecnica, Valencia, Spain; 9Department of Biocompatibility K.St. 4232, Beiersdorf AG, D-20253 Hamburg 20, Germany; 10Comparative Toxicology Laboratory, Instituto Superiorè di Sanita, I-00161 Rome, Italy; 11Department of Transplantology, Institute for Biostructure, University Medical School, Warsaw 02-004, Poland

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

Address for correspondence: Dr med. Horst Spielmann, Direktor und Professor, ZEBET, BgVV, Bereich Marienfelde, Diedersdorfer Weg 1, D-12277 Berlin, Germany, Tel. ++49 30 7236 2270; Fax ++49 30 7236 2958

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

Preface

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

The workshop on In Vitro Phototoxicity Testing was held in Angera, Italy, from 13-17 December 1993, under the co-chairmanship of Horst Spielmann and Will Lovell. The participants at the workshop evaluated the present status of in vivo and in vitro methods for photosensitisation testing, with emphasis on achieving regulatory acceptance of validated in vitro methods in order to replace the current animal tests. The aspects of phototoxicity testing which were discussed in detail included: appropriate terminology, mechanisms of phototoxicity, the quality of both animal and clinical data from phototoxicity testing, in vitro methods for screening purposes and for mechanistic studies, technical aspects of photobiology studies (including UV exposure and light sources), the selection of reference chemicals, and the development of a strategy for photosensitisation testing.

An essential requirement for validation is the availability of a set of reference chemicals of known activities, which covers a spectrum of potencies and physico-chemical properties (e.g. lipophilicity) and, with respect to phototoxicity testing, includes chemicals which act by various photochemical mechanisms. The participants at the workshop have produced a list of 50 reference chemicals which are supported by high quality data from clinical studies in humans. An expert working group should review all the available data on these chemicals.

A sequential approach for testing for chemical-induced photosensitisation has been proposed. It is suggested that, if results from properly validated in vitro phototoxicity tests are negative, then human testing may proceed, subject to a satisfactory outcome relating to assessment of the general toxicological effects of the test material. However, if in vitro testing indicates that the test material may be phototoxic, then animal tests may be necessary to determine a no-observed-effect level (NOEL) prior to undertaking human studies, if these are considered to be appropriate.

Introduction

The current toxicological procedures for investigating the acute phototoxic effects of chemicals on the skin are animal tests employing guinea pigs, rabbits, rats or mice. Although a standard protocol for determining the phototoxic potentials of topically applied substances in animals has recently been recommended by an OECD expert committee (2), a test guideline has not yet been produced because the updating of other guidelines has higher priority. As a first step, a sequential approach for phototoxicity testing was recommended by the OECD, according to which in vitro tests should be carried out before considering any animal testing. Thus, in 1992, the European Cosmetic Toiletry and Perfumery Association (COLIPA) and DGXI of the European Commission (EC) agreed to conduct a joint validation project on in vitro phototoxicity tests. During this study it became apparent that many promising in vitro phototoxicity test systems, covering several mechanisms of phototoxicity, have been developed, and that patented human cutaneous models are also being developed for in vitro phototoxicity testing. Several mechanisms of chemical-induced phototoxicity have been identified, all of which should be considered when selecting in vitro tests for regulatory purposes (3).

A critical aspect of validation studies is the quality of the in vivo data which are available for the test chemicals selected. This problem immediately became apparent during the EC/COLIPA validation study on in vitro phototoxicity tests, with regard to both human and animal data. The procedures used in animal testing and in clinical studies for evaluating the phototoxic potentials of chemicals have not been standardised sufficiently. High quality human phototoxicity data are, with a few exceptions, difficult to find. Moreover, the quality of the data from experiments conducted in humans and in laboratory animals is adversely affected by differences in the application and dosing regimens of the test chemicals, light exposure conditions, and species specificity. Not surprisingly, in photobiology, as in many other areas of biomedicine, there are differences in chemical photosensitivity between animals and humans.

Terminology

It is essential to define the terminology used in the area of toxicology which relates to the effects of UV and visible light. Light-induced genotoxic effects, such as photomutagenicity and photocarcinogenicity, will not be discussed in this report. Thus, the terminology outlined below and in Figure 1 is restricted to definitions of the toxicological and technical terms for light-induced acute toxicity (4-6).

Figure 1: Terminology

Toxicological Terms

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) which causes another substance (the substrate) to be changed by 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 reaction which can be caused by a single treatment with a chemical and UV or visible radiation. In vivo, the reaction can be evoked in all subjects provided that the concentration of chemical and dose of light are appropriate. 'Acute' includes both immediate and delayed (e.g. after 48 hours) reactions. The term photoirritation is used to describe phototoxic reactions in skin which are produced with topically applied substances following exposure to light.

Photoallergy is an acquired immunological reactivity. The skin reaction does not occur on first treatment with chemical and light. Rather, an induction period of one or two weeks is required before skin reactivity can be demonstrated.

Technical Terms

Dose of light is defined as the quantity of UV or visible radiation incident on a surface, measured in Joules per square metre (J/m2).

Irradiance: J/(s x m2) = W/m2

Wavebands: UVA, 320-400 nm; UVB, 280-320 nm; UVC, >280 nm. The border between UVA and UVB is placed at 315 nm by the Commission Internationale de l'Eclairage (CIE), but the divisions are arbitrary. When considering photobiological phenomena, account should be taken of the emission spectrum of the lamp, and of the action spectrum of the phenomenon in question. Simple summation of radiation within a waveband is often misleading.

Mechanisms of Photosensitisation Reactions

Chemical photosensitisation as an adverse reaction may be induced by a broad spectrum of industrial or therapeutic agents, which may enter the body by ingestion, inhalation, injection or by topical application to the skin. If photosensitivity is to occur, absorption and distribution must be efficient and detoxification and excretion less so. Photosensitisation reactions are mostly associated with absorption of UVA (or, with dyes, absorption of visible light). Exogenous photosensitisers are not usually competitive with endogenous UVB chromophores, although benoxaprofen is a significant exception. Erythrocytes are less sensitive than nucleated cells to UVB, and so larger doses of UVB may be used when desired.

The chemistry and biology of photosensitisation reactions have been reviewed by Spikes (6) and Foote (7). The first step is the absorption by a molecule of a photon of UV or visible radiation, resulting in the formation of a short-lived electronically excited species. Photosensitisers have two electronically excited states, the singlet and the triplet. The singlet is the initial product of light absorption, whereas the triplet is usually much longer lived. With very few exceptions, photosensitisation oxidations proceed by way of the triplet photosensitiser; thus, effective photosensitisers are usually those which result in the formation of a high yield of a long-lived triplet.

In simple systems, the triplet photosensitiser can participate in two major oxygen-dependent reactions (8; Figure 2): by electron or hydrogen transfer (free radical) processes (Type I reactions), which may or may not require oxygen; or by energy transfer to oxygen to form excited state singlet oxygen (Type II reactions). The relative involvement of the Type I and Type II processes depends upon the chemical nature of the photosensitiser and the substrate, the reaction conditions (e.g. solvent, pH, and concentrations of photosensitiser, substrate, and oxygen) and, in some cases, on whether the photosensitiser absorbs light into its first or second absorption band. The essential feature governing whether free radical or singlet oxygen reactions occur is competition between the substrate and oxygen for the triplet photosensitiser (9).

Figure 2: Mechanisms of Photosensitisation

Singlet oxygen is strongly electrophilic and reacts only with electron-rich substrates (e.g. histidine). Substrates which are readily reduced or oxidised (e.g. cysteine) favour free radical reaction. Unsaturated fatty acids and cholesterol react via both mechanisms, with cholesterol giving characteristic products (the 5a-hydroperoxide with singlet oxygen, and a mixture of products, including the 7a- and 7β-hydroperoxides, from free radical attack). Of the nucleic acid bases, guanine residues are preferentially photooxidised and studies suggest that both Type I and Type II reactions may be involved (8,9). Triplet photosensitiser, singlet oxygen, and most free radicals are short-lived, having sub-millisecond lifetimes, and so must be produced in the immediate vicinity of the substrate if reaction is to occur. Dark association of photosensitiser with cellular macromolecules may favour covalent photobinding to the particular macromolecule; for example, the covalent binding of the photoallergen tetrachlorosalicylanilide to protein (10), and of the photomutagen psoralen to DNA (11).

The light-excited chemical may simply degrade. Stable photodegradation products of some photosensitisers may act as ordinary toxins (12,13), or as photosensitisers in their own right (14,15). Metabolites of some systemic drugs have been reported to be photosensitisers (8,16). Biological targets for photosensitisation oxidations include the plasma membrane, cytoplasmic organelles and the nucleus, depending on the uptake and localisation of the photosensitiser. Complement activation by photosensitising chemicals has also been suggested to be involved in phototoxic reactions, although phototoxic effects were only observed in vivo following exposure to UVB (8,16,17).

The potency of photosensitisation reactions would appear to depend at least as much on factors affecting the association between the photosensitiser and the cell prior to irradiation, as on its photochemical properties (6,8,16). Its octanol-water partition coefficient, solubility, ionic nature, and molecular volume are likely to be the main physico-chemical factors affecting the association of the photosensitiser and the cell.

Photoallergy is considered to be a delayed-type hypersensitivity reaction which is mediated by the formation of a photosensitiser-protein conjugate (10). Therefore, binding to skin proteins is an essential feature of a photoallergen.

Animal Data

Prediction of the phototoxic potentials of xenobiotics by conducting studies in laboratory animals has been used for many years and the sensitivities of the methods found to be acceptable. The main problem is that it is very difficult to compare the results presented in the published literature to evaluate the differing sensitivities of the various animal models. These difficulties arise, in part, from two major technical problems. Firstly, there is no generally accepted, standardised, procedure. Several of the animal test procedures which have been developed are summarised in Table I, which demonstrates the diversity of species, light sources, and UV doses employed. Secondly, many factors influence the outcome of a particular test (Table II), and not all of these are sufficiently well described in the literature.

Table I: Animal Test Procedures

Reference Species Route of exposure Irradiation light source radiance
(mW/cm2)		 Time Dose
(J/cm2)
Ison & Blank (85) mouse intraperitoneal FBLa 0.32 48 hours 51.8
Marzulli & Maibach (86) rabbit
man		 epidermal Hanovia 3 20-30 mins
40 mins		 3.6-5.4
7.2
Gloxhuber (87) mouse intraperitoneal mercuryb 24-72 hours
Sambuco & Forbes (88) mouse epidermal xenon
FBL		 1.7 136
Maurer (89) mouse epidermal
oral		 xenon
metal-halide
10
8-60 mins		 2
5-40
Gerberick & Ryan (90) mouse epidermal xenon 6-10 UVA, 0.30 UVB
Beiersbergen van Henegouwen (91) rat epidermal
Intraperitoneal		 mono-
chromatic		 3 1 hour 10.8
Lovell & Sanders (92) guinea-pig epidermal FBLa 1.8 140 mins 15
Nilsson et al. (2) guinea-pig
mouse		 epidermal
oral		 various 10

afluorescent backlight.
bmedium pressure mercury vapour arc.

Factors Affecting In Vivo Phototoxicity

The factors which influence the outcome of animal tests (Table II) are not equally significant. The incorporation of proper controls is most important. The irradiation control group indicates whether the total light dose and the filters have been selected correctly. An irritation control group confirms that the concentrations of the test compound were selected correctly. Phototoxic reactions induced by the appropriate combination of light and test chemical can only be classified correctly if these controls are negative. Information on the concentration of test substance per unit area is of greater importance than that on the concentration given as a percentage, since light absorption is dependent on the former parameter. The positive control group should confirm the responsiveness (or sensitivity) of the method. Animal data on moderately phototoxic compounds, rather than just on strong phototoxins such as 8-methoxypsoralen, are particularly valuable.

Table II: Factors Affecting Phototoxicity Testing in Animals

Two general types of light source are typically used: those with a limited emission spectrum, such as the fluorescent tubes for UVA or UVB, or mercury arc lamps; and those with a solar light simulating emission spectrum. Examples of the latter are xenon lamps, which emit a high proportion of infrared (IR), and the more modern metal-halide lamps, which emit more UV and less IR than xenon lamps. Since the biological effects of UV depend on the wavelength, and since energy increases with decreasing wavelength, information about the emission spectrum and filtering of the light source is very important, as is proper measurement of the UVA and UVB irradiance. Due to the strong, direct, effect of small doses of UVB, special care should be taken to monitor the UVB proportion of the spectrum.

The species used most frequently are mice, guinea pigs, and rabbits. Hairless strains of rats, mice, and guinea pigs are available, the use of which avoids the problem of hair removal. The choice of species depends upon the intended route of exposure; for example, guinea pigs are not generally used for i.v. injections. Higher priority should be given to the careful selection of suitable control groups, rather than to the selection of a particular species.

The route of administration and pharmacokinetics of the test substance will determine the appropriate time interval between its application and subsequent irradiation of the animals, which should take place at a time point when bioavailability of the substance is guaranteed. This time point may also depend on the vehicle used.

Experimental experience is necessary for the evaluation of any phototoxic reactions induced. The type of reaction depends on the species used. For example, skin fold thickness measurements are performed in mice, since they show strong oedema formation, whereas erythema reactions are more pronounced in guinea pigs.

The mouse tail swelling and ear swelling methods

The mouse tail swelling method (26-28) appears to give the most consistent results for systemic drug-induced phototoxicity (Table III). However, the tails have to be forcibly removed for the endpoint measurement (increase in weight), and the dose of chemical required to elicit a positive response appears to be much higher for mice than for humans. The mouse ear swelling method used by Wagai and Tawara (29) is possibly more versatile, but also requires high doses. It enables the detection of phototoxic responses following repeated exposure to physiologically relevant doses of chemical and irradiation, which may not be observed with a single exposure (30).

Table III: Phototoxicity of Systemic Drugs Determined Using the Mouse Tail Method

Positive chemicals test Mouse daily dose
(mg/kg)		 Human daily dose
(mg/kg)		 Negative chemicals test Mouse
(mg/kg)a
Amiodarone 400 Chloroquine 60
Benoxaprofen 50 Chlorotetracycline 200
Carprofen 50 Chlorothiazide 1200
Chlordiazepoxide 20 0.2 Clomocycline 200
Chlorpromazine 2.5 7 - 8 Cyclamate 1600
Diclofenac 100 Fenoprofen 200
Diflunisal 100 Flurbiprofen 200
Dimethylchloro-
tetracycline		 100 4 - 5 Ibuprofen 200
Doxycycline 50 1.5 Indoprofen 200
Griseofulvin 200 7 - 8 Ketoprofen 200
8-Methoxypsoralen 10 0.6 Methacycline 400
Nalidixic acid 50 15 - 30 Minocycline 200
Naproxen 200 7 - 8 Oxytetracycline 200
Tiaprofenic acid 100 4 - 5 Quinine 150
Sulphatisodimidine 2400
Tetracycline 200
Tolbutamide 600

Data are from the studies undertaken by Ljunggren and colleagues (26-28).

amaximum dose tested.

Standardisation of regulatory animal test procedures

It is accepted that phototoxicity and photoallergenicity are important side effects of some chemicals, for example, certain cosmetics and drugs. Therefore, general recommendations for conducting phototoxicity studies are included in some regulatory guidelines. However, a particular method is not usually specified. For example, Japanese guidelines for testing medicines include a list of methods for photoallergy testing, but do not include methods for phototoxicity testing.

A few years ago, Nilsson initiated the establishment of an international working group to develop a proposal for an OECD guideline for phototoxicity and photoallergy testing. In 1991, at an OECD meeting in Paris, the proposed animal procedure for phototoxicity testing was discussed. Subsequently, a preliminary validation trial of the phototoxicity protocol was conducted, the results of which were published in 1993 (2). A test for photoallergenicity was also proposed, but this was neither published nor discussed at the OECD meeting.

Recommendations

A group of experts should evaluate the data available on selected photosensitisers, with particular reference to the experimental in vivo models employed. This group should determine whether test substances are/are not photosensitisers, and whether their biological effects are caused via phototoxic or photoallergic mechanisms. Where possible, the relative potencies of the photosensitisers should also be estimated. Investigations of chemical mechanisms, conducted using both in vivo and in vitro techniques, should be reviewed. The data should then be compared with human data. The objective of this exercise should be to add scientific weight and validity to any subsequent selection of test chemicals used in the evaluation of in vitro test protocols.

Human Data

Human data are derived from three sources: anecdotal case reports, photopatch testing in patients with suspected photosensitivities, and studies on systemic photosensitisation in selected patients or healthy volunteers by systemic provocative phototest.

Case reports

Photodermatitis induced by an exogenous photosensitiser is a rare occurrence. Consequently, only isolated cases or observations in a few patients are published in the scientific literature. It seems appropriate to establish a database containing details of all such reports, to document the clinical relevance of photosensitisers. This would enable the identification of substances with low photosensitising potentials, as well as the stronger photosensitisers. The former could then be further evaluated by topical (photopatch test) or systemic (provocative phototest) testing procedures.

Photopatch tests

There are numerous isolated examples of this kind of approach in the literature (31-35). For example, a simple trial with solar simulator irradiation of skin exposed to suspected photosensitisers for 6 hours under occlusive covering gave positive results with fragrance materials, some halogenated phenolic bacteriostatic compounds, and one dyestuff. Other dyes were positive when applied to scarified skin (36; Table IV). The considerably larger multicentre studies from the Scandinavian countries (37) and from Austria, Germany, and Switzerland (38) appear to provide the most satisfactory data. These investigations were mostly concerned with the identification of topically applied photoallergens. However, phototoxic agents are also detected as long as they are able to penetrate into the skin. The test procedure may not be adequate to detect systemic photosensitisers, which rarely give positive results.

Table IV: Substances Which Are Phototoxic When Applied to the Skin

Fragrances Angelica root oil (at high doses)
Bergamot oil (negative when rectified)
Cumin oil
Lemon oil
Lime oil
Oil of bitter orange
Rue oil
Dyes Disperse blue 35
Eosin (when applied to scarified skin>)
Methylene blue (when applied to scarified skin)
Rose bengal (when applied to scarified skin)
Bacteriostatics Bithionol
Jadit
3,3',4',5-Tetrachlorosalicylanilide

Information is from Kaidbey & Kligman (36).

Of the bacteriostatics tested, 3,4',5-tribromosalicylanilide and hexachlorophene were negative.

The methodology for photopatch testing has not been standardised internationally (39). Within each collaborative study the procedure employed was fairly uniform, but there were considerable variations between the different studies. Currently, the following method is employed by the Austrian, German, and Swiss groups in their on-going trial. Test materials are applied via Finn chambers (aluminium chambers) to the skin on the back of the subject for 24 hours. Test sites are then irradiated at a dose of 10 J/cm2 (320-400 nm, peak 355 nm; Phillips TL09 fluorescent tube light source). Readings are taken immediately and after 24, 48, and 72 hours. The evaluation score differentiates between erythema, infiltration, papulovesicles, blisters, and erosions. According to the reaction pattern, it may be possible to distinguish between phototoxic and photoallergic mechanisms. An unirradiated patch test, as well as UV irradiation on untreated skin, serve as controls.

Summaries of the results of the photopatch tests undertaken by Hölzle et al (38), Menz et al (32) and Thune et al (37) are given in Table V (photoallergic reactions) and Table VI (phototoxic reactions). It is important to note that there is a considerable overlap of the chemicals listed in these two tables, demonstrating the ability of many substances to induce both photoallergic and phototoxic reactions in humans. The rank orders represent the frequency of the reactions observed in large groups of patients and, therefore, to a certain extent reflect the photosensitising potencies of the substances.

Table V: Rank Order of Substances According to the Frequency of Photoallergic Reactions in Standardised Human Patch Testing

Substance Frequency of photoallergic reactions (%)
(A) Austrian, German, and Swiss Study
(n=1129; ref. 38)
1. Tiaprofenic acid 3.15
2. Fenticlor 1.60
3. Carprofen 1.34
4. 4-Isopropyldibenzoylmethane 0.89
5. 2-Hydroxy-4-methoxybenzophenone 0.71
6. Promethazine 0.71
7. Tetrachlorosalicylanilide 0.54
8. Musk ambrette 0.54
9. Perfume mix 0.37
10. Musk mix 0.36
11. Chlorpromazine 0.36
12. Triclosan 0.27
13. Bithionol 0.27
14. 6-Methylcoumarin 0.18
15. Hexachlorophene 0.18
16. Compositae mix 0.10
17. Cyclamate 0.09
18. Quinidine 0.09
19. Buclosamide 0.09
20. Wood tar 0.09
21. p-Aminobenzoic acid 0.09
22. Balsam of Peru 0.00
23. Tolbutamide 0.00
24. 3-(4-Methylbenzylidene)-camphor 0.00
25. Colophony 0.00
26. Bromosalicylchloranilide 0.00
27. Saccharin 0.00
28. Chlorothiazide 0.00
29. Sulphanilamide 0.00
30. Thiourea 0.00
31. Furosemide 0.00
32. Tribromosalicylanilide 0.00
(B) Scandinavian Multicentre Study*
(n=1993; ref. 37)
1. Musk ambrette 2.7
2. p-Aminobenzoic acid 2.1
3. Promethazine 1.8
4. Chlorpromazine 1.6
5. Fenticlor 1.2
6. Balsam of Peru 0.7
7. Tetrachlorosalicylanilide 0.6
8. Bithionol 0.5
(C) Mayo Clinic Study*
(n=70; ref. 32)
1. Chlorpromazine 18
2. Musk ambrette 13
3. Promethazine 11
4. Fenticlor 7
5. 6-Methylcoumarin 7
6. Hexachlorophene 6

*Distinction between photoallergic and phototoxic reactions could not be made with certainty.

Table VI: Rank Order of Substances According to the Frequency of Phototoxic Reactions in Standardised Human Patch Testing

Substance Frequency of phototoxic reactions (%)
1. Tiaprofenic acid 58
2. Promethazine 34
3. Carprofen 12.5
4. Chlorpromazine 10
5. Fenticlor 9.5
6. Hexachlorophene 6
7. Perfume mix 5.5
8. 4-Isopropyldibenzoylmethane 4.5
9. Balsam of Peru 4
10. Wood tar 3.5
11. Tetrachlorosalicylanilide 3
12. 2-Hydroxy-4-methoxybenzophenone 2.5
13. Bithionol 2.5
14. 6-Methylcoumarin 2.5
15. Colophony 2
16. Musk mix 2
17. Buclosamide 2
18. Triclosan 2
19. Bromosalicylchloranilide 1.5
20. p-Aminobenzoic acid 1.5
21. Cyclamate 1.5
22. Tribromosalicylanilide 1.5
23. Musk ambrette 1.5
24. 3-(4-Methylbenzylidene)-camphor 1.5
25. Compositae mix 1.5
26. Tolbutamide 1
27. Quinidine 1
28. Sulphonamide 1
29. Saccharin 1
30. Thiourea 0.5
31. Chlorothiazide 0.5
32. Furosemide 0.5

Data are from the Austrian, German, and Swiss multicentre study (38).

A variation of the photopatch test has been developed to try to detect systemic phototoxic agents. The test substance is injected intradermally prior to exposure to irradiation (40,41). The test appears to work quite well (Table VII), but there may be problems with local reactions to the test substance itself, and metabolism is not taken into account. Other modifications of the photopatch testing procedure include the application of material to the surface of the skin, and scarification of the skin prior to application (41). However, the results obtained have not been entirely satisfactory.

Table VII: Evaluation of the Phototoxicity of Chemicals Applied Systemically by Intradermal Testing in Humans

Positive with solar spectrum (i.e. including UVB)
Chlorothiazide
Sulphapyridine
Sulphonamide
Vinblastine

Positive with UVA
Chlorothiazide
Chlorpromazine
Dimethylchlorotetracycline
Furosemide
Griseofulvin
Nalidixic acid
Prochlorperazine
Quinidine
Tetracycline

Information is from Kaidbey & Kligman (40).

In addition to the investigations mentioned already, specific studies on the phototoxic effects of psoralens, dyestuffs, and polycyclic hydrocarbons have been undertaken (31, 42-45). The results obtained vary according to the concentrations of substances and the vehicles used, the method of application (i.e. open, occluded, or to scarified skin), time of contact with the skin, time between application and irradiation, the radiation dose and spectrum of the irradiation source, and the evaluation procedure (time frame of readings, grading system). A distinction between phototoxic and photoallergic reactions cannot always be made with certainty, and the clinical relevance of test results for the patients is frequently lacking.

Systemic provocative phototests

A number of methods for detecting systemic photosensitisers have been published (39-41, 46-55). They vary from phototesting subjects who are being treated for a certain disease, through relatively uncontrolled testing with volunteers, to controlled double-blind testing procedures which include positive and placebo controls.

Two similar standardised protocols have been established. One of these involves the administration of a single dose of test compound, with phototesting of different skin sites before and at varying times after dosing (39,54). Phillips TL09 fluorescent tube lamps (320-400 nm, peak 355 nm) are used as the radiation source. Irradiations are performed, with a UVA dose of 10 J/cm2, before and 1, 3, and 6 hours after treatment with the test material. Readings are made during the first six hours, and at 24, 48, and 72 hours, for erythema and any abnormal morphological changes in the skin. The method can be modified to determine threshold responses and wavelength dependencies.

The other protocol involves repeated treatment with the test material and the determination of minimal erythema doses (MED), using different wavebands of UV through to visible radiation, both before and after treatment. A monochromator (wavebands 305+5 nm, 335+30 nm, 365+30 nm, 400+30 nm, and further into visible light if required) is employed as the radiation source. The MED is determined for each waveband before treatment and phototesting is repeated, at the time of peak plasma levels, after seven days of daily dosing. Readings (MED and abnormal morphological responses) are made during the first six hours, and then at 24, 48, and 72 hours. If abnormal responses are obtained, testing is repeated until the response returns to normal. An ideal trial includes two dose levels of the suspected photosensitiser, positive control, and placebo. The protocol enables phototoxic effects resulting from repeated exposure, as well as from a single treatment, to be detected. It also provides information about the wavelength dependency for drug-induced phototoxicity by employing the relatively sophisticated monochromator as the radiation source.

In studies employing variations of these two protocols, the following substances have been shown to cause systemic photosensitivity: 8-methoxypsoralen, 5-methoxypsoralen, quinidine, hydrochlorothiazide, chlorpromazine, benoxaprofen, amiodarone, tetracycline, doxycycline, and quinolones (ciprofloxacin, norfloxacin).

Systemic provocative phototesting is the method of choice to detect potential systemic photosensitisers. It is highly sensitive and specific, can distinguish between phototoxic and photoallergic mechanisms, and serves as a good diagnostic tool in patients. Used as a screen with human volunteers, the phototoxic properties of systemically administered compounds can be evaluated, with each volunteer serving as his/her own control of baseline photosensitivity. In this way, abnormal photosensitivity in an individual may be demonstrated without showing sensitivity greater than that at the extreme of a normal population. However, the method is limited to testing a few compounds and subjects, since only a single substance can be tested in each subject at any given time.

Recommendations

The collection of human data on photosensitisers is essential since they serve as the 'gold standard' for the validation of in vitro methods and animal procedures. Human data derive from three sources: anecdotal case reports, photopatch tests to evaluate topical photosensitisers, and systemic photosensitisation studies by systemic provocative phototesting. The further elaboration of each of these sources of data is recommended.

  1. A database of anecdotal case reports should be created, which would help in detecting substances with low photosensitising potentials. Newly identified photosensitisers could then be subjected systematically to screening, either in photopatch tests or in systemic provocative phototests, depending on their route of administration.
  2. The method of choice for evaluating topical photosensitisers on human skin is the photopatch test. The ongoing large multicentre trial (38) should be continued and expanded. It is important to obtain further epidemiological information on topical photosensitisers by expanding the multicentre trials to other European countries. Grants should be made available to enable current studies to continue and to fund new trials in other countries.
  3. Systemic photosensitisers can be detected by systemic provocative phototesting, for which there are two similar, relatively standardised, procedures. Studies are needed to further optimise and validate these procedures. Multicentre studies should be initiated to collect reliable data on systemic photosensitisers in different European countries. Although systemic provocative phototesting is complex, it seems to be the only procedure suitable for evaluating systemic agents; results obtained from photopatch tests with systemic photosensitisers have been unreliable and disappointing.

In Vitro Methods

Numerous in vitro methods have been developed to assess the phototoxic potentials of chemicals by both academic and industrial laboratories. These can be assigned to two general groups: (a) those using cells and tissues for screening purposes; and (b) tests focusing on a specific mechanism of phototoxicity. A broad spectrum of cell and tissue culture systems, including tests which employ non-mammalian cells, has been developed for assessing the phototoxic potentials of chemicals (Table VIII). Several specific mechanisms which are responsible for chemical-induced phototoxicity have been identified; tests which are predominantly based on subcellular and molecular mechanisms of phototoxicity are also listed in Table VIII.

Table VIII: In Vitro Methods for Phototoxicity Testing

Type of test Protocol Priority
Methods for screening purposes
3T3 NRU PT assay P I
SKIN2 TM PT assay P II
Human keratinocytes UD III
Hepatocytes UD III
Candida yeast test Ex- III
Human lymphocytes Ex- III
Jurkat human lymphoma cells UD III
SOLATEX-PITM assay UD III
Methods for evaluating mechanisms
RBC haemolysis P I
Haemoglobin photooxidation P I
Histidine photooxidation P I
Photobinding to protein (HSA) P I
Linoleic acid peroxidation UD II
Complement PT assay UD III

NRU - neutral red uptake; PT - phototoxicity; RBC - red blood cell; HSA - human serum albumin

P: standard protocol available; UD: protocol under development, not yet standardised; Ex-: examined and dismissed

I: highest priority, test recommended for validation; II: high priority, test should be considered for validation; III: low priority, test not sufficiently developed

Methods for screening purposes

3T3 Fibroblast neutral red uptake phototoxicity assay (3T3 NRU PT assay)
In a recent EC/COLIPA validation trial of in vitro methods for phototoxicity testing, the NRU growth inhibition assay using BALB/c 3T3 fibroblasts to determine cytotoxicity was adapted for phototoxicity testing in the following way: Balb/c 3T3 cells, clone 31 (ICN-Flow), were cultured in 96-well microtitre plates (56-58). After 24 hours, the culture medium (DMEM) was removed, the cells were washed twice in Earle's balanced salt solution (EBSS), and eight concentrations of the test chemicals (dissolved in EBSS) were added. Insoluble test chemicals were dissolved in dimethylsulphoxide (DMSO) prior to use, and were added at a maximum concentration of 1% DMSO in EBSS. After preincubation with test chemicals for 1 hour, the cells were exposed to UVA (1.67 mW/cm2) for 50 minutes (= 5 J/cm2). During this period, a second set of plates, in which the cells had been treated with the same chemicals, was kept in the dark. The EBSS was then replaced with DMEM (without any test chemical) and NRU was determined 24 hours later, as described by Spielmann et al (59).

Data from the 3T3 NRU PT assay were analysed in the following way. Cytotoxic concentrations resulting in a 50% reduction in cell viability (IC50) with and without UV irradiation were compared by calculating the ratio of the IC50 values: UV factor = IC50(-UV)/IC50(+UV). The cut-off value of this factor for distinguishing phototoxic from non-phototoxic chemicals has been calculated using discriminant analysis of all the factors which were determined for all of the test chemicals in each of the seven laboratories participating in the EC/COLIPA study. The results for the 20 chemicals studied (11 phototoxins, 5 UV-absorbing non-phototoxins, and 4 non-UV-absorbing non-phototoxins, for which sufficient in vivo data were available), which were selected by a COLIPA task force, are given in Table IX. Surprisingly, the simple 3T3 NRU PT assay was able to identify the phototoxic as well as the non-phototoxic chemicals (56-58).

Table IX: EC/COLIPA Validation Study of In Vitro Phototoxicity Assays - 3T3 NRU PT Assay

Substance IC50-UV
(µg/ml)		 IC50+UV
(µg/ml)		 Ratio
(-UV/+UV)		 Classification
Class I UVA-absorbing, phototoxic
Promethazine 45.9 0.8 78.5 +
Chlorpromazine 24.6 0.6 46.6 +
6-Methylcoumarin *a 32.7   +
Tetrachlorosalicylanilide 19.8 0.4 55.6 +
Doxycycline 1182 6.4 255 +
8-Methoxypsoralen *a 14.7   +
Tetracycline 1916 16.8 374 +
Amiodarone 24.3 4.1 6 +
Bithionol 13.9 3.9 7 +
Neutral red *a 0.01   +
Rose bengal 4.2 0.2 70.2 +
Class II UVA-absorbing, non-phototoxic
Piroxicam *a *b   -
Cinnamic aldehyde 32.8 10.6 3.6 -
Chlorhexidine 61.5 74.4 1.5 -
Uvinul MS 40 15960 11580 1.4 -
p-Aminobenzoic acid 10460 9780 1 -
Class III non-UVA-absorbing, non-phototoxic
Penicillin G 53910 49760 1.1 -
L-Histidine *a *c   -
Thiourea 17650 16940 1 -
Lauryl sulphate 35.6 24.2 1.2 -

Data are from Liebsch et al (56).

Values are arithmetic means of 5-13 determinations.

+ = phototoxic; - = non-phototoxic

*a = IC50 could not be determined (insufficiently cytotoxic); *b = highest concentration tested was 2.4 mg/ml
*c = highest concentration tested was 46.4 mg/ml

The 3T3 NRU PT assay is easy to perform and seems very promising as a screening test. However, it requires further study and the data presented in Table IX do not appear to bear any relationship to phototoxic potency in vivo. Validation of the 3T3 NRU PT assay will continue in 1994, in an international blind trial jointly sponsored by ECVAM and COLIPA.

SKIN2 TM phototoxicity assay
The SKIN2 TM dermal model, ZK 1300/ZK 1350, is a three-dimensional human skin model which has dermal, epidermal and corneal layers. Neonatal fibroblasts are seeded onto an inert nylon mesh and are grown into dermal tissue. Keratinocytes are seeded on top of this tissue; these differentiate into an epidermis, including a multilayered stratum corneum. This model is available as 11 mm2 (ZK 1300) or 9 mm2 (ZK 1350) pieces (Advanced Tissue Sciences, Inc., California). The model has been used successfully for skin irritation testing. A UVA protocol for testing for phototoxic potential has been developed, which is being evaluated in the EC/COLIPA validation study. The test is based on comparison of the cytotoxic effects of a chemical with and without exposure to a non-toxic dose of UVA. Cytotoxicity is determined by measuring chemical-induced inhibition of mitochondrial activity (MTT reduction assay) compared to that for solvent/vehicle controls (56,57).

On the day of receipt the tissues are placed into Millicell plates (Millipore) containing maintenance medium. To perform the assay, this maintenance medium is replaced with serum-free assay medium. Test chemical (50 µl), either dissolved in water or homogenised in corn oil, is then applied to the tissues with a paper patch. Three concentrations of each chemical are tested, with three replicates per concentration. After preincubation for 24 hours, the patch is removed and the plates are irradiated (3 J/cm2, i.e. 1.67 mW/cm2 for 30 min). Control plates are kept in the dark. After a further incubation for 30 minutes, the tissues are washed with phosphate-buffered saline (PBS), transferred to Millicell plates, and cell viability is assessed using the MTT reduction assay. The cut-off point for classifying a chemical as phototoxic is a reduction in cell viability of more than 30% compared with the corresponding non-irradiated controls (56).

Initial results obtained with the SKIN2 TM phototoxicity assay are quite encouraging (Table X; 56-58). These results have been confirmed independently in studies carried out at the manufacturers' laboratories. The advantages of this test are that it is the only standardised semi-organised three-dimensional human skin model which is available for testing purposes, aqueous and lipid-soluble substances, as well as solid materials, can be tested, and strict quality control procedures are employed during the manufacturing process. The disadvantages include the high cost of the system, the problems of air transportation, and the higher penetration rate of substances through the stratum corneum of the SKIN2 TM model compared with through that of normal human skin.

Table X: EC/COLIPA Validation Study of In Vitro Phototoxicity Assays - SKIN2 TM (Model ZK 1350) Phototoxicity Assay

Substance Classification
Class I UV-absorbing, phototoxic
Promethazine +
Chlorpromazine +
6-Methylcoumarin -
Tetrachlorosalicylanilide +
Doxycycline +
8-Methoxypsoralen +
Tetracycline +
Amiodarone +
Bithionol -
Neutral red +
Rose bengal +
Class II UV-absorbing, non-phototoxic
Piroxicam -
Cinnamic aldehyde -
Chlorhexidine -
Uvinul MS 40 -
p-Aminobenzoic acid -
Class III non-UV-absorbing, non-phototoxic
Penicillin G -
L-Histidine -
Thiourea -
Lauryl sulphate -

Data are from Liebsch et al (56).

Human keratinocyte cultures
The effects of irradiation on xenobiotic-induced toxicity in human keratinocytes and in fibroblasts (primary cells or cell lines) have been compared, to study light-induced toxicity in different systems (60,61). However, the light sources and spectral distribution were different. Xenon arc lamps and fluorescent tubes were used, with and without UVB radiation. If UVB was included, the irradiance and dose had to be decreased to avoid direct cytotoxic reactions which may overlap with the phototoxic effects induced by certain xenobiotics.

Screening of photosensitisers in primary human keratinocyte cultures has not shown any obvious advantage compared with using permanent fibroblastic cell lines, such as BALB/c 3T3 mouse fibroblasts, which have already been used in validation trials (58,62,63). These studies support the observation of Maier and co-workers (61) that keratinocytes (either primary cultures or permanent cell lines) are less sensitive than fibroblasts to irradiation.

A standard protocol for using primary keratinocytes or appropriate keratinocyte lines for investigating the phototoxic effects of chemicals has not yet been developed. However, in organotypic tests and for certain specific purposes it is preferable to use keratinocytes, since in vivo they are the first target cell type exposed to sunlight. Nevertheless, in several in vitro phototoxicity tests, keratinocytes have been shown to be less sensitive than fibroblasts to the effects of UV light.

Further standardisation of in vitro phototoxicity tests using human keratinocytes is essential. Results should be compared with those obtained using other cell types (e.g. A431 human epidermal cell line, 3T3 mouse fibroblast cell line, and V79 hamster fibroblast cell line). Human keratinocytes appear to be less useful than fibroblasts for screening purposes. They may, however, be preferable for incorporating into organotypic models.

Hepatocyte cultures
Exposure of hepatocyte cultures to test chemicals and subsequent irradiation (e.g. using a 400 W mercury lamp with a Pyrex filter) may cause cytotoxic effects, which can be evaluated using numerous different criteria, including the assessment of mitochondrial integrity (e.g. the MTT reduction assay) and of cell membrane integrity (e.g. lactate dehydrogenase leakage). Using such a test system, tiaprofenic acid and suprofen (10-4 - 10-3 M) were found to be phototoxic according to both endpoint measurements, while ibuprofen produced no significant effect (64,65). This is in agreement with the reported in vivo data.

Hepatocytes are an important model for investigative purposes (66), for example, for determining the effects of metabolism on the phototoxic potentials of chemicals. However, they cannot be used as a simple screening procedure, since the preparation of hepatocytes is complex and they are not robust in culture. If a hepatocyte cell line maintaining the metabolic properties of freshly-isolated primary cells became available commercially, this could provide a very valuable in vitro screen. At present, however, the influence of metabolism is not a primary concern with respect to photosensitisation testing.

Candida yeast test
The yeast test using Candida albicans, as developed by Daniels (67), is perhaps the most simple in vitro phototoxicity test (68-72). Test material, either as a solid or as a solution applied to filter paper discs and air dried, is placed on Sabouraud's dextrose agar plates, freshly seeded with Candida. Alternatively, the non-pathogenic C. utilis (68), or even brewer's yeast, may be used. For best results, a standard 4 mm thickness of agar should be used. Seeding from a sterile water suspension of the yeast (5 ml), made up with a standard bacterial loop from a 48 hour culture, should produce a damp, but not wet, surface to the agar. Duplicate plates are set up, one serving as a dark control, the other being exposed to UVA or visible radiation from an array of fluorescent tubes at an irradiance of about 0.2 mW/cm2 for 48 hours. Phototoxicity is indicated by a clear zone around the test material in the irradiated cultures, while growth in the dark controls is unaffected.

The Candida assay is a very simple method, which works well with linear furocoumarins (psoralens [72; Table XI]), providing quantitative data and information about wavelength dependency. However, many photosensitisers are not detected by the assay (Table XI). From a technical point of view, there is the potential for yeast contamination in the tissue culture laboratory. Although the Candida yeast phototoxicity assay is well established and is supported by a large database, it is not suitable or necessary as a screening test, since simple and reproducible mammalian in vitro tests have been developed which show better predictivities.

Table XI: Phototoxicity Testing In Vitro with Candida Albicans

Type of substance Positive Negative
Plant extracts Umbelliferae, Giant Hogweed, Cow Parsnip Compositae:
Chrysanthemum, Tansy, Yarrow, Sagebrush, Ragweed, Dahlia, Cornflower, Coreopsis, Aster, Burweed, Wild Feverfew, Lettuce, Chicory, Golden Rod, Black Eyed Susan, Burdock, Sunflower, Tocklebur, Dog Fennel, Sneezeweed,
Chemicals 8-Methoxypsoralen, 5-Methoxypsoralen, Trimethylpsoralen Alantolactone, Isoalantolactone
Fragrance materials Oil of bergamot, 6-Methylcoumarin, Costus root oil, Cinnamaldehyde, Cinnamyl alcohol, Amylcinnamaldehyde, Jasmine mix, Laurel leaf oil, Colophony Musk ambrette, Oak moss, Balsam of Peru, Atranorin, Usnic acid, Isoeugenol, Hydroxycitronella, Eugenol, Geraniol, Benzylbenzoate, Benzylsalicylate, Benzylalcohol, Methylsalicylate
Dyestuffs Benzanthrone, Eosin, Methylene blue, Toluidine blue Anthraquinone
Drugs Chlorpromazine, Benoxaprofen Amiodarone, Azapropazone, Carbamazepine, Cimetidine, Dimethylchlorotetracycline, Diflunisal, Doxycycline, Griseofulvin, Imipramine, Hydrochlorthiazide, Methyl-DOPA, Nalidixic acid, Oxytetracycline, Piroxicam, Propranalol, Protriptyline, Sulphamethoxazole, Sulphapyridine, Tetracycline, Trimethoprim,

Data are from Johnson et al (72).

Human lymphocytes
This method, using human peripheral blood lymphocytes, is essentially a standard immunocytochemical method for determining the response of lymphocytes to the mitogen, phytohaemaglutinin (PHA), as adapted by Scherer et al (73) to study psoralen phototoxicity and as used by Morison et al (74) as a phototoxicity screening test. Phototoxicity is determined by assessing the inhibition of PHA-induced DNA synthesis (measured by [3H]-thymidine uptake) in lymphocyte cultures.

In outline, heparinised blood (10 ml) and PBS (10 ml) are centrifuged (Ficoll/Hypaque gradient; 400 g, 30 min), and the resulting lymphocyte preparation is washed twice in PBS, diluted to a density of about 106 cells/ml, and then plated out (50 µl/well) in 96 well microtitre plates. The test material is added and the cells are irradiated at the appropriate wavelength. Nutrient medium containing PHA (10 µl/ml) is added to the lymphocyte cultures, which are then incubated at 37°C for 72 hours prior to the addition of [3H]-thymidine (0.5°Ci/well) and further incubation for 24 hours. The cells are then harvested and counted using a scintillation counter.

The use of medium containing HEPES buffer avoids the need for special incubation conditions. The method detects most known photosensitisers (72,74; Tables XII & XIII) and, using 96 well plates, dose-dependent (test material and radiation) studies can be rapidly undertaken. The disadvantages of the method include the use of human blood and radioisotopes, the need for complex and expensive instrumentation, and the significant intra- and inter-individual variation in the sensitivity of lymphocytes to UVA, both by itself and in the presence of medium constituents. In conclusion, the human peripheral lymphocyte test has not reached the standard which is essential for appropriate screening tests; simpler cell culture techniques should be able to replace this test.

Table XII: Phototoxicity Assay using Human Peripheral Lymphocytes

Compound Concentration
(mg/ml)		 ID50a
(J/m2 x 10-5)
8-Methoxypsoralen 1.0 0.0019
0.1 0.08
Chlorpromazine 1.0 0.1
Dimethylchlorotetracycline 20 0.09
2.0 0.14
Retinoic acid 3.0 0.21
Nalidixic acid 30 0.32
Hydrochlorothiazide 1.0 0.46
Tetrachlorosalicylanilide 1.0 0.48
6-Methylcoumarin 10 0.63
Musk ambrette 20 0.92
Chlorpropamide 25 4.00

Data are from Morison et al (74).

aID50: dose of UVA causing 50% inhibition of PHA-induced DNA synthesis.

Hexachlorophene, sulphanilamide, ethyl alcohol, acetylsalicylic acid, caffeine, and paracetamol were negative in this assay.

Table XIII: Phototoxicity Testing In Vitro Using Lymphocytes

Type of Substance Positive Negative
Chemicals Angelicin, 3-Carbethoxypsoralen, 5-Methoxypsoralen, 8-Methoxypsoralen
Dyestuffs Benzanthrone
Drugs Amiodarone, Azapropazone, Dimethylchlorotetracycline, Doxycycline, Frusemide, Griseofulvin, Hydrochlorothiazide, Methyl-DOPA, Nalidixic acid, Oxytetracycline, Protriptyline, Sulphapyridine, Tetracycline, Trimethoprim, *Benoxaprofen, *Ciprofloxacin, *Etretin, *Fleroxacin, *Norfloxacin, *Ofloxacin, *Quinine, *Tretinoin Carbemazepine, Chloropropamide, Diflunisal, Minocycline, Propanolol, Sulphamethoxazole, *Etretinate

(*) Data are from Johnson et al (72) and subsequent publications of clinical and laboratory studies.

Jurkat human lymphoma cell line phototoxicity assay
As part of the EC/COLIPA validation study of in vitro phototoxicity assays, the Jurkat E6 cell line (human lymphoma), which is available from the European Collection of Animal Cell Cultures (ECACC), was used in the FRAME laboratory (Nottingham, England) to develop an in vitro phototoxicity test. The MTT reduction assay was used to determine cytotoxicity in UV-exposed and unexposed cells in the presence of photosensitisers. For coloured chemicals, the MTT assay may be performed on Multiscreen filtration plates (75). Since both UVA and UVB radiation (UVA: 1.14 mW/cm2, UVB: 0.087-0.103 mW/cm2; Sylvana GTE fluorescent tubes, 20 W; 30 minutes exposure) were used in the study, it is difficult to compare the photosensitivity of the Jurkat cell line to that of other mammalian cell lines, such as BALB/c 3T3 fibroblasts. The data obtained with the Jurkat E6 cell line (56; Table XIV) were less predictive than those obtained using the 3T3 NRU PT assay for the same group of test chemicals.

Table XIV: EC/COLIPA Validation Study of In Vitro Phototoxicity Assay - Jurkat Human Lymphoma Cell Line Phototoxicity Assay

Substance Classification
Class I UV-absorbing, phototoxic
Promethazine +
Chlorpromazine +
6-Methylcoumarin +
Tetrachlorosalicylanilide -
Doxycycline +
8-Methoxypsoralen +
Tetracycline -
Amiodarone +
Bithionol -
Neutral red +
Rose bengal nt
Class II UV-absorbing, non-phototoxic
Piroxicam -
Cinnamic aldehyde -
Chlorhexidine -
Uvinul MS 40 -
p-Aminobenzoic acid -
Class III non-UV-absorbing, non-phototoxic
Penicillin G -
L-Histidine -
Thiourea -
Lauryl sulphate -

Data are from Liebsch et al (56).

+ = phototoxic, - = non-phototoxic; nt = not tested

The Jurkat human lymphoma cell line assay may be a useful screen for potential photosensitisers, which could possibly replace the use of freshly isolated human lymphocytes in phototoxicity testing. The database is limited and evaluation of the assay should continue. However, it is probable that other mammalian cell lines would be better suited for in vitro phototoxicity testing.

SOLATEX-PITM assay
This assay is a modification of a commercial system developed for the detection of skin irritants, known as SKINTEXTM (In Vitro International, California). Both the SKINTEXTM and SOLATEX-PITM assays can be used with soluble and insoluble materials, and they do not require highly trained technical staff or special tissue culture facilities. The principle of the SOLATEX-PITM system is the quantification of an enhanced response upon exposure to UV light in a two-compartment, physico-chemical model of dermal irritation. Controls which are not exposed to UV radiation provide the background irritation response. Two calibrators are analysed in each assay, to ensure standardisation and reactivity of the barrier matrix and the reagent. Two concentrations of a known phototoxin are included to verify the test performance.

The SOLATEX-PITM assay has to be used with a special UV source provided by the manufacturer (UVA fluorescent bulbs, type F20T12/BL, General Electric). In the EC/COLIPA validation study, a standard protocol was developed for the SOLATEX-PITM assay at the ZEBET laboratory (Berlin, Germany), in collaboration with the manufacturer. It was found that unusually high doses of UV (70-80 J/cm2) had to be used (56,57).

The toxic response is quantified as a change either in a biomembrane barrier or in a highly ordered macromolecular matrix. Neutral red, attached to the biomembrane barrier (compartment 1), is released if the integrity of this barrier is affected. Chemical irritants can also affect the matrix (compartment 2) to produce turbidity. The total response is measured spectrophotometrically at 400 nm. Those materials which, following exposure to UV light, exhibit an increase in optical density of 21-40% above the background reading are classified as borderline photoirritants; materials which give an increase greater than 40% are classified as potential photoirritants (56,57).

According to the results obtained with the test chemicals used in the EC/COLIPA validation study (56,57; Table XV), the predictive value of the SOLATEX-PITM assay was lower than that for the other in vitro phototoxicity assays evaluated, such as the 3T3 NRU PT assay (Table IX) and the SKIN2TM phototoxicity assay (Table X). The mechanistic basis of the assay still remains obscure, and further development is necessary to demonstrate the usefulness of the test kits.

Table XV: EC/COLIPA Validation Study of In Vitro Phototoxicity Assay - SOLATEX-PITM Assay

Substance Classification
Class I UV-absorbing, phototoxic
Promethazine +
Chlorpromazine +
6-Methylcoumarin +
Tetrachlorosalicylanilide +
Doxycycline +
8-Methoxypsoralen -
Tetracycline +
Amiodarone +
Bithionol +/-
Neutral red +
Rose bengal +
Class II UV-absorbing, non-phototoxic
Piroxicam -
Cinnamic aldehyde -
Chlorhexidine -
Uvinul MS 40 +
p-Aminobenzoic acid -
Class III non-UV-absorbing, non-phototoxic
Penicillin G -
L-Histidine -
Thiourea nq
Lauryl sulphate -

Data are from Liebsch et al (56).

+ = phototoxic; - = non-phototoxic; nq = not qualified.

Methods for evaluating mechanisms

Red blood cell (RBC) phototoxicity assays
Photohaemolysis is one of the oldest and simplest in vitro techniques for the screening of putative photosensitisers and, as such, was proposed almost a hundred years ago (76). Many different protocols, using different light sources and erythrocytes from various species, have been reported. In addition to screening for possible photosensitisers, RBCs have been widely used for mechanistic studies and, in particular, for investigating oxygen-dependent membrane damage. Phototoxicity due to effects on DNA cannot be detected in this model. Based on practical experience and on the work of others (77,78), Hetherington and Johnson (79) published their experimental approach, which was intended to serve mainly as a method for determining the relative phototoxic potentials of a wide range of drugs and environmental chemicals.

Also of interest are the free radical production and oxidative reactions of haemoglobin, as summarised by Winterbourn (80). Since methaemoglobin (metHb) formation is often induced by xenobiotics, especially phototoxins, a RBC phototoxicity test has been proposed recently which investigates the effects of chemicals on both photohaemolysis and haemoglobin (Hb) oxidation (62,63). This simple approach enables the screening of chemicals while also providing some mechanistic information on the effects of putative photosensitisers at the cellular level.

Modifications of the protocol, such as the use of a 500 W sun simulator with reduced UVB irradiance (1 J/cm2 UVB, 15 J/cm2 UVA and visible light), and the spectrophotometric measurement of metHb formation at 630 nm, allow the differentiation between intra- and extracellular Hb oxidation in the presence of irradiated xenobiotics, which may have also induced photohaemolysis (as determined at 525 nm). Results from a recent study are given in Table XVI (62,63). In some cases, more complex processes may lead to Hb denaturation, resulting in Heinz body formation (80). Haemoglobin often seems to be a target in Type I reactions. However, some chemicals, such as Rose bengal, seem to attack erythrocytes mainly from the outside, without causing metHb formation, and thereby act predominantly by a Type II mechanism (62,63).

Table XVI: RBC Phototoxicity Assay

Substance Photohaemolysis Hb Photooxidation
Class I UV-absorbing, phototoxic
Promethazine + +++
Chlorpromazine ++ +++
6-Methylcoumarin + ++
Tetrachlorosalicylanilide + ++
Doxycycline - ++
8-Methoxypsoralen - +
Tetracycline - ++
Amiodarone + ++
Bithionol ++ ++
Neutral red ++ ++
Rose bengal +++ +
Class II UV-absorbing, non-phototoxic
Piroxicam - -
Cinnamic aldehyde - +
Chlorhexidine + -
Uvinul MS 40 - -
p-Aminobenzoic acid - -
Class III non-UV-absorbing, non-phototoxic
Penicillin G - -
L-Histidine - -
Thiourea - -
Lauryl sulphate - -

Data are from Pape et al (62,63).

+++ = strong; ++ = medium; + = weak; - = no effect

The combined RBC photohaemolysis/haemoglobin photooxidation assay is a useful test both for screening chemicals and for mechanistic investigations. In particular, the Hb photooxidation assay has shown high predictivity. Since chemicals which react with DNA cannot be detected using RBCs, additional tests have to be performed to cover all mechanisms of phototoxicity.

Histidine photooxidation test
The photosensitised oxidation of histidine may be used as a mechanistic-type test (72,81). Solutions of the test substance and histidine are prepared in aqueous solution (pH > 7), or in a mixture (1:1) of organic solvent and water to ensure that lipid-soluble test substances dissolve properly. The mixtures are irradiated with UVB, UVA or visible light, depending on the absorbance of the test substance. The histidine remaining is determined colorimetrically by a modified Pauly reaction. Controls of non-irradiated mixture, irradiated histidine alone and photosensitiser alone are incorporated. The quantum efficiency (Q) of the reaction may be computed, if desired: Q = molecules of histidine degraded/photons absorbed by test substance.

Histidine is likely to react via the singlet oxygen pathway under most conditions, but its destruction does not constitute formal mechanistic proof. For this, additional studies are required, such as enhancement of the singlet oxygen reaction in deuterated solvents, or the use of inhibitors such as azide or 1, 4-diazabicyclo(2,2,2)octane (DABCO). Photodegradation of the test chemical has to be taken into account when measuring photooxidation (81; Table XVII).

Table XVII: Photooxidation of Histidine

Photosensitiser Quantum Efficiency (%)
Photooxidation of histidine Photodegradation of photosensitiser
Rose bengal 50 0.01
Anthracene 27 0.4
Acridine 26 7.7
8-Methoxypsoralen 0.8 0.1
Musk ambrette 0.6 0.2
Tetrachlorosalicylanilide ns 20

Data are from Lovell & Sanders (81).

ns - not significant

The histidine photooxidation assay has enabled the identification of possible photosensitisers from a broad spectrum of chemicals which are used in the pharmaceutical and cosmetic industries (72; Table XVIII). It is a useful mechanistic test for detecting photosensitisers which act via the singlet oxygen pathway. Both aqueous and lipid-soluble materials may be tested.

Table XVIII: Histidine Photooxidation Test

Type of substance Positive Negative
Plant extracts Compositae: Dahlia, Yarrow, Cocklebur, Dog fennel, Chicory,, Ragweed, Chrysanthemum, Golden rod, Lettuce, Tansy, Sneezeweed, Sagebrush
Chemicals Kynurenic acid, Potassium dichromate, Tetrachlorosalicylanilide 8-Methoxypsoralen
Fragrance materials 6-Methylcoumarin, Oak moss, Amylcinnamaldehyde, Musk ambrette, Jasmines, Colophony Cinnamaldehyde, Cinnamylalcohol, Eugenol, Isoeugenol, Geraniol, Hydroxycitronella, Benzylsalicylate, Benzylalcohol, Benzylbenzoate, Methylsalicylate, Costus root oil, Laurel oil
Dyestuffs Methylene blue
Drugs Benoxaprofen, Diflusinal, Nalidixic acid, Oxytetracycline, Propranolol, Protryptyline Amiodarone, Azapropazone, Carbamazepine, Cimetidine, Doxycycline, Ethinyloestradiol, Griseofulvin, Imipramine, Methyl-DOPA, Minocycline, Piroxicam, Sulphamethoxazole, Sulphapyridine, Tetracycline, Trimethoprim

Data are from Johnson et al (72).

Photobinding to protein
Photoallergy is considered to be a delayed-type hypersensitivity reaction which is mediated by the formation of a photosensitiser-protein conjugate. Although the primary reaction occurs in the skin, in vitro photobinding of the test substance to human serum albumin (HSA) has been proposed as a screening test for photoallergens (82).

Mixtures of the test substance and HSA are exposed to UVA or visible light, and the irradiated mixture is then passed through a Sephadex G-10 gel filtration column to separate the photoconjugate from unreacted test substance. The process can be monitored by UV/visible absorption spectroscopy. When photochemical binding has occurred, the pre- and post-irradiation spectra are different, but the post-irradiation spectrum is identical to the post-gel filtration spectrum. Passage of the non-irradiated mixture through the Sephadex column results in quantitative removal of the test substance from HSA, unless there is strong, non-covalent, dark association of the chemical and protein. In such cases, the ionic strength of the eluting solvent may need to be increased.

All known photoallergens which have been tested show photobinding to HSA (Table XIX). In contrast, 2-ethylhexyl-p-methoxycinnamate, a sunscreen filter which absorbs UV but is not a photoallergen, does not bind photochemically to protein. Similarly, anthracene and acridine, which are both phototoxic but are not photoallergens, do not photobind to HSA (W.W. Lovell & D.J. Sanders, unpublished observations).

Table XIX: Photobinding to Protein of Known Photoallergens

Chemical Photobinding to: References
Serum Albumin -Globulin Insulin Poly-L-Lysine
Bithionol + + + + 82,83
Buclosamide + + - + 83,84
Chlorpromazine + nr nr nr 85
Fentichlor + + + + 82,86
6-Methylcoumarin + + + + 82,83
Musk ambrette + nr nr nr 82
Omadine (sodium salt) + + + + 82,83
Promethazine + nr nr nr 82
Quinoxaline-1,4-dioxide + nr nr nr 87
Sulphanilamide + nr nr nr 88,89
Tetrachlorosalicylanilide + + + nr 90-94
Triacetyldiphenolisatin + + nr + 95
Tribromosalicylanilide + + + - 83

+ = phototoxic; - = non-phototoxic; nr = no reaction.

Thus, the correlation between photoallergens and their ability to photobind to HSA appears to be excellent. However, further evaluation and validation are required to establish the specificity of the test, in particular using chemicals which are phototoxic but are not photoallergenic.

Linoleic acid photoperoxidation
Due to the detrimental effect of photoperoxidation on cell membranes, this process is thought to play an important role in skin phototoxicity. Linoleic acid (10-3 M solution in PBS) can be used as a model system to quantify photodynamic lipid peroxidation in vitro. Thus, tiaprofenic acid and suprofen (10-7-10-5 M) are able to photosensitise the formation of linoleic acid hydroperoxides (measured spectrophotometrically as the absorbance of the conjugated diene chromophore at 233 nm), while ibuprofen is not active in this test system (96,97). These results correlate well with the reported in vivo photosensitising potentials of the drugs. Data obtained by Castell et al. (96) are summarised in Table XX.

Table XX: Photoperoxidation of Linoleic Acid

Chemicals
Positive Negative
Dicloran
Fenofibric acid
2-Hydroxy-4-methoxy-benzophenone
Ketoprofen
Rose bengal
Suprofen
Tiaprofenic acid		 Fenoprofen
Flurbiprofen
Ibuprofen
Indoprofen
Naproxen

Data are from Castell et al (96).

Irradiation can conveniently be carried out using a 400 W mercury lamp (with a Pyrex filter). Since peroxidation is inhibited by radical scavengers, such as butylated hydroxyanisole and reduced glutathione, a Type I mechanism would appear to be involved. The photoperoxidation of linoleic acid is a very simple mechanistic test which can easily be standardised. Both aqueous and lipid-soluble materials can be tested.

Complement photoactivation assay
Complement activation is a mechanism which can lead to phototoxic reactions in human skin, and which cannot be evaluated in cell and tissue culture test systems. The role of complement in the reaction of the skin to light has been well described in relation to the photosensitivity associated with porphyrias (16,17,98,99). In animal models, complement products were detectable in the blood following irradiation of the skin in the presence of a photosensitising compound. Decomplementation prior to irradiation was shown to ameliorate the response (100).

The complement photoactivation assay is based on measurements of the specific complement fragments generated during activation of complement in vitro (101). The assay is performed using freshly isolated human plasma (centrifuged, heparinised blood). Plasma samples containing dissolved test compounds are irradiated, and the concentrations of the specific complement fragments which are generated are measured using ELISA test kits (Quidel, San Diego). Markers of both the classical and the alternative pathways have been measured. However, since photoactivation of complement has been shown to be mediated by the alternative but not by the classical pathway, a simplification of the test seems possible, in which only the products resulting from activation of the alternative pathway are determined.

Only very limited data have been published for the complement photoactivation assay (101). Since these were generated using a combined exposure to UVA and UVB, they are difficult to compare with results from other in vitro phototoxicity assays in which only UVA was used. The test may be used as an indicator of a specific mechanism of inflammation induced by photosensitisers. As this endpoint is not covered by other in vitro assays (both screening and mechanistic tests), it seems appropriate to determine chemical-induced complement photoactivation in vitro. However, the test is still at an early stage of development and needs standardising.

Technical Aspects of Phototoxicity Testing In Vitro

Light sources

Photosensitisation is initiated by the absorption of photons. Simulated sunlight is the optimum laboratory source for determining the potencies of photosensitisers, because the photon absorption would be similar to that from real sunlight. However, cells in culture are highly sensitive to the direct effects of the shorter wavelengths of sunlight (UVB). In addition, the spectral intensity of sunlight and the sensitivity of cells to UV both change rapidly, in opposite directions, in the UVB range. Attenuation of UVB in simulated sunlight may be advantageous for experiments with cells, since this removes a source of variation and also allows larger doses of UVA to be used.

Xenon arcs emit a continuum of light in the UV, visible and IR regions. A sharp cut-off UV filter of the Schott VG320 (2 mm) type may give a reasonable match to temperate summer sunlight. UVB may be further attenuated when desired using VG335 or VG345 filters. Borosilicate filters are available in larger sizes than Schott and allow illumination of larger areas. However, borosilicate has a shallow cut-off and gives a poorer match to sunlight. Furthermore, borosilicate exhibits changing transmission with exposure time (solarisation). All filters may exhibit batch-to-batch variation, and their transmission may vary reversibly with temperature. Attenuation of the IR emission of xenon arcs is required for biological applications (102,103). Glass IR filters also absorb UV, and are subject to solarisation. Dichroic mirrors which transmit IR and visible light, but reflect UV, are an alternative. Water or solutions of copper sulphate may be most satisfactory for the attenuation of IR.

Modern mercury metal-halide lamps are cheaper than xenon arcs, and they emit considerably less IR. However, borosilicate-type UV filters are usually fitted and these give a poor UV cut-off. Fluorescent UV tubes are cheap and stable sources. Different types of UVA tubes, with different emission spectra, are available. None of these match sunlight, but UVA tubes may be useful for studying the relative phototoxic potencies of chemicals, provided that the putative photosensitisers absorb in the same region of the spectrum. Monochromatic radiation may be useful for investigational purposes.

For simply comparing the phototoxic potentials of different substances, all that is really needed is for the amount of the substance and the light source to be specified (e.g. a comparison of drug A with drug B can be undertaken on a concentration basis, and if the lamp used is a UVA source or a solar simulator then this is stated and details are provided). The only complications here are those of relating the concentrations of the substance used in the test to those liable to occur in vivo.

Dosimetry

The spectral intensity of broad-band light sources should be measured using a spectroradiometer (104,105). Such instruments are expensive, and require careful calibration and selection of operating parameters. Broad-band radiometers are available for the measurement of UVA or UVB, but wide divergences of accuracy may occur unless the meters are calibrated, by radiospectrometry, against the specific sources they are to measure. The most severe discrepancies occur with measurement of UVB, because of the difficulty of measuring a small quantity of UVB in the presence of large quantities of UVA (104,105).

Thermophiles have the advantage of a flat frequency response and are capable of giving highly reproducible measurements. The measurement of a specific waveband is best achieved using two cut-off filters, for example at 320 nm and 400 nm for UVA. Band-pass filters for UVA are unsuitable because they also transmit IR, which would be detected by the thermophile. Chemical actinometry is an integrating system which measures the dose of light received within a vessel during a particular exposure (103,105). It is a relatively simple, accurate and reproducible technique, and is, therefore, the system of choice for solution photochemistry. Biological dosimeters may have advantages for some experiments, such as photo-mutagenicity studies, where close control of biologically effective UVB levels is a necessity.

Use of vehicles in phototoxicity testing in vitro

A common problem is the solubilisation of poorly water-soluble substances in aqueous media. Although there is evidence that solvents may influence the results of in vitro tests, it is often necessary to use them as vehicles. To facilitate comparison between results from different laboratories, it seems appropriate to propose standard procedures for solubilising test chemicals, and to address several technical aspects which are of importance to most of the procedures currently used for in vitro phototoxicity testing.

  1. The extent to which a specific solvent affects the test system and the endpoint to be measured must be investigated. Acceptable limits should be defined for each test system, since cells from different sources may exhibit different properties.
  2. The most commonly used solvents (e.g. ethanol, DMSO) should be able to solubilise the majority of 'difficult' chemicals, even after being diluted in the culture medium prior to incubation with cells or tissues.
  3. Special problems which are typical of photobiology (e.g. some photobiological reactions might be quenched or inhibited by a particular solvent) have to be taken into consideration.
  4. The use of a defined and generally agreed procedure for dissolving chemicals is recommended. To ensure minimum variation between the results from different laboratories, it is even necessary to use fixed final concentrations during range-finding experiments, which should similarly be performed according to standardised methods.
  5. Stock solutions (50%, 10%, 5% and 1%, if possible) should be prepared using the named solvent. Corresponding dilutions (e.g., 0.1, 1.0, 10, 100, and 1000 mg/l) of the test chemicals should be prepared for the range-finding experiment, to standardise the final concentrations following dilution with culture medium. This can easily be controlled by photometric analysis, which should include a calibration control.
  6. In cases where an EC50 value cannot be determined for the dark (i.e. unirradiated control) reaction, the highest concentration from the range-finding experiment can be used to calculate the ratio between the highest no-observed-effect concentration (NOEC) in the dark and either the highest NOEC following light exposure or the EC50 determined for irradiated samples.

Test Chemicals with Sufficient Human Data

In validation trials, the selection of test chemicals with high quality and scientifically sound in vitro data is one of the most critical tasks. As discussed earlier, acceptable animal data are not available, since the scientific community has not yet been able to agree upon a standardised animal test for phototoxicity. However, there are some chemicals for which there are good human data on their phototoxic effects.

Therefore, the participants of the ECVAM phototoxicity workshop recommend that chemicals to be used in evaluating phototoxicity test procedures are selected from the list given in Table XXI, since high quality human data are available for these particular compounds. A group of experts should evaluate all the available data on these chemicals, to produce a high quality reference database of compounds with known human effects and with supporting animal and in vitro experimental evidence as to their putative biological and chemical mechanisms of action.

Table XXI: List of Test Chemicals with Sufficient Human Data

<
Chemical (positive) Use in Humans Mechanism
Fibrates
Fenofibrate S II, metabolite
NSAIDs
Benoxaprofen S I, pp
Carprofen S
Ketoprofen S/T II, pp
Naproxen S I, II, pp
Suprofen S II, pp
Tiaprofenic acid S II, pp
Germicides
Bithionol T II
Buclosamide T II
Fenticlor T II
Hexachlorophene T
Tetrachlorosalicylanilide T II
Triclosane T
Tetracyclines
Demeclocycline S
Doxycycline S
Tetracycline S/T
Quinolones
Ciprofloxacin S
Fleroxacin S
Lomefloxacin S
Nalidixic acid S I, (II), tp
Norfloxacin S
Ofloxacin S
Coal Tar
Acridine T I
Anthracene T I
Psoralens
Bergamot oil S/T
5-Hydroxypsoralen S/T
Isopsoralen (angelicin) S/T
5-Methoxypsoralen S/T
8-Methoxypsoralen S/T
Trimethylpsoralen S/T (I), cb
Coumarins
6-Methylcoumarin T
UV Filters/Absorbers
2-Ethylhexyl-p-methoxy-cinnamate T
2-Hydroxy-4-methoxy-benzophenone T
4-Isopropyldibenzoylmethane T
3-(4-Methylbenzylidene)-camphor T
Miscellaneous
Amiodarone S II
Chlorpromazine S I, (II), cb, tp
Furosemide S/T II, tp
Musk ambrette T II
Neutral red T II
Promethazine S
Protoporphyrin IX S I
Rose bengal S I, II