Introduction

1. The International Conference on the Harmonisation of Technical Requirements for the Registration of Pharmaceuticals for human use (ICH) has agreed that the required bioassay data may be derived from only one species, i.e. the rat1 This would be supported by appropriate mutagenicity and pharmacokinetic data and also information that could come from newly proposed short-term in-vivo test models for assessment of potential carcinogenic activity in mice (in particular heterozygous p53+/- deficient Tg.AC model, and ras H2 models). Although these recommendations apply only to human medicines, any decision could have significant implications for other categories of chemical (e.g. food additives, pesticides, industrial chemicals etc). The key public health issue is whether the proposed short term tests in transgenic mice are appropriate adjuncts to the rat carcinogenicity bioassay in the identification of chemical carcinogens.

2. The COC has as part of its remit to advise on issues relating to chemical carcinogenesis and to present recommendations for testing strategy. The Committee was asked by the Department of Health in 1998 to consider the available literature from three research groups (namely NTP/NIEHS, ILSI/HESI and CIEA*) on the proposed short-term in-vivo tests for assessment of carcinogenic activity in mice, and specifically, on the transgenic mice models (heterozygous p53 +/- deficient, Tg.AC model, and ras H2 model). The Committee reached the following conclusions in 1998, which were published.2

-----------------------------------------------------------------------------------------------

* NTP National Toxicology Program U.S.A. NIEHS National Institute for Environmental Health Sciences U.S.A. ILSI International Life Sciences Institute. HESI Health and Environmental Sciences Institute U.S.A. CIEA Central Institute for Experimental Animals, Japan.

---------------------------------------------------------------------------------------------------

i) The Committee recognises that the three transgenic models considered in this paper (p53+/- , Tg.AC, ras H2) appear to be highly sensitive to carcinogens but questions whether data from such tests would add much to the information which can be derived from a well conducted in-vivo evaluation of mutagenic potential.

ii) Dose-response data from tests in transgenic animals might be useful but at present there is no way of interpreting these data and extrapolating them to humans.

iii) The Committee considers that the further development and validation of short-term in-vivo models to evaluate non-genotoxic carcinogenesis and tumour promoters may be valuable. However, there is likely to be less scope for the use of the proposed short-term animal models for other categories of chemicals, such as pesticides and industrial chemicals, where the available supporting information, such as the results of metabolism studies, is likely to be more limited. Hence the development of short-term carcinogenicity tests for these chemicals needs to be considered carefully.

iv) The Committee concludes that, in view of the lack of appropriate validation data, it would not be appropriate to use the data from short-term transgenic bioassays considered in this statement to support regulatory decisions at the present time.

Introduction to ILSI/HESI Programme in Alternative Cancer Models

3. The International Life Sciences Institute (ILSI) and the Health and Environmental Science Institute (HESI) have co-ordinated a multinational research programme, from 1996 –2000, on the use of alternative cancer models. The research involved input from over 50 industrial, governmental (USA, Denmark, Netherlands and Japan) and academic laboratories and cost around $35m.3

4. The data from the project along with a number of evaluation papers from independent experts and abstracts of additional work were published as a supplement to volume 29 of Toxicologic Pathology in November 2001 pp1-351. These papers were distributed first in draft form then as a final version to Regulatory Authorities and Advisory Committees, world wide, for comment. ILSI/HESI have co-ordinated a collaborative research project using 21 chemicals, including six known human carcinogens (three genotoxins, one immunosuppressant and two hormonal carcinogens), 12 rodent specific carcinogens (presumed on the basis of epidemiology and /or mechanism of action data) and three non-carcinogens. Chemical selection was targeted predominantly at non-genotoxic carcinogens in view of the need to examine specific mechanisms of chemical carcinogenicity in the animal models under consideration. In addition appropriate data were already available on a number of genotoxic carcinogens in some of these animal models. All chemicals used were readily accessible to test laboratories and certain core data were available; i.e. 2 year bioassay data in 2 species, established toxicology database, data on human exposure and effects.

5. The research programme was overseen by a Steering Committee of scientists drawn from academia and from pharmaceutical companies. The models under consideration were: p53+/-, ras H2+/- , Tg.AC, Xpa-/- , Xpa-/-/p53+/- double knockout, neonatal mouse, and Syrian Hamster Embryo (SHE) assay.

6. The protocols used were based on existing knowledge for each model. Positive control chemicals were used to demonstrate that each testing laboratory could undertake and report a positive assay for the model under test. Participating laboratories volunteered to act as compound co-ordinators, identifying sources of supply, co-ordinating the characterisation of chemicals and analytical methods for toxicokinetic studies. They also provided advice on the evaluation of 4-week dose range funding studies. It is noted that in practice the high dose level used equated to the Maximum Tolerated Dose (MTD). Assay Working Groups (AWGs) were formed for each assay, initially to refine protocols and to make recommendations on dose levels but eventually provided considerable assistance in resolving practical issues which arose during the research programme. AWGs also acted to collate data and to as a focal point for review of data and the application of the evaluation criteria. A Pathology Subcommittee and Statistics Subcommittee of the Alternative Cancer Test Committee were established to help set consistent criteria for evaluating studies.4 The AWG acted as peer-review for data assessment before the results of studies were entered into the ILSI Alternatives to Carcinogenicity Testing Database. The database will eventually be made publicly available. A workshop was held 1-3 November 2000 in Leesburg, Virginia, USA to review the data from the research programmes.

7. The Committee’s assessment was based predominantly on pre-publication reports submitted to the June 2001 COC and a brief consideration of the published results. The Committee’s comments focused on the proposal that the alternative cancer tests models under consideration could be used as replacements to a long-term carcinogenicity bioassay in the mouse. The Committee made a number of general comments on the strategy used by ILSI/HESI before considering the results of each model. The Committee agreed to consult the COM for additional advice on the conduct of the Syrian Hamster Embryo cell transformation assay.

General Comments on ILSI/HESI strategy

8. Members welcomed the opportunity to comment on the pre-publication papers and raw data from the AWGs. Members acknowledged the considerable administrative and practical problems that had confronted ILSI/HESI in co-ordinating this work. It was considered that the programme had provided a large amount of information on the evaluation of performance of these assays but the data were not sufficient to validate the use of any of the assays for regulatory testing. Members asked for a number of comments to be forwarded to the ILSI/HESI Alternative Cancer Test Committee for inclusion in the peer review process.

9. The Committee noted that one of the aims in the selection of test chemicals had been to expand the available data set to included non-genotoxic carcinogens as data were already available on a range of genotoxic carcinogens. A key aim was to examine the ability of the individual alternative cancer models to detect human carcinogens. The carcinogens selected by ILSI/HESI were considered to act by a range of mechanisms including immunosuppression, enzyme induction, cell proliferation, and receptor mediated. Members agreed the rationale proposed by the investigators but commented that the categorisation of some of the carcinogens based on mechanisms in rodents and epidemiology data was debatable. However, it was agreed that the categorisation as suggested by ILSI/HESI would be used in this statement.

10. The Committee agreed that it was important to have the results of tests for all of the 21 chemicals selected using all of the assays. Thus it was agreed that a good level of testing had been achieved with perhaps the exceptions being for some rodent carcinogens in the Tg.AC, Xpa, Xpa/P53 and neonatal mouse models. Members also considered that the inconsistent response of some positive control chemicals in some of the assays confounded the evaluation of the data. With regard to the test methods, Members agreed the rationale of using 3 dose levels and a transgenic control, but noted that there would be only a minimal reduction in animal usage if it proved necessary to also undertake additional concurrent studies with non-transgenic animals in order to provide adequate results for regulatory assessments of chemicals. Members also commented that the duration of testing required in the Xpa assay (39 weeks) and the duration of observation required in the neonatal mouse tests (1 year) were such that these two assays could not be called "short-term" assays.

Comments on Alternative Cancer Tests

11. The Committee then discussed the results from each of the assays included in the ILSI/HESI programme.

12. With regard to the p53+/- transgenic mouse model, Members confirmed their previous conclusion that there was a rationale for assuming that this model could identify genotoxic carcinogens. All 21 chemicals selected by ILSI/HESI had been tested.5,6 The Committee agreed that there were a number of queries regarding the results of some of the tests undertaken to be resolved before definite conclusions on assay performance could be reached. Members noted that a negative result had been obtained with phenacetin whereas a positive result should have been obtained. Members considered the positive result reported for cyclosporin but noted that there was little difference between the tumourigenicity observed in P53 +/- transgenic mice compared to wild type mice. Inconsistent results had been obtained with diethylstilbestrol and oestradiol whereas positive results should have been obtained. Members commented that the inclusion of hyperplasia as a positive result was not justified and overall diethylstilbestrol had, in their view, given a negative response in this assay. Equivocal responses had been found with chloroform and DEHP whereas negative responses should have been obtained. Members noted that there were inconsistencies between laboratories with regard to the performance of p-cresidine as a positive control in one study (negative result obtained) and the inadequate results obtained with benzene in one study. These data suggested a possible lack of reproducibility of the assay. Members confirmed their previous conclusion that the p53+/- mouse model could identify some genotoxic carcinogens.

13. With regard to the Tg.AC transgenic mouse model, Members confirmed their previous conclusion that there is a mechanistic rationale which could potentially support the use of this model to identify chemical carcinogens and potentially tumour promoters. It was noted that 14 out of the 21 chemicals selected by ILSI/HESI had been tested, and that data for only 6 out of the 13 rodent specific carcinogens had been presented.7,8 The incomplete testing with this model therefore limited the conclusions which could be reached from the ILSI/HESI project. The Tg.AC transgenic mouse model identified positive results for 5 out of the 6 human carcinogens tested (including those acting by genotoxic, immunosuppressant and hormonal mechanisms) when data for dermal and oral tests were considered together. However there were inconsistencies in the current trialsuch that the genotoxic carcinogens cyclophosphamide and mephalan gave equivocal results when tested dermally but positive results when tested by oral administration. Cyclosporin, diethylstilboestrol and oestradiol gave positive results in dermal tests and equivocal (cyclosporin) or negative results in oral tests. A negative result was obtained for phenacetin in both oral and dermal tests. With regard to the rodent specific carcinogens tested, the positive response with topically applied clofibrate and equivocal response with Wy-14, 643 needed further explanation. Taking all of the available data on the Tg.AC transgenic mouse model, Members agreed that further explanation of the results for glycidol (false negative) and resorcinol (false positive) were required before the utility of his model could be further considered. It was noted that the problems with non-responder phenotype reported in earlier studies with the Tg.AC transgenic mouse model had been overcome. However, Members were concerned that the rate of spontaneous tumours was significantly higher in the ILSI/HESI sponsored studies than in previous investigations using Tg.AC mice. Members were also concerned, for animal welfare reasons, at the sensitivity of these mice to audio induced seizures but were reassured to note in practice that such reactions were very rare. Members agreed that the available data on the Tg.AC transgenic mouse model showed that there were problems in consistently identifying human carcinogens which needed to be resolved. This suggested a need for further optimisation of methods, an understanding of the mechanisms underpinning differences between dermal and oral tests with the same chemical and a greater database before the performance of the model could be evaluated.

14. With regard to the Xpa-/- and Xpa-/- p53+/- transgenic mice models, Members observed that selection of the Xpa gene was only of relevance to the identification of bulky genotoxic carcinogens and possibly cross linking agents. Members agreed that there was no mechanistic rationale for producing a transgenic animal model with which was deficient for Xpa and heterozygous for p53 gene other than maximising the predisposition to detection of specific categories of genotoxic carcinogen such as cross linking agents. It was noted that 13 out of the 21 chemicals selected by ILSI/HESI had been tested in the Xpa transgenic mouse model.9,10 Negative results had been obtained with phenacetin and oestradiol in Xpa mice but this was not unexpected given the specificity of the transgenic modification used in this particular assay. The positive results obtained in the Xpa mouse for Wy-14,643 needed further explanation. An inconclusive result had been obtained for clofibrate. Members also noted that there was significant interlaboratory variation in results for the positive control chemical p-cresidine with a negative result reported for one laboratory. Fewer results were available for the Xpa/p53+/- transgenic mouse model, with results for only 10 out of the 21 ILSI/HESI selected chemicals available. Members noted that oestradiol had given a positive result in Xpa/p53+/- transgenic mice in contrast to the negative result with Xpa-/- and agreed that an explanation for the difference in results would be valuable. It was also noted that the peroxisome proliferators Clofibrate and Wy-14,643 had not been tested in Xpa-/- p53+/- which might have given some insight into the unexpected results with these two chemicals reported for Xpa-/- transgenic mouse model. Overall few conclusions could be drawn from such limited data with these two models. The Committee felt that a valid rationale for developing these two particular transgenic animal models for short-term testing of potential carcinogenicity had not been proposed.

15. With regard to the rasH2+/- transgenic mouse model, members reiterated their previous conclusion that there was uncertainty about the relevance of this model, which entailed the integration of multiple copies of the c-Ha-ras gene into the CB6F1mouse in respect of the relevance of the model to the carcinogenic process in humans. Data were available for 20 out of 21 test chemicals selected by ILSI/HESI.11,12 The study with Wy-14,643 was ongoing at the time of publication. Members noted that the immunosuppressant cyclosporin A and the hormonal human carcinogen oestradiol were negative in this model. Members considered that further explanation of the positive results with the peroxisome proliferators clofibrate and DEHP was required. It was noted that the papers supplied by ILSI contained the postulation that overexpression of the ras transgene followed by mutation of the transgene was the probable mechanism of carcinogenicity. Overall the Committee agreed that very little weight could be attached to results from this particular transgenic animal model given the proposed mechanism of carcinogenicity.

16. With regard to the neonatal mouse model, Members recalled the conclusion reached in 1998 that there was no evidence to support the use of either the neonatal rat or mouse bioassays as a part of regulatory testing strategies.13 The new information from the ILSI programme, where 13 out of the 21 selected test chemicals had been tested in neonatal mice, supported this view.14,15 Five out of 6 human carcinogens had been tested and a positive response had only been documented for cyclophosphamide and for oestradiol (in one out of three studies). Members reiterated their animal welfare concerns about the evidence of considerable animal mortality during these experiments. Overall there was no rationale for including the neonatal mouse model in carcinogenicity testing strategies.

17. With regard to the available data from the Syrian Hamster Embryo test16, the COC noted that full consideration of these data would be given by the COM and a separate statement published in due course.

Conclusion

18. The COC agreed an overall conclusion that none of the models used in the ILSI/HESI Alternative Cancer Test programme were suitable as a replacement for the mouse carcinogenicity bioassay (the primary purpose for the development of these models) and that further research should look to identify models with a greater relevance to mechanisms of carcinogenicity in humans. Of the animal models assessed there was evidence that p53+/- transgenic mouse model could identify some genotoxic carcinogens. There was insufficient data to suggest that the animal models under consideration (RasH2, Tg.AC, Xpa, Xpa/P53+/- and p53+/-) provide essentially similar results.

(A separate statement from the COM on the ILSI/HESI evaluation of the Syrian Hamster Embryo test would be published in due course).

April 2002

References

1. ICH, (2000) International Conference on Harmonisation of Technical Requirements for the Registration of Pharmaceuticals for human use (ICH). Harmonised Tripartite guideline, Testing for carcinogenicity of pharmaceuticals, 1997.

2. Blain PG, Battershill JM, Venitt S, Cooper CC and Fielder RJ (1998). Consideration of short-term carcinogenicity tests using transgenic mouse models. In Current Issues in Mutagenesis and Carcinogenesis, No 87, Mutation Research, 403, 259-263.

3. Robinson DE and MacDonald J (2001). Background and Framework for ILSI’s collaborative evaluation program on alternative models for carcinogenicity assessment. Toxicologic Pathology, 29 (supplement), 13-19.

4. Popp JA (2001). Criteria for the Evaluation of Studies in Transgenic Models. Toxicologic Pathology, 29 (supplement), 20-23.

5. French JE, Stoner R and Donehower LA (2001). The Nature of the heterzygous Trp53 Knockout model for the identification mutagenic carcinogens. Toxicologic Pathology, 29 (supplement), 24-29.

6. Storer RD et al (2001). p53+/- Hemizygous Knockout Mouse: Overview of available data. Toxicologic Pathology, 29 (supplement), 30-50.

7. Tennant RW, Stasiewicz S, Easton WC, Mennear JH, and Spalding JW (2001). The Tg.AC (v-Ha-ras) transgenic mouse: Nature of the model. Toxicologic Pathology, 29 (supplement), 51-59.

8. Eastin WC et al (2001). Tg.AC Genetically Altered Mouse: Assay Working Group Overview of Available data. Toxicologic Pathology, 29 (supplement), 60-80.

9. Van Steeg H, de Vries A, van Oostron CThM, Van Benthem, Beems RB and van Kreijl CF (2001). DNA Repair Deficient Xpa and Xpa/p53+/- Knockout-Out Mice: Nature of the models. Toxicologic Pathology, 29 (supplement), 109-117.

10. Coen F et al (2001). Xpa and Xpa/p53+/- Knockout-Out Mice: Overview of Available Data. Toxicologic Pathology, 29 (supplement), 117-127.

11. Tamaoki N (2001). The rasH2 Transgenic mouse. Nature of model and mechanistic study of tumourigenesis. Toxicologic Pathology, 29 (supplement), 81-89.

12. Usui T et al (2001). CB6F1-rasH2 mouse: Overview of Available Data. Toxicologic Pathology, 29 (supplement), 90-108.

13. Department of Health (1998). 1998 Annual eport of the Committees on Toxicity, Mutagenicity and Carcinogenicity of Chemicals in Food, Consumer Products and the Environment. Published Department of health. 21145 1P 1-5K Mar 00 (OAK).

14. Van Tungeln LS, Beland FA, Casciano DA, Kadlubar FK and Fu PP. Neonatal mouse model for short term carcinogenesis testing. Toxicologic Pathology, 29 (supplement), 198-199.

15. McClain R et al (2001). Neonatal Mouse Model: Review of Methods and Results. Toxicologic Pathology, 29 (supplement), 128-137.

16. Maulth RJ, Gibson DP, Bunch RJ and Caster L (2001). The Syrian Hamster Embryo (SHE) cell transformation Assay. Review of the methods and results from ILSI/HESI program on Alternative Cancer Testing. Toxicologic Pathology, 29 (supplement), 138-146.

Top

copyright: © | updated: 29 October 2002