The unique role of rodents in the detection of possible human carcinogens and mutagens

Some of the probable reasons underlying the observation that not all chemicals shown to be genotoxic in vitro are capable of eliciting tumours in rodents or humans are discussed using appropriate examples. It is suggested that a substantial proportion of the resources currently available for conducting rodent carcinogenicity bioassays should be employed in the short-term evaluation in vivo of some of the many hundreds of chemicals recently defined as genotoxic in vitro, rather than in the protracted evaluation of a few chemicals, often of unknown activity in vitro, for carcinogenicity. A decision tree approach to the evaluation of chemicals for human mutagenic/carcinogenic potential is presented which is at variance with the construction and philosophy of many of the current legislative guidelines. The immediate need for the adoption of one of the available short-term in vivo liver assays, and/or the development of a short-term in vivo rodent assay capable of concomitantly monitoring different genetic end-points in a range of organs or tissues is emphasized.     

Application of in vitro cell transformation assays in regulatory toxicology for pharmaceuticals, chemicals, food products and cosmetics

Two year rodent bioassays play a key role in the assessment of carcinogenic potential of chemicals to humans. The seventh amendment to the European Cosmetics Directive will ban in 2013 the marketing of cosmetic and personal care products that contain ingredients that have been tested in animal models. Thus 2-year rodent bioassays will not be available for cosmetics/personal care products. Furthermore, for large testing programs like REACH, in vivo carcinogenicity testing is impractical. Alternative ways to carcinogenicity assessment are urgently required. In terms of standardization and validation, the most advanced in vitro tests for carcinogenicity are the cell transformation assays (CTAs). Although CTAs do not mimic the whole carcinogenesis process in vivo, they represent a valuable support in identifying transforming potential of chemicals. CTAs have been shown to detect genotoxic as well as non-genotoxic carcinogens and are helpful in the determination of thresholds for genotoxic and non-genotoxic carcinogens. The extensive review on CTAs by the OECD (OECD (2007) Environmental Health and Safety Publications, Series on Testing and Assessment, No. 31) and the proven within- and between-laboratories reproducibility of the SHE CTAs justifies broader use of these methods to assess carcinogenic potential of chemicals.     

Chemical and biological characterization of hazardous industrial waste. II. Eukaryotic bioassay of a wood-preserving bottom sediment

The eukaryotic haploid and diploid forms of Aspergillus nidulans were used to detect gene mutations and various types of chromosome damage, respectively, in the acid, base and neutral fractions of a wood-preserving bottom sediment. The corresponding response to prokaryotic mutagenicity assays and major chemical constituents of the 3 waste fractions were described by Donnelly et al. (1987). The haploid methionine system detected genotoxic compounds in all 3 primary waste fractions without metabolic activation. With metabolic activation, the maximum response observed in the gene mutation assay was induced by the base fraction. In the diploid assay without metabolic activation, the acid fraction induced the maximum number of major chromosome abnormalities, while the base fraction induced the maximum number of minor deletions or insertions. These results appear to reflect the different composition of the waste fractions since each fraction induced a different type of genetic damage in the two bioassays employed. Alternately, because exposure in the diploid assay was during a growth stage, the results may reflect a varying response at different points of the cell division cycle. The results obtained using eukaryotic bioassays indicate that the wood preserving waste contains compound(s) capable of inducing point mutations, chromosome damage, recombination, and compound(s) acting as spindle poisons.     

Unscheduled DNA synthesis tests. A report of the U.S. Environmental Protection Agency Gene-Tox Program

The utility of unscheduled DNA synthesis (UDS) testing for screening potentially hazardous chemicals was evaluated using the published papers and technical reports available to the UDS Work Group. A total of 244 documents were reviewed. Based on criteria defined in advance for evaluation of the results, 169 were rejected. From the 75 documents accepted, results were reviewed for 136 chemicals tested using autoradiographic approaches and for 147 chemicals tested using liquid scintillation counting (LSC) procedures; 38 chemicals were tested by both approaches to measure UDS. Since there were no documents available that provided detailed recommendations of UDS screening protocols or criteria for evaluating the results, the UDS Work Group presents suggested protocols and evaluation criteria suitable for measuring and evaluating UDS by autoradiography in primary rat hepatocytes and diploid human fibroblasts and by the LSC approach in diploid human fibroblasts. UDS detection is an appropriate system for inclusion in carcinogenicity and mutagenicity testing programs, because it measures the repair of DNA damage induced by many classes of chemicals over the entire mammalian genome. However, for this system to be utilized effectively, appropriate metabolic activation systems for autoradiographic measurements of UDS in human diploid fibroblasts must be developed, the nature of hepatocyte-to-hepatocyte variability in UDS responses must be determined, and the three suggested protocols must be thoroughly evaluated by using them to test a large number of coded chemicals of known in vivo mutagenicity and carcinogenicity.     

Animal carcinogenicity studies: 2. Obstacles to extrapolation of data to humans

Due to limited human exposure data, risk classification and the consequent regulation of exposure to potential carcinogens has conventionally relied mainly upon animal tests. However, several investigations have revealed animal carcinogenicity data to be lacking in human predictivity. To investigate the reasons for this, we surveyed 160 chemicals possessing animal but not human exposure data within the US Environmental Protection Agency chemicals database, but which had received human carcinogenicity assessments by 1 January 2004. We discovered the use of a wide variety of species, with rodents predominating, and of a wide variety of routes of administration, and that there were effects on a particularly wide variety of organ systems. The likely causes of the poor human predictivity of rodent carcinogenicity bioassays include: 1) the profound discordance of bioassay results between rodent species, strains and genders, and further, between rodents and human beings; 2) the variable, yet substantial, stresses caused by handling and restraint, and the stressful routes of administration common to carcinogenicity bioassays, and their effects on hormonal regulation, immune status and predisposition to carcinogenesis; 3) differences in rates of absorption and transport mechanisms between test routes of administration and other important human routes of exposure; 4) the considerable variability of organ systems in response to carcinogenic insults, both between and within species; and 5) the predisposition of chronic high dose bioassays toward false positive results, due to the overwhelming of physiological defences, and the unnatural elevation of cell division rates during ad libitum feeding studies. Such factors render profoundly difficult any attempts to accurately extrapolate human carcinogenic hazards from animal data.     

Systems and results of tests for chemical induction of mitotic malsegregation and aneuploidy in Aspergillus nidulans

In Aspergillus several types of test systems have been developed for detection of chemicals which induce aneuploidy and/or malsegregation of chromosomes. Results from 23 papers were reviewed in which numerical data for 42 chemicals had been reported. The test systems fall into two groups. One group includes all purely genetic tests that detect euploid mitotic segregants from heterozygous diploids and identify these either as products of malsegregation of chromosomes or as products of crossing-over (13 papers, several reviewed in detail previously; K?fer et al. (1982) and Scott et al. (1982)). The other group includes tests that treat haploid or diploid strains and detect aneuploids as unstable abnormally growing segregants which can be identified as specific disomics or trisomics by their characteristic phenotypes. In addition, such tests characterize abnormal segregants from heterozygous diploids by correlating phenotypes with patterns of genetic segregation in spontaneous euploid sectors. This analysis makes it possible to distinguish between induced primary aneuploidy of whole chromosomes and partial tri- or monosomy resulting from chromosome breakage and secondary spontaneous malsegregation (10 papers). Based on results of both types of tests, it is postulated that chemicals which cause increases of euploid malsegregants, but not of crossovers, normally induce aneuploids as primary products (as shown for 7 of the 14 cases). These include compounds which damage spindles or membranes (especially the well-known haploidizing agents) and generally are effective only when growing cells are exposed. (8 chemicals that may belong in this category could not be classified for certain, because information was insufficient.) On the other hand, chemicals which cause increases of all types of euploid segregants (11 cases), mostly induce drastic mutations and aberrations as primary effects and cause spontaneous malsegregation or crossing-over only as secondary events (as demonstrated for radiation-induced abnormals). In addition, a few chemicals were negative, because they increased only crossing-over or showed no increased segregation at all at concentrations which reduced survival or growth rate (9 cases). Recommendations are made for standardization of methods and protocols. New tester strains and specific procedures are outlined which should be useful for conclusive tests of chemicals that may induce aneuploidy.     

The mouse spot test. Evaluation of its performance in identifying chemical mutagens and carcinogens

The published results on 60 chemicals and X-rays investigated in the mouse spot test were compared with data on the same chemicals tested in the bacterial mutation assay (Ames test) and lifetime rodent bioassays. The performance of the spot test as an in vivo complementary assay to the in vitro bacterial mutagenesis test reveals that of 60 agents, 38 were positive in both systems, 6 were positive only in the spot test, 10 were positive only in the bacterial test and 6 were negative in both assays. The spot test was also considered as a predictor of carcinogenesis; 45 chemicals were carcinogenic of which 35 were detected as positive by the spot test and 3 out of 6 non-carcinogens were correctly identified as negative. If the results are regarded in sequence, i.e. that a positive result in a bacterial mutagenicity test reveals potential that may or may not be realized in vivo, then 48 chemicals were mutagenic in the bacterial mutation assay of which 38 were active in the spot test and 31 were confirmed as carcinogens in bioassays. 12 chemicals were non-mutagenic to bacteria of which 6 gave positive responses in the spot test and 5 were confirmed as carcinogens. These results provide strong evidence that the mouse coat spot test is an effective complementary test to the bacterial mutagenesis assay for the detection of genotoxic chemicals and as a confirmatory test for the identification of carcinogens. The main deficiency at present is the paucity of data from the testing of non-carcinogens. With further development and improvement of the test it is probable that the predictive performance of the assay in identifying carcinogens should improve, since many of the false negative responses may be due to inadequate testing.     

Reliability of the hepatocyte primary culture/DNA repair test in testing of coded carcinogens and noncarcinogens

The hepatocyte primary culture/DNA repair test was evaluated for its reliability using a series of coded samples. Among the 30 chemicals tested, 15 were general reference compounds and 15 were chemicals that had been tested for carcinogenicity in the U.S. National Cancer Institute Bioassay Program. The latter group were from the same lot that had been used for the in vivo testing and had also been tested for mutagenicity in the Ames test. From the group of 15 reference compounds, 5 were positive for DNA repair and all 5 were carcinogens. Of the 10 samples scored as negative, 4 were noncarcinogens and 6 were carcinogens. Among the 6 carcinogens were 3 compounds whose carcinogenicity probably does not involve the production of DNA damage. From the 15 coded chemicals that were tested for carcinogenicity by the NCI in long-term animal studies, 7 were scored as positive. 5 of these were judged carcinogenic in the in vivo bioassays and the other 2, which were also mutagenic in Salmonella, showed some indication of carcinogenicity. Of the 8 compounds that were scored as negative, 5 were noncarcinogenic. Among the 3 carcinogens that were not detected, there was at least one whose carcinogenicity probably does not involve DNA damage. Thus, the results of this study indicate that positive results in the hepatocyte primary culture/DNA repair test are highly specific for carcinogens and that the test is also highly sensitive in the detection of DNA-damaging genotoxic carcinogens.     

Mathematical parameters for quantification of mutational responses in bacteria

This paper introduces a new parameter, derivable from dose-response data for induced mutagenesis in bacteria, that can be used to quantify mutational responses in short-term tests. We called this parameter the mutational response of the bipartite experimental system (agent plus cells). We define it as being jointly proportional to the efficiency of the mutagen and the sensitivity of the test. We show how this quantity can be used to rank order chemical carcinogens on the basis of their mutagenicity and to determine the strength of any quantitative correlation that may exist between mutagenicity in bacteria and carcinogenicity in rodents. We find that this particular measure of mutational response for 10 direct-acting monofunctional alkylating agents correlates remarkably well with the rodent carcinogenicity of these chemicals measured in terms of their reciprocal TD₅₀ values.     

International Commission for Protection against Environmental Mutagens and Carcinogens. ICPEMC working paper No. 5. Genotoxicity tests as predictors of carcinogens: an analysis

Differences between the results of numerical validation studies comparing in vitro and in vivo genotoxicity tests with the rodent cancer bioassay are leading to the perception that short-term tests predict carcinogenicity only with uncertainty. Consideration of factors such as the pharmacokinetic distribution of chemicals, the systems available for metabolic activation and detoxification, the ability of the active metabolite to move from the site of production to the target DNA, and the potential for expression of the induced lesions, strongly suggests that the disparate sensitivity of the different test systems is a major reason why numerical validation is not more successful. Furthermore, genotoxicity tests should be expected to detect only a subset of carcinogens, namely genotoxic carcinogens, rather than those carcinogens that appear to act by non-genetic mechanisms. Instead of relying primarily on short-term in vitro genotoxicity tests to predict carcinogenic activity, these tests should be used in a manner that emphasizes the accurate determination of mutagenicity or clastogenicity. It must then be determined whether the mutagenic activity is further expressed as carcinogenicity in the appropriate studies using test animals. The prospects for quantitative extrapolation of in vitro or in vivo genotoxicity test results to carcinogenicity requires a much more precise understanding of the critical molecular events in both processes.     

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.     

Genetic activity profiles and pattern recognition in test battery selection

Computer-generated genetic activity profiles and pairwise matching procedures may aid in the selection of the most appropriate short-term bioassays to be used in test batteries for the evaluation of the genotoxicity of a given chemical or group of chemicals. Selection of test batteries would be based on a quantitative comparative assessment of the past performance of similar tests applied to other chemicals of the same structural group. The information potentially available for test-battery selection through the use of this pattern-recognition technique is considerably greater than the qualitative results obtained from individual short-term tests. Application of the method should further our understanding of the relationships between chemical properties and genotoxic responses obtained in short-term bioassays and also may contribute to our knowledge of the mechanisms of complex processes such as carcinogenesis. This approach to battery selection should be augmented by careful consideration of established principles of genetic toxicity testing; that is, a chemical should be evaluated in a battery of tests representing the full range of relevant genetic endpoints.     

Mutagenicity examination of several non-steroidal anti-inflammatory drugs in bacterial systems

The mutagenicity of 6 marketed non-steroidal anti-inflammatory drugs (aspirin, flufenamic acid, diclofenac sodium, indomethacin, naproxen and chloroquine) as well as 2 new anti-inflammatory drugs (tenoxicam and carprofen) was examined by using in vitro bacterial systems (repair test and reversion test). None of them was mutagenic on Ames' reversion test. However, they differed in their responses to repair tests. Tenoxicam, carprofen, aspirin, flufenamic acid and naproxen were not mutagenic in either rec- or pol-assays, whereas chloroquine only showed positive results in the pol-assay system. Indomethacin and diclofenac sodium exhibited a slightly stronger inhibitory activity against B. subtilis rec- mutant than against its rec+ counterpart in rec-assay, which was much weaker than AF-2. Thus their mutagenicity was questionable. These results confirm the usefulness of DNA-repair assays as a complementary endpoint to gene mutation in assessing the genotoxic potential of environmental compounds.     

A comparative study on selected chemical carcinogens for chromosome malsegregation, mitotic crossing-over and forward mutation induction in Aspergillus nidulans

10 "false negative" chemical carcinogens, i.e. ineffective in bacterial mutagenicity assays, were thoroughly investigated for their genotoxic activity in the mould Aspergillus nidulans. Forward mutations (methionine suppressors), mitotic crossing-over and chromosome malsegregation were the end-points scored. Positive results were obtained in tests for the induction of mitotic segregation with benzene, ethylenethiourea and urethane, which increased the frequency of abnormal presumptive aneuploid colonies with euploid sectors showing whole chromosome segregation (i.e. non-disjunctional diploids and haploids). The same compounds were ineffective in increasing the frequency of mitotic crossing-over or forward mutations. The other chemical carcinogens investigated, namely acetamide, amitrole, dieldrin, heptachlor epoxide, nitrilotriacetic acid, p,p'-DDT and thiourea were ineffective both as inducers of forward mutations and mitotic segregation.     

Animal carcinogenicity studies: implications for the REACH system

The 2001 European Commission proposal for the Registration, Evaluation and Authorisation of Chemicals (REACH) aims to improve public and environmental health by assessing the toxicity of, and restricting exposure to, potentially toxic chemicals. The greatest benefits are expected to accrue from decreased cancer incidences. Hence the accurate identification of chemical carcinogens must be a top priority for the REACH system. Due to a paucity of human clinical data, the identification of potential human carcinogens has conventionally relied on animal tests. However, our survey of the US Environmental Protection Agency's (EPAs) toxic chemicals database revealed that, for a majority of the chemicals of greatest public health concern (93/160, i.e. 58.1%), the EPA found animal carcinogenicity data to be inadequate to support classifications of probable human carcinogen or non-carcinogen. A wide variety of species were used, with rodents predominating; a wide variety of routes of administration were used; and a particularly wide variety of organ systems were affected. These factors raise serious biological obstacles that render accurate extrapolation to humans profoundly difficult. Furthermore, significantly different International Agency for Research on Cancer assessments of identical chemicals, indicate that the true human predictivity of animal carcinogenicity data is even poorer than is indicated by the EPA figures alone. Consequently, we propose the replacement of animal carcinogenicity bioassays with a tiered combination of non-animal assays, which can be expected to yield a weight-of-evidence characterisation of carcinogenic risk with superior human predictivity. Additional advantages include substantial savings of financial, human and animal resources, and potentially greater insights into mechanisms of carcinogenicity.     

Animal carcinogenicity studies: 3. Alternatives to the bioassay

Conventional animal carcinogenicity tests take around three years to design, conduct and interpret. Consequently, only a tiny fraction of the thousands of industrial chemicals currently in use have been tested for carcinogenicity. Despite the costs of hundreds of millions of dollars and millions of skilled personnel hours, as well as millions of animal lives, several investigations have revealed that animal carcinogenicity data lack human specificity (i.e. the ability to identify human non-carcinogens), which severely limits the human predictivity of the bioassay. This is due to the scientific inadequacies of many carcinogenicity bioassays, and numerous serious biological obstacles, which render profoundly difficult any attempts to accurately extrapolate animal data in order to predict carcinogenic hazards to humans. Proposed modifications to the conventional bioassays have included the elimination of mice as a second species, and the use of genetically-altered or neonatal mice, decreased study durations, initiation-promotion models, the greater incorporation of toxicokinetic and toxicodynamic assessments, structure-activity relationship (computerised) systems, in vitro assays, cDNA microarrays for detecting changes in gene expression, limited human clinical trials, and epidemiological research. The potential advantages of non-animal assays when compared to bioassays include the superior human specificity of the results, substantially reduced time-frames, and greatly reduced demands on financial, personnel and animal resources. Inexplicably, however, the regulatory agencies have been frustratingly slow to adopt alternative protocols. In order to decrease the enormous cost of cancer to society, a substantial redirection of resources away from excessively slow and resource-intensive rodent bioassays, into the further development and implementation of non-animal assays, is both strongly justified and urgently required.     

The alkaline single cell gel electrophoresis assay with mouse multiple organs: results with 30 aromatic amines evaluated by the IARC and U.S. NTP

The genotoxicity of 30 aromatic amines selected from IARC (International Agency for Research on Cancer) groups 1, 2A, 2B and 3 and from the U.S. NTP (National Toxicology Program) carcinogenicity database were evaluated using the alkaline single cell gel electrophoresis (SCG) (Comet) assay in mouse organs. We treated groups of four mice once orally at the maximum tolerated dose (MTD) and sampled stomach, colon, liver, kidney, bladder, lung, brain, and bone marrow 3, 8 and 24 h after treatment. For the 20 aromatic amines that are rodent carcinogens, the assay was positive in at least one organ, suggesting a high predictive ability for the assay. For most of the SCG-positive aromatic amines, the organs exhibiting increased levels of DNA damage were not necessarily the target organs for carcinogenicity. It was rare, in contrast, for the target organs not to show DNA damage. Organ-specific genotoxicity, therefore, is necessary but not sufficient for the prediction of organ-specific carcinogenicity. For the 10 non-carcinogenic aromatic amines (eight were Ames test-positive and two were Ames test-negative), the assay was negative in all organs studied. In the safety evaluation of chemicals, it is important to demonstrate that Ames test-positive agents are not genotoxic in vivo. Chemical carcinogens can be classified as genotoxic (Ames test-positive) and putative non-genotoxic (Ames test-negative) carcinogens. The alkaline SCG assay, which detects DNA lesions, is not suitable for identifying non-genotoxic carcinogens. The present SCG study revealed a high positive response ratio for rodent genotoxic carcinogens and a high negative response ratio for rodent genotoxic non-carcinogens. These results suggest that the alkaline SCG assay can be usefully used to evaluate the in vivo genotoxicity of chemicals in multiple organs, providing for a good assessment of potential carcinogenicity.     

Basic and neutral amino acid transport in Aspergillus nidulans.

Arginine and methionine transport by Aspergillus nidulans mycelium was investigated. A single uptake system is responsible for the transport of arginine, lysine and ornithine. Transport is energy-dependent and specific for these basic amino acids. The Km value for arginine is 1 X 10(-5) M, and Vmax is 2-8 nmol/mg dry wt/min; Km for lysine is 8 X 10(-6) M; Kt for lysine as inhibitor of arginine uptake is 12 muM, and Ki for ornithine is mM. On minimal medium, methionine is transported with a Km of 0-I mM and Vmax about I nmol/mg dry wt/min; transport is inhibited by azide. Neutral amnio acids such as serine, phenylalanine and leucine are probably transported by the same system, as indicated by their inhibition of methionine uptake and the existence of a mutant specifically impaired in their transport. The recessive mutant nap3, unable to transport neutral amino acids, was isolated as resistant to selenomethionine and p-fluorophenylanine. This mutant has unchanged transport of methionine by general and specific sulphur-regulated permeases.     

Testing of chemicals for genetic activity with Saccharomyces cerevisiae: a report of the U.S. Environmental Protection Agency Gene-Tox Program

The yeast Saccharomyces cerevisiae is a unicellular fungus that can be cultured as a stable haploid or a stable diploid . Diploid cultures can be induced to undergo meiosis in a synchronous fashion under well-defined conditions. Consequently, yeasts can be used to study genetic effects both in mitotic and in meiotic cells. Haploid strains have been used to study the induction of point mutations. In addition to point mutation induction, diploid strains have been used for studying mitotic recombination, which is the expression of the cellular repair activities induced by inflicted damage. Chromosomal malsegregation in mitotic and meiotic cells can also be studied in appropriately marked strains. Yeast has a considerable potential for endogenous activation, provided the tests are performed with appropriate cells. Exogenous activation has been achieved with S9 rodent liver in test tubes as well as in the host-mediated assay, where cells are injected into rodents. Yeast cells can be recovered from various organs and tested for induced genetic effects. The most commonly used genetic end point has been mitotic recombination either as mitotic crossing-over or mitotic gene conversion. A number of different strains are used by different authors. This also applies to haploid strains used for monitoring induction of point mutations. Mitotic chromosome malsegregation has been studied mainly with strain D6 and meiotic malsegregation with strain DIS13 . Data were available on tests with 492 chemicals, of which 249 were positive, as reported in 173 articles or reports. The genetic test/carcinogenicity accuracy was 0.74, based on the carcinogen listing established in the Gene-Tox Program. The yeast tests supplement the bacterial tests for detecting agents that act via radical formation, antibacterial drugs, and other chemicals interfering with chromosome segregation and recombination processes.     

The second National Toxicology Program comparative exercise on the prediction of rodent carcinogenicity: definitive results

Chemical carcinogenicity has been the target of a large array of attempts to create alternative predictive models, ranging from short-term biological assays (e.g. mutagenicity tests) to theoretical models. Among the theoretical models, the application of the science of structure-activity relationships (SAR) has earned special prominence. A crucial element is the independent evaluation of the predictive ability. In the past decade, there have been two fundamental comparative exercises on the prediction of chemical carcinogenicity, held under the aegis to the US National Toxicology Program (NTP). In both exercises, the predictions were published before the animal data were known, thus using a most stringent criterion of predictivity. We analyzed the results of the first comparative exercise in a previous paper [Mutat. Res. 387 (1997) 35]; here, we present the complete results of the second exercise, and we analyze and compare the prediction sets. The range of accuracy values was quite large: the systems that performed best in this prediction exercise were in the range 60-65% accuracy. They included various human experts approaches (e.g. Oncologic) and biologically based approaches (e.g. the experimental transformation assay in Syrian hamster embryo (SHE) cells). The main difficulty for the structure-activity relationship-based approaches was the discrimination between real carcinogens, and non-carcinogens containing structural alerts (SA) for genotoxic carcinogenicity. It is shown that the use of quantitative structure-activity relationship models, when possible, can contribute to overcome the above problem. Overall, given the uncertainty linked to the predictions, the predictions for the individual chemicals cannot be taken at face value; however, the general level of knowledge available today (especially for genotoxic carcinogens) allows qualified human experts to operate a very efficient priority setting of large sets of chemicals.