Toxicology Letters, 6465 (1992) 173-182 0 1992 Elsevier Science Publishers B.V., All rights reserved 03784274/92/$5.99
173
Classification of chemicals for carcinogenic and mutagenic properties P.G.N. Kramersa and H. Roelfzemab “National Institute ofPublic Health and Environmental Protection, Bilthooen and %iinistry ofHealth, Welfare and Cultural Affairs, R&wck (The Netherlands) Key words: Classification of mutagens and carcinogens; Labelling; Carcinogenic chemicals; Mutagenic chemicals
SUMMARY In the framework of the EC labelling guide, three categories were devised for carcinogenic as well as for mutagenic chemicals. In short, category 1 is for compounds shown to produce these effects in man, whereas category 2 is meant to contain substances that should be regarded as such, generally on the basis of sufficient experimental data. Category 3 is for compounds that cause concern but for which the evidence is not sufficient for 1 or 2. These general statements have been implemented, both for carcinogenicity and mutagenicity, by a set of criteria and considerations of the group of “specialized experts” who are supporting the “national experts” in deciding upon the classification of individual compounds. Essentially, this classification relies on the strength of the experimental or epidemiological evidence rather than on considerations of risk. In other words, it is based on proven intrinsic properties of the substances. Compared to the IARC classification for carcinogens, the EC criteria rely more on mechanistic data and on qualitative considerations concerning the relevance for man, and less on a strict count of the amount of positive experimental evidence. Difficulties of this (and any) classification system include: (1) the problem of classifying with limited data and (2) the fundamental problem involved in forcing sets of multi-facetted, and often unsatisfactory, scientific data into a simple one-dimensional scheme that may have huge economic consequences.
CLASSIFICATION AND LABELLING WITHIN THE EC
The EC (European Community) labelling guide provides regulations on how to classify and label dangerous chemicals, for the protection of workers, the general public, and the environment. It also provides a classification scheme on which this labelling should be based. With respect to the carcinoCorrespondence to: P.G.N. Kramers, National Institute of Public Health and Environmental Protection, P.O. Box 1,372O BA Bilthoven, The Netherlands.
genie, mutagenic and teratogenic (C/M/T~ properties of substances, the EC Working Group “Classification and Labelling of Dangerous Substances” (C/M/T) is charged with the assignment of individual chemicals to categories. Recently, the term “teratogenic” has been replaced by “toxic to reproduction”, and the issue covers a wider scope. Accordingly, “C/M/T” has been changed into YYhQB”. The agenda of the Working Group’s meetings is determined by lists of substances which are submitted for classification by member states, and which may be selected by these for a variety of reasons. This Working group C/M/R (sometimes called the “national experts”) consists mostly of government officials from the member states. The Working group is assisted by a group of “specialized experts”, for the development of criteria and for advice on cases in which the “national experts” cannot come to a conclusion concerning the classification of a specific chemical. This group of “‘specialized experts” consists mostly of scientists from the EC member states. The following sections will firstly cover the basic classification scheme, as set up by the EC in the early 1980s. Then the detailed’criteria for classification will be discussed. This paper will deal with the classification for carcinogenic and mutagenic substances. For the reproduction toxicity see Sullivan Ill. BASIC CLASSIFICATION SCHEME
The basic scheme for the classification of carcinogens and mutagens, as given in the Labelling Guide, is printed below. Evidently, a further specification of these criteria is needed before they can be applied in practice. Such specifications have been worked out by the group of “specialized experts” as shown in the following sections. Criteria for the classification corresponding risk phrases.
of carcinogenic
substances,
and the
Category 1: Substances known to be carcinogenic to man. There is sufficient evidence to establish a causal association between human exposure to a substance and the development of cancer. Category 2: Substances which should be regarded as if they are carcinogenic to man. There is sufficient evidence to provide a strong presumption that human exposure to a substance may result in the development of cancer, generally on the basis of: - appropriate long-term animal studies - other relevant i~or~tion.
175
Category 3: Substances which cause concern for man owing to possible carcinogenic effects but in respect of which the available information is not adequate for making a satisfactory assessment. There is some evidence from appropriate animal studies, but this is insufficient to place the substance in category 2. Risk phrases: Categories 1 and 2: May cause cancer. Category 3: Possible risk of irreversible effects. Criteria for the classification corresponding risk phrases.
of mutagenic
substances,
and the
Category 1: Substances known to be mutagenic to man. There is sufficient evidence to establish a causal association between human exposure to a substance and heritable genetic damage. Category 2: Substances which should be regarded as if they are mutagenic to man; there is sufficient evidence to provide a strong presumption that human exposure to the substance may result in the development of heritable genetic damage, generally on the basis of: - appropriate animal studies -other relevant information. Category 3: Substances which cause concern for man owing to possible mutagenic effects; there is evidence from appropriate mutagenicity studies, but this is insufficient to place the substance in category 2. Risk phrases: Categories 1 and 2: May cause heritable genetic damage. Category 3: Possible risk of irreversible effects.
SPECIFICATION OF THE CLASSIFICATION CRITERIA FOR CARCINOGENIC PROPERTIES
The criteria for the classification of compounds for their carcinogenic properties have been worked out during 1985-1988. They have been published in the Official Journal of the European Communities E!l and are printed below:
176
Comments regarding the categorization substances
of carcinogenic
The placing of a substance into category 1 is done on the basis of epidemiological data; placing into categories 2 and 3 is based primarily on animal experiments.
For classification as a category 2 carcinogen either positive results in two animal species should be available or clear positive evidence in one species, together with supporting evidence such as genotoxicity data, metabolic or biochemical studies, induction of benign tumours, structural relationship with other known carcinogens, or data from epidemiological studies suggesting an association. Category 3 actually comprises 2 sub-categories: (a) substances which are well investigated but for which the evidence of a tumour-inducing effect is insufficient for classification in category 2. Additional experiments would not be expected to yield further relevant information with respect to classification; (b) substances which are insufficiently investigated. The available data are inadequate, but they raise concern for man. This classification is provisional; further experiments are necessary before a final decision can be made. For a distinction between categories 2 and 3 the arguments listed below are relevant which reduce the significance of experimental tumour induction in view of possible human exposure. These arguments, especially in combination, would lead in most cases to classification in category 3, even though tumours have been induced in animals: -
carcinogenic effects only at very high dose levels exceeding the “maximal tolerated dose”. The maximal tolerated dose is characterized by toxic effects which, although not yet reducing lifespan, go along with physical changes such as about 10% retardation in weight gain;
-
appearance of tumours, especially at high dose levels, only in particular organs of certain species known to be susceptible to a high spontaneous tumour formation;
-
appearance of tumours, only at the site of application, in very sensitive test systems (e.g., i.p. or S.C.application of certain locally active compounds), if the particular target is not relevant to man;
-
lack of genotoxicity in short-term tests in vivo and in vitro;
-
existence of a secondary mechanism of action with the implication of a practical threshold above a certain dose level (e.g., hormonal effect on target organs or on mechanisms of physiological regulation, chronic stimulation of cell proliferation);
177
-
existence of a species-specific mechanism of tumour formation (e.g. by specific metabolic pathways) irrelevant for man,
For a distinction between category 3 and no classification arguments are relevant which exclude a concern for man: -
a substance should not be classified in any of the categories if the mechanism of experimental tumour formation is clearly identified, with good evidence that this process cannot be extrapolated to man.
-
ifthe only available tumour data are liver tumours in certain sensitive strains of mice, without any other supplementary evidence, the substance may not be classified in any of the categories.
-
particular attention should be paid to cases where the only available tumour data are the occurrence of neoplasms at sites and in strains where they are well known to occur spontaneously with a high incidence.
It is interesting to compare these criteria with those formulated by IARC for their classification system [31. A major difference is that the IARC classification has never been meant to serve as a direct basis for regulatory action, whereas the EC classification has the purpose to serve as an immediate basis for, e.g., labelling, and thus potentially has great social (economic) impact. Another difference is that IARC sometimes classifies an exposure to a mixture, whereas the EC classification is strictly linked to a chemical entity. In contrast to the EC system, the IARC criteria are of a two-step type. In the first step the available data in humans, in animals, and the other relevant data are each considered separately. Within the first two, the data are evaluated as “sufficient”, “limited” or “inadequate”, or “negative” evidence. Combinations of these lead to a classification in group 1 (carcinogenic to humans), group 2A (probably carcinogenic to humans), group 2B (possibly carcinogenic to humans), group 3 (not classifiable) and group 4 (probably not carcinogenic). Roughly, the EC and IARC categories I contain the same substances. For IARC group 4 there is no EC equivalent. For the partitioning among the other groups the IARC criteria place somewhat more weight on quantitative aspects (number of animal species tested, number of independent studies), whereas the EC criteria are more explicit on mechanistic and qualitative considerations suggesting less relevance for man of an effect seen in animals. From the wording of the criteria, it would be expected that the EC category 2 would include IARC groups 2A and (partly) 2B, that EC category 3 would include substances from IARC groups 3 as well as 2B, and that IARC group 3 would also contain some substances not classified by EC. A quick survey of about 140 substances classified in both systems until 1992
178
shows that indeed the bulk of EC-2 substances are about equally divided between IARC-2A and IARC-2B. Similarly, EC-3 substances fall about equally in IARC-2B and IARC-3. Considering how IARC-2B substances are distributed among EC-2 and EC-3, it is clear that the considerations given above under the heading “Comments...” did play a role. Especially the nature (and the presence) of the genotoxicity data has been given considerable weight. This distinction shows most clearly the larger weight of mechanistic considerations in the EC system, as compared to the IARC scheme. The EC-3/IARC-3 group contains quite some substances that were. borderline, and were put in EC-3 after long deliberations within the group of “specialized experts”. SPECIFICATION OF THE CLASSIFICATION CRITERIA FOR MUTAGENIC PROPERTIES
The criteria for the classification of substances for mutagenic properties were worked out by the specialized experts during the period 1989-1990. The final text is given below: Category 1: To place a substance in category 1, positive evidence from human mutation epidemiology studies will be needed. Examples of such substances are not known to date. It is recognized that it is extremely difficult to obtain reliable information from studies on the incidence of mutations in human populations, or on possible increases in their frequencies. Category 2: To place a substance in category 2, positive results are needed from assays showing (a) mutagenic effects, or (b) other cellular interactions relevant to mutagenicity, in germ cells of mammals in duo, or (c) mutagenic effects in somatic cells of mammals in uivo in combination with clear evidence that the substance or a relevant metabolite reaches the germ cells. With respect to placement in category 2, at present the following methods are appropriate: 2a.
in uivo germ cell mutagenicity assays: - specific locus mutation test; -heritable translocation test; - dominant lethal mutation test.
These assays actually demonstrate the appearance of affected progeny or a defect in the developing embryo. 2b.
in uiuo assays showing relevant interaction with germ cells (usually
DNA):
179 -
assays for chromosomal abnormalities, as detected by cytogenetic analysis, including aneuploidy, caused by malsegregation of chromosomes; -test for sister chromatid exchanges (SCEs); - test for unscheduled DNA synthesis (UDSI; - assay of (covalent) binding of mutagen to germ cell DNA; - assaying other kinds of DNA damage. These assays provide evidence of a more or less indirect nature. Positive result in these assays would normally be supported by positive results from in viuo somatic cell mutagenicity assays, in mammals or in man (see under Category 3, preferably methods as under 3a). 2c.
in uiuo assays showing mutagenic effects in somatic cells of mammals (see under 3a), in combination with toxicokinetic methods, or other methodologies capable of demonstrating that the compound or a relevant metabolite reaches the germ cells.
For 2b and 2c, positive results from host-mediated assays or the demonstration of unequivocal effects in in vitro assays can be considered as supporting evidence. Category 3: To place a substance in category 3, positive results are needed in assays showing (a) mutagenic effects or (b) other cellular interaction relevant to mutagenicity, in somatic cells in mammals in uiuo. The latter especially would normally be supported by positive results from in vitro mutagenicity assays. For effects in somatic cells in uiuo at present the following
methods are
appropriate: 3a.
in -
uiuo somatic cell mutagenicity assays: bone marrow micronucleus test or metaphase analysis; metaphase analysis of peripheral lymphocytes; mouse coat color spot test.
3b.
in -
uiuo somatic cell DNA interaction assays: tests for SCEs in somatic cells; tests for UDS in somatic cells; assay for the (covalent) binding of mutagen to somatic cell DNA; assay for DNA damage, e.g. by alkaline elution, in somatic cells.
Substances showing positive results only in one or more in vitro mutagenicity assays should normally not be classified. Their further investigation using in uiuo assays, however, is strongly indicated. In exceptional cases, e.g. for a compound showing pronounced responses in several in vitro assays, for which no relevant in vivo data are available, and which shows resemblance to known mutagens/carcinogens, classification in category 3 could be considered.
180
There is a degree of consistency between the carcinogenicity and the mutagenicity criteria. For both, category 1 addresses the observation of the effect in man, category 2 addresses the clear demonstration of the effect (turnours versus inherited damage) in experimental animals, and category 3 involves the more indirect or uncertain types of evidence. For category 3, the carcinogenic and mutagenic hazards come together: the type of evidence leading to classification in 3 for mutagenicity is considered relevant for cancer risk, although by itself it would not be sufficient for classification as a category 3 carcinogen. Therefore it is logical that the risk phrase is the same for category 3 carcinogens and mutagens (although it is extraordinarily non-informative). In fact, an in vivo somatic cell mutagen is probably more relevant as a potential carcinogen than as a potential germ-cell mutagen. An important issue of discussion has been whether or not to classify substances that have only been shown mutagenic in in vitro test systems, with or without results from in vivo assays. Here a crucial difference shows up between a classification system based on experimental results, on the one hand, and a sequential test strategy, on the other. The former is based on available positive evidence, whereas the latter is to a large extent directed by negative evidence, i.e., a series of effects have to be excluded before a substance is considered safe (see e.g. [4,51). Therefore, the conclusion that an in vitro positive result would generally warrant further testing or research but would not be sufficient as a base of far-reaching regulatory actions such as EC-wide labelling, is not illogical. Another aspect of the criteria that reflect a process of negotiations between member states is the addition of 2c: This item is beyond the strict requirement of demonstrating the effect of interest in the tissue of interest, but combines data that make it very plausible that the effect could occur. The issue has given rise to questions by a member state indicating that “the compound reaching the germ cells” could be interpreted as “the compound reaching the systemic circulation”. In fact, the editors of the text have clearly intended to imply that the presence of the compound or a relevant metabolite should be demonstrated within the germ cells. Until now, a limited number of “existing” chemicals has been classified, or proposed for classification, for mutagenicity. Among these are, for category 2: hexamethyl phosphoramide, ethylene imine, ethylene oxide, 1,2dibromo-3-chloropropane, diethyl sulphate, and benzo(a)pyrene; and for category 3: epichlorohydrin, dimethyl sulphate, 2-nitropropane, arsenic compounds, chromates, benomyl, carbendazim, thiophanate-methyl, and atrazin. The first five of these are classified as category 2 carcinogens, which overrules their classification as category 3 mutagens. This small and relatively random collection of mutagens reflects the fact that only exceptionally a member state submits a substance for classification as a mutagen.
181
For “new” chemicals there is a special situation. The “base set” of mutagenicity assays that are required within the context of the notification of new chemicals under the 6th Amendment to Directive 67/548/EEC is meant to serve as a basis for classification and labelling. However, this base set very often contains only in vitro assays which would, according to the criteria given above, normally not lead to any classification. In order to overcome this problem, a strategy is under discussion in which for every possible outcome of the base set assays a follow-up is indicated. This scheme shows close resemblance to the sequential test strategy laid down by the Mutagenesis Committee in the U.K. [51, and its outcomes should allow classification according to the criteria given above. A strategy paper is expected to be adopted during 1992. Current discussions focus on the role of specific assay systems within the scheme, and on the use and dangers of “provisional” labelling. FINAL REMARKS
The issue of classification of chemicals on the basis of scientific evidence has several problems. Firstly, it is an almost impossible task to translate a multi-facetted collection of experimental data, in combination with the lack of certain crucial data, into a discrete one-dimensional scale of three or four entities. The criteria are in fact a series of weighing factors combining at least two dimensions, i.e. the seriousness of the effect found, and the solidity of the data. Secondly, the classification is meant to be done on the intrinsic properties of a substance, as shown by experimental evidence, without taking account of quantitative risk considerations. Here lies a central dilemma: A scientific forum is asked to make a classification on the basis of available data, according to a preset scheme. However, knowing that their “scientific” decision will in most cases be automatically followed by placing a risk phrase on a label, there is a temptation to take risk considerations into account. This point is especially important because there is so much uncertainty with respect to the relevance of experimental data to actual human hazard. Especially relevant to category three is the issue of “inflation” in labelling: from uncertainty, people tend to be over-cautious and print a warning on everything, whereas the strength of a labelling system is its discriminating power, i.e., its ability to sort out the more severe hazards. A final point is the irony of spending hours of deliberation to decide whether or not to place the risk phrase “possible risk of irreversible effects”, which to the general public is either just frightening or non-informative, and most likely both. To conclude, the EC system for the classification of carcinogens and mutagens provides a sensible framework with sufficient flexibility to ac-
182
count for unexpected aspects and new scientific insights. At the same time, this flexibility leaves room for the existing variety of interpretations concerning the uncertainties in assessing the significance of experimental findings to man. Yet, we think this is preferable to a more rigid checklisttype approach. It does, however, rely heavily on the quality and the continuity of the expertise involved in the procedure. REFERENCES 1 2 3 4
5
Sullivan, F.M. (1992) The European Community classification of chemicals for reproductive toxicity. Toxicol. Lett. 64165, Classification on the basis of specific effects to human health (1991). Ofticial Journal of the European Communities, No. L 180 (8-7-91). L4RC (1991) Monograph on the Evaluation of Carcinogenic Risks to Humans. Vol. 53. Kramers, P.G.N., Knaap, A.G.A.C., van der Heijden, CA., Taalman, R.D.F.M. and Mohn, G.R. (1991) Role of genotoxicity assays in the regulation of chemicals in The Netherlands: considerations and experiences. Mutagenesis 6,487-493. Department of Health (1989) Guidelines for the testing of chemicals for mutagenicity. Report on Health and Social Subjects (no. 35) from the Committe on Mutagenicity of Chemicals in Food, Consumer Products and the Environment, London.
173
Classification of chemicals for carcinogenic and mutagenic properties P.G.N. Kramersa and H. Roelfzemab “National Institute ofPublic Health and Environmental Protection, Bilthooen and %iinistry ofHealth, Welfare and Cultural Affairs, R&wck (The Netherlands) Key words: Classification of mutagens and carcinogens; Labelling; Carcinogenic chemicals; Mutagenic chemicals
SUMMARY In the framework of the EC labelling guide, three categories were devised for carcinogenic as well as for mutagenic chemicals. In short, category 1 is for compounds shown to produce these effects in man, whereas category 2 is meant to contain substances that should be regarded as such, generally on the basis of sufficient experimental data. Category 3 is for compounds that cause concern but for which the evidence is not sufficient for 1 or 2. These general statements have been implemented, both for carcinogenicity and mutagenicity, by a set of criteria and considerations of the group of “specialized experts” who are supporting the “national experts” in deciding upon the classification of individual compounds. Essentially, this classification relies on the strength of the experimental or epidemiological evidence rather than on considerations of risk. In other words, it is based on proven intrinsic properties of the substances. Compared to the IARC classification for carcinogens, the EC criteria rely more on mechanistic data and on qualitative considerations concerning the relevance for man, and less on a strict count of the amount of positive experimental evidence. Difficulties of this (and any) classification system include: (1) the problem of classifying with limited data and (2) the fundamental problem involved in forcing sets of multi-facetted, and often unsatisfactory, scientific data into a simple one-dimensional scheme that may have huge economic consequences.
CLASSIFICATION AND LABELLING WITHIN THE EC
The EC (European Community) labelling guide provides regulations on how to classify and label dangerous chemicals, for the protection of workers, the general public, and the environment. It also provides a classification scheme on which this labelling should be based. With respect to the carcinoCorrespondence to: P.G.N. Kramers, National Institute of Public Health and Environmental Protection, P.O. Box 1,372O BA Bilthoven, The Netherlands.
genie, mutagenic and teratogenic (C/M/T~ properties of substances, the EC Working Group “Classification and Labelling of Dangerous Substances” (C/M/T) is charged with the assignment of individual chemicals to categories. Recently, the term “teratogenic” has been replaced by “toxic to reproduction”, and the issue covers a wider scope. Accordingly, “C/M/T” has been changed into YYhQB”. The agenda of the Working Group’s meetings is determined by lists of substances which are submitted for classification by member states, and which may be selected by these for a variety of reasons. This Working group C/M/R (sometimes called the “national experts”) consists mostly of government officials from the member states. The Working group is assisted by a group of “specialized experts”, for the development of criteria and for advice on cases in which the “national experts” cannot come to a conclusion concerning the classification of a specific chemical. This group of “‘specialized experts” consists mostly of scientists from the EC member states. The following sections will firstly cover the basic classification scheme, as set up by the EC in the early 1980s. Then the detailed’criteria for classification will be discussed. This paper will deal with the classification for carcinogenic and mutagenic substances. For the reproduction toxicity see Sullivan Ill. BASIC CLASSIFICATION SCHEME
The basic scheme for the classification of carcinogens and mutagens, as given in the Labelling Guide, is printed below. Evidently, a further specification of these criteria is needed before they can be applied in practice. Such specifications have been worked out by the group of “specialized experts” as shown in the following sections. Criteria for the classification corresponding risk phrases.
of carcinogenic
substances,
and the
Category 1: Substances known to be carcinogenic to man. There is sufficient evidence to establish a causal association between human exposure to a substance and the development of cancer. Category 2: Substances which should be regarded as if they are carcinogenic to man. There is sufficient evidence to provide a strong presumption that human exposure to a substance may result in the development of cancer, generally on the basis of: - appropriate long-term animal studies - other relevant i~or~tion.
175
Category 3: Substances which cause concern for man owing to possible carcinogenic effects but in respect of which the available information is not adequate for making a satisfactory assessment. There is some evidence from appropriate animal studies, but this is insufficient to place the substance in category 2. Risk phrases: Categories 1 and 2: May cause cancer. Category 3: Possible risk of irreversible effects. Criteria for the classification corresponding risk phrases.
of mutagenic
substances,
and the
Category 1: Substances known to be mutagenic to man. There is sufficient evidence to establish a causal association between human exposure to a substance and heritable genetic damage. Category 2: Substances which should be regarded as if they are mutagenic to man; there is sufficient evidence to provide a strong presumption that human exposure to the substance may result in the development of heritable genetic damage, generally on the basis of: - appropriate animal studies -other relevant information. Category 3: Substances which cause concern for man owing to possible mutagenic effects; there is evidence from appropriate mutagenicity studies, but this is insufficient to place the substance in category 2. Risk phrases: Categories 1 and 2: May cause heritable genetic damage. Category 3: Possible risk of irreversible effects.
SPECIFICATION OF THE CLASSIFICATION CRITERIA FOR CARCINOGENIC PROPERTIES
The criteria for the classification of compounds for their carcinogenic properties have been worked out during 1985-1988. They have been published in the Official Journal of the European Communities E!l and are printed below:
176
Comments regarding the categorization substances
of carcinogenic
The placing of a substance into category 1 is done on the basis of epidemiological data; placing into categories 2 and 3 is based primarily on animal experiments.
For classification as a category 2 carcinogen either positive results in two animal species should be available or clear positive evidence in one species, together with supporting evidence such as genotoxicity data, metabolic or biochemical studies, induction of benign tumours, structural relationship with other known carcinogens, or data from epidemiological studies suggesting an association. Category 3 actually comprises 2 sub-categories: (a) substances which are well investigated but for which the evidence of a tumour-inducing effect is insufficient for classification in category 2. Additional experiments would not be expected to yield further relevant information with respect to classification; (b) substances which are insufficiently investigated. The available data are inadequate, but they raise concern for man. This classification is provisional; further experiments are necessary before a final decision can be made. For a distinction between categories 2 and 3 the arguments listed below are relevant which reduce the significance of experimental tumour induction in view of possible human exposure. These arguments, especially in combination, would lead in most cases to classification in category 3, even though tumours have been induced in animals: -
carcinogenic effects only at very high dose levels exceeding the “maximal tolerated dose”. The maximal tolerated dose is characterized by toxic effects which, although not yet reducing lifespan, go along with physical changes such as about 10% retardation in weight gain;
-
appearance of tumours, especially at high dose levels, only in particular organs of certain species known to be susceptible to a high spontaneous tumour formation;
-
appearance of tumours, only at the site of application, in very sensitive test systems (e.g., i.p. or S.C.application of certain locally active compounds), if the particular target is not relevant to man;
-
lack of genotoxicity in short-term tests in vivo and in vitro;
-
existence of a secondary mechanism of action with the implication of a practical threshold above a certain dose level (e.g., hormonal effect on target organs or on mechanisms of physiological regulation, chronic stimulation of cell proliferation);
177
-
existence of a species-specific mechanism of tumour formation (e.g. by specific metabolic pathways) irrelevant for man,
For a distinction between category 3 and no classification arguments are relevant which exclude a concern for man: -
a substance should not be classified in any of the categories if the mechanism of experimental tumour formation is clearly identified, with good evidence that this process cannot be extrapolated to man.
-
ifthe only available tumour data are liver tumours in certain sensitive strains of mice, without any other supplementary evidence, the substance may not be classified in any of the categories.
-
particular attention should be paid to cases where the only available tumour data are the occurrence of neoplasms at sites and in strains where they are well known to occur spontaneously with a high incidence.
It is interesting to compare these criteria with those formulated by IARC for their classification system [31. A major difference is that the IARC classification has never been meant to serve as a direct basis for regulatory action, whereas the EC classification has the purpose to serve as an immediate basis for, e.g., labelling, and thus potentially has great social (economic) impact. Another difference is that IARC sometimes classifies an exposure to a mixture, whereas the EC classification is strictly linked to a chemical entity. In contrast to the EC system, the IARC criteria are of a two-step type. In the first step the available data in humans, in animals, and the other relevant data are each considered separately. Within the first two, the data are evaluated as “sufficient”, “limited” or “inadequate”, or “negative” evidence. Combinations of these lead to a classification in group 1 (carcinogenic to humans), group 2A (probably carcinogenic to humans), group 2B (possibly carcinogenic to humans), group 3 (not classifiable) and group 4 (probably not carcinogenic). Roughly, the EC and IARC categories I contain the same substances. For IARC group 4 there is no EC equivalent. For the partitioning among the other groups the IARC criteria place somewhat more weight on quantitative aspects (number of animal species tested, number of independent studies), whereas the EC criteria are more explicit on mechanistic and qualitative considerations suggesting less relevance for man of an effect seen in animals. From the wording of the criteria, it would be expected that the EC category 2 would include IARC groups 2A and (partly) 2B, that EC category 3 would include substances from IARC groups 3 as well as 2B, and that IARC group 3 would also contain some substances not classified by EC. A quick survey of about 140 substances classified in both systems until 1992
178
shows that indeed the bulk of EC-2 substances are about equally divided between IARC-2A and IARC-2B. Similarly, EC-3 substances fall about equally in IARC-2B and IARC-3. Considering how IARC-2B substances are distributed among EC-2 and EC-3, it is clear that the considerations given above under the heading “Comments...” did play a role. Especially the nature (and the presence) of the genotoxicity data has been given considerable weight. This distinction shows most clearly the larger weight of mechanistic considerations in the EC system, as compared to the IARC scheme. The EC-3/IARC-3 group contains quite some substances that were. borderline, and were put in EC-3 after long deliberations within the group of “specialized experts”. SPECIFICATION OF THE CLASSIFICATION CRITERIA FOR MUTAGENIC PROPERTIES
The criteria for the classification of substances for mutagenic properties were worked out by the specialized experts during the period 1989-1990. The final text is given below: Category 1: To place a substance in category 1, positive evidence from human mutation epidemiology studies will be needed. Examples of such substances are not known to date. It is recognized that it is extremely difficult to obtain reliable information from studies on the incidence of mutations in human populations, or on possible increases in their frequencies. Category 2: To place a substance in category 2, positive results are needed from assays showing (a) mutagenic effects, or (b) other cellular interactions relevant to mutagenicity, in germ cells of mammals in duo, or (c) mutagenic effects in somatic cells of mammals in uivo in combination with clear evidence that the substance or a relevant metabolite reaches the germ cells. With respect to placement in category 2, at present the following methods are appropriate: 2a.
in uivo germ cell mutagenicity assays: - specific locus mutation test; -heritable translocation test; - dominant lethal mutation test.
These assays actually demonstrate the appearance of affected progeny or a defect in the developing embryo. 2b.
in uiuo assays showing relevant interaction with germ cells (usually
DNA):
179 -
assays for chromosomal abnormalities, as detected by cytogenetic analysis, including aneuploidy, caused by malsegregation of chromosomes; -test for sister chromatid exchanges (SCEs); - test for unscheduled DNA synthesis (UDSI; - assay of (covalent) binding of mutagen to germ cell DNA; - assaying other kinds of DNA damage. These assays provide evidence of a more or less indirect nature. Positive result in these assays would normally be supported by positive results from in viuo somatic cell mutagenicity assays, in mammals or in man (see under Category 3, preferably methods as under 3a). 2c.
in uiuo assays showing mutagenic effects in somatic cells of mammals (see under 3a), in combination with toxicokinetic methods, or other methodologies capable of demonstrating that the compound or a relevant metabolite reaches the germ cells.
For 2b and 2c, positive results from host-mediated assays or the demonstration of unequivocal effects in in vitro assays can be considered as supporting evidence. Category 3: To place a substance in category 3, positive results are needed in assays showing (a) mutagenic effects or (b) other cellular interaction relevant to mutagenicity, in somatic cells in mammals in uiuo. The latter especially would normally be supported by positive results from in vitro mutagenicity assays. For effects in somatic cells in uiuo at present the following
methods are
appropriate: 3a.
in -
uiuo somatic cell mutagenicity assays: bone marrow micronucleus test or metaphase analysis; metaphase analysis of peripheral lymphocytes; mouse coat color spot test.
3b.
in -
uiuo somatic cell DNA interaction assays: tests for SCEs in somatic cells; tests for UDS in somatic cells; assay for the (covalent) binding of mutagen to somatic cell DNA; assay for DNA damage, e.g. by alkaline elution, in somatic cells.
Substances showing positive results only in one or more in vitro mutagenicity assays should normally not be classified. Their further investigation using in uiuo assays, however, is strongly indicated. In exceptional cases, e.g. for a compound showing pronounced responses in several in vitro assays, for which no relevant in vivo data are available, and which shows resemblance to known mutagens/carcinogens, classification in category 3 could be considered.
180
There is a degree of consistency between the carcinogenicity and the mutagenicity criteria. For both, category 1 addresses the observation of the effect in man, category 2 addresses the clear demonstration of the effect (turnours versus inherited damage) in experimental animals, and category 3 involves the more indirect or uncertain types of evidence. For category 3, the carcinogenic and mutagenic hazards come together: the type of evidence leading to classification in 3 for mutagenicity is considered relevant for cancer risk, although by itself it would not be sufficient for classification as a category 3 carcinogen. Therefore it is logical that the risk phrase is the same for category 3 carcinogens and mutagens (although it is extraordinarily non-informative). In fact, an in vivo somatic cell mutagen is probably more relevant as a potential carcinogen than as a potential germ-cell mutagen. An important issue of discussion has been whether or not to classify substances that have only been shown mutagenic in in vitro test systems, with or without results from in vivo assays. Here a crucial difference shows up between a classification system based on experimental results, on the one hand, and a sequential test strategy, on the other. The former is based on available positive evidence, whereas the latter is to a large extent directed by negative evidence, i.e., a series of effects have to be excluded before a substance is considered safe (see e.g. [4,51). Therefore, the conclusion that an in vitro positive result would generally warrant further testing or research but would not be sufficient as a base of far-reaching regulatory actions such as EC-wide labelling, is not illogical. Another aspect of the criteria that reflect a process of negotiations between member states is the addition of 2c: This item is beyond the strict requirement of demonstrating the effect of interest in the tissue of interest, but combines data that make it very plausible that the effect could occur. The issue has given rise to questions by a member state indicating that “the compound reaching the germ cells” could be interpreted as “the compound reaching the systemic circulation”. In fact, the editors of the text have clearly intended to imply that the presence of the compound or a relevant metabolite should be demonstrated within the germ cells. Until now, a limited number of “existing” chemicals has been classified, or proposed for classification, for mutagenicity. Among these are, for category 2: hexamethyl phosphoramide, ethylene imine, ethylene oxide, 1,2dibromo-3-chloropropane, diethyl sulphate, and benzo(a)pyrene; and for category 3: epichlorohydrin, dimethyl sulphate, 2-nitropropane, arsenic compounds, chromates, benomyl, carbendazim, thiophanate-methyl, and atrazin. The first five of these are classified as category 2 carcinogens, which overrules their classification as category 3 mutagens. This small and relatively random collection of mutagens reflects the fact that only exceptionally a member state submits a substance for classification as a mutagen.
181
For “new” chemicals there is a special situation. The “base set” of mutagenicity assays that are required within the context of the notification of new chemicals under the 6th Amendment to Directive 67/548/EEC is meant to serve as a basis for classification and labelling. However, this base set very often contains only in vitro assays which would, according to the criteria given above, normally not lead to any classification. In order to overcome this problem, a strategy is under discussion in which for every possible outcome of the base set assays a follow-up is indicated. This scheme shows close resemblance to the sequential test strategy laid down by the Mutagenesis Committee in the U.K. [51, and its outcomes should allow classification according to the criteria given above. A strategy paper is expected to be adopted during 1992. Current discussions focus on the role of specific assay systems within the scheme, and on the use and dangers of “provisional” labelling. FINAL REMARKS
The issue of classification of chemicals on the basis of scientific evidence has several problems. Firstly, it is an almost impossible task to translate a multi-facetted collection of experimental data, in combination with the lack of certain crucial data, into a discrete one-dimensional scale of three or four entities. The criteria are in fact a series of weighing factors combining at least two dimensions, i.e. the seriousness of the effect found, and the solidity of the data. Secondly, the classification is meant to be done on the intrinsic properties of a substance, as shown by experimental evidence, without taking account of quantitative risk considerations. Here lies a central dilemma: A scientific forum is asked to make a classification on the basis of available data, according to a preset scheme. However, knowing that their “scientific” decision will in most cases be automatically followed by placing a risk phrase on a label, there is a temptation to take risk considerations into account. This point is especially important because there is so much uncertainty with respect to the relevance of experimental data to actual human hazard. Especially relevant to category three is the issue of “inflation” in labelling: from uncertainty, people tend to be over-cautious and print a warning on everything, whereas the strength of a labelling system is its discriminating power, i.e., its ability to sort out the more severe hazards. A final point is the irony of spending hours of deliberation to decide whether or not to place the risk phrase “possible risk of irreversible effects”, which to the general public is either just frightening or non-informative, and most likely both. To conclude, the EC system for the classification of carcinogens and mutagens provides a sensible framework with sufficient flexibility to ac-
182
count for unexpected aspects and new scientific insights. At the same time, this flexibility leaves room for the existing variety of interpretations concerning the uncertainties in assessing the significance of experimental findings to man. Yet, we think this is preferable to a more rigid checklisttype approach. It does, however, rely heavily on the quality and the continuity of the expertise involved in the procedure. REFERENCES 1 2 3 4
5
Sullivan, F.M. (1992) The European Community classification of chemicals for reproductive toxicity. Toxicol. Lett. 64165, Classification on the basis of specific effects to human health (1991). Ofticial Journal of the European Communities, No. L 180 (8-7-91). L4RC (1991) Monograph on the Evaluation of Carcinogenic Risks to Humans. Vol. 53. Kramers, P.G.N., Knaap, A.G.A.C., van der Heijden, CA., Taalman, R.D.F.M. and Mohn, G.R. (1991) Role of genotoxicity assays in the regulation of chemicals in The Netherlands: considerations and experiences. Mutagenesis 6,487-493. Department of Health (1989) Guidelines for the testing of chemicals for mutagenicity. Report on Health and Social Subjects (no. 35) from the Committe on Mutagenicity of Chemicals in Food, Consumer Products and the Environment, London.
