Expectations for Transgenic Rodent Cancer Bioassay Models

Transcription

1 TOXICOLOGIC PATHOLOGY, vol 29(Suppl.), pp , 2001 Copyright C 2001 by the Society of Toxicologic Pathology Expectations for Transgenic Rodent Cancer Bioassay Models JOHN ASHBY Syngenta Central Toxicology Laboratory, Alderley Park, Maccles eld, Cheshire, SK10 4TJ, United Kingdom ABSTRACT The results of the present study have advanced dramatically the database on transgenic mouse abbreviated carcinogenicity bioassay models. As such, it will provide a secure foundation for future evaluations of these assays and for their eventual validation as models for the prediction of possible human carcinogens. Based upon the results derived from the present study, it is suggested that 5 areas require discussion as a prelude to the further evaluation of existing models and the future evaluation of new models. First, there is the need to agree a standard list of calibration chemicals to be studied and to derive agreement on optimal bioassay group sizes, statistical methods, and exposure periods. Second, general agreement must be reached regarding the classes/types of known rodent carcinogens so that it is acceptable for the new models to nd negative, by implication, those rodent carcinogens considered not to pose a carcinogenic hazard to humans. Third, current understanding of mechanisms of carcinogenesis should be integrated into the evaluation of new bioassay models. Fourth, any changes made to the standard rodent carcinogenicity bioassay protocol will require compromises being made, and these should be commonly owned between interested parties in order to reduce the number of regional/agencyspeci c carcinogenicity testing schemes. Fifth, a mechanism needs to be developed by which assays can be adopted or rejected for use in the routine bioassay of chemicals. In the absence of such initiatives the increasing number of new bioassay models will come to exist along side of the standard 2-species bioassay, and this may potentially lead to confusion regarding the true future role of these assays. Keywords. Carcinogen; cancer bioassay; cancer model evaluation. DISCUSSION The development of genetically modi ed rodents having shorter latent periods for the chemical induction of cancer represents a milestone in cancer research and a potentially critical development in chemical safety assessment. The number of mouse models genetically modi ed to enhance susceptibility to carcinogenesis continues to grow, and the development of rat models is imminent. The present evaluation exercise used several of the currently most promising mouse models (13). A pressing need to emerge from the meeting of investigators was for a means to assess the potential value/advantages of individual models, given that new models are appearing at a faster rate than existing models can be evaluated using current methods. In the absence of a solution to this problem, inadequate assessments of assays may lead to insecure revisions being made to existing carcinogenicity testing practices, which themselves have yet to be fully rationalized. The data generated during the present evaluation exercise have clearly focused the key issues faced when evaluating a new assay, and this should help the future establishment of robust assay validation criteria. The following discussion is based on a plenary lecture delivered by the author to the meeting of investigators, and has been divided into subsections, each of which explores the major assay assessment criteria discussed at the meeting. The design and results of the present collaborative evaluation exercise are assumed in this paper. Integration of New Bioassay Models With the Standard Rodent Bioassay One approach to the evaluation of new bioassay models would be to study all of the known human carcinogens, and Address correspondence to: J. Ashby, Syngenta Central Toxicology Laboratory, Alderley Park, Maccles eld, Cheshire, SK10 4TJ, United Kingdom; John.Ashby@Syngenta.com. if these are found to be positive, to accept the new model(s) as reliable predictors of human carcinogens. This approach would also require that noncarcinogens gave a negative response while nding the human carcinogens positive. To date, no human noncarcinogens have been de ned, so reliance would have to be placed on those rodent noncarcinogens de- ned by the US National Toxicology Programme (NTP). An alternative approach is to regard 2-species rodent carcinogens as the reference point for possible human carcinogens. Even so, the NTP rodent bioassay protocol has never been fully justi ed in terms of the species/strains of animals used, or in terms of the duration or route of chemical exposure. The approach adopted for the present study was a compromise between these two extremes to use a small number of human carcinogens, together with a range of rodent carcinogens considered not to pose a hazard to humans. The advent of transgenic mouse models has also posed the problem of whether one or more of them should be regarded as replacements for the B6C3F1 mouse in the NTP bioassay, or as additions to the existing NTP bioassay. Further, compromises have been made to the test protocols used in the new assays compared with that used in the NTP bioassay, including the use of reduced group sizes and reduced durations of chemical exposure. In addition, with one of the assays (the TgAC mouse), topical application of the test chemical has been accepted as the standard route of exposure, a rarely used route in NTP bioassays. These changes make it dif cult to relate responses given by chemicals in the NTP bioassay with those given by the new bioassay models. There is also discussion regarding whether the new models provide individually different information regarding the possible mechanism of carcinogenic action of the chemicals selected for study. These several problems make it dif cult to conduct a formal validation of the new assays at this stage, which was why that was not attempted. In fact, it was for these reasons that the present study was described as an evaluation, /01$3.00 $0.00

2 178 ASHBY TOXICOLOGIC PATHOLOGY as opposed to a validation. However, given the large number of existing new bioassay models, and the prospect for the development of an increasing number of additional new models, the issue of assay validation must be taken seriously in the near future. Before that can happen, however, it will be necessary to agree the future role of these assays. Perhaps the rst step to answering that question will be to derive further data, mindful that the underlying need for some form of assay validation must eventually be faced. The following discussion is concerned primarily with the steps necessary to expedite the eventual acceptance of rejection of individual, or combinations, of the new bioassay models as routine aids to the assessment of the potential human carcinogenicity of chemicals. Comparisons of the relative costs of the NTP and the newer bioassays ( nancial costs, numbers of animals used, and possible effects on human safety assessments) must await a clearer presentation of the function and performance characteristics expected of the new assays. Required Performance Characteristic of New Bioassay Models Any attempt to change the protocol/test species of the NTP carcinogenicity bioassay necessitates a clear exposition of the sensitivity, speci city, and accuracy for rodent carcinogens required of any modi ed bioassay model. This is made dif cult by the absence of general agreement on which classes/types of rodent carcinogens should be detected for purposes of human hazard assessment of chemicals (3, 9). This uncertainty is illustrated by the choice of test chemicals for the present evaluation (13). Thus, in addition to the human carcinogens and the rodent noncarcinogens selected for study, there was a group of rodent carcinogens considered not to pose a carcinogenic hazard to humans by virtue of their proposed mechanism of action, knowledge that the effects in rodents were only observed at doses never encountered by humans, or because of knowledge that they do not cause cancer in exposed populations. The expected response of the new rodent models to this last group of carcinogens was not stated at the start of the study, and this allows post hoc rationalizations of the data generated. An example of this is provided by the data derived during this study for the 3 peroxisome proliferator rodent liver carcinogens shown in Figure 1. In the absence of clear expectations of assay per- FIGURE 1. Results for 3 of the peroxisome proliferator carcinogens studied. The different constellations results observed (arrows) are each capable of rationalisation, as discussed in the text. FIGURE 2. The 4 major classes of mammalian carcinogens from which chemicals can be selected for assay validation. *Probably more ef ciently detected by other toxicity assays. HC human carcinogen; GT genotoxic; NC noncarcinogen. formance for these chemicals, the divergent sets of results observed (arrows in Figure 1) are each open to rationalization. For example, the SHE cell transformation assay could be said to have correctly detected the most potent of these rodent carcinogens, and the XPA mouse model to have correctly detected only the most potent rodent carcinogen of the three. In contrast, the Hras 2 assay found all three to be positive, consistent with their rodent liver carcinogenicity. On the other hand, the P53 model and the neonatal mouse models could be said to have correctly failed to detect any of these rodent carcinogens, consistent with the presumed status of these chemicals as nongenotoxic carcinogens. Clearly, as the evaluation of new bioassay models proceeds, it will be necessary to de ne more precisely the performance characteristics required of them. As a prelude to this, the 4 major categories of mammalian carcinogens are shown in Figure 2 and discussed below. Human Carcinogens. This class of carcinogen is an obvious priority for detection by any new bioassay model. However, it is suggested that the critical activities of relevance for human risk assessment of the hormonal and immunosupressive agents could be detected more ef ciently by standard toxicity assays than by the conduct of rodent carcinogenicity bioassays. Thus, the rst criterion for the assessment of a modi ed bioassay will be its ability to detect as positive reference genotoxic human carcinogens, preferably within a shortened time scale and perhaps also with enhanced statistical clarity as compared with that provided by the standard rodent bioassay. If this is not achieved, possible weak human carcinogens cannot be expected to be detected by these assays. In fact, many of the assays found only weak responses for the 3 genotoxic carcinogens studied (melphalan, cyclophosphamide, and phenacetin); this is a cause for concern, possibly indicating the need to increase either test group sizes and/or the duration of the bioassays. However, that would increase the costs of the new bioassays, currently claimed to be one of their main advantages over the standard NTP bioassay.

3 Vol. 29(Suppl.), 2001 RODENT CANCER BIOASSAY MODELS 179 Genotoxic Rodent Carcinogens. No reference genotoxic carcinogens were evaluated in the present study because most of the assays evaluated had detected several of the most potent of these carcinogens in earlier studies, albeit usually using slightly different test protocols than those used in the present study. The evaluation of historically important reference rodent carcinogens, such as 2-acetylamino uorene, benzo[a]pyrene, and diethylnitrosamine, and their weaker analogues, should probably form a key part of future evaluations of the new bioassay models. Noncarcinogens. The evaluation of rodent noncarcinogens is required to assess the speci city of those assays shown to have appropriate carcinogen sensitivity. In general, the assays evaluated performed well with the 3 noncarcinogens studied ampicillin, mannitol, and sul soxazole. However, offsetting this success is the fact that the sensitivity of the assays to the human carcinogens studied was quantitatively low. In this connection, it is interesting to note that at the outset of the study, a general concern was expressed that all of the assays evaluated would suffer from high carcinogen sensitivity and low noncarcinogen speci city, the reverse of that seen. Rodent Carcinogens Assessed to be Noncarcinogenic to Humans. The most pointed dilemma encountered in the present study related to those rodents carcinogens classi- ed as not presenting a hazard to humans (Figure 2). These selective carcinogens were considered to operate either by a high-dose and rodent-speci c mechanism, or to be incapable of eliciting cancer in humans based upon the results of the available epidemiological observations. Most of the current discussions regarding the relevance to humans of rodent carcinogenicity data concern compounds such as these, and the inclusion of such a high proportion of them in the present study complicated evaluation of the results. In general, most assays failed to detect these 12 rodent carcinogens, a nding that again must be taken within the context of the low levels of activity recorded by most of the assays for the human carcinogens studied. It was perhaps premature in the evaluation process to study such a large number of these selective carcinogens. FIGURE 3. Differing responses of the 2-year bioassay to dichlorvos depending upon the route of administration used (1, 2). S stomach tumors. FIGURE 4. Differing responses to DDT dependent upon the strain of mouse used (10, 11). Variables in the Study Beyond the Genetic Background of the Mice In the present study, differences in the genetic backgrounds of the various transgenic mice was the implicit key variable compared to the NTP rodent bioassay protocol, yet the route of administration and the duration of dosing of the test chemical, and the species/strain of the test animals used, were concomitantly varied, and such variations can themselves in uence the outcome of a carcinogenicity bioassay. The data shown in Figures 3 to 5 illustrate that carcinogenicity is not always an absolute property of chemicals, but may be in uenced by the conditions of the cancer bioassay. Thus, dichlorvos is a route-dependent carcinogen (Figure 3; 1, 2) and DDT is a strain-speci c mouse liver carcinogen (Figure 4; 10, 11). Such activity pro les underline the importance of gaining general agreement on the expectations for these new models and of ensuring, as far as is possible, that the only variable under study is the genetic makeup of the test species. FIGURE 5. Lifetime bioassay results for pyridine,togetherwith the responses seen in the hemizygous p53 and the TgAC mouse bioassays (14).

4 180 ASHBY TOXICOLOGIC PATHOLOGY The Problem Posed by Single-Species/Single-Tissue Carcinogens One of the enduring problems in cancer risk assessment has been how to extrapolate carcinogenic effects seen only in a single tissue of a single strain of rodent to humans. This problem is often reduced to the question of mouse liver speci c carcinogens, but the real picture is more interesting. Thus, pyridine is a tissue-speci c/strain-speci c carcinogen to both rats and mice (Figure 5) and these NTP bioassay data have recently been supplemented by negative bioassay results for pyridine in the TgAC and the p53 mouse models (14). This overall pro le of activities has been argued to indicate a carcinogen of no carcinogenic hazard to humans (14). If there were to be general agreement on this proposal, it would simplify the evaluation of the new bioassay models, but there is no such agreement. Further, such agreement will be complicated by the fact that the liver tumors induced by pyridine in both sexes of mouse consist of a mixture of hepatocarcinomas and the very rare hepatoblastoma tumor type. Reanalysis of earlier analyses of the 301 chemicals evaluated for carcinogenicity by the NTP by 1991 (4, 5) reveals that 30% of the carcinogens affected only a single tissue of a single species 15% being speci c to the rat and 15% to the mouse (Figures 6 to 8). In the rat, the affected tissues are those that have been studied for rodent-speci c mechanisms of carcinogenic action, and include primarily the kidney, bladder, thyroid gland, and haemopoietic system (Figure 7). In the mouse, the main tissue is the liver; a situation heightened by the mouse liver also being the single mouse tissue affected for 24 of the trans-species rodent carcinogens (Figure 8). A wide selection of tissue-speci c NTP carcinogens exist and have been discussed elsewhere (4, 12), illustrating that selective carcinogenic responses are a reality beyond the mouse liver, and beyond the 12 carcinogens of this type evaluated in the present study. Whether or not genetically modi ed rodents should detect such carcinogens has yet to be agreed upon by FIGURE 7. Reanalysis of earlier (2, 3) analyses of the US NTP carcinogen database. The types of rat-speci c carcinogens are thereby revealed. the wider scienti c/regulatory community. Until such time the apparent carcinogen sensitivity gures of these assays will be dependent on the proportion of selective carcinogens employed in the assays validation. The 2 mouse Model Testing Proposal Made by Tennant et al Tennant et al (15, 16) have proposed the use of a combination of the P53 mouse (oral gavage administration of the test agent) and the TgAC mouse (skin application) to provide a screen for chemical carcinogens (Figure 9). Although the present study was not formally predicated on those proposals (15, 16), the results derived from this study will inevitably be considered within that context, and the topic is therefore worthy of consideration here. An implication of this testing scheme proposed by Tennant et al (15, 16) is that the classical 2-year bioassay result is not the primary correlate. Thus, some selective rodent, carcinogens can be found to be inactive by the 2 mouse models without incurring a penalty to their FIGURE 6. Reanalysis of earlier (4, 5) analyses of the US NTP carcinogen database. The proportion of species-speci c carcinogens is thereby revealed. FIGURE 8. Reanalysis of earlier (4, 5) analyses of the US NTP carcinogen database. The types of mouse-speci c carcinogens are thereby revealed.

5 Vol. 29(Suppl.), 2001 RODENT CANCER BIOASSAY MODELS 181 FIGURE 9. Shematic illustration of the two-mouse limited-carcinogenicity bioassay testing proposal made by Tennant et al (15, 16). overall performance pyridine being one such example discussed above (14). The drawback with this concept is that the speci c carcinogens that are allowed to remain undetected by the 2 mouse strains have not been listed or rationalized. This leaves the door open to post hoc rationalizations of test data. A deeper problem is the implication that carcinogens found to be active in the P53 model are thereby de ned as genotoxic carcinogens, while those only found to be active in the TgAC assay are thereby de ned as being nongenotoxic carcinogens. The data supporting this proposal have not been marshalled and discussed, and the matter is clouded by the occasional recovery of chemically induced P53 mouse tumors bearing the original single copy of the intact P53 gene. However, the ultimate problem with the 2 mouse model testing scheme is that it leaves no room for the other assays evaluated in the present study, or for the many additional genetically modi- ed mouse and rat models that will inevitably be developed in the future. The 2 mouse model testing scheme proposed by Tennant et al (15, 16) should act as a catalyst to re nement of the NTP bioassay protocol, rather than as a ready-made solution to currently perceived weaknesses of it. Absence of a 6-Month Database for Chemical Carcinogens The decision to limit the exposure period to 6 months for most of the available transgenic mouse models was arbitrary, but was essential if these models are to present a clear advantage over the standard 2-year bioassay protocol. To date, the only data-based suggestion for a shortening of the bioassay exposure period has been that 18 months may suf ce without loss of overall carcinogen sensitivity (6), but even this relatively moderate proposal has been contested (8). A particular problem to adoption of a 6-month bioassay exposure period (or 9 months in the case of the XPA model) is that we do not have knowledge regarding how many 2-year rodent carcinogens would be still be active at this earlier time. Thus, when a rodent carcinogen is found to be negative in one of the new models after 6 months of exposure, it is not clear whether this is an attribute of the genetics of the new model or of the shortened period of the bioassay. This dilemma was recognized by Yamamoto et al (17) when they reported a negative result after 6 months of exposure of the Hras2 mouse to the mouse hepatocarcinogen 1,1,1-trichloroethane. In the discussion section of that paper, the authors concluded either that the rodent carcinogenicity of this chemical may have no relevance for humans, or, that liver tumors may have developed had the bioassay been continued for a longer period. Such unresolved questions will bedevil the future use of these assays until answered speci cally. Although not a permanent solution to this problem, the current practice of most investigators of including in the bioassay the wild-type mouse from which the transgenic mouse was developed is to be encouraged. For example, in the present study, some of the genotoxic human carcinogens elicited a weak effect after 6 months of exposure of the wild-type mice, and a stronger response in the transgenic mice. Such data con rm that the transgenic model under study has superior sensitivity to the wild-type mouse, and this engenders con dence in the test results. However, when a negative result is observed for a 2-year rodent carcinogen in both the wild-type and the transgenic mouse after 6 months, little can be learned from the results. The only solution to this problem is to establish clearly the sensitivity of the new models at the 6-month time point to carcinogens that must be detected, agents such as the genotoxic human carcinogens and reference genotoxic rodent carcinogens. This will involve consideration of optimum group sizes, statistical assessment criteria (7), and exposure periods of the bioassay. CONCLUSIONS AND THE WAY FORWARD The evaluation of transgenic rodents as accelerated carcinogenicity bioassay models should continue, but greater clarity is required concerning the future role of these assays and their required performance characteristics. Based upon the results of the present study, it is suggested that 5 areas require discussion as a prelude to the further evaluation of existing models and the future evaluation of new models. First, there is the need to agree a standard list of calibration chemicals to be studied and to derive agreement on optimal bioassay group sizes, statistical methods, and exposure periods. Second, general agreement must be reached regarding the classes/types of known rodent carcinogen that it is acceptable for the new models to nd negative, by implication, those rodent carcinogens considered not to pose a carcinogenic hazard to humans. Third, current understanding of mechanisms of carcinogenesis should be integrated into the evaluation of new bioassay models. Fourth, any changes made to the standard rodent carcinogenicity bioassay protocol will require compromises being made, and these should be commonly owned between interested parties in order to reduce the number of regional/agency-speci c carcinogenicity testing schemes. Fifth, a mechanism needs to be developed by which assays can be adopted or rejected for use in the routine bioassay of chemicals. In the absence of such initiatives, the increasing number of new bioassay models will come to exist along side of the standard 2-species bioassay, and this potentially can lead to confusion regarding the true future role of these assays. REFERENCES 1. Anonymous (1977). Bioassay of dichlorvos for possible carcinogenicity. NCI Carcinogenesis TR Anonymous(1989). Toxicologyand carcinogenesis studies of dichlorvosin F344/N rats and B6C3F1 mice (gavage studies). NTP Carcinogenesis TR Ashby J (1997). Identifying potential human carcinogens the role of genetically altered rodents. Toxicol Pathol 25:

6 182 ASHBY TOXICOLOGIC PATHOLOGY 4. Ashby J, Paton D (1995). Chemicals for evaluating the sensitivity and speci- city of reduced/transgenic rodent cancer bioassay protocols. Mut Res 331: Ashby J, Tennant RW (1991). De nitive relationships among chemical structure, carcinogenicity and mutagenicity for 301 chemicals tested by the U.S. NTP. Mut Res 257: Davies TS, Lynch BS, Monro AM, Munro IC, Nestmann ER (2000). Rodent carcinogenicity tests need be no longer than 18 months: an analysis based on 210 chemicals in the IARC monographs. Food Chem Toxicol 38: Dunson DB, Haseman JK, van Birgelen APJM, Stasiewicz S, Tennant RW (2000). Statistical analysis of skin tumor data from Tg.AC mouse bioassays. Toxicol Sci 55: Haseman J, Melnick R, Tomatis L, Huff J (2001). Carcinogenesis bioassays: study duration and biological relevance. Food Chem Toxicol 39: Huff J (1999). Long-term chemical carcinogenesis bioassays predict human cancer hazards. Issues, controversies, and uncertainties. Ann N Y Acad Sci 895: IARC (1974). Monographs on the Evaluation of the Carcinogenic Risks of Chemicals to Man, 5. Some Organochlorine Pesticides. IARC, Lyon, France. 11. IARC (1991). Monographs on the Evaluation of Carcinogenic Risks to humans, 53. Occupational Exposures in Insecticide Application, and Some Pesticides. IARC, Lyon, France. 12. IARC (1999). Species Differences in Thyroid, Kidney and Urinary Bladder Carcinogenesis. Publication No IARC, Lyon, France. 13. Robinson D (1998). The International Life Sciences Institute s role in the evaluation of alternative methodologies for the assessment of carcinogenic risk. Toxicol Pathol 26: Spalding JW, French JE, Stasiewicz S, Furedi-Machacek M, Conner F, Tice RR, Tennant RW (2000). Responses of transgenic mouse lines p53 / and Tg AC to agents tested in conventional carcinogenicity bioassays. Toxicol Sci 53: Tennant RW, French JE, Spalding JW (1995). Identifying chemical carcinogens and assessing potential risk in short-term bioassays using transgenic mouse models. Environ Health Perspect 103: Tennant RW, Spalding J, French JE (1996). Evaluation of transgenic mouse bioassays for identifying carcinogens and non-carcinogens. Mut Res 365: Yamamoto S, Urano K, Nomura T (1998). Validation of transgenic mice harboring the human prototype c-ha-ras gene as a bioassay model for rapid carcinogenicity testing. Toxicol Lett :