Evaluation of the Toxicity Forecasting Capability of EPA's ToxCast Phase I Data: Can ToxCast In Vitro Assays Predict Carcinogenicity?

Long-term rodent bioassays have played a central role in protecting human health from carcinogens; for ethical and practical reasons their use is decreasing whereas genotoxicity testing has taken a pivotal role. However, this strategy—as presently implemented—is not sensitive enough to detect all genotoxic carcinogens, and cannot detect nongenotoxic carcinogens. Among the alternative approaches under study there is the ToxCast/Tox21 project. Following a previous study from our laboratory, here we present a new, more extensive analysis of ToxCast Phase I results, indicating that at the present state-of-art this approach is not able to predict the carcinogenicity of chemicals. Possible reasons for this mediocre performance are discussed, and opinions on ways to tune up the project in the next phases are presented.     

Investigating human cancer etiology by DNA lesion footprinting and mutagenicity analysis

Many genotoxic carcinogens are known to leave unique signatures on cancer-related genes. The signature of carcinogens is manifested by the induction of characteristic mutations at distinctive nucleotide positions along oncogenes and/or tumor suppressor genes. Often, the nucleotide positions, wherein mutations occur, co-localize with the sites of initial DNA damage induced by the respective carcinogens. Thus, DNA damage-targeted mutation can be a predictor of carcinogenicity of genotoxins. Today, genomic sequencing technologies for investigating human cancer etiology are based on DNA-lesion footprinting in conjunction with mutagenicity analysis of genotoxic carcinogens. In this review article, we discuss the ligation-mediated PCR and terminal transferase-dependent-PCR, two versatile DNA-lesion footprinting techniques. We highlight the in vitro shuttle vector-based mutation systems for investigating site-specific mutagenicity of carcinogens and the in vivo transgenic rodent mutation systems for exploring DNA damaging and mutagenic properties of carcinogens. We present examples of application of each of these methodologies to human cancer etiology, and provide prospective views on investigations using these technologies for carcinogenicity testing.     

CACE对诱变性和非诱变性致癌物分子结构的识别

CACE(Computer Automated Comprehensive Evaluation)计算机自动综合评价系统,是根据有机化学物质分子结构,预测诱变性,致癌性的计算机工程系统.1个未知活性的化学物质分子结构式输入计算机后,即产生各种可能的局部片段,并与已知的片段以及其他有关信息比较,经构效关系分析后,对该物质的诱变性、致癌性及其强度作出判断,指出与活性有关的片段所在部位,以及这些片段在整个分子活性中所起作用的大小。本文报道已知的102种遗传毒性物质和42种非遗传毒性物质致癌性的CACE结构分析结果。获得与致癌性有关的分子片段98。其中56.1％片段仅在遗传毒性致癌物中出现,而在2种致癌物中均有的片段只占21.4％。该结果表明:遗传毒性和非遗传毒性致癌物的分子结构是有差异的,而CACE系统对这种结构差异具有一定的识别能力。     

Short-term tests for defining mutagenic carcinogens.

The results of short-term tests for mutagenicity were first included in the IARC Monographs in the mid-1970s on the basis of the observation that most carcinogens are also mutagens, although not all mutagens are carcinogens. The experimental evidence at that time showed a strong correlation between mutagenicity and carcinogenicity and indicated that the short-term tests were useful for predicting carcinogenicity. Although the correlations have become weaker over the past 20 years, and with them the predictive value of short-term tests, such tests still provide vital information for identifying and understanding mechanisms involved in carcinogenicity. The results of short-term tests compiled in the US Environmental Protection Agency-IARC Genetic Activity Profile database over the past 12 years are summarized and reviewed here in relation to the classification of agents for carcinogenicity within the system used at IARC. The role of the information from short-term tests in making overall classifications of specific compounds in recent Monographs is discussed. The usefulness of data on three genetic end-points, gene mutation, chromosomal aberrations and aneuploidy, and the criteria for mutagenicity and lack of mutagenicity based on a 'defining set' of test results are examined. Recommendations are made for assessing chemicals on the basis of the strength of the evidence from short-term tests, and the implications of this approach for identifying putative mutational mechanisms of carcinogenicity are discussed.     

Screening assays for carcinogenic agents and mixtures: an appraisal based on data in the IARC Monograph series

To determine whether genotoxic and non-genotoxic carcinogens contribute similarly to the cancer burden in humans and which types of short-term test are more relevant for predicting human hazards, an analysis was performed on agents that were evaluated in IARC Monographs Supplements 6 and 7 for their carcinogenic effects in humans and animals and for activity in short-term genotoxicity tests. The prevalence of genotoxicity among four groups of agents, consisting of established human carcinogens (group 1, n = 30), probable human carcinogens (group 2A, n = 37), possible human carcinogens (group 2B, n = 113) and agents with limited evidence of carcinogenicity in animals (a subset of group 3, n = 66) was determined. Each of the groups 1, 2A and 2B contained a high proportion (80-90%) of genotoxic carcinogens, which were also multi-species or multi-tissue carcinogens. The distribution of carcinogenic potency in rodents did not reveal any specific characteristic of the human carcinogens in group 1 that would differentiate them from agents in groups 2A, 2B and many in group 3. Although limited by the data-base available through the Monographs series, this analysis implies that genotoxic carcinogens add more to the human cancer burden than non-genotoxic carcinogens. Thus, the continued use of in vitro/in vivo short-term tests, involving as endpoints DNA chromosomal or mutational damage, to identify genotoxic carcinogens or in the isolation of carcinogenic components in complex mixtures is fully justified. It is concluded that (a) an agent or complex mixture with unknown carcinogenic potential showing sufficient evidence of activity in genotoxicity assays in vitro or in vivo is likely to represent a hazard to humans and (b) an agent or complex mixture showing lack of activity in this spectrum of genotoxicity assays should undergo evaluation for carcinogenicity for rodent bioassay, in view of the present lack of validated short-term tests for non-genotoxic carcinogens.     

Prediction of carcinogenic potential by a toxicogenomic approach using rat hepatoma cells

The long-term rodent bioassay is the standard method to predict the carcinogenic hazard of chemicals for humans. However, this assay is costly, and the results take at least two years to produce. In the present study, we conducted gene expression profiling of cultured cells exposed to carcinogenic chemicals with the aim of providing a basis for rapid and reliable prediction of carcinogenicity using microarray technology. We selected 39 chemicals, including 17 rat hepatocarcinogens and eight compounds demonstrating carcinogenicity in organs other than the liver. The remaining 14 were non-carcinogens. When rat hepatoma cells (MH1C1) were treated with the chemicals for 3 days at a non-toxic dose, analysis of gene expression changes with our in-house microarray allowed a set of genes to be identified differentiating hepatocarcinogens from non-carcinogens, and all carcinogens from non-carcinogens, by statistical methods. Moreover, optimization of the two gene sets for classification with an SVM and LOO-CV resulted in selection of 39 genes. The highest predictivity was achieved with 207 genes for differentiation between non-hepatocarcinogens and non-carcinogens. The overlap between the two selected gene sets encompassed 26 genes. This gene set contained significant genes for prediction of carcinogenicity, with a concordance of 84.6% by LOO-CV SVM. Using nine external samples, correct prediction of carcinogenicity by SVM was 88.9%. These results indicate that short-term bioassay systems for carcinogenicity using gene expression profiling in hepatoma cells have great promise.     

Carcinogenic potential of some pesticides in a medium-term multi-organ bioassay in rats

The carcinogenic potential of 5 pesticides was analyzed using a medium-term multi-organ bioassay for carcinogenicity. Male F344 rats were initially treated with 3 known carcinogens (diethylnitrosamine, N-methyl-N-nitrosourea and N-bis(2-hydroxypropyl)nitrosamine) during a period of 4 weeks to induce neoplastic changes in a variety of organs, and then given one of 5 pesticides in the diet for a further 16 weeks. Neoplastic and pre-neoplastic lesions were found in the thyroid, kidney and urinary bladder with propineb, in the forestomach, kidney and thyroid with captan and folpet. The number of glutathione S-transferase placental-form-positive liver-cell foci was significantly increased in the captan- and phosmet-treated groups. Based on these findings, captan and propineb can be considered as carcinogens and carcinogenicity is suspected for folpet and phosmet. These results are in concordance with reported long-term carcinogenicity for captan, folpet and propineb. Daminozide was considered not to be carcinogenic. Thus, the present assay of 20 weeks' duration is useful for the prediction of potential carcinogens.     

Genetic counseling program in familial breast cancer: analysis of its effectiveness, cost and cost-effectiveness ratio

Women with a family history of breast cancer are at increased risk for developing this neoplasm. Starting surveillance more frequently at a younger age than the general population and the possibility of undergoing genetic testing are options for their medical management. We analyzed the benefits and costs of our clinical program in familial breast cancer (FBC) and carried out a cost-effectiveness analysis of such procedure. The benefits and costs of performing genetic counseling and a screening program in FBC based on 143 high-risk families registered in our database between June 1995 and December 2001 were analyzed. A decision tree was constructed to estimate the survival benefit and cost-effectiveness of the clinical genetic counseling program compared with the strategy of not performing any screening protocol. We estimated that the prevalence of a BRCA mutation in an unaffected relative of our high-risk cohort was 10% and that 53% of the mutations are found in the BRCA1 gene. We assigned a 58.5% lifetime risk of breast cancer for a 30-year-old mutation carrier according to the SEER data. The effectiveness of the screening was obtained from our experience and data for estimating survival were derived from other studies with longer follow-up. We used our local payment data to calculate the costs of the program. A mutation in the BRCA1 or BRCA2 genes was identified in 20% of the probands. Seventy primary breast cancer cases were recorded since the onset of the program. Thirty percent of the tumors were diagnosed through the screening program and 71% of them were lymph node-negative compared to 49% of the tumors diagnosed outside the program (p=0.1). The cost-effectiveness ratio of our FBC genetic counseling and screening program was 4,294 euros per life-year gained. The model was sensitive to the prevalence of mutation carriers, the lifetime risk of breast cancer and the effectiveness of the screening. In our setting and according to our model, this analysis suggests that a program of genetic testing and screening for breast cancer in a high-risk population may be cost-effective. These results need to be confirmed as more effective interventions for cancer prevention and screening are being implemented.     

P450 and Human Cancer

Most of the chemical carcinogens in our environment are activated mainly by a restricted number of cytochrome P450 species, P450 1A1,1A2, 2E1, and 3A. This metabolic activation of procarcinogens is a crucial part of the initial host response to the environmental exposure, since most chemical carcinogens do not show any carcinogenicity by themselves. Inter-individual variability in the metabolic activity may thus be a key host factor to explain the differences in susceptibility to chemical carcinogenesis among individuals. Recent studies on P450s in cancer etiology have provided some valuable insights into this problem.     

P450 and human cancer.

Most of the chemical carcinogens in our environment are activated mainly by a restricted number of cytochrome P450 species, P450 1A1, 1A2, 2E1, and 3A. This metabolic activation of procarcinogens is a crucial part of the initial host response to the environmental exposure, since most chemical carcinogens do not show any carcinogenicity by themselves. Inter-individual variability in the metabolic activity may thus be a key host factor to explain the differences in susceptibility to chemical carcinogenesis among individuals. Recent studies on P450s in cancer etiology have provided some valuable insights into this problem.     

Medium-term bioassays for carcinogens.

Medium-term bioassay systems in rats have been developed for rapid detection of carcinogenic agents and have been introduced for practical use. There are two major systems: N-nitrosodiethylamine and partial hepatectomy in liver and multiorgan models. The first model takes eight weeks; it consists of an initial intraperitoneal injection of N-nitrosodiethylamine, six weeks' administration of test chemical(s) beginning two weeks later, partial hepatectomy at week 3, and analysis of immunohistochemically identified glutathione S-transferase placental form-positive liver-cell foci. Using this model, the potency of carcinogenic agents can be predicted satisfactorily, at least for those of which the liver is a target organ. The multi-organ models include wide-spectrum organ initiation by N-methyl-N-nitrosourea or combined, short treatment with several carcinogens sequentially and subsequent administration of test chemical(s) for 12-20 weeks. Although these systems require a longer time, their accuracy as organotropism-independent testing systems has been validated by whole-body histological examination. These approaches promise precise, rapid detection of carcinogenic agents and should bridge the gap between established methods for screening in vitro and long-term carcinogenicity testing, with their inherent disadvantages. Other medium-term bioassay methods are discussed in addition to our two systems.     

化学品致癌性评分方案的比较研究

致癌性分类和分级是致癌危害性评定的主要步骤,对管理毒理学具有重要的意义。本文利用含47种化学品的致癌性数据库,比较了6种致癌性分类和评分方案。结果发现IARC分类和机理分类与致癌强度无显著的相关。TD50和多因素致癌性评分法具有某些局限性。对于遗传毒致癌物和非致癌物,多因素致癌性评分与遗传毒性联合评分有较好的相关(r=0.7872)。致癌性的综合评分方案应定量评价对人和动物的致癌性,还应考虑致癌作用机理、毒物动力学及生物学标志等研究资料。     

The contribution of molecular epidemiology to the identification of human carcinogens: current status and future perspectives

BackgroundThe use of biological-based markers of exposure, intermediate effect, outcome, and susceptibility has become standard practice in cancer epidemiology, which has contributed to identification of several carcinogenic agents. Nevertheless, with the exception of biological agents, this contribution, in terms of providing sufficiently strong evidence as required by the International Agency for Research on Cancer (IARC) monographs, has been modest.Materials and methodsWe discuss the overall contribution of molecular epidemiology to identification of carcinogens, with focus on IARC monographs.ResultsFor many carcinogens, valid biological markers of exposure and mechanisms of actions are not available. Molecular markers are usually assessed in single biological samples, which may not represent the actual exposure or biological events related to carcinogens. The contribution of molecular epidemiology to identification of carcinogens has mainly been limited to the carcinogens acting through a genotoxic mechanism, i.e. when carcinogens induce DNA damage. A number of factors, including certain hormones and overweight/obesity, may show carcinogenic effects through nongenotoxic pathways, for which mechanisms of carcinogenicity are not well identified and their biomarkers are sparse.ConclusionLongitudinal assessment of biomarkers may provide more informative data in molecular epidemiology studies. For many carcinogens and mechanistic pathways, in particular nongenotoxic carcinogenicity, valid biological markers still need to be identified.     

Evaluation of the carcinogenic risk of environmental chemicals

Many carcinogens have been detected in our environment, and chemicals which have been shown to induce cancers in experimental animals have been largely eliminated. However, there are still many chemicals which have not been studied in rodent carcinogenicity tests. In this paper, we review the evaluation of carcinogenic risk of environmental chemicals to humans. Carcinogenic risk to humans is evaluated primarily in three different phases: carcinogenicity studies in animal experiments, in vitro short-term screening studies, and finally human epidemiological studies. There are 30 chemicals, processes or industries which have been subject to these three phases and determined to be carcinogenic to humans by the International Agency for Research on Cancer (IARC). Carcinogenic risk of chemicals should be evaluated quantitatively by their threshold doses, TD50 and virtually safe dose (VSD). Furthermore, the target organ of the carcinogen should also be taken into account, which has been emphasized by the demonstration of carcinogenicity of BHA in the forestomach of rats. Summational, synergistic, antagonistic and inhibitory effects of chemicals should also be discussed. It is important to establish a guideline for detecting and eliminating environmental carcinogens in order to prevent cancer in humans.     

Carcinogenicity of mutagens: predictive capability of the Salmonella mutagenesis assay for rodent carcinogenicity

A total of 224 chemicals that have been tested in long-term studies for carcinogenicity in rats and mice by the National Cancer Institute and the National Toxicology Program were tested for mutagenicity in Salmonella typhimurium. Correlations between mutagenicity and carcinogenicity were examined. The influences of chemical structure, rodent species and organ responses, and bacterial strain responses on the carcinogenesis/mutagenesis correlations were also examined. Not all carcinogens induced tumors in both rats and mice. A clear mutagenic or equivocal mutagenic response in Salmonella was predictive for 77% of the carcinogens or equivocal carcinogens, although only 54% of the 149 carcinogens or equivocal carcinogens were mutagens, and 58% of the nonmutagens were carcinogens or equivocal carcinogens. The proportion of mutagens and equivocal mutagens that were not carcinogenic or equivocal was 23%. There was no apparent way to distinguish the mutagenic carcinogens from the mutagenic noncarcinogens by the responses of the specific Salmonella strains. The proportions of different chemical classes in the data base strongly affected the correlations; 40% of the chlorinated carcinogens were mutagens, whereas 75% of the amines and 100% of the nitro-containing carcinogens were mutagens. Because 29% of the chemicals (30% of the carcinogens) were chlorinated, the poor correlation of this class was reflected in the overall correlation. It is concluded that the use of the Salmonella mutagenicity assay is warranted for the identification of carcinogens, but not for noncarcinogens. The proportion of carcinogens detected as mutagens is dependent on the specific classes of chemicals tested and on the rodent species used to define the carcinogens.     

Thiophene analogues of the carcinogens benzidine and 4-aminobiphenyl: evaluation in vitro.

A biologically active molecule with one or more aromatic rings often retains its activity when one of these rings is replaced by an isosteric and/or isoelectronic aromatic ring. Consideration has been given to whether this effect can be expected to apply to aromatic organic carcinogens. The literature relevant to this topic has been reviewed and the thiophene analogues of the carcinogens benzidine and 4-aminobiphenyl have been synthesized and evaluated for potential carcinogenicity. The compounds prepared were 5-p-acetamidophenyl-2-thiophenamine hydrochloride (XIII), 5-phenyl-2-thiophenamine hydrochloride (XIV), N-(5-p-acetamido-phenylthiophen-2-yl)acetamide (XV) and N-(5-phenylthiophen-2-yl)-acetamide (XVI) (see Chart for structures). Each compound was evaluated in the Salmonella reverse-mutation assay of Ames and the cell-transformation assay of Styles. The activity profiles observed for these compounds in vitro were consistent with their known chemistry, and indicate potential carcinogenicity. However, their overall chemical and biological behaviour casts doubt upon whether they would be capable of eliciting tumours in vivo. Because it is important to establish the degree of reliance which can be placed upon in vitro predictions of potential carcinogenicity generated for structurally new compounds, one of the thiophene derivatives, N-(5-phenylthiophen-2-yl)acetamide ((XVI), is currently being evaluated for carcinogenicity in mice.     

Structure-activity relationships for carcinogens with different modes of action.

The application of structure-activity concepts to the elucidation of the action of chemical carcinogens may proceed by two approaches: the hypothesis and the knowledge-based approaches. The former, exemplified by the 'structural alerts' devised by Ashby and associates, derives from the recognition of the electrophilic nature of carcinogens that damage DNA. The latter approach does not assume an a priori mechanism of action but derives information from the establishment of relationships between structural features and carcinogenicity. Indeed, the 'structural alerts' of Ashby et al. are recognized by such an approach; however, if structural features are associated with the activity of 'nongenotoxic' carcinogens, they would also be recognized by the knowledge-based approach. Obviously, the recognition of new (nonelectrophilic) structural features associated with carcinogenicity will lead to testable hypotheses.     

The initiator tRNA acceptance assay as a short-term test for carcinogens. 5. Results with 42 cytostatic drugs

The activity of 42 cytostatic drugs used for the treatment of human cancer was tested by the initiator tRNA acceptance assay for carcinogens. Of 17 drugs carcinogenic for rodents, 16 (94.1%) gave a positive response in the assay and six (85.7%) out of seven non-carcinogens showed no activity. The predictive value of the test for cytostatics was 91.7%. Treatment of tRNA with several cytostatics resulted in an inhibition of its acceptance for L-methionine. Cyclophosphamide, dibromdulcitol, 5-deoxy-5-fluorouridine and vincristine also inhibited, in addition to this, the charging of unfractionated tRNA from rat liver with L-alanine, L-lysine, L-phenylalanine and L-valine. Some drugs apparently react with the same target nucleoside which is common for all species of tRNA (probably the terminal adenosine residue that is esterified with amino acids). Such compounds do not yield reliable results in the initiator tRNA acceptance assay since this inhibitory effect interferes with the stimulating effect characteristic for carcinogens. However, results of the present study agree well with those obtained earlier with different classes of compounds (N-nitroso compounds, mycotoxins, etc.) and indicate that this newly developed assay may be a useful alternative also for the testing of carcinogenicity of cytostatic drugs.     

Response of the ke test to NCI/NTP-screened chemicals. III. Complementary value of ke in screening for carcinogens.

The value of using a physico-chemical carcinogen-screening test, the ke test, in conjunction with the Salmonella typhimurium/microsome assay (the Ames test) and/or structural alerts of reactivity (the S/A test), is analyzed on the basis of the response of the three tests to 171 chemicals of known rodent carcinogenicity. The Ames test is widely used to screen chemicals for potential carcinogenicity; however, its relatively low sensitivity (proportion of true positives among carcinogens tested) has prompted a search for complementary tests that increase sensitivity without an unacceptable decrease in specificity (proportion of true negatives among non-carcinogens tested). The S/A test is a structural analysis based on recognition of chemicals groups likely to react with DNA. The S/A test does not complement the Ames test well, because of the high similarity of responses (dependence) between these two tests. The ke test measures the affinity of a test chemical for electrons, and has a sensitivity and specificity comparable to the Ames test. The ke test is shown in this work to complement both the Ames test and the S/A test. Addition of the ke test to either the Ames test or the S/A test results in a substantial decrease in false negatives and an approximately equal increase in false positives, which is a trade-off that would be desirable in all but the least risk averse situations. The S/A and ke battery has a sensitivity of > 0.9, and could be applied to untested chemicals without any biological testing. In view of these observations, it is proposed that the ke test be considered in developing future strategies to optimize the screening of potential carcinogens in the most cost-effective manner.     

DNA binding and its relationship to carcinogenesis by different polycyclic hydrocarbons.

Five different polycyclic hydrocarbons with different degrees of carcinogenicity in vivo were tested for their metabolism to water-soluble products and their binding to DNA, RNA, and protein in normal embryonic hamster and BHK cells. The compounds studied were 7, 12-dimethylbenz(a)anthracene, benzo(a)pyrene, 20-methyl-cholanthrene, dibenz(a,h)anthracene and dibenz(a,c)anthracene. All five compounds were metabolized to water-soluble produces in both types of cells and treatment of cells with aminophylline enhanced this metabolism. After and not before this enhancement of metabolism by aminophylline, there was a relationship between the degree of carcinogenicity and binding to DNA. There was no such relationship with binding to RNA or protein. The results, indicating a relationship between the degree of carcinogenicity and binding to DNA under appropriate conditions of metabolism, support the suggestion that DNA is the target for carcinogenesis by such carcinogens.