As indicated in the original statement of task to this committee, the purpose of this report is to examine the occurrence, toxicologic data, mechanisms of action, and potential role of natural carcinogens in the causation of cancer, including relative risk comparisons with synthetic carcinogens, and a consideration of anticarcinogens. The committee was also asked to develop a strategy for selecting additional natural substances for toxicological testing.

This subject is immense because by far the major source of exposure to naturally occurring chemicals is the diet, and the concern that dietary factors might play a major role in human cancer causation. Therefore, this committee addressed its charge by focusing its attention on naturally occurring and synthetic dietary chemicals with carcinogenic potential. This report does not consider noncancer effects or broader aspects of environmental exposure, such as air, water (other than drinking water), or occupational hazards; nor are the carcinogenic effects of radiation or physical agents covered in this report. Some of the latter aspects have been covered in previous NRC reports.

The committee was also asked to assess the impact of naturally occurring chemicals on the initiation, promotion, and progression of tumors. In this report, these stages of carcinogenesis were considered from a mechanistic point of view, but most carcinogenicity data on the compounds reviewed do not provide precise information on the specific stage or stages of the multistage process at which these compounds act. Because the terms ''initiator," "promoter," and "progressor" are especially difficult to apply to specific

agents, particularly as they pertain to human carcinogenesis, the committee chose to use the more contemporary classification of agents as genotoxic and nongenotoxic. Although the committee recognized the importance of the legal and regulatory aspects of the subject of naturally occurring dietary carcinogens, it was felt that these aspects were beyond the purview of its task. A strengthening of the scientific base of the subject of naturally occurring carcinogens will provide a better foundation for regulatory policy and actions.

The purpose of this chapter is to review the committee's major conclusions and recommendations. Several specific principles emerged from the deliberations, including 1) that the great majority of individual naturally occurring and synthetic chemicals in the diet are present at levels below a significant biologic effect, so low that they are unlikely to pose an appreciable cancer risk and 2) that the macronutrients and excess total calories present the greatest dietary cancer risk in the United States. Nevertheless, it was apparent that the existing database and methods are insufficient to identify the precise roles of individual naturally occurring dietary chemicals in human cancer causation and prevention. Therefore, this chapter concludes with a set of proposals for future directions of research, including fundamental mechanistic studies and development of more sophisticated methods that can be applied in laboratory assays and in human populations. Advances in these areas of research will be greatly enhanced by taking advantage of powerful new concepts and methods in analytical and structural chemistry, cellular biology, and molecular genetics.

Several broad perspectives emerged from the committee's deliberations. First, the committee concluded that based upon existing exposure data the great majority of individual naturally occurring and synthetic chemicals in the diet appear to be present at levels

below which any significant biologic effect is likely, and so low that they are unlikely to pose an appreciable cancer risk. The NRC report Diet and Health (NRC 1989) concluded that macronutrients and excess calories are likely the greatest contributors to dietary cancer risk in the United States. It is not clear the degree to which, in aggregate, naturally occurring and synthetic chemicals present an appreciable risk. It is apparent that existing concentration and exposure data and current cancer risk assessment methods are insufficient to definitively address the aggregate roles of naturally occurring or synthetic dietary chemicals in human cancer causation and prevention.

The committee's other major conclusions are listed below. They address the complexity and variability of the human diet, cancer risk from the diet, mechanisms and properties of synthetic vs. naturally occurring carcinogens, the role of anticarcinogens, and models for identifying dietary carcinogens and anticarcinogens.

• The human diet is a highly complex and variable mixture of naturally occurring and synthetic chemicals. Of these, the naturally occurring far exceed the synthetic in number and quantity. The naturally occurring chemicals include macronutrients (fat, carbohydrate, and protein), micronutrients (vitamins and trace metals), and non-nutrient constituents. Only a small number of specific carcinogens and anticarcinogens have been identified in the human diet (e.g., aflatoxins have been identified in the human diet, however it seems unlikely that important carcinogens are yet to be identified). In part, this may reflect the limited number of studies performed.
• Human epidemiologic data indicate that diet contributes to an appreciable portion of cancer, but the precise components of diet responsible for increased cancer risk are generally not well understood.

• Current epidemiologic evidence suggests the importance of protective factors in the diet, such as those present in fruits and vegetables.
• Current evidence suggests that the contribution of excess macronutrients and excess calories to cancer causation in the United States outweighs that of individual food microchemicals, natural and synthetic. This is not necessarily the case in other parts of the world.
• Epidemiologic data indicate that alcoholic beverages (specifically ethanol, the macronutrient found in alcoholic beverages) consumed in excess are associated with increased risk for specific types of cancer.
• Given the greater abundance of naturally occurring substances in the diet, the total exposure to naturally occurring carcinogens (in addition to excess calories and fat) exceeds the exposure to synthetic carcinogens. The committee reviewed data, including those generated by the Department of Agriculture and the Department of Health and Human Services through the Nationwide Food Consumption Surveys, the National Health and Nutrition Examination Surveys, and other related databases, regarding dietary exposure. However, data are insufficient to determine whether the dietary cancer risks from naturally occurring substances exceeds that for synthetic substances (e.g., databases do not include concentration data on many of the potential carcinogenic constituents found in foods). Indeed, at the present time quantitative statements cannot be made about cancer risks for humans from specific dietary chemicals, either naturally occurring or synthetic.
• Current regulatory practices have applied far greater stringency to the regulation of synthetic chemicals in the diet than to naturally occurring chemicals. The committee reviewed IARC, NTP, and other data to ascertain the status of carcinogenicity testing of naturally occurring versus synthetic chemicals. Only a very small fraction of naturally occurring chemicals has been tested for carcinogenicity.

• Naturally occurring dietary chemicals known to be potent carcinogens in rodents include agents derived through food preparation, such as certain heterocyclic amines generated during cooking, and the nitrosamines and agents acquired during food storage, such as aflatoxins and some other fungal toxins.
• The human diet also contains anticarcinogens that can reduce cancer risk. For example, the committee evaluated relevant literature on antioxidant micronutrients, including vitamins A, C, E, folic acid, and selenium, and their suggested contributions to cancer prevention. Human diets that have a high content of fruits and vegetables are associated with a reduced risk of cancer, but the specific constituents responsible for this protective effect and their mechanisms of action are not known with certainty. The vitamin and mineral content of fruits and vegetables may be important factors in this relationship. In addition, fruits and vegetables are dietary sources of many non-nutritive constituents, such as isoflavonoids, isothiocyanates and other sulfur-containing compounds, some of which have inhibited the carcinogenic process in experimental animal studies. Foods high in fiber content are associated with a decreased risk of colon cancer in humans, but it is not yet clear that fiber per se is the component responsible for this protective effect.
• Carcinogens and anticarcinogens present in the diet can interact in a variety of ways that are not fully understood. This makes it difficult to predict overall dietary risks based on an assessment of the risks from individual components based on uncertainties associated with rodent to human extrapolation and high-dose to low-dose extrapolation. It is likely that there is also considerable interindividual variation in susceptibility to specific chemicals or mixtures due to either inherited or acquired factors.

• Overall, the basic mechanisms involved in the entire process

• of carcinogenesis—from exposure of the organism to expression of tumors—are qualitatively similar, if not identical, for synthetic and naturally occurring carcinogens. The committee concluded that there is no notable biologic difference(s) between synthetic and naturally occurring carcinogens. To assess relative potency, the committee compiled and analyzed data on over 200 carcinogens—65 of which were naturally occurring. The data set included agents identified by IARC as having sufficient evidence of carcinogenicity in humans or animals, or by the NTP as known or reasonably anticipated to be human carcinogens. Based in part on this limited sample, the committee concluded that there is no clear difference between the potency of naturally occurring and synthetic carcinogens that may be present in the human diet. In general, both types of chemicals have similar mechanisms of action, similar positivity rates in rodent bioassay tests for carcinogenicity, and encompass similar ranges of carcinogenic potencies. Consequently, naturally occurring and synthetic chemicals can be evaluated by the same epidemiologic or experimental methods and procedures.
• Although there are differences between specific groups of synthetic and naturally occurring chemicals with respect to properties such as lipophilicity, degree of conjugation, resistance to metabolism, and persistence in the body and environment, it is unlikely that information on these properties, if available, will enable predictions to be made of the degree of carcinogenicity of a naturally occurring or synthetic chemical in the diet. Both categories of chemicals—naturally occurring and synthetic—are large and diverse. Predictions based on chemical or physical properties are problematic, due to the likely overlap of values between the categories.

• The committee evaluated the current methods for assessing carcinogenicity, and concluded that current strategies for identifying

• and evaluating potential carcinogens and anticarcinogens are essentially the same. They can be grouped into epidemiologic studies, in vivo experimental animal models, and in vitro systems. The committee recognized the value and limitations of these tests, specifically for identifying dietary carcinogens and anticarcinogens.
• In its assessment of traditional epidemiologic approaches to identifying dietary carcinogens and anticarcinogens, the committee concluded that these can be beneficially expanded by incorporating into research designs biochemical, immunologic, and molecular assays that utilize human tissues and biologic fluids. Furthermore, incorporating the identification of biologic markers into these approaches may provide early indicators of human carcinogenicity—long before the development of tumors.
• The committee analyzed the applicability of rodent bioassays—specifically the long-term bioassays conducted by the National Toxicology Program—for identifying dietary carcinogens and anticarcinogens. The committee concluded that, despite their limitations, rodent models (involving high-dose exposures) have served as useful screening tests for identifying chemicals as potential human carcinogens and anticarcinogens. Concerns about the use of data generated from these models for predicting the potential carcinogenicity and anticarcinogenicity of chemicals in our food are due to the fact that they do not mimic human exposure conditions, i.e., we are exposed to an enormous complex of chemicals, many at exceedingly low quantities, in our diet.

The committee's recommendations, derived from the conclusions just presented, are listed below. The committee was charged to examine the risk from naturally occurring versus synthetic components of the diet on human cancer. Numerous and extensive gaps in the current knowledge base became apparent. These gaps are so

large—and resources are so limited—that careful prioritization of further research efforts is essential. The following recommendations emphasize the need for expanded epidemiologic studies, more human exposure data, improved and enhanced testing methods, more detailed data on dietary components, and further mechanistic studies, if these gaps are to be filled. Research might prove feeble, however, when the complexity and variability of diets and food composition, as well as human behavior, are considered.

• Improved methods are needed to enable the incorporation of relevant cellular and molecular markers of exposure, susceptibility, and preneoplastic effects (DNA damage, etc.) into epidemiologic studies.
  
  

  While existing markers are useful, additional molecular markers of exposure and susceptibility need to be developed, and their relevance and predictivity to the carcinogenic process evaluated. These markers should then be incorporated into traditional epidemiologic studies. In particular, methods are needed to identify high- and low-risk populations. Biologic markers for both genotoxic and nongenotoxic agents need to be developed and validated.

• Additional data on the concentrations and human exposures to naturally occurring and synthetic chemicals in foods are needed.
  
  

  In order to determine the exposures to specific dietary chemicals, it is necessary to know the concentration of a specific chemical in individual food commodities as well as the consumption of those food commodities. At present, the concentrations are known for relatively few chemicals. In addition, more information is needed on the factors that modify these concentrations.

  Current methods for assessing food consumption based on recall or food diaries have limitations; they may entail a substantial degree

• of error and lack of reproducibility. Furthermore, the sample sizes of existing food consumption surveys are limited, particularly when subpopulations such as infants and children or the elderly are considered. In order to minimize the resources needed to acquire these data, consideration should be given to building on other large population-based studies, such as the Women's Health Initiative Study currently supported by the National Institutes of Health.

• Improved bioassay screening methods are needed to test for carcinogens and anticarcinogens in our diet.
  
  

  The rodent bioassay currently used in screening chemicals for potential carcinogenicity or anticarcinogenicity has major problems and uncertainties, especially in providing quantitative estimates of dietary cancer risk to humans or the magnitude of protection by anticarcinogens. These uncertainties relate to the variability of the composition and caloric content of the human diet and the bioassay's inability to mimic this range of variability. In addition, human exposure levels to individual naturally occurring or synthetic chemicals are far lower than experimental test conditions. (The committee recognized that of the NTP bioassays netting positive results, only 6% were from levels exclusively at the maximum tolerated dose.) Uncertainties also result from variation in responses among species. The factors causing these and other uncertainties should be further evaluated and minimized wherever possible. New methods are needed for assessing complex mixtures such as those present in food. Because some chemicals may produce or prevent cancer in animals by mechanisms not relevant to humans, or do so only at high doses, information on the mechanisms of action of chemical carcinogens and anticarcinogens is crucial to improving the science of human risk assessment.

• Further testing of naturally occurring chemicals present in the food supply for carcinogenic and anticarcinogenic potential should be conducted on a prioritized basis.
  
  

  At present, only a limited number of naturally occurring substances present in the human diet have been subjected to testing for carcinogenic and anticarcinogenic potential. Selected additional substances should be subjected to appropriate testing in order to develop a more comprehensive database on which to base comparisons of the potential cancer risks or protective effects of naturally occurring and synthetic chemicals in the diet. Because resources for toxicological testing are limited and because there is a vast number of naturally occurring dietary chemicals, further testing of appropriately selected naturally occurring food chemicals requires the establishment of selection criteria. For potential carcinogens, priority should be assigned to those suspected naturally occurring non-nutritive chemicals that occur at relatively high concentrations in commonly consumed foods, and/or those whose consumption is associated with diets or life styles known to be deleterious. Research should only be conducted when there is substantial evidence that an important problem exists and when there is a reasonable expectation of a meaningful result. Unless a suspected carcinogen or anticarcinogen occurs at high and measurable levels in a diet, its risk to humans cannot be predicted using present methods (experimental animal studies or human epidemiologic investigations).

  Additional criteria should be based on our knowledge of known carcinogens and anticarcinogens. For example, naturally occurring chemicals could also be accorded a higher priority for testing if they 1) fall in the same chemical class as known chemical carcinogens or anticarcinogens, 2) contain chemical groups also found in known chemical carcinogens or anticarcinogens, 3) are likely, based on structural comparisons with known chemical carcinogens or anticarcinogens, to form reactive intermediates, in vivo, 4) are known to be mutagenic and/or to bind to DNA, 5) share biologic effects similar to those of known nongenotoxic carcinogens, or 6)

• are likely, based on structural comparisons with known chemicals, to be unusually stable (i.e., long lasting) in vivo.
  
  

  High priority for identifying potential anticarcinogens might be given additional consideration, in view of the fact that they do offer the possibility of new approaches to cancer control and prevention.

• To help fill the data gaps on the cancer risk of dietary constituents, improved short-term screening tests for carcinogenic and anticarcinogenic activity should be developed, especially for detecting nongenotoxic effects that are relevant to carcinogenesis.
  
  

  Existing short-term screening tests, usually employing cell culture systems, often provide useful information, but new methods need to be developed and validated. Emphasis should be placed on developing systems that use human genes, enzymes, cells, or tissues. Since most of the present short-term tests detect DNA reactive compounds, new methods are needed for screening chemicals for nongenotoxic endpoints such as cell proliferation, hormonal effects, receptor mediated events, and effects on cell-cell interactions, gene expression, differentiation, and apoptosis (programmed cell death). Great promise exists for the use of transgenic mice.

• The risk of cancer from excess calories and fat should be further delineated vis-à-vis naturally occurring and synthetic dietary chemicals.
  
  

  There is considerable evidence that excessive calorie (energy) intake (i.e., in excess of body needs and including fat) is associated with increased cancer risk for several sites. In rodents, and especially in humans, mammary cancer is associated both with excess calories and with high proportion of calories as fat. The mechanism(s) responsible for this effect have not been clearly identified. Possible mechanisms that have been implicated include: increased

• cell proliferation, decreased cell death, changes in hormonal status, and alterations in the activity of enzymes which metabolize endogenous and environmental agents, and increased oxidative stress. Further studies are needed to elucidate the precise mechanisms and to better define what is optimal or excess caloric intake. Dietary fat has also been associated with increased risk of some forms of cancer, but it is not clear if this is related to the high caloric contribution of fat, to specific constituents in foods high in saturated fats (such as specific fatty acids or other lipid oxidation products), or heterocyclic amines produced in cooking. These relationships and the underlying mechanisms need further study and clarification.
• The specific chemicals that provide the protective effects of vegetables and fruits should be identified and their protective mechanisms delineated.
  
  

  The consumption of diets rich in fruits and vegetables is associated with reduced incidence of several forms of human cancer. The specific factors accounting for this relationship are not known with certainty and require further investigation. A number of vitamins, minerals and non-nutritive components of fruits and vegetables have been identified which may contribute to the protective properties of these foods. Further research is needed on the independent and interactive effects of these compounds and on the identification of additional protective components. At present, a sound recommendation for cancer prevention is to increase fruit and vegetable intake. Concerning specific plant derived chemicals, we do not have adequate information to recommend supplementation beyond the recommended daily requirements for particular vitamins or other nutrients.

It is apparent from the above list of recommendations that new research approaches are required to elucidate the precise roles of

naturally occurring and synthetic dietary chemicals in human cancer causation and prevention. These needs merge with those related to the broader issue of the roles that chemicals in the general environment may play in cancer causation and prevention. The following section provides a broad outline of options available for future directions of research in this area.

The conclusions and recommendations of this report indicate that a better understanding is needed of the role of specific dietary naturally occurring and synthetic chemicals in cancer causation and prevention. This is an extremely complex task. Progress in this area will require expertise and extensive further research in several disciplines, including food chemistry, analytical chemistry, toxicology, nutrition, carcinogenesis, biochemistry, molecular biology, and epidemiology. It would be highly desirable, therefore, to develop multi-disciplinary teams to conduct research in this broad area and to develop new resources. One approach would be to develop specific programs committed to this field of research. It is also apparent from this report that answers to these questions will not come simply by further analyzing the existing database, since it is often fragmentary and inconclusive. In addition, because of the vast number of chemicals present in a typical diet, new concepts and methods must be developed to address this problem, and extensive research is required so that results obtained in various model systems can be extrapolated with greater validity to human populations. It is essential, therefore, that additional intellectual and physical resources be developed to meet these needs.

A major limitation in this field that we do not understand the

precise mechanisms that underlie the multistage process of carcinogenesis and how chemicals in the diet are able to alter this process. Recent advances in the cellular biology and molecular genetics of cancer are beginning to provide insight into the details. These findings indicate that the carcinogenic process is often associated with activation of dominant-acting cellular oncogenes and/or the inactivation of recessive tumor suppressor genes that normally influence growth. Evidence also indicates that the carcinogenic process is associated with abnormalities in cyclin and cyclin-related genes that control the cell cycle and proliferation and inhibit the processes of apoptosis and differentiation. Advances are also being made in identifying abnormalities in genes responsible for tumor invasion and metastasis, i.e., genes that control cell-cell interactions, cell locomotion, extracellular proteases, and angiogenesis. In addition, genetically-based variations in susceptibility to cancer are being identified in human populations.

Progress is also being made in identifying the ways in which various chemicals interact with critical cellular targets either to increase or decrease the process of carcinogenesis. Most of the progress relates to chemicals that act by binding directly to DNA (genotoxic agents), including mechanisms involved in metabolic activation or deactivation of these chemicals and the protective role of DNA repair. In addition, the roles of endogenous processes of DNA damage, including oxidative damage, deamination, depurination, and the formation of exocyclic adducts, are being identified.

Furthermore, research is studying the interaction of chemicals with cellular components other than DNA that might be involved in carcinogenesis or its prevention. In this group are the interactions of chemicals with cellular receptors, membranes, protein kinases, and transcription factors, as well as alterations in the metabolism of endogenous hormones and growth factors. Also being identified are alterations in pathways of signal transduction, the cascade of cytoplasmic events involving numerous protein kinases that eventually trigger alterations in gene expression in the nucleus by

affecting the function of specific transcription factors. Elucidation of the specific pathways altered by various chemicals, particularly those that do not interact directly with DNA, should provide not only insight into the carcinogenic process but also knowledge necessary for extrapolating from results obtained at high doses in rodents or in cell culture systems to the human situation, which usually involves low-dose exposures.

As discussed in Chapter 4, a number of dietary intervention studies are in progress or are being developed, as well as chemoprevention studies that employ micronutrients or related compounds. It would be desirable to incorporate specific analytic assays and biologic markers into these studies to provide mechanistic insights and more sensitive indicators of response. More detailed information on the actual exposure to the substances being evaluated and on cellular responsiveness to various stimuli would also be useful.

The advantages and limitations of various types of epidemiologic studies, including ecologic, case-control, cohort, intervention, and molecular epidemiology studies are discussed in Chapter 4. Since results from these studies have suggested that a major fraction of human cancer is due to dietary factors, it is appropriate to extend this approach to identify with greater certainty the specific factors and types of cancer involved and populations affected.

Molecular epidemiology holds considerable promise in this regard. As discussed in detail in Chapter 4, laboratory procedures are now available for measuring a variety of biologic markers related to exposure and susceptibility in samples obtained from relatively large populations. These include markers that are related predominantly to DNA-altering chemicals. The markers provide information on 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. Future progress in this field requires improvements in the sensitivity, specificity, reproducibility, and predictive value of the assays for such biologic markers. The procedures for collecting specimens must be minimally invasive.

In addition, there is a critical need to develop markers related to cell proliferation, apoptosis, receptors, protein kinases, and altered gene expression, since, as mentioned above, certain carcinogens might exert their effects through these mechanisms.

To assess exposure, previous epidemiologic studies on diet, nutrition, and cancer relied largely on dietary recall methods and, in some instances, on the levels of various nutrients in blood, tissues, and urine. These approaches have proved useful in identifying both potential risk factors and protective factors for cancer. Biologic markers that better reflect long-term intakes are needed, since currently used assays of nutrient levels in blood and urine generally reflect only recent consumption patterns (NRC 1993a). Such markers could be particularly helpful in further assessing the role of dietary fat in cancer causation. In addition, since dietary recall is subject to considerable error, methods are needed that will improve the accuracy of these assessments or correct for errors when they are nondifferential (e.g., Rosner and Willett 1988).

To further explore the hypothesis that excessive caloric intake per se may be a major risk factor for human cancer, biologic markers that assess specific effects of excess or limited caloric intake need to be developed. These might include assays related to carcinogen metabolism, oxidative damage, signal transduction and gene expression, cell proliferation, and apoptosis.

Biologic markers related to oxidative damage may prove to be of considerable importance, since several antioxidants in the diet (e.g., vitamin C, vitamin E, certain carotenoids) have been implicated as protective factors. Such markers may include 1) urinary levels of oxidized DNA bases, 2) analyses of DNA samples for strand breaks

or oxidized bases (thymine glycol, 8-hydroxyguanine, etc.), 3) blood and tissue levels of malondialdehyde, an oxidation product of lipids, and 4) markers of enzymes that detoxify activated forms of oxygen, such as catalase and superoxide dismutase (Cerutti and Trump 1991, Pryor 1993, Teebor et al. 1988).

Aside from histopathology, there are few markers for detecting preneoplastic lesions (e.g., colonic adenomatous polyps). The further development of such markers would make cohort and intervention studies more feasible and enhance the assessment of various hypotheses related to diet and cancer, since such markers could greatly shorten the required interval for follow-up and also improve intervention strategies.

A number of intervention studies based on dietary modification or the use of chemopreventive agents (vitamin A and related retinoids, beta carotene, vitamins E and C, calcium etc.) are in progress or are being developed (see Chapter 4). It would be desirable to incorporate some of the above biologic markers into these studies, in order to refine the assessment of biologic responses in individuals and to provide mechanistic insights.

Rodent bioassays, short-term tests in rodents, and in vitro assays, including recently developed cellular and molecular biology methods, are summarized in Chapter 4. The committee agreed that although many of these assays provide useful information, it is important to stress that they serve only as screening tests (NRC 1993b). Better extrapolation of the results obtained to the human situation requires new approaches, particularly those that provide insights into the underlying mechanisms. Experimental animals are being developed, and numerous other potentially useful models are being developed. These may provide better screening tests and also provide mechanistic insight.

The current standard NTP rodent bioassay employs only one

type of diet, although other fairly extensive studies have been conducted on the effects of dietary factors on carcinogenesis in rodents (see Chapters 2 and 4). Further studies in which dietary constituents, particularly caloric intake and fat content, are varied, either alone or in combination with other factors, are needed. These studies should also have a mechanistic approach, should incorporate some of the types of biologic markers mentioned above, and should test specific and novel hypotheses. A recent approach, interesting although fraught with difficulties, is to feed rodents a homogenate of an actual daily human Western-type diet, to assess the total effects of the human diet on cancer induction, when testing alone or in combination with specific natural or synthetic chemicals (Rozen et al. 1996).

The above considerations also apply to the use of rodent carcinogenicity assays to detect naturally occurring and synthetic dietary chemicals that inhibit tumor formation, i.e., anticarcinogens. A major difficulty with current protocols is that they require large doses of potent carcinogens, which frequently necessitates the use of high doses of anticarcinogens. Assays more realistically approximating human exposure to dietary chemicals are necessary.

It is of critical importance to obtain greater insight into interactions among multiple chemicals, since food is a highly complex mixture of agents that can increase or decrease the risk of carcinogenesis. The study of complex mixtures has been limited, largely because examining such interactions is difficult and expensive (NRC 1988). Several complex interactions between agents in the human diet (and other types of agents) have been identified, e.g., the interaction of aflatoxin and hepatitis B and C viruses, in the causation of liver cancer; nitrosamines and the bacterium Helicobacter pylori in the causation of stomach cancer; cigarette smoke and asbestos in the causation of lung cancer; and others. Nevertheless, more complexities exist in the diet, and these need to be evaluated. In particular, the interaction between carcinogenic and anticarcinogenic components of the diet needs to be better delineated,

including interactions affecting metabolism and other types of cellular responses to chemicals. As described in previous chapters of this report, several components of the diet have been identified that can increase or decrease the risk of developing cancer in experimental animal or in vitro assays. Determining with greater certainty which of these components are significant contributors to cancer risk in humans is necessary so that an optimum diet can be recommended to the American public (NRC 1982). Obviously, intensive research is required on this subject.

A critical parameter in carcinogenesis is the number of cell divisions in a target tissue. This number can be increased by a variety of mechanisms that either increase cell births or decrease cell deaths. Greater understanding of these mechanisms is needed, as are better assays, especially in humans, for quantitating cell division, apoptosis, differentiation and other processes known to affect carcinogenesis. Assays for these processes are needed that do not require that human tissue be removed to assess the effects of various dietary factors. Quantitative assays in rodent tissues are being developed, but more sensitive and specific assays are needed.

The use of human tissues and cell systems for studies on carcinogenesis, including mechanistic studies, is strongly encouraged. Powerful new tools have been provided by the advent of epithelial human cell lines and the ability to genetically engineer derivatives of these cells that display increased or decreased expression of specific genes (for example, genes that encode drug metabolizing enzymes and genes that alter oxidative stress). In addition, the development of specific strains of transgenic mice or of mice in which specific genes have been inactivated also provide new approaches to carcinogen testing and mechanistic studies. It is hoped that these new approaches will provide greater insight into the carcinogenic process in both rodents and humans, as well as provide a more rational approach to our understanding of chemical interactions in carcinogenesis and anticarcinogenesis and of their relevance to humans.

Rapid advances are being made in analytical methods and in methods for elucidating the complex structure of natural products. These methods should be applied to the analysis of various human dietary constituents. Advances in this general area, coupled with powerful computer-based methods for cataloging and analyzing chemical structures and activities, should also accelerate progress in the field of structure-activity analyses to predict carcinogenicity. To date, these methods have been applied largely to carcinogens that are genotoxic. It is essential to expand this approach to nongenotoxic carcinogens and also it to the detection and development of anticarcinogenic substances.

Processing has long been used to remove or reduce unwanted constituents, and to introduce, increase, or restore desirable constituents, including essential nutrients. As we identify those naturally occurring constituents that either enhance or inhibit human cancer, and define the conditions and concentrations that govern their actions, we can expect processing to play a large role in increasing the protective potential of diet, as it has previously played a role in improving nutritional quality.

Over many centuries plants, and to a lesser degree animals, have been optimized as human food sources by selection and breeding for such desirable characteristics as safety, size, color, flavor, yield, and resistance to disease. During this century, scientists have been able to use the principles of Mendelian genetics to expedite this process. With the exception of flavor and toxic constituents in foods, the chemical compounds responsible for the improved characteristics have not been well characterized, nor has such information usually been required.

The recent advances in genetic engineering and biotechnology (see Chapter 2) should facilitate further improvements in the quality of the food supply with respect to cancer prevention. Specifically, as naturally occurring dietary chemicals that either enhance or inhibit human cancer risk are identified, these chemicals will be candidates for appropriate modification through food biotechnology. However, in contrast to earlier approaches, this new technology requires much greater knowledge of the specific compounds involved, together with knowledge of their biosynthesis in the source organism. Depending on the compound involved, such information may or may not now be available in the biochemical literature, although it can be acquired through technology now or soon to be available.

At the present time, cancers are the second leading cause of mortality in the United States, resulting in over 500,000 deaths per year. Unless current trends are reversed, cancer will be the major cause of death in the early part of the 21^st century. Conventional wisdom states that dietary factors play an important role in the causation of a major fraction of these cancers. Although synthetic chemicals present in the diet cannot be ignored as potential carcinogenic risks, it seems likely that it is the naturally occurring compounds in our diet, together with excess fat and total calories, that have the greatest effect on cancer causation and prevention. This subject is, therefore, of major relevance to public health protection and disease prevention. However, the assessment by this committee indicates several inadequacies in 1) the specific naturally occurring chemicals (or mixtures) that are involved in cancer causation; 2) the mechanisms by which they act; 3) which types of cancer they affect; and 4) the magnitude of these effects. New research approaches, at both the fundamental and applied levels, are urgently required to address this important problem.

Coupled with the requirement for research efforts in these areas is the need to better characterize the chemical composition of our diet and its variations in the American population. Advances in analytic and survey techniques should facilitate this endeavor.

Finally, as specific naturally occurring dietary chemicals are identified that either enhance or inhibit cancer risks in humans, it will be possible to better formulate specific dietary guidelines for the American public. It may also be possible to utilize this information to modify the composition of our food sources, through food processing, breeding methods, and genetic engineering and other advances in biotechnology, so as to optimize the quality of the diet with respect to cancer prevention. Above all, a major effort will be needed to educate the American public regarding appropriate life style modifications if we are to achieve these goals.

Cerutti, P.A., and B.F. Trump. 1991. Inflammation and oxidative stress in carcinogenesis. Cancer Cells 3:1-7.

NRC (National Research Council). 1982. Diet, Nutrition and Cancer. Washington, DC: National Academy Press.

NRC (National Research Council). 1988. Complex Mixtures: Methods for In Vivo Toxicity Testing. Washington, DC: National Academy Press.

NRC (National Research Council). 1989. Diet and Health: Implications for Reducing Chronic Disease Risk. Food and Nutrition Board, Committee on Diet and Health. Washington, DC:National Academy Press.

NRC (National Research Council). 1993a. Pesticides in the Diets of Infants and Children. Washington, DC: National Academy Press.

NRC (National Research Council). 1993b. Issues in Risk Assessment. Washington, DC: National Academy Press.

Pryor, W.A., 1993. Measurement of oxidative stress in humans. Cancer Epidemiology, Biomarkers and Prevention 2:289-292.

Rozen, P., V. Liberman. F. Lubin, S. Angel, R. Owen, N. Trostler, T. Skolnik, and D. Kritchevsky. 1996. A new dietary model to study colorectal carcinogenesis: Experimental design, food preparation and experimental findings. Nutrition and Cancer. in press.

Rosner, B., and W.C. Willett. 1988. Interval estimates for correlation coefficients corrected for within-person variation: Implications for study design and hypothesis testing . Am. J. Epidemiol. 127:377-386.

Teebor, G.W., R.J. Boorstein, and J. Cadet. 1988. The repairability of oxidative free radical mediated damage to DNA: A review. International J. Radiat. Res. 54:131-150.

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, B.S. Kramer, and D.L. Weed, ed. New York: Marcel Dekker, Inc.