Mutation Research, 250 (1991) 457-465 © 1991 Elsevier Science Publishers B.V. All rights reserved 0027-5107/91/$03.50 ADONIS 002751079100204N
457
MUT 02511
Use of in vivo genetic toxicity data for risk assessment M a r v i n S. L e g a t o r a n d J o n a t h a n B. W a r d Jr. The University of Texas Medical Branch at Galveston, Department of Preventive Medicine and Community Health, Division of Environmental Toxicology, 2.102 Ewing Hall, J-I0, Galveston, TX 77550 (U.S.A.) (Accepted 5 April 1991)
Keywords: Risk assessment; Epigenetic, Comprehensive Testing Protocol; Benzene
Summary Mutagenicity studies have been used to identify specific agents as potential carcinogens or other human health hazards; however, they have been used minimally for risk assessment or in determining permissible levels of human exposure. The poor predictive value of in vitro mutagenesis tests for carcinogenic activity and a lack of mechanistic understanding of the roles of mutagens in the induction of specific cancers have made these tests unattractive for the purpose of risk assessment. However, the limited resources available for carcinogen testing and large number of chemicals which need to be evaluated necessitate the incorporation of more efficient methods into the evaluation process. In vivo genetic toxicity testing can be recommended for this purpose because in vivo assays incorporate the metabolic activation pathways that are relevant to humans. We propose the use of a multiple end-point in vivo comprehensive testing protocol (CTP) using rodents. Studies using sub-acute exposure to low levels of test agents by routes consistent with human exposure can be a useful adjunct to methods currently used to provide data for risk assessment. Evaluations can include metabolic and pharmacokinetic endpoints, in addition to genetic toxicity studies, in order to provide a comprehensive examination of the mechanism of toxicity of the agent. A parallelogram approach can be used to estimate effects in non-accessible human tissues by using data from accessible human tissues and analogous tissues in animals. A categorical risk assessment procedure can be used which would consider, in order of priority, genetic damage in man, genetic damage in animals that is highly relevant to disease outcome (mutation, chromosome damage), and data from animals that is of less certain relevance to disease. Action levels of environmental exposure would be determined based on the lowest observed effect levels or the highest observed no effect levels, using sub-acute low level exposure studies in rodents. As an example, the known genotoxic effects of benzene exposure at low levels in man and animals are discussed. The lowest observed genotoxic effects were observed at about 1-10 parts per million for man and 0.04-0.1 parts per million in subacute animal studies. If genetic toxicity is to achieve a prominent role in evaluating carcinogens and in characterizing germ-cell mutagens, minimal testing requirements must be established to ascertain the risks associated with environmental mutagen exposure. The use of the in vivo approach
Correspondence: M.S. Legator, University of Texas Medical Branch, 2.102 Ewing Hall, J-10, Galveston, TX 77550 (U.S.A.).
458
described here should provide the information needed to meet this goal. In addition, it should allow truly epigenetic or non-genotoxic carcinogens to be distinguished from the genotoxic carcinogens that are not detected by in vitro methods.
Minimal data required for risk assessment Prediction of potential human risks from laboratory studies is an imprecisc process and risk assessment is especially complex for chronic disease outcomes such as cancer. Mutagenicity studies arc used to a limited extent in risk assessment and have played a marginally supportive role in evaluating potential carcinogens. Mutagenicity studies per se have not played a critical role in determining the permissible levels for human exposure. Utilizing in vitro mutagenicity data, even for qualitatively identifying a carcinogen, is questionable. Several papers, including Rinkus and Legator (1979), have addressed the limitations of in vitro studies and the fallacy of the correlation between microbial tests and animal carcinogens. However, it was not until 1986 that the data was available to challenge the validity of these shortterm assays (Zeiger and Tennant, 1986; Zeiger, 1987). Tcnnant (1987) in his overall assessment of the field, came to the following conclusion: 'If current in vitro short-term tests for genetic toxicology are expected to replace long-term rodent studies for the identification of chemical carcinogens, that expectation should be abandoned.' The attractiveness of using mutagenicity data in risk assessment remains, however. The limited resources available for carcinogen testing, and the large number of chemicals which need to be examined, necessitate the incorporation of more efficient methods into the evaluation process. Mutagenicity testing offers one of the few available supplements to long-term rodent studies for carcinogenic risk evaluation. The major impediment is the questionable relevance of mutagenicity data when most of it is derived from in vitro studies. The in vivo approach to mutagenicity testing, in contrast to an in vitro one, is defensible because it incorporates metabolic activation pathways that are relevant to humans. It is difficult at the present time to identify a data base derived
from animal mutagenicity studies that is comprehensive enough to allow for risk assessment. Even in those instances where animal studies have been performed, they are usually acute studies using high concentrations and focusing on a single specific assay. If mutagenicity data is to be integrated into a risk-assessment procedure, subacute studies measuring multiple cndpoints, at realistic exposure levels, should be used. Towards this purpose, we describe a testing approach for evaluating mutagenic agents and a procedure for using this test data for risk assessment. This testing approach is a multi-endpoint procedure that we call the Comprehensive Testing Protocol (CTP) (Legator and Harper, 1988; Moslen and Legator, 1988; Au et al., 1988; Harper ct al., 1989; Legator and Harper, 1987; Legator et al., 1986). This procedure, utilizing a sub-acute exposure protocol in which animals are exposed to low levels of chemicals, will yield comprehensive data upon which to base a risk assessment. The CTP, in contrast to the traditional animal cancer bioassay, can be carried out in weeks rather than years. Since several endpoints arc measured concurrently, the C-TP maximizes thc amount of information acquired from a single experiment. The cost of a CTP is much less than the expense involved in carrying out individual mutation assays in separate experiments and will be a fraction of the cost of an animal cancer bioassay. An additional strength of the C-~P for risk assessment is that, for many agents, genotoxic responses can be observed at relatively low exposure levels. When animals can be treated at dose levels which are not far above those experienced by humans, concerns about the physiological relevance of exposure to man are reduced.
The comprehensive testing protocol (CTP) The CTP offers the unique opportunity to observe several steps in the mutagenesis process in the context of the full range of metabolic and
459
pharmacokinetic events in an intact animal. Steps in the mutagenesis process from the presence of mutagens in body fluids, to adduct formation, to DNA repair, to chromosome damage, to the final fixation of specific-locus mutation can be measured. In addition, it is now possible to determine the spectrum of mutations induced by specific exposure conditions. Cells from several tissues may be used as reporters of genetic damage. The conditions of exposure can be manipulated to evaluate duration, concentration, route or rate, and the influence of specific aspects of metabolism or pharmacokinetics can be evaluated by measurement of enzyme activities or levels, and distribution of the agent or its metabolites. The interactive effects of chemical mixtures can be studied or the influence of inducers or inhibitors of metabolic enzymes can be assessed. Thus, a comprehensive evaluation of the process of mutagenesis under exposure and metabolic conditions chosen by the investigator can be carried out. Not only can the potential of an agent to induce genetic damage be assessed, but the entire mutagenesis process and its relationship to exposure conditions and metabolic status can be evaluated. The CTP procedure is carried out by exposing mice or rats to a test agent under the desired conditions. Subsequently, urine can be collected and evaluated for the presence of mutagenic activity or specific metabolites. Protein a n d / o r DNA adducts may be quantitated in virtually any tissue from the animal. The distribution and concentrations of the agent or its metabolites may be determined and the effects of exposure on metabolic enzyme levels may be ascertained. DNA-repair processes may be detected in many tissues. Cytogenetic effects, observed as structural chromosome aberrations, micronucleus formation, or sister-chromatid exchange can be observed in a variety of tissues including bone marrow, spleen, thymus, testes and lung. The frequency of mutations at the hypoxanthine-guanine phosphoribosyl transferase locus is determined in T-lymphocytes from the spleen or thymus. Handling data from the CTP for risk assessment
The rapid advancements in physiology, molecular biology and genetics have provided direct
evidence for the multistage process that leads to malignancy. Specific mutations play a central role in this orderly and programmed process. Mutations in protooncogenes or normal genes that promote cellular growth lead to their activation as oncogenes. Mutations in tumor suppressor genes, which function in normal cells to halt proliferation, leads to the loss of the ability of the cell to constrain its own growth. Chromosomal mutations, including small and large deletions and translocations, as well as gene mutations, account for the specific alterations of both suppressor genes and protooncogenes. Therefore, animal studies that faithfully reflect the ability of specific metabolites to induce gene or chromosomal mutations in somatic cells should identify carcinogenic agents. The induction of chromosome damage a n d / o r gene mutation by a chemical or by radiation raises a high index of suspicion that the agent is a carcinogen. Because they are less closely associated mechanistically with carcinogenesis, adduct formation and induction of DNA repair are less relevant in classifying a chemical as a carcinogen or mutagen. There are several additional procedures, such as detection of urinary mutagens and induction of sister-chromatid exchanges (SCE), that are useful but alone cannot be considered sufficient for classification. It is assumed, as supported by extensive data, that the processes leading to cancer are essentially the same in animals and man. Therefore, the relative importance of each step in the process is the same in this multistage sequence whether it occurs in laboratory animals or in humans. Human risk for damage at a particular target tissue can be estimated by extrapolating from animals to humans using the parallelogram approach. A parallel study in humans and in experimental animals measures a common endpoint in both and also measures in the animal the endpoint in the desired target tissues inaccessible in man (Brewen and Preston, 1974; Sobels, 1982; Waters et al., 1986). For example, the genotoxic effects on human germinal cells, a cell type which is not readily accessible, can be estimated by determining the effects on lymphocytes from both humans and mice and on germinal cells from mice. The ratio of the effects in the two cell types
460
in the mouse can be used to extrapolate the germinal cell effects in man from the observed effects in human lymphocytes. The extrapolated data can be used to estimate genetic risk (from germinal cell effects) and carcinogenic potential (from bone-marrow cell effects). This technique is most amenable for use with cytogenetic assays (e.g., Au and Hsu, 1980; International Commission, 1983). Risk assessment from cancer bioassay studies can be quantitative (mathematical modeling), semi-quantitative (specific categories), or entirely qualitative (i.e., it is or it is not a carcinogen). The quantitative approach may not be of much greater actual value than the semi-quantitative approach that establishes categories of concern. A modified categorical ranking scheme based on chronic toxicity has been used by the U.S. Environmental Protection Agency (EPA) for over 8 years in the development of reportable quantities (RQs) of environmental pollutants (EPA, 1986). We believe that the CTP may be an appropriate adjunct to current procedures for establishing categories of concern. This categorical approach requires the delineation of an orderly and graded series of toxicological responses (EPA Draft Document on Risk Assessment, 1986). Our proposed categorical scheme will consider: (a) human a n d / o r animal studies, (b) endpoints that relate to the final disease outcomes, and (c) endpoints that are the most relevant in identifying a health hazard (see Table 1). For example, more weight will be given to especially relevant endpoints such as gene mutation and chromosome aberrations than to adduct formation or repair induction, as discussed earlier. Selection of the proper species for testing is critical, and one can argue that each chemical should be tested in both the rat and mouse as is now done in the standard cancer bioassay program. Brockman and DeMarini (1988) have indicated that the concordance between rat and mouse carcinogenicity studies is only 67%, which is roughly equivalent to the concordance between the rodent carcinogenicity studies and short-term genotoxic tests. In genetic toxicology, if we could show that by testing the two species we would identify with 100% accuracy all human carcino-
TABLE 1 C A T E G O R I E S F O R RISK A S S E S S M E N T H u m a n somatic mutagen
A
C h r o m o s o m e aberrations or gene mutations in h u m a n studies
Probable
BI
C h r o m o s o m e aberrations or gene mutations in animal studies and other positive endpoints. Limited h u m a n data C h r o m o s o m e aberrations or gene mutations in animal studies
B2
Possible
C
Positive animal endpoints other than chromosome aberrations or gene mutations
Insufficient data
D
Animal data lacking or deficient
Non-mutagen
E
Negative results in sufficient animal studies
a C h r o m o s o m e aberrations include the micronucleus test but not SCEs. ' O t h e r ' indicates studies, besides chromosome aberrations and gene mutations, such as adduct formation, repair induction, detection of urinary m u t a g e n s and SCEs.
gens, as is the case with the cancer bioassay, we would indeed be delighted. The figures of Brockman and DeMarini, however, may be misleading. Griesemer et al. (personal communication) have pointed out that the cancer bioassay is not constructed to test the concordance between the two rodent species. If a chemical is found to be positive in one species, and equivocal in another, it is not retested. The authors go on to point out that if equivocal tests were redone at higher dose levels the concordance with the chemicals tested to date in the bioassay may be much closer to 100%. Action level
A non-quantitative ranking procedure, as described above, is the most appropriate approach for categorizing chemicals on the basis of animal studies. There may be occasions, however, when it is desirable to generate a number indicating the need for immediate action. Action levels would
461 be determined by considering the lowest exposure at which genetic effects are observed or the highest levels at which no effects are seen. The routes and types (e.g. acute or chronic) of exposure would be consistent with typical patterns of human exposure. Pharmacokinetic data would be considered in order to determine whether the physiological context in which effects were observed was relevant to human exposure. Finally, safety factors of 1 : 10 or 1 : 100 would help assure an adequate margin of safety to compensate for synergistic effects of multiple agent exposures in man and to account for sensitivity among individuals. The ultimate objective would be to arrive at an action level which could form the basis for setting permitted levels of human exposure. In the above procedure, we have deliberately eliminated from consideration non-animal data. It is our belief that whole-animal studies must be the basis for risk assessment. Chemicals that may be carcinogenic through undefined epigenetic mechanisms are not considered. The term 'epigenetic', in the broadest sense, refers to a carcinogen that does not operate through any known, conventional genotoxic mechanism. It is our belief that many so called 'epigenetic' carcinogens are simply compounds that have not been evaluated properly in animal studies such as the c r P . The multi-endpoint CTP approach may offer, however, the best opportunity for identifying truly epigenetic agents. Although we have stressed somatic cell mutations and cancer in this article, a similar scheme could be worked out for germinal mutations. Risk assessment, at best, is an imprecise field where most of the assumptions are challengeable. Our present proposal for handling mutational data is obviously fraught with uncertainties. This approach, however, may be the most logical since it emphasizes the need for multi-endpoint testing (CTP) at sub-acute and low level exposures. This approach recognizes the differences in relative importance of various assays, and identifies the position in the total of the endpoints measured in the assays. Results from the most sensitive assay is determined or calculated. Finally, an action level can be determined that relates the most sensitive, relevant endpoint to actual human exposure.
Case study: benzene The categorical approach to risk assessment can be exemplified by examining the case for benzene. Benzene is a well established leukemogen in man (IARC, 1982) and has been shown to be a multipotent carcinogen in rats and mice (Huff et al., 1989; Maltoni et al., 1989). In numerous genotoxicity studies, benzene has been associated with increased frequencies of structural chromosome aberrations in lyrnphocytes from exposed humans (reviewed in ATSDR, 1987). The most informative study, for our purposes, is the report by Sarto et al. (1984). 22 workers from a benzene production facility were compared with the same number of matched subjects not exposed to benzene or other known genotoxins. Exposures were monitored with personal samplers and confirmed by measurement of alveolar benzene levels and urinary excretion of phenols. Exposures ranged from 1.1 to 12.4 ppm. Sarto et al. found a statistically significant increase of about 2-fold in the percentage of metaphase lymphocytes containing chromosometype structural aberrations. These findings are consistent with results of other cytogenetic monitoring studies of benzene exposed workers and establish a lowest observed effect level in the range of 1-10 ppm, which is near the current OSHA Permitted Exposure Limit of 1 ppm. An extensive literature exists on the genetic toxicity of benzene in rodents (ATSDR, 1987). Benzene has been found to consistently produce structural chromosome aberrations, micronuclei, and sister-chromatid exchanges in the bone marrow of mice and rats. The most relevant studies for risk evaluation are an acute inhalation study in rats and mice conducted by the Chemical Industry Institute of Toxicology (Erexson et al., 1986) and a sub-acute inhalation study conducted in our laboratory (Ward et al., 1989; Au et al., in press). In the study by Erexson et al., mice and rats were exposed to benzene by inhalation for periods of 6 h at concentrations of 0, 10, 100, 1000 ppm (male D B A / 2 mice) or 0, 0.1, 0.3, 1, 3, 10 or 30 ppm (male Sprague-Dawley rats). Induction of SCEs in peripheral blood lymphocytes and micronuclei in bone marrow erythrocytes were monitored. Significant increases in both mi-
462
"'" 10
o
8
O'~LO 6d
•
457
MUT 02511
Use of in vivo genetic toxicity data for risk assessment M a r v i n S. L e g a t o r a n d J o n a t h a n B. W a r d Jr. The University of Texas Medical Branch at Galveston, Department of Preventive Medicine and Community Health, Division of Environmental Toxicology, 2.102 Ewing Hall, J-I0, Galveston, TX 77550 (U.S.A.) (Accepted 5 April 1991)
Keywords: Risk assessment; Epigenetic, Comprehensive Testing Protocol; Benzene
Summary Mutagenicity studies have been used to identify specific agents as potential carcinogens or other human health hazards; however, they have been used minimally for risk assessment or in determining permissible levels of human exposure. The poor predictive value of in vitro mutagenesis tests for carcinogenic activity and a lack of mechanistic understanding of the roles of mutagens in the induction of specific cancers have made these tests unattractive for the purpose of risk assessment. However, the limited resources available for carcinogen testing and large number of chemicals which need to be evaluated necessitate the incorporation of more efficient methods into the evaluation process. In vivo genetic toxicity testing can be recommended for this purpose because in vivo assays incorporate the metabolic activation pathways that are relevant to humans. We propose the use of a multiple end-point in vivo comprehensive testing protocol (CTP) using rodents. Studies using sub-acute exposure to low levels of test agents by routes consistent with human exposure can be a useful adjunct to methods currently used to provide data for risk assessment. Evaluations can include metabolic and pharmacokinetic endpoints, in addition to genetic toxicity studies, in order to provide a comprehensive examination of the mechanism of toxicity of the agent. A parallelogram approach can be used to estimate effects in non-accessible human tissues by using data from accessible human tissues and analogous tissues in animals. A categorical risk assessment procedure can be used which would consider, in order of priority, genetic damage in man, genetic damage in animals that is highly relevant to disease outcome (mutation, chromosome damage), and data from animals that is of less certain relevance to disease. Action levels of environmental exposure would be determined based on the lowest observed effect levels or the highest observed no effect levels, using sub-acute low level exposure studies in rodents. As an example, the known genotoxic effects of benzene exposure at low levels in man and animals are discussed. The lowest observed genotoxic effects were observed at about 1-10 parts per million for man and 0.04-0.1 parts per million in subacute animal studies. If genetic toxicity is to achieve a prominent role in evaluating carcinogens and in characterizing germ-cell mutagens, minimal testing requirements must be established to ascertain the risks associated with environmental mutagen exposure. The use of the in vivo approach
Correspondence: M.S. Legator, University of Texas Medical Branch, 2.102 Ewing Hall, J-10, Galveston, TX 77550 (U.S.A.).
458
described here should provide the information needed to meet this goal. In addition, it should allow truly epigenetic or non-genotoxic carcinogens to be distinguished from the genotoxic carcinogens that are not detected by in vitro methods.
Minimal data required for risk assessment Prediction of potential human risks from laboratory studies is an imprecisc process and risk assessment is especially complex for chronic disease outcomes such as cancer. Mutagenicity studies arc used to a limited extent in risk assessment and have played a marginally supportive role in evaluating potential carcinogens. Mutagenicity studies per se have not played a critical role in determining the permissible levels for human exposure. Utilizing in vitro mutagenicity data, even for qualitatively identifying a carcinogen, is questionable. Several papers, including Rinkus and Legator (1979), have addressed the limitations of in vitro studies and the fallacy of the correlation between microbial tests and animal carcinogens. However, it was not until 1986 that the data was available to challenge the validity of these shortterm assays (Zeiger and Tennant, 1986; Zeiger, 1987). Tcnnant (1987) in his overall assessment of the field, came to the following conclusion: 'If current in vitro short-term tests for genetic toxicology are expected to replace long-term rodent studies for the identification of chemical carcinogens, that expectation should be abandoned.' The attractiveness of using mutagenicity data in risk assessment remains, however. The limited resources available for carcinogen testing, and the large number of chemicals which need to be examined, necessitate the incorporation of more efficient methods into the evaluation process. Mutagenicity testing offers one of the few available supplements to long-term rodent studies for carcinogenic risk evaluation. The major impediment is the questionable relevance of mutagenicity data when most of it is derived from in vitro studies. The in vivo approach to mutagenicity testing, in contrast to an in vitro one, is defensible because it incorporates metabolic activation pathways that are relevant to humans. It is difficult at the present time to identify a data base derived
from animal mutagenicity studies that is comprehensive enough to allow for risk assessment. Even in those instances where animal studies have been performed, they are usually acute studies using high concentrations and focusing on a single specific assay. If mutagenicity data is to be integrated into a risk-assessment procedure, subacute studies measuring multiple cndpoints, at realistic exposure levels, should be used. Towards this purpose, we describe a testing approach for evaluating mutagenic agents and a procedure for using this test data for risk assessment. This testing approach is a multi-endpoint procedure that we call the Comprehensive Testing Protocol (CTP) (Legator and Harper, 1988; Moslen and Legator, 1988; Au et al., 1988; Harper ct al., 1989; Legator and Harper, 1987; Legator et al., 1986). This procedure, utilizing a sub-acute exposure protocol in which animals are exposed to low levels of chemicals, will yield comprehensive data upon which to base a risk assessment. The CTP, in contrast to the traditional animal cancer bioassay, can be carried out in weeks rather than years. Since several endpoints arc measured concurrently, the C-TP maximizes thc amount of information acquired from a single experiment. The cost of a CTP is much less than the expense involved in carrying out individual mutation assays in separate experiments and will be a fraction of the cost of an animal cancer bioassay. An additional strength of the C-~P for risk assessment is that, for many agents, genotoxic responses can be observed at relatively low exposure levels. When animals can be treated at dose levels which are not far above those experienced by humans, concerns about the physiological relevance of exposure to man are reduced.
The comprehensive testing protocol (CTP) The CTP offers the unique opportunity to observe several steps in the mutagenesis process in the context of the full range of metabolic and
459
pharmacokinetic events in an intact animal. Steps in the mutagenesis process from the presence of mutagens in body fluids, to adduct formation, to DNA repair, to chromosome damage, to the final fixation of specific-locus mutation can be measured. In addition, it is now possible to determine the spectrum of mutations induced by specific exposure conditions. Cells from several tissues may be used as reporters of genetic damage. The conditions of exposure can be manipulated to evaluate duration, concentration, route or rate, and the influence of specific aspects of metabolism or pharmacokinetics can be evaluated by measurement of enzyme activities or levels, and distribution of the agent or its metabolites. The interactive effects of chemical mixtures can be studied or the influence of inducers or inhibitors of metabolic enzymes can be assessed. Thus, a comprehensive evaluation of the process of mutagenesis under exposure and metabolic conditions chosen by the investigator can be carried out. Not only can the potential of an agent to induce genetic damage be assessed, but the entire mutagenesis process and its relationship to exposure conditions and metabolic status can be evaluated. The CTP procedure is carried out by exposing mice or rats to a test agent under the desired conditions. Subsequently, urine can be collected and evaluated for the presence of mutagenic activity or specific metabolites. Protein a n d / o r DNA adducts may be quantitated in virtually any tissue from the animal. The distribution and concentrations of the agent or its metabolites may be determined and the effects of exposure on metabolic enzyme levels may be ascertained. DNA-repair processes may be detected in many tissues. Cytogenetic effects, observed as structural chromosome aberrations, micronucleus formation, or sister-chromatid exchange can be observed in a variety of tissues including bone marrow, spleen, thymus, testes and lung. The frequency of mutations at the hypoxanthine-guanine phosphoribosyl transferase locus is determined in T-lymphocytes from the spleen or thymus. Handling data from the CTP for risk assessment
The rapid advancements in physiology, molecular biology and genetics have provided direct
evidence for the multistage process that leads to malignancy. Specific mutations play a central role in this orderly and programmed process. Mutations in protooncogenes or normal genes that promote cellular growth lead to their activation as oncogenes. Mutations in tumor suppressor genes, which function in normal cells to halt proliferation, leads to the loss of the ability of the cell to constrain its own growth. Chromosomal mutations, including small and large deletions and translocations, as well as gene mutations, account for the specific alterations of both suppressor genes and protooncogenes. Therefore, animal studies that faithfully reflect the ability of specific metabolites to induce gene or chromosomal mutations in somatic cells should identify carcinogenic agents. The induction of chromosome damage a n d / o r gene mutation by a chemical or by radiation raises a high index of suspicion that the agent is a carcinogen. Because they are less closely associated mechanistically with carcinogenesis, adduct formation and induction of DNA repair are less relevant in classifying a chemical as a carcinogen or mutagen. There are several additional procedures, such as detection of urinary mutagens and induction of sister-chromatid exchanges (SCE), that are useful but alone cannot be considered sufficient for classification. It is assumed, as supported by extensive data, that the processes leading to cancer are essentially the same in animals and man. Therefore, the relative importance of each step in the process is the same in this multistage sequence whether it occurs in laboratory animals or in humans. Human risk for damage at a particular target tissue can be estimated by extrapolating from animals to humans using the parallelogram approach. A parallel study in humans and in experimental animals measures a common endpoint in both and also measures in the animal the endpoint in the desired target tissues inaccessible in man (Brewen and Preston, 1974; Sobels, 1982; Waters et al., 1986). For example, the genotoxic effects on human germinal cells, a cell type which is not readily accessible, can be estimated by determining the effects on lymphocytes from both humans and mice and on germinal cells from mice. The ratio of the effects in the two cell types
460
in the mouse can be used to extrapolate the germinal cell effects in man from the observed effects in human lymphocytes. The extrapolated data can be used to estimate genetic risk (from germinal cell effects) and carcinogenic potential (from bone-marrow cell effects). This technique is most amenable for use with cytogenetic assays (e.g., Au and Hsu, 1980; International Commission, 1983). Risk assessment from cancer bioassay studies can be quantitative (mathematical modeling), semi-quantitative (specific categories), or entirely qualitative (i.e., it is or it is not a carcinogen). The quantitative approach may not be of much greater actual value than the semi-quantitative approach that establishes categories of concern. A modified categorical ranking scheme based on chronic toxicity has been used by the U.S. Environmental Protection Agency (EPA) for over 8 years in the development of reportable quantities (RQs) of environmental pollutants (EPA, 1986). We believe that the CTP may be an appropriate adjunct to current procedures for establishing categories of concern. This categorical approach requires the delineation of an orderly and graded series of toxicological responses (EPA Draft Document on Risk Assessment, 1986). Our proposed categorical scheme will consider: (a) human a n d / o r animal studies, (b) endpoints that relate to the final disease outcomes, and (c) endpoints that are the most relevant in identifying a health hazard (see Table 1). For example, more weight will be given to especially relevant endpoints such as gene mutation and chromosome aberrations than to adduct formation or repair induction, as discussed earlier. Selection of the proper species for testing is critical, and one can argue that each chemical should be tested in both the rat and mouse as is now done in the standard cancer bioassay program. Brockman and DeMarini (1988) have indicated that the concordance between rat and mouse carcinogenicity studies is only 67%, which is roughly equivalent to the concordance between the rodent carcinogenicity studies and short-term genotoxic tests. In genetic toxicology, if we could show that by testing the two species we would identify with 100% accuracy all human carcino-
TABLE 1 C A T E G O R I E S F O R RISK A S S E S S M E N T H u m a n somatic mutagen
A
C h r o m o s o m e aberrations or gene mutations in h u m a n studies
Probable
BI
C h r o m o s o m e aberrations or gene mutations in animal studies and other positive endpoints. Limited h u m a n data C h r o m o s o m e aberrations or gene mutations in animal studies
B2
Possible
C
Positive animal endpoints other than chromosome aberrations or gene mutations
Insufficient data
D
Animal data lacking or deficient
Non-mutagen
E
Negative results in sufficient animal studies
a C h r o m o s o m e aberrations include the micronucleus test but not SCEs. ' O t h e r ' indicates studies, besides chromosome aberrations and gene mutations, such as adduct formation, repair induction, detection of urinary m u t a g e n s and SCEs.
gens, as is the case with the cancer bioassay, we would indeed be delighted. The figures of Brockman and DeMarini, however, may be misleading. Griesemer et al. (personal communication) have pointed out that the cancer bioassay is not constructed to test the concordance between the two rodent species. If a chemical is found to be positive in one species, and equivocal in another, it is not retested. The authors go on to point out that if equivocal tests were redone at higher dose levels the concordance with the chemicals tested to date in the bioassay may be much closer to 100%. Action level
A non-quantitative ranking procedure, as described above, is the most appropriate approach for categorizing chemicals on the basis of animal studies. There may be occasions, however, when it is desirable to generate a number indicating the need for immediate action. Action levels would
461 be determined by considering the lowest exposure at which genetic effects are observed or the highest levels at which no effects are seen. The routes and types (e.g. acute or chronic) of exposure would be consistent with typical patterns of human exposure. Pharmacokinetic data would be considered in order to determine whether the physiological context in which effects were observed was relevant to human exposure. Finally, safety factors of 1 : 10 or 1 : 100 would help assure an adequate margin of safety to compensate for synergistic effects of multiple agent exposures in man and to account for sensitivity among individuals. The ultimate objective would be to arrive at an action level which could form the basis for setting permitted levels of human exposure. In the above procedure, we have deliberately eliminated from consideration non-animal data. It is our belief that whole-animal studies must be the basis for risk assessment. Chemicals that may be carcinogenic through undefined epigenetic mechanisms are not considered. The term 'epigenetic', in the broadest sense, refers to a carcinogen that does not operate through any known, conventional genotoxic mechanism. It is our belief that many so called 'epigenetic' carcinogens are simply compounds that have not been evaluated properly in animal studies such as the c r P . The multi-endpoint CTP approach may offer, however, the best opportunity for identifying truly epigenetic agents. Although we have stressed somatic cell mutations and cancer in this article, a similar scheme could be worked out for germinal mutations. Risk assessment, at best, is an imprecise field where most of the assumptions are challengeable. Our present proposal for handling mutational data is obviously fraught with uncertainties. This approach, however, may be the most logical since it emphasizes the need for multi-endpoint testing (CTP) at sub-acute and low level exposures. This approach recognizes the differences in relative importance of various assays, and identifies the position in the total of the endpoints measured in the assays. Results from the most sensitive assay is determined or calculated. Finally, an action level can be determined that relates the most sensitive, relevant endpoint to actual human exposure.
Case study: benzene The categorical approach to risk assessment can be exemplified by examining the case for benzene. Benzene is a well established leukemogen in man (IARC, 1982) and has been shown to be a multipotent carcinogen in rats and mice (Huff et al., 1989; Maltoni et al., 1989). In numerous genotoxicity studies, benzene has been associated with increased frequencies of structural chromosome aberrations in lyrnphocytes from exposed humans (reviewed in ATSDR, 1987). The most informative study, for our purposes, is the report by Sarto et al. (1984). 22 workers from a benzene production facility were compared with the same number of matched subjects not exposed to benzene or other known genotoxins. Exposures were monitored with personal samplers and confirmed by measurement of alveolar benzene levels and urinary excretion of phenols. Exposures ranged from 1.1 to 12.4 ppm. Sarto et al. found a statistically significant increase of about 2-fold in the percentage of metaphase lymphocytes containing chromosometype structural aberrations. These findings are consistent with results of other cytogenetic monitoring studies of benzene exposed workers and establish a lowest observed effect level in the range of 1-10 ppm, which is near the current OSHA Permitted Exposure Limit of 1 ppm. An extensive literature exists on the genetic toxicity of benzene in rodents (ATSDR, 1987). Benzene has been found to consistently produce structural chromosome aberrations, micronuclei, and sister-chromatid exchanges in the bone marrow of mice and rats. The most relevant studies for risk evaluation are an acute inhalation study in rats and mice conducted by the Chemical Industry Institute of Toxicology (Erexson et al., 1986) and a sub-acute inhalation study conducted in our laboratory (Ward et al., 1989; Au et al., in press). In the study by Erexson et al., mice and rats were exposed to benzene by inhalation for periods of 6 h at concentrations of 0, 10, 100, 1000 ppm (male D B A / 2 mice) or 0, 0.1, 0.3, 1, 3, 10 or 30 ppm (male Sprague-Dawley rats). Induction of SCEs in peripheral blood lymphocytes and micronuclei in bone marrow erythrocytes were monitored. Significant increases in both mi-
462
"'" 10
o
8
O'~LO 6d
•
