NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Research Council (US) Committee on Comparative Toxicity of Naturally Occurring Carcinogens. Carcinogens and Anticarcinogens in the Human Diet: A Comparison of Naturally Occurring and Synthetic Substances. Washington (DC): National Academies Press (US); 1996.

Cover of Carcinogens and Anticarcinogens in the Human Diet

Carcinogens and Anticarcinogens in the Human Diet: A Comparison of Naturally Occurring and Synthetic Substances.

Show details

4Methods for Evaluating Potential Carcinogens and Anticarcinogens

Carcinogenic activity in rodents, following the oral administration of certain dyes, was first demonstrated in the early 1930s. Since then, numerous experimental studies have been conducted to identify carcinogens in the diets of humans. Such studies in the 1930s and 1940s were predominantly experimental and focused on food additives, especially colorants, contaminants, and carcinogens formed during food processing, cooking, and storage. Early experimental studies on the effects of malnutrition on carcinogenicity were also initiated during this period. Relatively few epidemiologic investigations were conducted until the midcentury. Although most investigations concentrated on cancer of the gastrointestinal tract and liver, it soon became clear that cancers at other sites could be induced by ingested chemicals. The oral route became widespread as a convenient method of administering any suspect carcinogen, irrespective of target organ, and a considerable database on chemicals tested for carcinogenicity was developed.

After World War II, results from experimental and epidemiologic studies reinforced the view that dietary patterns were significantly related to geographic variations in cancer incidence. However, in the absence of testable hypotheses and of well-conducted epidemiologic studies, the role of individual dietary components, including potential carcinogens, remained largely unclear for most organ sites, with few exceptions. Nonetheless, while most human studies concentrated on synthetic chemicals or dietary deficiency, the carcinogenic effect of natural carcinogens was not completely ignored. Thus, the senecio alkaloids, cycasin, and aflatoxin were all identified by the early 1960s. By 1970, the possible anticarcinogenic activity of vitamin A was being explored, as were the modifying effects of fruit, fiber, dairy products, and certain vegetables.

In 1969, the International Union Against Cancer (UICC) convened a committee to address issues of cancer testing. The committee held a workshop that focused on the major testing methods and priorities for carcinogenicity testing. In proceedings from the workshop, the committee concluded, ''there is general agreement that both (a) the extent to which man is exposed to a substance, and (b) the degree of suspicion with which the substance is regarded, must be considered. In many specific cases (a) or (b) will be clearly dominant. Both natural and synthetic substances must be considered for testing. There is a tendency to consider first substances of the latter category; however, an increasing number of natural products with carcinogenic activity are being found and substances suspected to be in this category deserve more attention." (UICC, 1970).

At about the same time, the National Research Council's Committee on Food Protection conducted a review of naturally occurring toxicants in foods, including carcinogens (NRC 1973), in response to growing public apprehension about the safety of the food supply. The committee's list comprised the major natural carcinogens and toxicants as we know them today, and emphasized that they should be further studied. Neither the UICC committee nor the National Research Council committee suggested that naturally occurring compounds posed any unique problems for testing, nor did they mention any qualitative differences between naturally occurring and synthetic carcinogens.

Most cancers suspected to be diet-related are likely to have a multifactorial origin. The human diet is a complex mixture of nutrients and chemicals that are notoriously difficult to measure in observational studies and many of which might be plausible confounders of the effect under study. Although a single factor might be examined in animals through dietary manipulation, this is rarely possible in humans unless the suspected agent is identifiable, discrete, and present at high levels, such as a mycotoxin. In the past, traditional epidemiologic methods have been effective in identifying exposures to ingested carcinogens, e.g., aflatoxin and arsenic, in the diet at relatively high levels and in raising plausible hypotheses about individual foods and macro- or micronutrients. Modification of one component in a diet is usually associated with a change in others. An increase in calories from fat, for instance, usually reflects a reduced percentage of calories from other sources. However, in studying the role of dietary factors, the problem is even more complex because micro- and macronutrients might behave differently qualitatively and quantitatively between humans and the animals in which they are often studied. Further, experimental diets often compare extreme dietary variations, possibly at toxicologic or pharmacologic levels, leading to inappropriate conclusions in humans, in whom variations are usually within a more modest range. Accordingly, it is necessary to discuss first those limitations that arise from a lack of sensitivity or specificity inherent in the methods used for detecting trivial or minimal exposures and their effects in humans or animals. We must also discuss the issues involved in study of complex mixtures in the presence of multiple plausible confounders.

When adequate human data are not available, it is often necessary to base opinions about human risk on results from experiments in animal models. Animal studies of suspected carcinogens are assumed, with some reservations, to provide qualitative predictions of human risk, especially where there is evidence of common mechanisms and endpoints. However, susceptibility to chemically induced carcinogenesis can show interspecies variability. This discordance results at least in part from differences, either hereditary or induced, among animal species in the steps involved in chemical carcinogenesis, particularly at the level of procarcinogen bioactivation and detoxification. Enzymes involved in bioactivation and detoxification of procarcinogens have now been identified and characterized in multiple animal species, including humans (Gonzalez and Gelboin 1994). There are many instances of interspecies differences in these enzymes, in terms both of catalytic specificity and of regulation (Wright and Stevens 1992). Hence, a given chemical can take divergent metabolic pathways, resulting in different health outcomes, depending on the species exposed.

Furthermore, susceptibility to carcinogenesis can vary significantly within a species. In humans, much of this variability appears to reflect genetic heterogeneity. For example, there are several inherited variations in xenobiotic metabolizing enzymes and in DNA repair enzymes that have been associated with susceptibility to certain malignancies. Genetic predisposition to cancer can also be influenced by inherited mutations in tumor suppressor genes, as illustrated by the Li-Fraumeni syndrome, in which patients inherit mutations in one allele of the p53 gene, and in hereditary retinoblastoma, which involves the RB gene. Interestingly, inherited mutations in either of these tumor suppressor genes increases the susceptibility of individuals to certain radiation-induced tumors (Frebourg and Friend 1992). Inheritance of specific polymorphic alleles of the ras oncogene (Weston et al. 1991) and of the p53 gene (Weston et al. 1992) have been linked to lung cancer risk, but the significance of this association is not known. Recent identification on chromosome 17 of the BRCA1 gene that is associated with familial breast cancer (Miki et al. 1994) might provide a clue as to which genetic factors influence breast cancer risk. In addition, nongenetic factors such as diet and hormones might substantially influence susceptibility to cancer in both humans and inbred laboratory rodents. For example, differences in susceptibility to chemical carcinogenesis have been demonstrated between well-fed and calorie-deprived rodents of the same species, possibly because of calorie-induced differences in the catalytic activities of xenobiotic metabolizing and DNA repair enzymes. Individuals in different age groups might also differ in their susceptibility to chemical carcinogenesis.

Biologic markers are being used to investigate individual susceptibility to various exogenous chemical agents. Cloning genes involved in the activation or detoxification of various xenobiotics and in the fidelity and efficiency of DNA repair, for example, will provide probes that may be used to identify and monitor interindividual variations. Currently used markers relate mainly to DNA-damaging (genotoxic) agents. However, because individuals might vary in their susceptibility to processes not directly involving DNA damage (nongenotoxic effects), markers specific for these changes are needed for routine use in molecular epidemiology studies.

Despite the differences between humans and animals, epidemiologic and experimental models need to be considered as ways to evaluate the potential carcinogenicity of naturally occurring chemicals. For example, epidemiologic data have been crucial in developing the association between cigarette smoking and lung cancer. In addition, such studies have consistently demonstrated the relationship between the consumption of alcoholic beverages and cancer. Further experimental studies in diverse animal species have indicated that alcohol induces cancer by nongenotoxic mechanisms.

In Chapter 4, the following questions are addressed:

  • What methods are currently being used to identify and evaluate chemicals as potential carcinogens?
  • Should the methods for testing naturally occurring potential carcinogens differ from those used for testing synthetic chemicals?
  • Are existing methods adequate?
  • How should naturally occurring compounds be prioritized for evaluation of carcinogenic potential?

Methods For Evaluating Chemical Carcinogenesis

Studies in Human Populations


Epidemiology, a science based on population measurements, can be described as the study of the distribution and determinants of diseases in human populations and the application of the results to disease prevention or control. Epidemiologic approaches to determining cancer risks from chemical constituents of foods require the assessment of exposure (diet) and outcome (disease). Exposure data can be classified as (1) general diet, such as patterns of consumption of macro- and micronutrients, certain non-nutrient constituents, and caloric intake; and (2) the identification, isolation, and biological activity of individual suspected carcinogens and anticarcinogens in the diet. The complexity of the human diet makes it difficult to assess retrospectively. Dietary intake data are often based on the use of food diaries or recall of recent or past diet. These methods have qualitative and quantitative limitations. Over the past 2 decades, laboratory techniques have been developed that attempt to address some of the problems associated with epidemiologic studies of diet and cancer. Biologic markers of intake, either of certain nutrients or of individual chemicals found in foods, might provide a better assessment than has been possible before now of the role of diet in human cancer. Some biologic markers with potential use in epidemiologic studies have recently been reviewed (Riboli et al. 1987), and their use is discussed in more detail in the section on "Molecular Epidemiology." Generally, epidemiologic research follows one of four study designs:

  • Ecologic Studies. These studies attempt to relate exposures to disease outcomes at a group level. Such studies suffer from several limitations: individual exposure data are not associated with individual outcome; investigators are unable to control for many potential confounders; and measures of exposure are crude. Because of these limitations, the primary value of such studies is in "hypothesis generation" (i.e., suggesting potentially important risk factors for study by methods based on individuals). On the other hand, such studies can often incorporate a broader range of exposures than can studies based on data from individuals. Thus, for weak risk factors, or for risks that occur only at extremes of exposure, this approach might be more useful in identifying or excluding etiologic factors than has traditionally been assumed (Prentice and Sheppard 1990).
    A common type of ecologic study in diet and cancer research has been international correlations of per capita food consumption with corresponding incidence or mortality rates from specific cancers (Armstrong and Doll 1975). Other studies have been carried out within national boundaries by the selection of distinctive subpopulations, such as ethnic groups (e.g., in Hawaii and South Africa), religious groups (e.g., Mormons and Seventh Day Adventists), or certain dietary cultures (e.g., vegetarians or abstainers from alcohol) (Lyon and Sorenson 1978, Kolonel et al. 1981). In such studies, the measure of exposure is often a very crude estimate of what individuals might actually be ingesting. Per capita food intakes, for example, use food production and import/export data to determine average exposures for individuals in the population. They do not account for food wastage or food fed to animals, nor for differences in intake by sex and age.
  • Case-Control Studies. These studies are based on individuals rather than groups, and overcome many of the limitations just cited. In these studies, persons who have the outcome of interest, e.g., breast cancer, are identified, and suitable controls are obtained for comparison. Variables thought to be potential confounders in the relationship can be overcome by matching during control selection or by statistical adjustment at the time of data analysis. Other advantages of this design are that rare diseases (like most cancers) can be studied, results can be obtained rather quickly, and the research is relatively cost-effective. Disadvantages include the fact that exposure data are obtained retrospectively (dietary recall), and that differential misclassification between cases and controls (bias) can occur, despite great care in designing the study and in collecting the data.
    Examples of such studies are (1) a comparison of exposure to aflatoxins in foods relative to hepatitis B virus status between persons with liver cancer and controls (Qian et al. 1994); and (2) a study comparing consumption of salted fish by persons with nasopharyngeal cancer and controls (Ning et al. 1990). Such studies depend on dietary recall methods, primarily on diet histories, in which individuals recall their intake of specific foods at some specified time period in the past. These recall methods are subject to errors in memory. Such errors might be random (nondifferential) or selective (bias). Nondifferential error generally leads to reduced relative risks, so that a true positive finding might be missed. Bias, however, can lead to a false conclusion from the data. Sources of variation in food consumption data are discussed in Chapter 5. Although large sample sizes can help to reduce some of the effects of random misclassification, the effects of bias cannot be dealt with so readily. However, the findings from many case-control studies, such as those on the effects of fruits and vegetables on cancer risk, have been remarkably consistent, attesting to the strength of this approach (Steinmetz and Potter 1991). Nonetheless, there is evidence for biased recall in some case-control studies of breast cancer (Giovannucci et al. 1993) and of colorectal cancer (Wilkens et al. 1992).
    In some instances, biological specimens (usually serum) have been collected from cases and controls, in an effort to obtain more exact information. Unfortunately, effects of the disease itself on serum levels, variability in serum levels over time, and other factors limit the value of this approach. Newer biologic marker approaches that overcome some of these limitations are discussed below.
  • Cohort Studies. These studies are generally preferred over case-control studies, because the potential for bias is less. In this design, healthy subjects are classified on exposures of interest prior to disease occurrence. The incidence of disease over time is then compared between the two groups. Since exposure data (e.g., diet histories) are obtained prospectively, recall bias is reduced, but the potential for substantial misclassification is nearly always present. However, prospectively assembled cohort studies are expensive, because very large samples are required, and the subjects must be followed for many years to accrue sufficient numbers of cases for meaningful statistical analysis. Such studies are not generally useful for very rare cancers. An alternative approach is to use pre-existing data sets. However, although this might be less costly, such data might not be ideal.
    An example of a cohort study in nutritional epidemiology research is a population of more than 100,000 U.S. nurses being monitored for breast, colon, and other cancers relative to antecedent dietary intakes (e.g., fat and red meat) (Willett et al. 1990, Willett 1994). Other diet-related cohorts in the U.S. include a population of 8,000 Japanese-American men in Hawaii (Heilbrun et al. 1984) and a sample of over 40,000 women in Iowa (Folsom et al. 1990). Cohort studies have often included biochemical measures, such as serum nutrient levels, since data collection occurs prior to the onset of disease. However, because of the lengthy period of follow-up, changes in dietary habits (e.g., fat intake) might occur in the participants, complicating the analyses.
    A multicenter, prospective cohort study designed to investigate the relationship of diet, nutritional status, various lifestyles and environmental factors, and the incidence of different forms of cancer is currently being conducted in Europe. The cohort of the European Prospective Investigation Into Cancer and Nutrition (EPIC) study, developed under the auspices of IARC (IARC 1993), will eventually total approximately 350,000 middle-aged men and women. Data on current diet are being collected by means of detailed dietary assessment. A standardized questionnaire is being used to obtain anthroprometric measurements, as well as information on physical activity, tobacco smoking, alcohol consumption, occupation, socio-economic status, reproductive history, contraception, use of hormone replacement therapy, previous illness, and current drug use. Blood samples are being collected that will be analyzed at a later date. Samples from subjects who develop cancer will be compared with appropriate disease-free control subjects.
    The range of analyses will depend on the type of cancer and availability of techniques.
    The EPIC study has several advantages, including the prospective approach, large sample size, and a wide range of dietary exposures. Short-term screening procedures for dietary modulators of cancer risk are also being developed. One problem that needed to be addressed was the necessity of collecting dietary samples from multiple countries in a comparable and standardized manner (Friedenreich et al. 1992).
  • Intervention Studies. Intervention studies (randomized trials) are theoretically the most desirable of the basic epidemiologic approaches to research. Because they resemble experiments, their results are potentially the most convincing. In intervention studies, individuals are randomly allocated to an experimental or a control group. The experimental arm receives the intervention of interest, while the control arm does not. Because of the randomized design, the potential for bias is minimal, and any differences in outcome between the two groups can be attributed with some confidence to the intervention itself.
    Intervention studies that involve dietary manipulation are particularly difficult to perform successfully and to interpret. For example, if the intervention involves decreasing a macronutrient, such as fat, then to maintain weight, protein or carbohydrate must be increased, or energy expenditure decreased. Thus, a change in outcome could be attributable to any of the altered variables, not just to fat. Even an intervention that does not focus on macronutrients could have an effect on total caloric intake. For example, increasing vegetable intake (which adds considerable bulk to the diet) could result in decreased consumption of higher calorie foods, thereby leading to inadvertent weight loss. Even well-designed intervention trials can founder on such obstacles.
    Examples of intervention studies related to dietary exposures include a trial of Image img00010.jpg -carotene supplements to lower risk of skin cancers (Greenberg et al. 1990); a trial of tocopherol and Image img00010.jpg -carotene supplements to reduce the incidence of lung and other cancers among male smokers (Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group 1994); The Carotene and Retinol Efficacy Trial (CARET; Thornquist et al. 1993, Omenn et al. 1994); the Physicians Health Study (PHS; Hennekens et al.); trials of calcium supplements and precursors of colon cancer (Vargas and Alberts 1992); and a trial of low fat intake and breast and colon cancer (the recently begun Women's Health Initiative) (IOM 1993). Under ideal conditions, one would always choose to conduct intervention trials. However, use of trials is limited by several considerations, including the following: (1) excessively large sample size requirements unless very high-risk (and therefore nonrepresentative) populations are selected for the trial; (2) substantial logistical difficulties, such as maintaining compliance to dietary change over extended time periods; (3) the possibility in dietary interventions that the controls might also change their habits on their own initiative, thereby reducing differences between the two groups; (4) very high costs that must be justified; and (5) ethical considerations that often preclude the study (only interventions that are likely to be beneficial and almost certainly not harmful can be tested). Thus, intervention studies can only be justified when substantial supporting evidence from other studies already exists.

Implementation of these four basic designs in epidemiologic research has been expanded in recent years by the incorporation of new discoveries in molecular genetics and advances in molecular biology techniques. This field has been referred to as molecular epidemiology and is discussed in detail below.

Molecular Epidemiology

Research on molecular mechanisms of carcinogenesis will likely provide additional methods for identifying human exposures to potential carcinogens, and mechanistic understanding of value in risk assessment. Conventional approaches in cancer epidemiology have supplied a wealth of information, but, as noted in the previous section, they have several limitations for identifying specific causal factors, particularly for those cancers that result from multifactor interactions. In addition, epidemiologic studies are largely retrospective, and unless very large numbers of individuals are studied, they are quite insensitive to relatively small increases in risk. Molecular epidemiology is an emerging field that combines traditional epidemiologic studies with biochemical, immunologic, and molecular assays of human tissues and biologic fluids. For example, one study in China is measuring DNA or protein-aflatoxin B1 adducts in individuals at risk for liver cancer. Biologic markers are also being used by NCI to establish efficacy in chemoprevention trials. The usefulness of biologic markers in epidemiologic studies will be determined by their sensitivity, specificity, and predictive value. As noted in the NRC report on Biologic Markers in Immunotoxicology (NRC 1992), the definitions of sensitivity and specificity, as related to epidemiologic studies differ from those used in laboratory studies. While laboratory sensitivity refers to the lowest level that can be reliably analyzed, sensitivity in population studies refers to the proportion of cases that the marker correctly identifies. Similarly, laboratory specificity refers to the ability of the technique to exclude identification of other substances, while specificity in population studies refers to the ability of the marker to identify a true negative correctly. Predictive value is determined by identifying an exposed individual in a population. Laboratory procedures are now available that can be used as biologic markers of factors related to the following: (1) genetic and acquired host susceptibility, (2) metabolism and tissue levels of carcinogens, (3) levels of covalent adducts formed between carcinogens and DNA or other macromolecules, and (4) early cellular responses to carcinogen exposure. Some of these biologic markers are briefly discussed below (for detailed reviews of this subject see Perera and Weinstein 1982, Harris 1986, Santella 1988, Griffith et al. 1989, Skipper and Tannenbaum 1990, and Weinstein et al. 1995).

Genetic Markers of Susceptibility

As mentioned previously, individual susceptibility to chemical carcinogens is influenced by variation in genes coding for enzymes that activate or detoxify carcinogens, as well as repair damage to DNA. A genetic predisposition to cancer may also be due to mutations in oncogenes and tumor suppressor genes. Once such genes have been identified, an individual's phenotype or genotype can be determined.

Biologic Markers of Internal Dose

Toxicant exposure is often assessed at the level of external source. This approach has limitations with respect to precision, reliability, and the extent to which it reflects internal dose, that is, the amount of compound found within the body following exposure. Highly sensitive analytic procedures and immunoassays now make it possible to measure the amounts of a chemical carcinogen or its metabolites in cells, tissues, or body fluids (saliva, blood, urine, or feces). These biologic markers of internal dose reflect individual differences in absorption or bioaccumulation of the compound in question and indicate the level of the compound within the body and in specific tissues or compartments. Examples of this type of marker include the following chemicals: cotinine in serum or urine resulting from cigarette smoke exposure; urinary 1-hydroxypyrene resulting from exposure to polycyclic aromatic hydrocarbons; aflatoxin in urine from dietary or endogenous sources; and DDT or PCBs in serum or adipose tissue biopsies from environmental contamination. Another example that is not specific to an individual chemical is the Ames Salmonella typhimurium mutagenesis assay, which can detect the presence of mutagens in urine that might reflect exposure to cigarette smoke or other genotoxic environmental agents.

Biologic Markers of Biologically Effective Dose

Although markers of internal dose are quite valuable, they do not indicate the extent to which a given compound has interacted with critical cellular targets. In contrast, assays of the biologically effective dose measure the amount of a compound that has reacted with cellular macromolecules, usually DNA, or with a protein such as hemoglobin in the blood (Skipper and Tannenbaum 1990). When DNA from a target tissue is not readily available, sometimes surrogate tissues can be used instead (e.g., placenta or peripheral blood cells). The relationship between the types and levels of adducts in surrogate samples to those in target tissues has not been well characterized in humans, but this relationship has been established for certain carcinogens in laboratory animals. Another limitation is that levels of carcinogen-DNA adducts generally reflect recent exposure rather than cumulative exposure over time and do not indicate if critical targets in DNA such as oncogenes or tumor suppressor genes are affected.

Several methods have been developed for detecting and quantitating carcinogen-DNA adducts in extracts of human peripheral blood cells and tissues. These include physical methods such as fluorescence spectroscopy and gas chromatography/mass spectrometry (GC/MS), the 32P-postlabeling procedure, immunoassays employing antisera to specific carcinogen-DNA adducts, and combinations of these methods (Santella 1988, Weinstein et al. 1995). These methods can detect one carcinogen-DNA adduct per about 107 to 109 nucleotides, which is equivalent to between 1 and 100 adducts per cell. The enzyme-linked immunoassay (ELISA) procedure has been the most widely used method.

Early Biological Responses and Gene Mutations

The next category of biologic markers in the multistep sequence of carcinogenesis comprises markers of very early cellular responses to carcinogen-DNA damage, especially responses thought to play a role in carcinogenesis. These effects can be measured in target tissues or more convenient surrogates, such as peripheral white blood cells. These biologic markers include DNA single- or double-strand breaks, mutations in various genes, and various cytogenetic effects, including sister chromatid exchange, micronuclei, and chromosomal aberrations.

Other Types of Biologic Markers

Several nongenotoxic chemicals, including such compounds as TPA, phenobarbital, TCDD, various PCBS, and hormones (including both natural and synthetic estrogens and androgens) can enhance carcinogenesis without forming covalent adducts with cellular DNA or proteins (Diamond 1987, Tomatis et al. 1987, Weinstein et al. 1995). However, there are currently no assays to determine the biologically effective doses of these agents. One approach would be to develop assays for biologic markers that assess occupancy rates of high affinity receptors for specific hormones or TCDD. Because carcinogenesis can involve disturbances in signal transduction and gene expression, the following assays could be incorporated into molecular epidemiology studies in the future: assays to evaluate levels of specific growth factors, growth factor receptors, second messengers (like cAMP or diacylglycerol) protein kinases, specific phosphoproteins, and the expression of genes related to cell proliferation and other nongenotoxic endpoints. Other assays that may prove to be useful include immunocytochemical assays of proliferating cell nuclear antigen (PCNA) associated with DNA replication and repair, or of other proteins associated with specific phases of the cell cycle, e.g., cyclins (Weinstein 1991, Weinstein et al. 1995).

Because many epidemiologic studies on diet, nutrition, and cancer have been limited by errors in dietary recall methods (see previous section), it is essential to identify objective biologic markers of exposure to specific dietary constituents. Assays have been used to measure the levels of various vitamins, minerals, and nutrients in human blood, tissues, and urine (Weinstein et al. 1995). It would be useful, in addition, to develop biologic markers that reflect the effects of various dietary factors in the intact individual and the relevance of these factors to the carcinogenic process. Biologic markers related to oxidative damage may prove to be of considerable importance for this purpose. These markers include: urinary levels of oxidized DNA bases; analyses of DNA samples for strand breaks or oxidized bases (thymine glycol, 8-hydroxyguanine, etc.); blood and tissue levels of malonaldehyde, an oxidized product of lipids; and markers of enzymes that detoxify activated forms of oxygen, such as catalase and superoxide dismutase (Teebor et al. 1988, Cerutti and Trump 1991, Pryor 1993).

Although biologic markers can provide useful information for evaluating human risk, molecular epidemiology has some limitations. There can be difficulties in correlating indicators of exposure, effect, or susceptibility with a disease. For example, an Institute of Medicine committee (1993) concluded that there was considerable uncertainty in the use of current TCDD serum levels as indicators of past dioxin exposures of Vietnam veterans. Discrepancies were noted and were attributed to the half-life of the biologic marker, leakage of sequestered material from adipose tissue, and accuracy of the determinations. The widespread use of current biologic markers of exposure is also generally limited to compounds whose structures have been identified. Such a limitation may create problems when the complex mixture of the human diet is investigated. However, as progress is made in understanding the mechanisms of carcinogenesis, additional, biologic markers that can be used in epidemiology will be developed.

Screening Tests in Model Systems

Frequently there are insufficient human data to evaluate the potential carcinogenicity of a chemical. Consequently, human risk must often be assessed using information from experimental models. A number of systems are currently available, including structure-activity analyses, short-term tests, and animal bioassays.

Structure-Activity Analyses

As more potential human and animal carcinogens have been evaluated, it has become apparent that certain structural features of these compounds are associated with the induction of tumors. This observation has led to the development of methods for performing structure-activity analyses to predict carcinogenicity. One approach tests major structural groupings associated with electrophilic carcinogens (Ashby 1985, Ashby and Paton 1993). Because of its reliance on DNA reactivity, this system has been most successful in identifying genotoxic carcinogens. A second approach evaluates the structures of chemicals known to induce tumors in humans or animals to determine functionalities associated with either the presence or the absence of biologic activity (Rosenkranz and Klopman 1990a,b; Rosenkranz 1992).

Using these analyses to evaluate naturally occurring chemicals as carcinogens or anticarcinogens in the diet requires that their structures be determined. It was predicted in one analysis of 98 naturally occurring compounds found in plants that 25% will be carcinogens when evaluated in rodent bioassays (Rosenkranz and Klopman 1990b), but this prediction remains to be confirmed.

A limitation of the structure-activity analyses approach is that it has not been well developed for nongenotoxic agents or agents that inhibit carcinogenesis.

Short-Term Tests

A variety of short-term assays are currently being used to evaluate the carcinogenic potential of chemicals. For the purposes of this discussion, short-term tests include both in vitro systems and in vivo systems that examine the effects of short-term exposures; the end-point is not the induction of cancer, but effects that are likely to be predictive of carcinogenicity. The initial observation that several carcinogens were also mutagens (McCann et al. 1975) provided the basis for developing many short-term tests, the majority of which evaluate genotoxicity. Commonly used endpoints for genotoxicity assays include gene mutation, chromosomal aberration, DNA damage, and mammalian cell transformation. Cell transformation assays can also detect certain nongenotoxic carcinogens. Another assay that can identify nongenotoxic carcinogens involves detecting inhibition of cell-to-cell communication. Generally, a battery of tests based on several endpoints and different cell types is appropriate, although in practice the exact nature of the battery varies. Considerations in choosing tests include the origin of the cells, i.e., bacterial or mammalian, and their capacity for biotransformation, in vivo versus in vitro exposure, as well as sensitivity and selectivity. Only tests that have been standardized should be used.

Several agencies that use short-term test data to evaluate potential carcinogenicity have provided recommendations on the types of assays that they consider appropriate. The U.S. Food and Drug Administration (FDA) suggests three tests, that of gene mutation in Salmonella typhimurium , gene mutation in mammalian cells (in vitro), and cytogenetic damage in vivo (FDA 1982). Supplemental tests include an in vitro mammalian cell transformation test and unscheduled DNA synthesis in rat hepatocytes (FDA 1982). The U.S. Environmental Protection Agency (EPA), Office of Pesticide Programs, recommends the same three initial tests as the FDA (Dearfield et al. 1991). The test scheme of the EPA Office of Toxic Substances indicates that a positive result in these assays should be followed by further testing to evaluate germ cell effects (Dearfield et al. 1991). International efforts are currently underway to standardize protocols for the performance of short-term assays and criteria for acceptance of data.

Short-term test data are considered, where relevant, by the International Agency for Research on Cancer (IARC) in evaluating the carcinogenic risk of chemicals to humans (IARC 1987). The endpoints considered are all types of DNA damage, mitotic recombination, gene mutation, sister chromatid exchange, micronuclei, chromosomal aberrations, aneuploidy, cell transformation, and the inhibition of intercellular communication. Particular end points can be detected in prokaryotes, in lower eukaryotes, and in animal or human cells in vitro as well as animal cells in vivo.

Traditionally, short-term tests have been conducted on single chemicals. Aflatoxin B1, for example, a naturally occurring carcinogen found in the diet, has been positive in a number of genotoxicity assays. However, current methods may pose a problem when investigators try to evaluate foods, which are complex mixtures of potential carcinogens as well as anticarcinogens. One useful approach is to study mixtures via bioassay-directed fractionation (NRC 1988). This method was been used to isolate heterocyclic aromatic amines such as 2-amino-3-methylimidazo [4,5-f] quinoline (IQ) from cooked foods. Organic extracts of the charred surfaces of fish or meat were assayed for mutagenic activity in Salmonella typhimurium.

The mutagenic agents were then characterized chemically, and sufficient quantities were synthesized for animal bioassays. Based on these studies, IQ was judged by the IARC to be an animal carcinogen and a probable human carcinogen (IARC 1993). Brewed coffee and tea have also been evaluated for genotoxicity using a variety of tests (IARC 1993). These examples indicate that short-term tests for naturally occurring dietary carcinogens in mixtures are technically feasible.

While current short-term assays should continue to be used in assessing carcinogenic potential, new assays, particularly those to identify carcinogens that are not DNA-reactive, need to be developed and validated. Short-term assays can provide useful information but, like all experimental models, their limitations need to be considered in evaluating test results. Although positive results in these assays suggest that a chemical may be a carcinogen, they are not sufficient to label a chemical as a human carcinogen, and further testing is often required to confirm these data.

Rodent Carcinogenicity Assays

The evaluation of the carcinogenic potential of chemicals is commonly conducted in rodent bioassays. Medium- and long-term exposures can be used.

Limited or medium-term bioassays provide an opportunity to evaluate the potential carcinogenicity of chemicals by exposing animals in vivo and examining site-specific changes associated with tumorigenesis. The animals generally used have an increased susceptibility to chemical carcinogens. These tests use preneoplastic lesions or benign tumors as markers of a neoplastic response. Preneoplastic lesions are defined as phenotypically altered cells that are not themselves neoplastic but that indicate an increased likelihood that benign or malignant neoplasms will occur (Bannasch 1986). Four main categories of changes have been identified: 1) enzyme content or activity, 2) accumulation of macromolecules, 3) alterations in cellular organelles, and 4) cell proliferation and nuclear changes (Bannasch 1986). A commonly used system is the rat liver foci assay (Pereira 1982, Ito et al. 1992). Assays for the induction of mouse skin papillomas (Slaga 1986) or mouse lung adenomas (Stoner and Shimkin 1982) are also used. Assays to detect tumor induction in specific target organs such as the mammary gland, urinary bladder, or stomach have also been developed in mice and rats (Ito et al. 1992). Some of these assays are based on the multi-stage theory of carcinogenesis, so that agents can be evaluated as affecting different stages. The advantages of these medium-term assays are that they take less time than the standard 2-year rodent bioassays and that they can provide useful mechanistic data. These assays can be performed using single agents as well as complex mixtures. However, these tests have limited sensitivity and tend to evaluate changes in a single tissue.

The long-term rodent bioassay involves exposing animals to the test compound and then determining tumor incidence. Testing is usually performed in male and female rats and mice for 18-24 months.

The U.S. National Toxicology Program (NTP) has defined the following protocol for rodent bioassays (Office of Technology Assessment Task Force 1988). F344 rats and B6C3F1 mice are used as the test strains. Fifty animals of each species and sex are used in the control and exposure groups. Doses are determined from a 90-day exposure study by identifying the highest concentration that causes minimal toxicity and little or no growth suppression. This is done so that animals do not die from non-neoplastic causes during the course of the study, thus ensuring appropriate numbers of control and exposed animals at the end of the study. This estimated maximum tolerated dose (MTD) is used as the highest exposure level. Two or more lower doses are also tested (of NTP tests with positive results, 94% are not MTD only). Animals are exposed to the test agent, beginning at approximately eight weeks of age for up to 104 weeks. Ideally, the route of administration should mimic human exposure; however, the one most commonly used is oral. At the end of the study, each animal is autopsied, gross and microscopic pathologic examinations are performed, and the incidence of tumors in control and experimental groups is compared. Carcinogenicity of a compound is then determined based on the incidence of malignant and benign tumors (Office of Technology Assessment Task Force 1988).

The purpose of the rodent bioassay is to identify compounds that induce tumors in an animal model. It is a qualitative test that alone is not sufficient for human risk assessment (NTP 1992). Consequently, the results should be used in combination with other types of data, to assess the likelihood that the substance in question poses a risk for cancer in humans. While current policy accepts that positive results in rodent bioassays are likely to be predictive of human risk, it has been suggested by the NTP Board of Scientific Counselors that hypothesis-driven mechanistic research be incorporated into the NTP bioassay to place these results in proper perspective (NTP 1992). To do so is particularly important because there are examples (e.g., induction of bladder tumors in rodents by saccharin and renal tumors in male rats by d-limonene) where species-specific responses in rodents can occur that might not be relevant to humans (see Chapter 3).

There are concerns about the design of the long-term rodent bioassay, one of which is the use of the maximum tolerated dose (MTD). The issue of the MTD has been reviewed by a committee convened by the National Research Council, Committee on Risk Assessment Methodology (NRC 1993). The reader is referred to that report for a detailed discussion of this issue. The majority of that committee recommended that the MTD should continue to be used as one of the test doses, although a minority suggested that the process of dose selection be modified. It should be noted that MTD/high-dose testing represents not just high-dose exposures, but also can introduce entirely different mechanisms of effect (salt crystals and bladder carcinogenesis; particle overload and lung cancers). In any case, if the compound in question is also positive for carcinogenicity when tested at doses lower than the MTD, then such data might have greater validity.

Although studies spanning the last 60 years have shown that tumor incidence could be altered by dietary modulation, including caloric intake (Kritchevsky 1995), only recently has the concern been raised about the current practice of allowing ad libitum feeding in the bioassays. It has been shown that dietary restriction increases survival, decreases the incidence of spontaneous tumors, and may alter susceptibility to chemical carcinogens. Calorie restriction results in a change in the expression of enzymes involved in the biotransformation of xenobiotics that may influence the formation or persistence of toxic products (Manjgaladze et al. 1993). In one instance, caloric restriction resulted in an increase in the formation of benzo(a) pyrene DNA-adducts, while a similar regimen decreased aflatoxin B1 DNA-adducts (Chou et al. 1993). Cells from animals maintained on a restricted number of calories also showed a reduction in c-H-ras oncogene expression compared to animals allowed to feed ad libitum (Hass et al. 1993). Dietary restriction has enhanced apoptosis of preneoplastic cells, as well as decreased cell replication; these results suggest that food restriction may provide protection from carcinogens (Grasl-Kraupp et al. 1994). Evaluation of tumor incidence in male and female B6C3F1 mice in 16 NTP bioassays suggests a correlation between tumor incidence and body weight (Turturro et al. 1993). When four chemicals were evaluated under the typical conditions of an NTP bioassay, as well as with dietary restriction protocols, the latter increased survival and decreased tumor incidence in both control and exposed animals (Kari and Abdo 1995). Different rates of tumorigenesis were noted in target organs, suggesting that the sensitivity of the bioassay may be altered by dietary manipulation. With two chemicals, dietary restriction altered the site of tumorigensis; however, when the ad libitum fed animals were compared to weight-matched controls, tumor sites identified under both protocols were detected. Moderate dietary restriction improves the health of animals, thus potentially improving the carcinogenicity bioassay (Keenan and Soper 1995). However, it has been suggested that using dietary restriction will both increase (Keenan and Soper 1995) and decrease (Kari and Abdo 1995) the sensitivity of the bioassay. These and other concerns about the rodent bioassay need to be addressed.

Comparison Of Methods For Evaluating Natural And Synthetic Carcinogens

No evidence to date indicates any consistent, fundamental differences between the known naturally occurring and synthetic carcinogens in terms of mechanisms of action (Chapter 3). Consequently, the potential carcinogenicity of both naturally occurring and synthetic compounds might be evaluated using the same methods, except for essential nutrients, which cannot be tested with a zero control. Either single agents or mixtures can be tested. To test mixtures is particularly relevant, since human exposure to both natural carcinogens and most synthetic carcinogens generally occurs as a result of exposure to mixtures of those agents with other chemicals; however, to evaluate the toxicity of chemical mixtures is problematic and generally avoided. In a 1988 NRC report, Complex Mixtures reviewed epidemiologic evidence of effects of exposure to chemical mixtures and proposed strategies for testing mixtures. Much of the epidemiologic evidence was derived from exposures to relatively high doses of substances in the workplace. This report concluded that detecting the effects of mixtures at low doses will require better methods for documenting relevant exposures and better ways to avoid misclassification of both exposures and outcomes. This problem is certainly germane to evaluating the potential effects of chemicals that occur in low concentrations in foods. Complex Mixtures concluded that although testing such mixtures in the laboratory presents a formidable scientific problem, ''reasonably standardized techniques that were developed for the testing of single chemicals can usually be adapted to study complex mixtures" (NRC 1988). In addition, when complex mixtures are tested, problems such as interspecies and high-to-low-dose extrapolation are no different from those encountered when testing single agents.

One of the central issues associated with testing mixtures, whether synthetic or naturally occurring, is identifying the causative agents. If a dietary component is suspected to increase or decrease cancer risk, bioassay-directed fractionation might identify the responsible agent. Complete chemical characterization of complex and diverse mixtures is unlikely to be "prudent or possible" (NRC 1988); however, when individual chemicals and their biologic effects are known, such information should be utilized. The activity of a specific chemical component of a food or other mixture, suspected to increase or decrease cancer risk when administered alone, might be altered when exposure occurs to the mixture. Alterations might result from interactions with the other components of the mixture that, for example, might change the component's structure, activity, dose-response relationship, bioavailability, metabolism, or biologic effects.

Criteria For Selecting And Testing


In 1984, the IARC identified general criteria for selecting agents for carcinogenicity evaluation (see Table 4-1). Although they were developed primarily to set priorities for testing synthetic compounds, these criteria are also appropriate, with minor modifications, for testing naturally occurring compounds (Table 4-1). For these compounds, the presence of an agent in the food and the amount consumed should be considered and might influence the ranking of priorities. For example, constituents present in food in low concentrations, particularly if the foods are consumed in small quantities, might rate a lower priority than those that are relatively abundant in foods and are eaten by a large segment of the population. Exceptions would be chemicals known to be highly potent in other assays, for example mutagenicity assays.

Table 4-1. Criteria for Selecting Agents for Evaluating Carcinogenic or Anticarcinogenic Potential.

Table 4-1

Criteria for Selecting Agents for Evaluating Carcinogenic or Anticarcinogenic Potential.

Dietary consumption patterns should also be considered since they are known to differ between children and adults. Geographic, racial, and ethnic patterns of food use also exist. In addition, important information can be gained by comparing the structure of the compound of interest, if known, with other naturally occurring and synthetic chemicals. Higher priority should be given to naturally occurring compounds that fall in the same chemical class as known carcinogens, that contain the same chemical groups, that are likely to form reactive intermediates, or are likely to be persistent. Short-term tests can play a role in prioritizing chemicals for long-term rodent bioassays, and the latter assays can help prioritize chemicals for epidemiologic studies. In turn, epidemiologic and molecular epidemiology studies might highlight dietary constituents that warrant further examination in short-term tests and rodent bioassays, thus providing further verification of their carcinogenic potential.

As noted earlier in this chapter, current methods for evaluating chemicals as carcinogens and anticarcinogens have limitations. Consequently, only agents that substantially meet these criteria should be considered for testing.


Two general approaches have been used to identify a chemical's potential to prevent cancer. The first involves assessing properties that have been associated with cancer prevention, e.g., antioxidant activity, induction of detoxification systems such as glutathione and superoxide dismutase, or the ability to block interaction of reactive species with cellular constituents. Such studies are often used as screening systems that help to identify components for further study. The second approach involves treatment with the potential cancer prevention agent before or after treatment with a carcinogenic agent. This approach is used in short-term, in vitro systems and in long-term, in vivo systems. Examples of this approach include adding of cancer prevention agents to mutagenesis assay systems before or after treatment with carcinogens or mutagens; treating animals with agents before or after treatment with a carcinogen; or measuring the impact of the agent on the metabolism of the carcinogen or on enzyme systems that are induced by the carcinogen. It should be noted that the problems associated with extrapolating results from rodent carcinogenicity studies to humans are also inherent in the experiments designed to assess anticarcinogenicity. The use of a high-dose carcinogen, high-dose treatment with the agent under study, and short-term observation periods all limit the application of these results to humans.

Studying the effect of anticarcinogenic agents on specific stages of cancer development has identified whether the agents modify genotoxic or nongenotoxic processes. For example, studies have evaluated the ability of cancer prevention agents to inhibit nongenotoxic effects such as cell proliferation, by applying these agents before or after treatment with the phorbol ester TPA following exposure to a genotoxic agent. A recent approach to assessing cancer prevention is the use of intermediate markers for tumor formation, such as aberrant crypts and hyperplasia. This approach permits the study of potential cancer prevention agents in humans.

In selecting agents to be evaluated as anticarcinogens, the criteria shown in Table 4-1 for establishing testing priorities can be applied to both synthetic and naturally occurring agents.

Summary And Conclusions

To limit the human risk of cancer, it is necessary to evaluate the carcinogenic potential of chemicals, whether they are synthetic or naturally occurring. Current strategies for identifying and evaluating potential naturally occurring carcinogens and anticarcinogens can be grouped into epidemiologic studies and those using experimental animal and cell models. The methods to assess carcinogenicity have been presented in this chapter and the following conclusions derived.

  • As stated in Chapter 3, there is no reason to assume that the mechanisms involved in the process of carcinogenesis differ between naturally occurring and synthetic carcinogens. Consequently, they can be evaluated by the same methods.
  • Current methods to identify potential human carcinogens, whether naturally occurring or synthetic, have limitations. Existing tests should be modified and coupled with new methods developed that reflect current understanding of the mechanisms of chemical carcinogenesis.
  • The value of traditional epidemiologic approaches to identifying dietary carcinogens would be expanded by incorporating into their research designs new biochemical, immunologic, and molecular assays based on human tissues and biologic fluids.
  • Despite their limitations, experimental models serve as important screening tests to identify potential human carcinogens. However, there are concerns about extrapolating the results from these models to humans, both with respect to carcinogenic risks and to risks at levels of human exposure. With respect to risk extrapolating, data from screening tests should be used in combination with mechanistic and other available information to predict more reliably the potential human carcinogenicity of a given substance. This is true for both synthetic and naturally occurring compounds.


  • Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group. 1994. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. N. Engl. J. Med. 330:1029-1035. [PubMed: 8127329]
  • Armstrong, B. and R. Doll. 1975. Environmental factors and cancer incidence and mortality in different countries, with special reference to dietary practices. Int. J. Cancer 15:617-631. [PubMed: 1140864]
  • Ashby, J. 1985. Fundamental structural alerts to potential carcinogenicity or noncarcinogenicity. Environ. Mutagen. 7:919-921. [PubMed: 4065064]
  • Ashby, J. and D. Paton. 1993. The influence of chemical structure on the extent and sites of carcinogenesis for 522 rodent carcinogens and 55 different human carcinogen exposures. Mutat. Res. 286:3-74. [PubMed: 7678908]
  • Bannasch, P. 1986. Preneoplastic lesions as end points in carcinogenicity testing. I. Hepatic preneoplasia. Carcinogenesis 7:689-695. [PubMed: 2870822]
  • Cerutti, P.A. and B.F. Trump. 1991. Inflammation and oxidative stress in carcinogenesis. Cancer Cells 3:1-7. [PubMed: 2025490]
  • Chou, M.W., J. Kong, K.T. Chung, and R.W. Hart. 1993. Effect of caloric restriction on the metabolic activation of xenobiotics. Mutation Research 295:223-235. [PubMed: 7507559]
  • Dearfield, K.L., A.E. Auletta, M.C. Cimino, and M.M. Moore. 1991. Considerations in the U.S. Environmental Protection Agency's testing approach for mutagenicity. Mutat. Res. 258:259-283. [PubMed: 1719404]
  • Diamond, L. 1987. Tumor promoters and cell transformation. Pp. 731-734 in Mechanisms of Cellular Transformation by Carcinogenic Agents. Grunberger, D., editor; and S. Goff, editor. , eds. Elmsford: Pergamon Press.
  • FDA (U.S. Food and Drug Administration). 1982. Toxicological Principles for the Safety Assessment of Direct Food Additives and Color Additives used in Food. U.S. Food and Drug Administration, Bureau of Foods.
  • Folsom, A.R., S.A. Kaye, R.J. Prineas, J.D. Potter, S.M. Gapstur, and R.B. Wallace. 1990. Increased incidence of carcinoma of the breast associated with abdominal adiposity in postmenopausal women. Am. J. Epidemiol. 131:794-803. [PubMed: 2138863]
  • Frebourg, T. and S. H. Friend. 1992. Cancer risks from germline p53 mutations. J. Clin. Invest. 90:1637-1641. [PMC free article: PMC443218] [PubMed: 1430194]
  • Friedenreich, C.M., N. Slimani, and E. Riboli. 1992. Measurement of past diet: review of previous and proposed methods. Epidemiol. Rev. 14:177-196. [PubMed: 1289112]
  • Giovannucci, E., M. J. Stampfer, G. A. Colditz, J. E. Manson, B. A. Rosner, M. Longnecker, F. E. Speizer, and W. C. Willett. 1993. A comparison of prospective and retrospective assessments of diet in the study of breast cancer. Am. J. Epidemiol. 137:502-511. [PubMed: 8465802]
  • Gonzalez, F.J., and H.V. Gelboin. 1994. Role of human cytochromes P450 in the metabolic activation of chemical carcinogens and toxins. Drug Metab. Rev. 26(1-2):165-183. [PubMed: 8082563]
  • Grasl-Kraupp, B., W. Bursch, B. Ruttkay-Nedecky, A. Wagner, B. Lauer, and R. Shulte-Hermann. 1995. Food restriction eliminates preneoplastic cells through apoptosis and antagonizes carcinogenesis in rat liver. Proc. Natl. Acad. Sci. U.S.A. 91:9995-9999. [PMC free article: PMC44944] [PubMed: 7937932]
  • Greenberg, E. R., J. A. Baron, T.A. Stukel, M. M. Stevens, J. S. Mandel, S. K. Spencer, P. M. Elias, N. Lowe, D. W. Nierenberg, G. Bayrd, J.C. Vance, D. H. Freeman, Jr., W. E. Clendenning, T. Kwan, and the Skin Cancer Prevention Study Group. 1990. A clinical trial of beta carotene to prevent basal-cell and squamous-cell cancers of the skin. N. Engl. J. Med. 323:789-795. [PubMed: 2202901]
  • Griffith, J., R.C. Duncan, and B.S. Hulka. 1989. Biochemical and biological markers: Implications for epidemiologic studies. Arch. Environ. Health 44:375-381. [PubMed: 2610525]
  • Hass, B.S., R.W. Hart, M.H. Lu, and B.D. Lyn-Cook. 1993. Effects of caloric restriction in animals on cellular function, oncogene expression and DNA methylation in vitro. Mutation Research 295:281-289. [PubMed: 7507563]
  • Harris, C.C., editor. , ed. 1986. Biochemical and Molecular Epidemiology of Cancer. New York: Alan R. Liss. 473 p.
  • Heilbrun, L.K., A.M. Nomura, and G.N. Stemmermann. 1984. Dietary cholesterol and lung cancer risk among Japanese men in Hawaii. Am. J. Clin. Nutr. 39:375-379. [PubMed: 6695837]
  • IARC (International Agency for Research on Cancer). 1984. Chemicals and Exposures to Complex Mixtures Recommended for Evaluation in IARC Monographs and Chemicals and Complex Mixtures Recommended for Long-Term Carcinogenicity Testing. IARC Intern. Tech. Rep. No. 84/002. Lyon, France: IARC.
  • IARC (International Agency for Research on Cancer). 1987. IARC monographs on the evaluation of carcinogenic risks to humans. Genetic and related effects: an updating of selected IARC monographs from volumes 1-42. Supplement 6. [PubMed: 3504843]
  • IARC (International Agency for Research on Cancer). 1993. Biennial Report, 1992-1993. For the period 1 July 1991 to 30 June 1993. Lyon, France: IARC. Pages 44-49.
  • IARC (International Agency for Research on Cancer). 1993. IARC monographs on the evaluation of carcinogenic risks to humans. Vol. 56.
  • IOM (Institute of Medicine). 1993. An Assessment of the NIH Women's Health Initiative. Thaul, S., editor; and D. Hoftra, editor. , eds. National Academy of Sciences. Washington, DC: National Academy Press. 142 pp.
  • IOM (Institute of Medicine). 1993. Veterans and Agent Orange: health effects of herbicides used in Vietnam. National Academy of Sciences. Washington, DC: National Academy Press.
  • Ito, N., T. Shirai, and R. Hasegawa. 1992. Medium-term bioassays for carcinogens. Pp. 353-388 in Mechanisms of Carcinogenesis and Risk Identification. Vainio, H., P., editor; N. Magee, editor; , D. B. McGregor, editor; , and A. J. McMichael, editor. , eds. IARC Scientific Publications 116.
  • Kari, F.W., and K.A. Abdo. 1995. The sensitivity of the NTP bioassay for carcinogen hazard evaluation can be modulated by dietary restriction. Pp. 63-78 in Dietary Restriction: Implications for the Design and Interpretation of Toxicity and Carcinogenicity Studies. R.W. Hart, editor; , D.A. Neumann, editor; , and R.T. Robertson, editor. , eds. Washington, D.C.: ILSI Press.
  • Keenan, K.P., and K.A. Soper. 1995. The effects of ad libitum overfeeding and moderate dietary restriction on Sprague-Dawley rat survival, spontaneous carcinogenesis, chronic disease and the toxicologic response to pharmaceuticals. Pp. 99-126 in Dietary Restriction: Implications for the Design and Interpretation of Toxicity and Carcinogenicity Studies. R.W. Hart, editor; , D.A. Neumann, editor; , and R.T. Robertson, editor. , eds. Washington, D.C.: ILSI Press.
  • Kolonel, L.N., J.H. Hankin, J. Lee, S.Y. Chu, A.M.Y. Nomura, and M.W. Hinds. 1981. Nutrient intakes in relation to cancer incidence in Hawaii. Br. J. Cancer 44:332-339. [PMC free article: PMC2010777] [PubMed: 7284230]
  • Kritchevsky, D. 1995. Fat, calories, and cancer. Pp. 155-165 in Dietary REstriction: Implications for the Design and Interpretation of Toxicity and Carcinogenicity Studies. R.W. Hart, editor; , D.A. Neumann, editor; , and R.T. Robertson, editor. , eds. Washington, D.C.: ILSI Press.
  • Lyon, J.L. and A.W. Sorenson. 1978. Colon cancer in a low-risk population. Am. J. Clin. Nutr. 31:S227-S230. [PubMed: 707378]
  • Manjgaladze, M., S. Chen, L.T. Frame, J.E. Seng, P.H. Duffy, R.J. Feurers, R.W. Hart, and J.E.A. Leakey. 1993. Effects of caloric restriction on rodent drug and carcinogen metabolizing enzymes: implications for mutagenesis and cancer. Mutat. Res. 295:201-222. [PubMed: 7507558]
  • McCann, J., E. Choi, E. Yamasaki, and B. N. Ames. 1975. Detection of carcinogens as mutagens in the Salmonella/microsome test: assay of 300 chemicals. Proc. Natl. Acad. Sci. (USA) 72:5135-5139. [PMC free article: PMC388891] [PubMed: 1061098]
  • Miki, Y., J. Swensen, D. Shattuck-Eidens, P.A. Futreal, K. Harshman, S. Tavtigian, Q. Liu, C. Cochran, L.M. Bennett, W. Ding, R. Bell, J. Rosenthal, C. Hussey, T. Tran, M. McClure, C. Frye, T. Hattier, R. Phelps, A. Haugen-Strano, H. Katcher, K. Yakumo, Z. Gholami, D. Shaffer. S. Stone, S. Bayer, Christian Wray, R. Bogden, P. Dayananth, J. Ward, P. Tonin, S. Narod, P.K. Bristow, F.H. Norris, L. Helvering, P. Morrison, P. Rosteck, M. Lai, J.C. Barrett, C. Lewis, S. Neuhausen, L. Cannon-Albright, D. Goldgar, R. Wiseman, A. Kamb, M.H. Skolnick. 1994. A strong candidate for the breast and ovarian cancer susceptibility gene BRCA1. Science 266(5182):66-71.
  • NRC (National Research Council). 1973. Toxicants occurring naturally in foods, 1st edition, 1966; 2nd edition, 1973. Committee on Food Protection. Washington, DC: National Academy Press.
  • NRC (National Research Council). 1988. Complex Mixtures. Washington, DC: National Academy Press.
  • NRC (National Research Council). 1992. Biologic Markers in Immunotoxicology. Washington, DC: National Academy Press.
  • NRC (National Research Council). 1993. Issues in Risk Assessment. Washington, DC: National Academy Press.
  • NTP (National Toxicology Program). 1992. Final Report of the Advisory Review by the NTP Board of Scientific Counselors; Request for Comments. Federal Register 57:312721-31730.
  • Ning, J.-P., M. C. Yu, Q.-S. Wang, and B. E. Henderson. 1990. Consumption of salted fish and other risk factors for nasopharyngeal carcinoma (NPC) in Tianjin, a low-risk region for NPC in the People's Republic of China. J. Natl. Cancer Inst. 82:291-296. [PubMed: 2299678]
  • Office of Technology Assessment Task Force. 1988. Identifying and regulating Carcinogens. Michigan: Lewis Publishers.
  • Pereira, M. A. 1982. Rat liver foci bioassay. J. Amer. Coll. Toxicol. 1:101-117.
  • Perera F.P. and I. B. Weinstein. 1982. Molecular epidemiology and carcinogen-DNA adduct detection: new approaches to studies of human cancer causation. J. Chronic Dis. 35:581-600. [PubMed: 6282919]
  • Prentice, R. L. and L. Sheppard. 1990. Dietary fat and cancer: Consistency of the epidemiologic dats, and disease prevention that may follow from a practical reduction in fat consumption. Cancer Causes and Control 1:81-97. [PubMed: 2102280]
  • Pryor, W. A. 1993. Measurement of oxidative stress status in humans. Cancer Epidemiol. Biomarkers Prev. 2:289-292. [PubMed: 8318882]
  • Qian, G.S., R.K. Ross, M.C. Yu, J.-M. Yuan, Y.-T. Gao, B.E. Henderson, G.N. Wogan, and J.D. Groopman. 1994. A follow-up study of urinary markers of aflatoxin exposure and liver cancer risk in Shanghai, People's Republic of China. Cancer Epidemiol. Biomarkers and Prevention 3:3-10. [PubMed: 8118382]
  • Riboli, E., H. Ronnholm, and R. Saracci. 1987. Biological markers of diet. Cancer Surveys 6:685-718. [PubMed: 2838169]
  • Rosenkranz, H. S. 1992. Structure-activity relationships for carcinogens with different modes of action. Pp. 271-277 in Mechanisms of Carcinogenesis and Risk Identification. Vainio, H., editor; , P. Magee, editor; , D. B. McGregor, editor; , and A. J. McMichael, editor. , eds. IARC Scientific Publication 116.
  • Rosenkranz, H. S. and G. Klopman. 1990. a. Structural basis of carcinogenicity in rodents of genotoxicants and non-genotoxicants. Mutat. Res. 228:105-124. [PubMed: 2300064]
  • Rosenkranz, H. S. and G. Klopman. 1990. b. Natural pesticides present in edible plants are predicted to be carcinogenic. Carcinogenesis 11:349-353. [PubMed: 2302762]
  • Santella, R. M. 1988. Application of new techniques for the detection of carcinogen adducts to human population monitoring. Mutat. Res. 205:271-282. [PubMed: 3285191]
  • Skipper, P. L. and S. R. Tannenbaum. 1990. Protein adducts in the molecular dosimetry of chemical carcinogens. Carcinogenesis 11:507-518. [PubMed: 2182215]
  • Slaga, T. J. 1986. SENCAR mouse skin tumorigenesis model versus other strains and stocks of mice. Environ. Health Persp. 68:27-32. [PMC free article: PMC1474258] [PubMed: 3096709]
  • Steinmetz, K. A. and J. D. Potter. 1991. Vegetables, fruit, and cancer. I. Epidemiology (Review). Cancer Causes Control 2:325-357. [PubMed: 1834240]
  • Stoner, G. D. and M. B. Shimkin. 1982. Strain A mouse lung tumor bioassay. J. Amer. Coll. Toxicol. 1:145-169.
  • Teebor, G. W., R. J. Boorstein, and J. Cadet. 1988. The repairability of oxidative free radical mediated damage to DNA: a review. Int. J. Radiat. Biol. 54:131-150. [PubMed: 2900272]
  • Tomatis, L., A. Aitio, J. Wilbourn, and L. Shuker. 1987. Human carcinogens so far identified. Jpn J. Cancer Res. 80:795-807. [PMC free article: PMC5917851] [PubMed: 2513295]
  • Turturro, A., P.H. Duffy, and R.W. Hart. 1993. Modulation of toxicity by diet and dietary macronutrient restriction. Mutat. Res. 295:151-164. [PubMed: 7507554]
  • Vargas, P. A. and D. S. Alberts. 1992. Primary prevention of colorectal cancer through dietary modification. Cancer (Suppl) 70:1229-1235. [PubMed: 1324781]
  • Weinstein, I. B. 1988. The origins of human cancer: molecular mechanisms of carcinogenesis and their implications for cancer prevention and treatment. Cancer research 48:4135-4143. [PubMed: 3292040]
  • Weinstein, I. B. 1991. Cancer Prevention: recent progress and future opportunities. Cancer Res. (suppl.) 51:5080S-5085S. [PubMed: 1884384]
  • Weinstein, I.B., R.M. Santella, and F. Perera. 1995. Molecular Biology and Epidemiology of Cancer. Pp. 83-110 in Cancer Prevention and Control, P. Greenwald, editor; , B.S. Kramer, editor; , and D.L. Weed, editor. , ed. New York: Marcel Dekker, Inc.
  • Weston, A., L.S. Perrin, K. Forrester, R. N. Hoover, B. F. Trump, C. C. Harris, and N. E. Caporaso. 1992. Allelic frequency of a p53 polymorphism in human lung cancer. Cancer Epidemiol. Biomarkers Prev. 1:481-483. [PubMed: 1302561]
  • Weston, A., P. Vineis, N. E. Caporaso, T. G. Krontiris, J. A. Lovergaw, and H. Sugimura. 1991. Racial variation in the distribution of H-ras-1 alleles. Mol. Carcinog. 4:265-268. [PubMed: 1714739]
  • Wilkens, L. R., J. H. Hankin, C. N. Yoshizawa, L. N. Kolonel, and J. Lee. 1992. Comparison of long-term dietary recall between cancer cases and noncases. Am. J. Epidemiol. 136:825-835. [PubMed: 1442748]
  • Willett, W.C. 1994. Diet and health: What should we eat? Science 264:532-537. [PubMed: 8160011]
  • Willett, W.C., M.J. Stampfer, G.A. Colditz, B.A. Rosner, and F.E. Speizer. 1990. Relation of meat, fat, and fiber intake to the risk of colon cancer in a prospective study among women. N. Engl. J. Med. 323:1664-1672. [PubMed: 2172820]
  • Wright, S.A., and J.C. Stevens. 1992. The human hepatic cytochromes P450 involved in drug metabolism. Crit. Rev. Toxicol. 22(1):1-21. [PubMed: 1616599]
Copyright 1996 by the National Academy of Sciences. All rights reserved.
Bookshelf ID: NBK232624


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (2.5M)

Related information

  • PMC
    PubMed Central citations
  • PubMed
    Links to PubMed

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...