The carcinogenicity prediction and battery selection (CPBS) method: a Bayesian approach

Recently, a large number of relatively inexpensive in vitro short-term tests have been developed to help predict the carcinogenicity of chemicals. The carcinogenicity prediction and battery selection (CPBS) method utilizes the results of such short-term tests to screen for chemicals that are most likely to cause cancer. The method is an integrated approach for analyzing large, often sparsely filled, data bases containing short-term test results, which often have only marginal representation of known non-carcinogens. The CPBS method is developed for the purpose of (i) determining the reliability and predictive capability of individual and batteries of short-term tests, and (ii) developing a strategy for formulating and selecting optimally preferred batteries of short-term tests for screening chemicals for further testing. The term 'optimally preferred' connotes the best acceptable combination of tests in terms of trade-offs among the multiple attributes of each test and resulting battery (e.g., cost, sensitivity, specificity, etc). The CPBS method consists of 5 major tasks: (1) data consolidation, (2) parameter estimation, (3) predictivity calculation, (4) battery selection and (5) risk assessment. Although there is a great need for more research and improvement, the CPBS method at its present stage should add an important method to the maze of the thousands of new chemicals that are introduced into drugs, foods, consumer goods and to the environment every year. This method should also provide an enhanced identification procedure for classifying chemicals more accurately as suspected carcinogens or non-carcinogens.     

Cluster analysis in predicting the carcinogenicity of chemicals using short-term assays

Cluster analysis can be a useful tool for exploratory data analysis to uncover natural groupings in data, and initiate new ideas and hypotheses about such groupings. When applied to short-term assay results, it provides and improves estimates for the sensitivity and specificity of assays, provides indications of association between assays and, in turn, which assays can be substituted for one another in a battery, and allows a data base containing test results on chemicals of unknown carcinogenicity to be linked to a data base for which animal carcinogenicity data are available. Cluster analysis was applied to the Gene-Tox data base (which contains short-term test results on chemicals of both known and unknown carcinogenicity). The results on chemicals of known carcinogenicity were different from those obtained when the entire data base was analyzed. This suggests that the associations (and possibly the sensitivities and specificities) which are based on chemicals of known carcinogenicity may not be representative of the true measures. Cluster analysis applied to the total data base should be useful in improving these estimates. Many of the associations between the assays which were found through the use of cluster analysis could be 'validated' based on previous knowledge of the mechanistic basis of the various tests, but some of the associations were unsuspected. These associations may be a reflection of a non-ideal data base. As additional data becomes available and new clustering techniques for handling non-ideal data bases are developed, results from such analyses could play an increasing role in strengthening prediction schemes which utilize short-term tests results to screen chemicals for carcinogenicity, such as the carcinogenicity and battery selection (CPBS) method (Chankong et al., 1985).     

Low Efficiency of Short-Term Tests in the Assessment of the Potential Mutagenic Hazard of Chemical Compounds to Mammals

A new method of the efficiency assessment of testing mutagenicity chemical pollutants is proposed. The method is based on the selective information criterion and allows one to compare the prognostic significance of results obtained in both individual tests and test batteries. The efficiency of mutagen detection in mammals was estimated in Ames' test, the in vivo test for cytogenetic abnormalities in rodent bone-marrow cells, and the battery combining both these tests. The level of evidence for mutagenicity was determined for chemicals analyzed in these tests. Based on information obtained during the trials, a low efficiency of the analyzed tests and their battery was inferred.     

Low efficiency of short-term tests in the assessment of the potential mutagenic hazard of chemical compounds to mammals

A new method of the efficiency assessment of testing mutagenicity chemical pollutants is proposed. The method is based on the selective information criterion and allows one to compare the prognostic significance of results obtained in both individual tests and test batteries. The efficiency of mutagen detection in mammals was estimated in Ames' test, the in vivo test for cytogenetic abnormalities in rodent bone-marrow cells, and the battery combining both these tests. The level of evidence for mutagenicity was determined for chemicals analyzed in these tests. Based on information obtained during the trials, a low efficiency of the analyzed tests and their battery was inferred.     

Knowledge-based battery design of short-term tests based on dose information

A construction of batteries of short-term tests (STTs) is described which is based on a classification of 73 chemicals in regard to their carcinogenicity. The 73 chemicals were studied within the U.S. National Toxicology Program (Ashby and Tennant, 1988). The batteries are validated using the classification of 35 additional chemicals. They are defined by logically structured combinations of rules. The single rules are defined by the z-scores of the logarithmic values of the limiting doses obtained from the 4 in vitro STTs used in the study by Ashby and Tennant. The limiting dose is defined as the lowest effective dose or the highest ineffective dose (Waters et al., 1987). The batteries are constructed by minimizing the number of disagreements with the classification by Ashby and Tennant. Compared with the results obtained from single STTs, 2 batteries of 3 STTs have higher concordances with the carcinogenicity data, namely 70% for the NTP data and 74-77% for the independent test data. In addition, a theoretical result shows that the proposed battery design, for a large enough learning set of chemicals, leads to results which are replicated with high probability on a large enough validation set. Based on the first results obtained with a limited number of chemicals it is concluded that the knowledge-based battery design is worth further development.     

Principles of formalizing a quantitative evaluation of the genetic danger of chemical compounds for man

Specific characteristics of the mutagenic effect of chemicals, which must be taken into account in developing the test system to assess the potential genetic risk caused by chemical substances, are considered. The organizational principles of the procedures currently available for testing and ranking chemicals by their mutagenic and carcinogenic hazard to humans are discussed. The use of selective information suggested by Wiener and Shannon as an efficiency measure of testing and estimating the potential genetic hazard of chemical substances is substantiated. The feasibility of this approach was demonstrated by testing the efficiency of the battery of two short-term in vitro tests as an example. It was shown that selective information is able to serve as an integral universal criterion of the efficiency of testing, if either one test or the test battery were used.     

Efficiency of the prediction of carcinogenic activities of chemical substances based on scoring somatic mutations in the soybean Glycine max (L.) Merrill

The efficiency of scoring somatic mutations in soybean (Glycine max (L.) Merrill) leaves as a test for carcinogenic activity of chemical substances in rodents has been evaluated. The efficiency of the test used alone or as part of a battery of tests has been estimated. The mutagenic activities of some chemical substances estimated using the soybean test are presented. Selective information on the carcinogenic activities of substances obtained in special carcinogenicity tests has been used as a quantitative measure of the efficiency of the tests with soybean leaves. To estimate the weight of evidence for the presence of this activity in the tested substances, a special function has been used whose values are uniquely related to the complete information, which is the sum of a priori information and the information obtained after testing. In general, the results have shown that the somatic mutation score test using soybean leaves is at least as efficient as the well-known tests that are generally used now, such as the Ames test and the chromosome aberration score test using mammalian cells in vitro. This test may be promising for the formation of efficient short-term test batteries.     

Efficiency of the prediction of carcinogenic activities of chemical substances based on scoring somatic mutations in the soybean <Emphasis Type="Italic">Glycine max</Emphasis> (L.) Merrill

The efficiency of scoring somatic mutations in soybean (Glycine max (L.) Merrill) leaves as a test for carcinogenic activity of chemical substances in rodents has been evaluated. The efficiency of the test used alone or as part of a battery of tests has been estimated. The mutagenic activities of some chemical substances estimated using the soybean test are presented. Selective information on the carcinogenic activities of substances obtained in special carcinogenicity tests has been used as a quantitative measure of the efficiency of the tests with soybean leaves. To estimate the weight of evidence for the presence of this activity in the tested substances, a special function has been used whose values are uniquely related to the complete information, which is the sum of a priori information and the information obtained after testing. In general, the results have shown that the somatic mutation score test using soybean leaves is at least as efficient as the well-known tests that are generally used now, such as the Ames test and the chromosome aberration score test using mammalian cells in vitro. This test may be promising for the formation of efficient short-term test batteries.     

The genetic toxicology of Gene-Tox non-carcinogens

The Gene-Tox Program has identified 61 chemicals that have been tested in chronic rodent carcinogenesis bioassays and found to be inactive. The genetic toxicology data of these 61 non-carcinogens is reviewed and summarized. A large proportion of these chemicals have been tested to a limited extent in genetic toxicity bioassays: 32 in 2 tests or less. Of the remaining 29 chemicals, 28% have been tested in 9 or more tests which encompass a range of genetic endpoints: gene mutation, chromosomal effects, other genetic endpoints, and cell transformation. The genetic toxicity of 12 chemicals with sufficient data is discussed in detail: benzoin, caffeine caprolactam, ethanol, halothane, hycanthone methanesulfonate, malathion, maleic hydrazide, methotrexate, 1-naphthylamine, 4-nitro-o-phenylenediamine, and p-phenylenediamine. A new technique for the evaluation of multiple test data, the "genetic activity profile", has been applied to 6 of these chemicals, allowing the qualitative and quantitative information to be compared collectively. In the evaluation of the genotoxicity effects of these non-carcinogens, a number of discrepancies between the results from genetic toxicity bioassays and chronic rodent bioassays have been uncovered. These discrepancies are discussed in light of current knowledge on the strengths and weaknesses of both genetic toxicity bioassays and chronic rodent bioassays.     

Prediction of a carcinogenic potential of rat hepatocarcinogens using toxicogenomics analysis of short-term in vivo studies

The carcinogenic potential of chemicals is currently evaluated with rodent life-time bioassays, which are time consuming, and expensive with respect to cost, number of animals and amount of compound required. Since the results of these 2-year bioassays are not known until quite late during development of new chemical entities, and since the short-term test battery to test for genotoxicity, a characteristic of genotoxic carcinogens, is hampered by low specificity, the identification of early biomarkers for carcinogenicity would be a big step forward. Using gene expression profiles from the livers of rats treated up to 14 days with genotoxic and non-genotoxic carcinogens we previously identified characteristic gene expression profiles for these two groups of carcinogens. We have now added expression profiles from further hepatocarcinogens and from non-carcinogens the latter serving as control profiles. We used these profiles to extract biomarkers discriminating genotoxic from non-genotoxic carcinogens and to calculate classifiers based on the support vector machine (SVM) algorithm. These classifiers then predicted a set of independent validation compound profiles with up to 88% accuracy, depending on the marker gene set. We would like to present this study as proof of the concept that a classification of carcinogens based on short-term studies may be feasible.     

An evaluation of the L5178Y TK+/− mouse lymophoma forward mutation assay using 42 chemicals

The L5178YTK^+/? mouse lymphoma assay (MLA) has been utilized in several laboratories as a short-term test for chemical-induced forward mutation in cultured mammalian cells. In order to evaluate several technical modifications to the MLA, 42 chemicals representing 9 chemical classes were tested and the results were compared with those published elsewhere as well as with findings in a genetic toxicology test battery currently used in this laboratory. A positive response for the induction of TK^+/? mutants was obtained for 26 chemicals. With the exception of p-aminophenol, all of these compounds were recognized mutagens or carcinogens and were represented of direct-acting and activation-dependent genotoxins. 16 compounds did not induce IK^?/? mutanants and among these were 5 compounds that were considered to be mutagens or carcinogens. A comparison of the results of this study with those published elsewhere revealed a strong agreement among findings for this test irrespective of minor technical variations. It was concluded that th MLA is a useful system for identifying chemical mutagens in mammalian cells and can serve as a valuabel component in a genetic toxicology test battery.     

Application of a cellular test battery in the decision point approach to carcinogen identification

The assessment of the potential carcinogenicity of a chemical requires a systematic approach taking into account various types of data. Important information on the DNA reactivity and other genetic effects of chemicals can be obtained from a battery of cellular tests. A battery is described which includes DNA repair in hepatocytes, mutagenesis in Salmonella typhimurium, mutagenesis, chromosome alterations, and transformation in mammalian cells. The interpretation of findings in this battery for the identification of potential carcinogenicity of chemicals is discussed.     

Cost-Effectiveness Analysis of Chemical Testing for Decision-Support: How to Include Animal Welfare?

Toxicity testing for regulatory purposes raises the question of test selection for a particular endpoint. Given the public's concern for animal welfare, test selection is a multi-objective decision problem that requires balancing information outcome, animal welfare loss, and monetary testing costs. This paper demonstrates the applicability of cost-effectiveness analysis as a decision-support tool for test selection in a regulatory context such as, for example, the new European chemicals legislation (REACH). We distinguish different decision-making perspectives, in particular the regulator's and chemical industry's perspectives, and we discuss how cost-effectiveness analysis can be applied to test selection from both perspectives. Furthermore, we show how animal welfare goals can be included in cost-effectiveness analyses, and we provide a three-dimensional extension to the standard cost-effectiveness analysis if animal welfare loss cannot be valued in monetary terms. To illustrate how cost-effectiveness analysis works in different settings, we apply our model to a simple case of selecting short-term tests for mutagenicity. We demonstrate that including sufficiently high values for animal welfare induces cost-effective replacements of animal testing. Furthermore, we show that the regulator and chemical companies face different tradeoffs when selecting tests. This may lead to different choices of tests or testing systems.     

Implications for genetic toxicology of the chromosomal breakage syndromes

The activation of oncogenes and our knowledge of the chromosome breakage syndromes show that both intragenic mutations and chromosomal aberrations are important in carcinogenesis. Each suggests that an agent could produce genetic changes in a tissue without producing cancer there, if the types of genetic change do not match: chromosomal aberrations may be irrelevant in the mammary epithelium but be very significant in the bone marrow, and vice versa. This has vital implications for genetic toxicology: (1) both gene mutations and chromosomal aberrations should be measured, and (2) carcinogens may be mutagenic in tissues in which they are not carcinogenic. One might therefore expect in vivo assays for mutagenicity to correlate rather well with cancer bioassays; unfortunately, the bioassays themselves seem faulty. If cancer bioassays are valid, they would be reproducible. If bioassays are reproducible, they would be internally consistent. The information supplied by Tennant et al. (1987) for their validation of in vitro assays gives data from both sexes in rats and mice for 70 chemicals. When the data are analyzed site-by-site, positive results were not replicated in the other sex or in the other species much of the time: in half the cases the other sex does not give the same result; in two-thirds of the cases the other species does not give the same result. There are 3 potential explanations for these differing results: (1) genuine sex-specific carcinogens are common, (2) genuine species-specific carcinogens are common, or (3) the bioassay does not replicate well, i.e., is erratic. The third possibility best explains the data. The apparent inability of short-term in vitro tests to discriminate well between carcinogens and non-carcinogens may be more a reflection of the cancer bioassays that were used to determine which chemicals were carcinogenic than any defect in the assays. In this situation in vivo assays can scarcely be expected to do better even if they are better.     

The genetic toxicity of human carcinogens and its implications

23 chemicals and chemical combinations have been designated by the International Agency for Research on Cancer (IARC) as causally associated with cancer in humans. The literature was searched for reports of their activity in the Salmonella mutagenicity assay and for evidence of their ability to induce chromosome aberrations or micronuclei in the bone marrow of mice or rats. In addition, the chemical structures of these carcinogens were assessed for the presence of electrophilic substituents that might be associated with their mutagenicity and carcinogenicity. The purpose of this study was to determine which human carcinogens exhibit genetic toxicity in vitro and in vivo and to what extent they can be detected using these two widely employed short-term tests for genetic toxicity. The results of this study revealed 20 of the 23 carcinogens to be active in one or both short-term tests. Treosulphan, for which short-term test results are not available, is predicted to be active based on its structure. The remaining two agents, asbestos and conjugated estrogens, are not mutagenic to Salmonella; asbestos is not likely to induce cytogenetic effects in the bone marrow and the potential activity of conjugated estrogens in the bone marrow is difficult to anticipate. These findings show that genetic toxicity is characteristic of the majority of IARC Group 1 human carcinogens. If these chemicals are considered representative of human carcinogens, then two short-term tests may serve as an effective primary screen for chemicals that present a carcinogenic hazard to humans.     

Testing strategies in mutagenicity and genetic toxicology: an appraisal of the guidelines of the European Scientific Committee for Cosmetics and Non-Food Products for the evaluation of hair dyes

The European Scientific Committee on Cosmetics and Non-Food Products (SCCNFP) guideline for testing of hair dyes for genotoxic/mutagenic/carcinogenic potential has been reviewed. The battery of six in vitro tests recommended therein differs substantially from the batteries of two or three in vitro tests recommended in other guidelines. Our evaluation of the chemical types used in hair dyes and comparison with other guidelines for testing a wide range of chemical substances, lead to the conclusion that potential genotoxic activity may effectively be determined by the application of a limited number of well-validated test systems that are capable of detecting induced gene mutations and structural and numerical chromosomal changes. We conclude that highly effective screening for genotoxicity of hair dyes can be achieved by the use of three assays, namely the bacterial gene mutation assay, the mammalian cell gene mutation assay (mouse lymphoma tk assay preferred) and the in vitro micronucleus assay. These need to be combined with metabolic activation systems optimised for the individual chemical types. Recent published evidence [D. Kirkland, M. Aardema, L. Henderson, L. Müller, Evaluation of the ability of a battery of three in vitro genotoxicity tests to discriminate rodent carcinogens and non-carcinogens. I. Sensitivity, specificity and relative predictivity, Mutat. Res. 584 (2005) 1-256] suggests that our recommended three tests will detect all known genotoxic carcinogens, and that increasing the number of in vitro assays further would merely reduce specificity (increase false positives). Of course there may be occasions when standard tests need to be modified to take account of special situations such as a specific pathway of biotransformation, but this should be considered as part of routine testing. It is clear that individual dyes and any other novel ingredients should be tested in this three-test battery. However, new products are formed on the scalp by reaction between the chemicals present in hair-dye formulations. Ideally, these should also be tested for genotoxicity, but at present such experiences are very limited. There is also the possibility that one component could mask the genotoxicity of another (e.g. by being more toxic), and so it is not practical at this time to recommend routine testing of complete hair-dye formulations as well. The most sensible approach would be to establish whether any reaction products within the hair-dye formulation penetrate the skin under normal conditions of use and test only those that penetrate at toxicologically relevant levels in the three-test in vitro battery. Recently published data [D. Kirkland, M. Aardema, L. Henderson, L. Müller, Evaluation of the ability of a battery of three in vitro genotoxicity tests to discriminate rodent carcinogens and non-carcinogens. I. Sensitivity, specificity and relative predictivity, Mutat. Res. 584 (2005) 1-256] suggest the three-test battery will produce a significant number of false as well as real positives. Whilst we are aware of the desire to reduce animal experiments, determining the relevance of positive results in any of the three recommended in vitro assays will most likely have to be determined by use of in vivo assays. The bone marrow micronucleus test using routes of administration such as oral or intraperitoneal may be used where the objective is extended hazard identification. If negative results are obtained in this test, then a second in vivo test should be conducted. This could be an in vivo UDS in rat liver or a Comet assay in a relevant tissue. However, for hazard characterisation, tests using topical application with measurement of genotoxicity in the skin would be more appropriate. Such specific site-of-contact in vivo tests would minimise animal toxicity burden and invasiveness, and, especially for hair dyes, be more relevant to human routes of exposure, but there are not sufficient scientific data available to allow recommendations to be made. The generation of such data is encouraged.     

Scientific and cost-effectiveness criteria in selecting batteries of short-term tests

The scientific and cost-effectiveness criteria introduced in this paper can be applied to published datasets and current and proposed batteries of short-term tests. The reports in the current volume will provide a wealth of additional material for such evaluations, but more systematically obtained information will be necessary to assess both the internal and external validity of these tests. Individual tests and batteries of tests should be standardized, employ positive controls, generate results capable of quantitative analyses that may make dichotomous classification as "positive" and "negative" obsolete, be interpreted in light of mechanisms of action, and be cost-effective on a grand scale. For regulatory purposes our long-term goal should be to replace the whole animal lifetime bioassay with an appropriate and cost-effective set of short-term tests.     

Artificial intelligence and Bayesian decision theory in the prediction of chemical carcinogens

Two procedures for predicting the carcinogenicity of chemicals are described. One of these (CASE) is a self-learning artificial intelligence system that automatically recognizes activating and/or deactivating structural subunits of candidate chemicals and uses this to determine the probability that the test chemical is or is not a carcinogen. If the chemical is predicted to be carcinogen, CASE also projects its probable potency.

The second procedure (CPBS) uses Bayesian decision theory to predict the potential carcinogenicity of chemicals based upon the results of batteries of short-term assays. CPBS is useful even if the test results are mixed (i.e. both positive and negative responses are obtained in different genotoxic assays). CPBS can also be used to identify highly predictive as well as cost-effective batteries of assays.

For illustrative purposes the ability of CASE and CPBS to predict the carcinogenicity of a carcinogenic and a non-carcinogenic polycyclic aromatic hydrocarbon is shown. The potential for using the two methods in tandem to increase reliability and decrease cost is presented.     

Deployment of short-term assays for the detection of carcinogens; genetic and molecular considerations

The deployment of short-term assays for the detection of carcinogens inevitably has to be based on the genetic alterations actually involved in carcinogenesis. This paper gives an overview of oncogene activation and other mutagenic events connected with cancer induction. It is emphasized that there are indications of DNA alterations in carcinogenicity, which are not in accordance with "conventional" mutations and mutation frequencies, as measured by short-term assays of point mutations, chromosome aberrations and numerical chromosome changes. This discrepancy between DNA alterations in carcinogenicity and the endpoints of short-term assays in current use include transpositions, insertion mutations, polygene mutations, gene amplifications and DNA methylations. Furthermore, tumourigenicity may imply an induction of a genetic instability, followed by a cascade of genetic alterations. The evaluation of short-term assays for carcinogenesis mostly involves two correlations that is, between mutation and animal cancer data on the one hand and between animal cancer data and human carcinogenicity on the other. It should be stressed that animal bioassays for cancer in general imply tests specifically for the property of chemicals to function as complete carcinogens, which may be a rather poor reflection of the actual situation in human populations. The primary aim of short-term mutagenicity assays is to provide evidence as to whether a compound can be expected to cause mutations in humans, and such evidence has to be considered seriously even against a background of negative cancer data. For the evaluation of data from short-term assays the massive amount of empirical data from different assays should be used and new computer systems in that direction can be expected to provide improved predictions of carcinogenicity.     

Invited contribution: An objective approach to the development of short-term tests predictive of carcinogenicity

The Carcinogenicity Prediction and Battery Selection procedure was developed to address two problems: (1) the identification of highly predictive, yet cost-effective, batteries of short-term tests and (2) the objective prediction of the potential carcinogenicity of chemicals based upon the results of short-term tests even when a mixture of positive and negative results is obtained. In the present report the usefulness of the Carcinogenicity Prediction and Battery Selection procedure is demonstrated using benzo[a]pyrene, benzoin and diethylstilbestrol as examples. In addition, its applicability in the analysis of all the possible outcomes of a battery is illustrated together with an analysis of the worth of additional testing.Abbreviations B[a]P benzo[a]pyrene - CASE Computer-Automated Structure Evaluation - CPBS Carcinogenicity Prediction and Battery Selection - DEHP diethylhexylphthalate - DES diethylstilbestrol - NTA nitrilotriacetate - TCDD 2,3,7,8-tetrachlorodibenzo-p-dioxin