Fd Chem. Toxic. Vol. 30, No. 4, pp. 327-332, 1992
0278-6915/92 $5.00 +0.00 Copyright © 1992 Pergamon Press Ltd
Printed in Great Britain. All rights reserved
Review Section C A R C I N O G E N I C CHEMICALS IN FOOD: E V A L U A T I N G THE HEALTH RISK* P. J. ABBoTt Chemicals Safety Unit, Department of Health, Housing & Community Services, PO Box 9848, Canberra, ACT 2601, Australia
(Accepted 18 December 1991) Summary--The presence of a low level of potentially harmful chemicals in food continues to be a concern to many individuals. A major concern is that these chemicals, which can be synthetic or naturally occurring, may be a causative factor in human cancer. Synthetic chemicals in food may be present either as specific additives or as contaminants derived from environmental or agricultural chemicals. Food also contains a variety of naturally occurring chemicals derived from vegetables or other plants. These may in some cases be considered as contaminants, and are occasionally used as specific additives. New chemicals can also be formed during the cooking or preserving processes. The capacity of any of these chemicals to induce cellular damage and mutation is minimized by natural defence systems such as an efficient cellular detoxification system and DNA repair. The factors influencing turnout formation in humans are numerous and interrelated and exposure to minor dietary chemicals needs to be considered in this context. Thus, the results of animal carcinogenicity assays on individual chemicals need to be interpreted with care, taking into account the mechanisms by which mutagenic and other chemicals initiate cancer, as well as the level of human exposure to these chemicals. Further research is necessary to determine the role, if any, of minor dietary components in tumour formation. Meanwhile, there needs to be a more holistic approach to the multitude of factors, including total diet, that may influence human cancer incidence. In this way, the relative risk of dietary chemicals may be given a more meaningful perspective for health professionals and consumers alike.
Introduction
The current interest in 'natural' foods has raised numerous questions regarding the safety of the many chemicals, both synthetic and naturally occurring, that are present at low levels in food. Synthetic chemicals found in food either serve a specific purpose (e.g. flavour or colouring agent, preservative, stabilizer, etc.) or are present as contaminants (e.g. pesticide residues, metals or other environmental chemicals). The levels of such chemicals are generally low to very low. However, their presence is a health concern for m a n y individuals and therefore the potential risk needs to be addressed. Another source of potentially harmful chemicals in food is the naturally occurring chemicals of plant and fungal origin. These may occur at varying levels and in some cases may be classed as contaminants. Some naturally occurring chemicals are also used as specific additives. The role of this wide spectrum of natural chemicals remains largely u n k n o w n , although some may act as natural pesticides in plants. M a n y have been shown to have toxicological properties in animals and humans and
*The views expressed in this paper do not necessarily reflect those of the National Health & Medical Research Council of Australia.
some to induce cancer in rodents under certain conditions (Cheeke, 1989). A third group of chemicals in food is the novel chemicals formed either during cooking or during the preserving process. There is a vast array of such chemicals to which humans are exposed at low levels. The most closely studied have been the pyrolysis products of cooked meat, some of which have been shown to be mutagenie (Larsen and Poulsen, 1987; Overvik and Gustafsson, 1990). When either the synthetic, naturally occurring or newly formed chemicals in food are shown to induce cancer in laboratory animals, there is understandable concern regarding h u m a n safety. The possibility of a lifetime of low level exposure to these agents increases such concerns. The current process for assessing potential h u m a n carcinogenicity relies on an extrapolation of the results from the relatively high doselevel studies in laboratory animals to the real-life situation of low-level h u m a n exposure. Given the paucity of our knowledge of the effects of low-level exposure, the assumptions inherent in this process are crucial to an overall appraisal of potential risk. The question of whether the potentially hazardous chemicals found at low levels in food have a significant role in the aetiology of h u m a n cancer is not easily answered. However, significant progress has
327
328
P.J. ABaOTT
occurred both in understanding the mechanisms involved in carcinogenesis and also in identifying the natural defence systems that protect humans from low-level exposure to hazardous chemicals. This paper seeks to explore some of these issues in the light of more recent developments in carcinogenic risk assessment with the aim of arriving at a more balanced view of food safety, particularly with regard to carcinogenic risk. Chemicals and human cancer
Although diet is established as a factor in human cancer, the extent of its contribution remains unknown (Davis, 1989; Doll, 1988; Doll and Peto, 1981). The majority of the dietary components that lead to the variation in cancer incidence between human populations have not been identified. The presence of potentially hazardous chemicals in food, albeit at very low levels, has often been implicated as one of the causative factors, but their contribution is unknown and may not be possible to determine by epidemiologicai investigations. Historically, those chemicals that have been identified by epidemiological methods as human carcinogens occurred in situations of significant human exposure, usually occupational, over prolonged periods (IARC, 1987). Evaluating the potential human carcinogenicity of some of these same chemicals when present at very low levels in food requires different criteria for assessing hazard and will require a better understanding of the multitude of factors that interact to produce cancer. Indeed, it is possible that the presence of these chemicals could be incidental to the overall process of carcinogenesis in relation to diet. The desire to identify specific dietary chemicals as the causative agents in cancer formation is understandable but may be an over-simplification of what is now known to be a complex multistage process. This process may involve multiple mutations, hyperplasia, inhibition of intercellular communication, oncogene activation, deactivation of tumour suppressor genes, clonal expansion and tumour progression (Farber, 1984; Weinstein, 1988). Dietary constituents, both macro- and micro-components, may influence specific steps in this process in either a positive or negative manner. Of the micro-components, those chemicals that can cause DNA damage and mutation, such as nitrosamines, are considered most likely to be associated with cancer initiation. Mutation can, however, also occur spontaneously from errors in replication or DNA repair (Mullaart et al., 1990) or from oxidative damage (Ames, 1989), thus obscuring the influence of dietary mutagens. Other micro-components may influence the carcinogenic process subsequent to its initiation. For example, in animals, levels of antioxidants such as butylated hydroxytoluene and butylated hydroxyanisole are known to influence cancer incidence (Ito et al., 1987).
Whether the association between consumption of vegetables and a reduced cancer incidence in humans (Le Marchand et al., 1989) can be linked with the level of antioxidants in vegetables remains obscure. Other so-called 'anti-carcinogens', such as the retinoids, have been identified in food and their role in influencing human cancer is currently being investigated (Lippman and Meyskens, 1988). It may, however, be premature to speak of 'anti-carcinogens' to specific dietary components (Ames, 1983) when the exact modulating role of the macro-components such as fat, cholesterol, salt, fibre and carbohydrate intake is yet to be resolved (Davis, 1989). Both dietary micro- and macro-components may also influence the rate of cell turnover during either normal growth or regenerative hyperplasia, and this has been shown to influence cancer incidence in both animals (Cohen and Ellwein, 1990) and humans (Preston-Martin et al., 1990). Consideration of the positive and negative influence of chemicals on all factors relevant to cancer development may be necessary to elucidate the role of dietary chemicals in human cancer. Defining the term 'carcinogen'
A complicating factor when discussing carcinogenic chemicals is that whereas our understanding of the carcinogenic process and its modulating factors has improved enormously, the evolution of appropriate terminology to describe the process has been less impressive. Thus, the classic definition of a carcinogen (i.e. any substance that can cause an increase in the incidence or rate of onset of malignant tumour formation) takes no account of factors such as potency, route of exposure, or other confounding factors such as diet, stress and general health. If the evidence for carcinogenicity comes from animal studies, species differences must also be considered. The assumption behind this definition of a carcinogen is that a chemical that is labelled as a 'carcinogen' on the basis of epidemiological evidence or animal bioassay is hazardous to human health given appropriate exposure levels and circumstances. Unfortunately, this has been misinterpreted by some as meaning all exposure levels and all circumstances. Use of animal bioassays to identify carcinogens began initially with the assumption that carcinogenic chemicals could be clearly separated from noncarcinogenic chemicals and thus eliminated from the human environment, or at least controlled. Although this has been effective for the more potent carcinogens, it has become clear that many chemicals are capable of inducing turnouts in animals under particular experimental conditions that may not be relevant to human exposure. This has led to a reappraisal of how such results should be interpreted. Whether or not one should classify such chemicals as potential human carcinogens is a matter of judgement, taking into account the conditions under which tumours were induced in the animal studies and the
Carcinogenic chemicals in food dose levels used, as well as the expected route and level of exposure in humans (Abbott, 1990). Use of the term 'carcinogen' without qualification can no longer be justified, particularly in relation to the low-level chemicals in food. Many more factors need to be recognized as influencing potential human carcinogenicity, and the conditions under which a chemical may be justifiably suspected of human carcinogenic activity need to be clearly identified.
Natural and synthetic carcinogens in food Regulation of human exposure to chemicals in food has been, and continues to be, almost exclusively directed towards synthetic chemicals. This is a result of limited toxicological information on chemicals that occur naturally in food and difficulties in controlling their levels, together with the public perception that chemicals that are natural must be harmless. Although the acutely toxic components of many plants have been identified, subtle aspects of the toxicity of these and other components are not as well understood, particularly their potential for longterm effects, including cancer (Cheeke, 1989). Because of these limited data on naturally occurring chemicals, it is difficult to make a fair comparison with synthetic chemicals. Pesticides, for example, are extensively tested in animals before use and the levels permissible in crops or food products is controlled. Many of the naturally occuring chemicals in plants act as natural pesticides (Ames et al., 1990a; Beier, 1990): these include coumarins, glycoalkaloids, isoflavonoids, stilbenes and terpenoids, and their level can vary with different plant varieties and at different phases of growth. Selection for disease-resistant plants can, in some cases, be directly related to the level of naturally occurring pesticides. The advent of biotechnology may allow the possibility of altering the levels of naturally occurring plant constituents, either increasing the levels of toxic components to provide enhanced insect resistance, or decreasing these levels to reduce toxicity (IFBC, 1990). Hence the need for more information on the toxicology of these natural components. The limited animal testing that has been performed has confirmed the potential carcinogenicity of a number of naturally occurring compounds, such as caffeic acid, benzylacetate, safrole and allyl isothiocyanate (Ames et al., 1990a; Rosenkranz and Klopman, 1990) and has led to some questioning of the relatively greater emphasis placed on the toxicity testing of synthetic chemicals (Ames et al., 1990b).
Natural defence systems Animals, including humans, have evolved effective defence systems against potentially harmful dietary chemicals at the whole-animal, cellular and molecular levels. These systems operate equally well for both naturally occurring and synthetic chemicals. For FCT 30/4---E
329
those chemicals with high acute toxicity, the most important and effective barriers are the more general mechanisms of emesis and shedding of surface cells in the oesophagus, stomach and gastro-intestinal tract. For the majority of chemicals in food, however, the major defence mechanism is the inducible pathways of the liver. The oxidation reactions, which are catalysed by the numerous cytochrome P-450 enzymes, can operate on a wide range of substrates. For the majority of chemicals, oxidation reactions catalysed by P-450 enzymes result in detoxification and excretion. For some chemicals, however, oxidation reactions can result in a more highly reactive molecule capable of covalent binding to DNA to form pro-mutagenic lesions. Some chemicals may also enhance DNA damage through the formation of reactive oxygen species such as superoxide, hydroxyl, peroxyl and alkoxyl radicals (Ames, 1983; Epe et al., 1990) although the extent to which DNA is damaged in this way is unknown. Cells contain enzymes such as superoxide dismutase and glutathione peroxidase, which protect against oxidative damage. Dietary antioxidants such as r-carotene, selenium, ascorbic acid and glutathione have also been shown to reduce cell damage and, under certain circumstances, to reduce tumour incidence (Ames, 1989). DNA damage, regardless of the origin, can be repaired by a variety of DNA-repair enzymes that recognize small or bulky lesions, DNA-strand breaks or gaps. The importance of D N A repair in reducing mutation rate and maintaining cell integrity can be observed in humans who have a genetic deficiency associated with repair of damage caused by ultraviolet light (Xeroderma pigmentosum patients); such individuals have a significantly increased incidence of skin cancer. Further understanding of D N A repair mechanisms may enable the identification of individuals with increased cancer risk. DNA damage, if not repaired, may result in a mutational event, but although there is substantial evidence that mutation is an essential step for tumour induction, it is only one of many steps and is likely to be a relatively common event, compared with the occurrence of tumours. In comparing the toxicity of natural versus synthetic chemicals, a common argument is that, in terms of detoxification systems, humans are biologically better equipped to handle naturally occurring chemicals in food. This argument is not particularly convincing when we consider, first, the incredibly wide variety of chemical components in food and the enormous metabolic diversity of the liver, and secondly, that many of the foods we now consume are relatively new in evolutionary terms and for which we could never have evolved specific detoxification systems. A third point is that humans consume very different regional diets, which clearly influence cancer incidence. There is some evidence that naturally occurring dietary components may be linked with this variation (Lu et al., 1986).
330
P.J. ABBOTT
Current testing for carcinogenic chemicals Identifying those chemicals that may cause an increase in cancer incidence in humans has proved to be a difficult task for a variety of reasons. Most chemicals classified by the International Agency for Research on Cancer (IARC) as proven human carcinogens have been identified from epidemiological data and subsequently supported by experiments in laboratory animals. Indeed, the close correlation between the response of animals and humans to the potent carcinogens that were first identified has been the driving force to use animals to test other chemicals for carcinogenicity. This extensive use of animals for testing, although providing a large quantity of data, has raised new questions regarding interpretation of the results, particularly the difficulty of making species and dose extrapolations without reference to the mechanism of carcinogenesis and the route and level of human exposure (Clayson, 1987). This uncertainty is reflected in the large number of chemicals that have been classified by IARC as 'probable' or 'possible' human carcinogens. The major difficulty in using animal data has been extrapolation of dose. Humans are exposed only to low or very low dose levels whereas the same chemicals are tested in animals at relatively high dose levels. Although the high dose levels improve the statistical power of the assay, they also lead to complications that may influence the outcome. Secondary factors such as tissue damage and regenerative hyperplasia, hormone imbalance, mineral deposition and saturation of metabolic pathways can alter both the rate of tumour development and possibly the initiation of new tumours. Consideration of these factors is necessary when interpreting the results of animal bioassays. In many cases, carcinogenic mechanisms have been identified in animal studies that have no relevance at lower dose levels or in the human species. The induction of ~-2-microglobulin in male rat kidney is an example of a carcinogenic mechanism that is irrelevant to humans (Alden, 1986).
A threshold dose level for carcinogens? The question of a threshold dose for carcinogens has been a controversial one and stems from the close correlation between carcinogenicity, mutagenicity and D N A damage for those chemicals first identified as human carcinogens. The potential for mutation from a single D N A lesion, together with the clear mutagenicity of most of the potent human carcinogens, led to the idea that no safe dose existed for carcinogens. Despite advances in our understanding of carcinogenesis and the significant flaws in this argument, this theory underlies much of the fear regarding carcinogenic chemicals. The relationship between mutagenicity and carcinogenicity has undergone close examination as the number of chemicals undergoing extensive rodent carcinogenicity testing has risen. Whereas those
chemicals that are clear mutagens have generally continued to be shown to be carcinogens, the reverse relationship has not proved so reliable. Approximately half the chemicals that have given rise to tumours in either mice or rats in studies conducted by the US National Toxicology Program (NTP) were negative in a range of mutagenicity assays (Ashby and Tennant, 1988). Although there may have been some selection bias by the NTP towards non-mutagenie chemicals, the results indicate that other than mutagenic routes to cancer initiation can exist. These so-called 'non-genotoxic' carcinogens were generally positive in animal bioassays only at the higher dose levels of the range used. The mechanism of tumour formation may vary for individual chemicals but appears to involve a sustained disruption to tissue homoeostasis (Clayson, 1989). At dose levels below which such disruption occurs, no tumour formation is evident. Although D N A mutation may still occur at a later phase as part of the carcinogenic process induced by these chemicals, it does not represent the initial step in the process, and a clear threshold for the induction of cancer can be demonstrated. Examples of such chemicals that can occur in food include contaminants such as the pesticide amitraz, oestrogenic agents such as diethylstilboestrol, goitrogenic agents such as amitrole, peroxisome proliferators such as phthalates, and the artificial sweetener saccharin. For the majority of such chemicals, human exposure occurs at levels far below the demonstrated threshold for tumour induction. For chemicals with demonstrated genetic activity in the somatic cells of animals, a cautious approach is to assume a propensity to induce cancer in humans. This is based on the premise that genotoxic carcinogens have a stochastic mode of action, that is, the probability of occurrence of cancer depends only on the dose, with no threshold dose level. Given the considerable barriers associated with physical absorption, distribution, metabolism and D N A repair, it seems more likely that a threshold dose level for mutation will occur in the target organ. Establishing such a threshold, however, may be difficult, although recent advances in measuring low-level mutagenic lesions in D N A may provide some progress in determining the dose-response relationship at the low dose levels to which humans are normally exposed (Swenberg et aL, 1990). In time, it may be possible to determine a virtually safe dose for mutagenic chemicals, based on the level and rate of accumulation of D N A lesions in target organs. Evidence for the existence of thresholds for either genotoxic or nongenotoxic carcinogenic chemicals may be able to provide a more scientific basis for establishing safe exposure levels (Clayson and Arnold, 1991).
Identifying carcinogenic risk factors Diet is one of the factors that can influence human cancer levels, and individual components can have
Carcinogenic chemicals in food either a positive or negative influence. Of the potential dietary risk factors, considerable public attention to date has focused on synthetic chemicals such as additives and pesticide residues; the natural constitutents, both micro-components, including antioxidants and mutagens, and macro-components, such as fat, fibre, salt and carbohydrate intake, have received less attention. In a consideration of carcinogenic risk factors, the micro-components (both synthetic and natural) may be divided into either mutagenic or non-mutagenic chemicals. The non-mutagenic chemicals in food are unlikely to be consumed at a level capable of causing a disruption to tissue homoeostasis (as discussed previously for 'non-genotoxic' carcinogens) and thus to initiate cancer by this secondary route. Exceptions to this generalization include some of the natural fungal toxins such as the ochratoxins, the levels of which must be carefully monitored. The mutagenic chemicals in food, which include polycyclic aromatic hydrocarbons, nitrosamines and aflatoxins, are considered to be a carcinogenic risk factor, although at present it is generally not possible to establish their activity in target organs. Their potential to cause mutation must also be considered in relation to the spontaneous mutation rate. Given that spontaneous mutation may also lead to cancer initiation, it may be difficult to establish the contribution of a particular chemical to the mutagenic burden. A better understanding of spontaneous D N A damage and mutation in relation to cancer incidence may improve this situation. Progress in establishing the real risks associated with low levels of mutagenic chemicals in the diet may come from a better understanding of the capacity of the natural defence systems and also from the development of better techniques for measuring the potential D N A or protein lesions in target organs. The current dependence on long-term animal testing of individual chemicals seems unlikely to provide meaningful answers to assist risk estimation. Individual chemicals may also need to be considered in the context of the entire diet, as well as taking into account other cellular controls on cancer development. Despite much speculation, there is little evidence that the synthetic chemicals found in food are linked with human cancer. This is perhaps not surprising given that, for any chemical suspected of being carcinogenic to humans, the risk is minimized by strict measures to control exposure. On the other hand, the levels of human cancer remain high despite these controls, perhaps because other major influences have not been recognized. Indeed, it seems likely that the greatest potential for any further reduction in human cancer levels lies in a better understanding of the influence of the macrocomponents of the diet. Epidemiological evidence (Weisburger, 1987) is beginning to support the view that dietary imbalance is a risk factor for human
331
cancer. However, given the number of confounding variables, the mechanism by which such components modulate cancer incidence may take some time to elucidate. Future directions
The next decade is likely to see a greater appreciation of the multitude of complex factors that influence cancer incidence in both animals and humans. This could lead to some changes in current procedures used to identify those dietary chemicals suspected of carcinogenic activity in humans. It is to be hoped that emphasis will shift further from a study of the carcinogenicity of individual chemicals to a consideration of the interaction between dietary components in relation to their modulation of cancer incidence. Other changes that are beginning to occur include closer attention to mechanistic studies, which may identify some of the early changes in carcinogenesis. Further research on biological monitoring of exposure at the tissue and D N A level may also allow for better risk assessment through the identification of biological thresholds. Finally, improvement in communicating the scientific basis for establishing risk factors associated with human cancer will, it is hoped, lead to a better appreciation of the relative carcinogenic risk of both the natural and the synthetic components of the diet. REFERENCES
Abbott P. J. (1990) Identifying human carcinogens: the need for a multidirectional approach. Australian Cancer Society, Cancer Forum. 14, 37-40. Alden C. L. (1988) A review of unique male rat hydrocarbon nephropathy. Toxicologic Pathology 14, 109-111. Ames B. N. (1983) Dietary carcinogens and anticarcinogens: oxygen radicals and degenerative diseases. Science, New York 221, 1256-1264. Ames B. N. (1989) Endogenous DNA damage as related to cancer and aging. Mutation Research 214, 41-46. Ames B. N., profet M. and Gold L. S. (1990a) Dietary pesticides (99.99% all natural). Proceedings of the National Academy of Sciences of the U.S.A. 87, 7777-7781. Ames B. N., Profet M. and Gold L. S. (1990b) Nature's chemicals and synthetic chemicals: comparative toxicology. Proceedings of the National Academy of Sciences of the U.S.A. 87, 7782-7786. Ashby J. and Tennant R. W. (1988) Chemical structure, Salmonella mutagenicity, and extent of carcinogenicity as indices of genotoxic carcinogens among 222 chemicals tested in rodents by the US NCI/NTP. Mutation Research 204, 17-115. Beier R. C. (1990) Natural pesticides and bioactive components in foods. Reviews of Environmental Contamination and Toxicology 113, 47-137. Cheeke P. R. (Editor) (1989) Toxicants of Plant Origin. Vols 1-4. CRC Press Inc., Boca Raton, FL. Clayson D. B. (1987) The need for biological risk assessment in reaching decisions about carcinogens. Mutation Research 185, 243-269. Clayson D. B. (1989) Can a mechanistic rationale be provided for non-genotoxic carcinogens identified in rodent bioassays? Mutation Research 221, 53~7.
332
P. J. AeeoTT
Clayson D. B. and Arnold D. L. (1991) The classification of carcinogens identified in the rodent bioassay as potential risks to humans: what type of substance should be tested next? Mutation Research 257, 91-106. Cohen S. M. and Ellwein L. B. (1990) Cell proliferation in carcinogenesis. Science, New York 249, 1007-1011. Davis D. L. (1989) Natural anticarcinogens, carcinogens, and changing patterns in cancer: some speculation. Environmental Research 50, 322-340. Doll R. (1988) Epidemiology and the prevention of cancer: some recent developments. Journal of Cancer Research and Clinical Oncology 114, 447-458. Doll R. and Peto R. (1981) The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today. Journal of the National Cancer Institute 66, 1191-1308. Epe B., Hegler J. and Wild D. (1990) Identification of ultimate DNA damaging oxygen species. Environmental Health Perspectives 88, 11 I-115. Farber E. (1984) The multistep nature of cancer development. Cancer Research 44, 4217-4223. IARC (1987) IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Supplement 7. Overall evaluation of carcinogenicity: an updating of IARC Monographs volumes 1 to 42. International Agency for Research on Cancer. International Food Biotechnology Council (1990) Biotechnologies and food: assuring the Safety of foods produced by genetic modification. Regulatory Pharmacology and Toxicology 12, 1-129. Ito N., Fukushima S. and Hirose M. (1987) Modification of the carcinogenic response by anti-oxidants. In Toxicological Aspects of Food. Edited by K. Miller. pp. 253-293. Elsevier, London. Larsen G. and Poulsen E. (1987) Mutagens and carcinogens in heat-processed food. In Toxicological Aspects of Food. Edited by K. Miller. pp. 205-252. Elsevier, London.
Le Marchand L., Yoshizawa C. N., Kolonel L. N., Hankin J. H. and Goodman M. T. (1989) Vegetable consumption and long cancer risk: a population-based case control study in Hawaii. Journal of the National Cancer Institute 81, 1158-1164. Lippman S. M. and Meyskens F. L. (1988) Retinoids as anticancer agents. Progress in Clinical and Biological Research 259, 229-244. Lu S. H., Montesano R., Zhang M. S., Feng L., Luo F., Chui S. X., Umbenhauer D., Saffhill R. and Rajewsky M. F. (1988) Relevance of N-nitrosamines to eosophageal cancer in China. Journal of Cellular Physiology, Suppl. 4, 51-58. Mullaart E., Lohman P. H., Berends F. and Vijg J. (1990) DNA damage metabolism and aging. Mutation Research 237, 189-210. Overvik E. and Gustafsson J.-A. (1990) Cooked food mutagens: current knowledge of formation and biological significance. Mutagenesis 5, 437-446. Preston-Martin S., Pike M. C., Ross R. K., Jones P. A. and Henderson B. E. (1990) Increased cell division as a cause of human cancer. Cancer Research 50, 7415-7421. Rosenkranz H. S. and Klopman G. (1990) Natural pesticides present on edible plants are predicted to be carcinogenic. Carcinogenesis 11, 349-353. Swenberg J. A., Fedtke N., Fennell T. R. and Water V. E. (1990) Relationship between carcinogen exposure, DNA adducts and carcinogenesis. In Progress in Predictive Toxicology. Edited by D. B. Clayson, I. R. Munro, P. Shubik and J. A. Swenberg. pp. 161-184. Elsevier, London. 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. Weisburger J. H. (1987) On the mechanisms relevant to nutritional carcinogenesis. Preventive Medicine 16, 586-591.
0278-6915/92 $5.00 +0.00 Copyright © 1992 Pergamon Press Ltd
Printed in Great Britain. All rights reserved
Review Section C A R C I N O G E N I C CHEMICALS IN FOOD: E V A L U A T I N G THE HEALTH RISK* P. J. ABBoTt Chemicals Safety Unit, Department of Health, Housing & Community Services, PO Box 9848, Canberra, ACT 2601, Australia
(Accepted 18 December 1991) Summary--The presence of a low level of potentially harmful chemicals in food continues to be a concern to many individuals. A major concern is that these chemicals, which can be synthetic or naturally occurring, may be a causative factor in human cancer. Synthetic chemicals in food may be present either as specific additives or as contaminants derived from environmental or agricultural chemicals. Food also contains a variety of naturally occurring chemicals derived from vegetables or other plants. These may in some cases be considered as contaminants, and are occasionally used as specific additives. New chemicals can also be formed during the cooking or preserving processes. The capacity of any of these chemicals to induce cellular damage and mutation is minimized by natural defence systems such as an efficient cellular detoxification system and DNA repair. The factors influencing turnout formation in humans are numerous and interrelated and exposure to minor dietary chemicals needs to be considered in this context. Thus, the results of animal carcinogenicity assays on individual chemicals need to be interpreted with care, taking into account the mechanisms by which mutagenic and other chemicals initiate cancer, as well as the level of human exposure to these chemicals. Further research is necessary to determine the role, if any, of minor dietary components in tumour formation. Meanwhile, there needs to be a more holistic approach to the multitude of factors, including total diet, that may influence human cancer incidence. In this way, the relative risk of dietary chemicals may be given a more meaningful perspective for health professionals and consumers alike.
Introduction
The current interest in 'natural' foods has raised numerous questions regarding the safety of the many chemicals, both synthetic and naturally occurring, that are present at low levels in food. Synthetic chemicals found in food either serve a specific purpose (e.g. flavour or colouring agent, preservative, stabilizer, etc.) or are present as contaminants (e.g. pesticide residues, metals or other environmental chemicals). The levels of such chemicals are generally low to very low. However, their presence is a health concern for m a n y individuals and therefore the potential risk needs to be addressed. Another source of potentially harmful chemicals in food is the naturally occurring chemicals of plant and fungal origin. These may occur at varying levels and in some cases may be classed as contaminants. Some naturally occurring chemicals are also used as specific additives. The role of this wide spectrum of natural chemicals remains largely u n k n o w n , although some may act as natural pesticides in plants. M a n y have been shown to have toxicological properties in animals and humans and
*The views expressed in this paper do not necessarily reflect those of the National Health & Medical Research Council of Australia.
some to induce cancer in rodents under certain conditions (Cheeke, 1989). A third group of chemicals in food is the novel chemicals formed either during cooking or during the preserving process. There is a vast array of such chemicals to which humans are exposed at low levels. The most closely studied have been the pyrolysis products of cooked meat, some of which have been shown to be mutagenie (Larsen and Poulsen, 1987; Overvik and Gustafsson, 1990). When either the synthetic, naturally occurring or newly formed chemicals in food are shown to induce cancer in laboratory animals, there is understandable concern regarding h u m a n safety. The possibility of a lifetime of low level exposure to these agents increases such concerns. The current process for assessing potential h u m a n carcinogenicity relies on an extrapolation of the results from the relatively high doselevel studies in laboratory animals to the real-life situation of low-level h u m a n exposure. Given the paucity of our knowledge of the effects of low-level exposure, the assumptions inherent in this process are crucial to an overall appraisal of potential risk. The question of whether the potentially hazardous chemicals found at low levels in food have a significant role in the aetiology of h u m a n cancer is not easily answered. However, significant progress has
327
328
P.J. ABaOTT
occurred both in understanding the mechanisms involved in carcinogenesis and also in identifying the natural defence systems that protect humans from low-level exposure to hazardous chemicals. This paper seeks to explore some of these issues in the light of more recent developments in carcinogenic risk assessment with the aim of arriving at a more balanced view of food safety, particularly with regard to carcinogenic risk. Chemicals and human cancer
Although diet is established as a factor in human cancer, the extent of its contribution remains unknown (Davis, 1989; Doll, 1988; Doll and Peto, 1981). The majority of the dietary components that lead to the variation in cancer incidence between human populations have not been identified. The presence of potentially hazardous chemicals in food, albeit at very low levels, has often been implicated as one of the causative factors, but their contribution is unknown and may not be possible to determine by epidemiologicai investigations. Historically, those chemicals that have been identified by epidemiological methods as human carcinogens occurred in situations of significant human exposure, usually occupational, over prolonged periods (IARC, 1987). Evaluating the potential human carcinogenicity of some of these same chemicals when present at very low levels in food requires different criteria for assessing hazard and will require a better understanding of the multitude of factors that interact to produce cancer. Indeed, it is possible that the presence of these chemicals could be incidental to the overall process of carcinogenesis in relation to diet. The desire to identify specific dietary chemicals as the causative agents in cancer formation is understandable but may be an over-simplification of what is now known to be a complex multistage process. This process may involve multiple mutations, hyperplasia, inhibition of intercellular communication, oncogene activation, deactivation of tumour suppressor genes, clonal expansion and tumour progression (Farber, 1984; Weinstein, 1988). Dietary constituents, both macro- and micro-components, may influence specific steps in this process in either a positive or negative manner. Of the micro-components, those chemicals that can cause DNA damage and mutation, such as nitrosamines, are considered most likely to be associated with cancer initiation. Mutation can, however, also occur spontaneously from errors in replication or DNA repair (Mullaart et al., 1990) or from oxidative damage (Ames, 1989), thus obscuring the influence of dietary mutagens. Other micro-components may influence the carcinogenic process subsequent to its initiation. For example, in animals, levels of antioxidants such as butylated hydroxytoluene and butylated hydroxyanisole are known to influence cancer incidence (Ito et al., 1987).
Whether the association between consumption of vegetables and a reduced cancer incidence in humans (Le Marchand et al., 1989) can be linked with the level of antioxidants in vegetables remains obscure. Other so-called 'anti-carcinogens', such as the retinoids, have been identified in food and their role in influencing human cancer is currently being investigated (Lippman and Meyskens, 1988). It may, however, be premature to speak of 'anti-carcinogens' to specific dietary components (Ames, 1983) when the exact modulating role of the macro-components such as fat, cholesterol, salt, fibre and carbohydrate intake is yet to be resolved (Davis, 1989). Both dietary micro- and macro-components may also influence the rate of cell turnover during either normal growth or regenerative hyperplasia, and this has been shown to influence cancer incidence in both animals (Cohen and Ellwein, 1990) and humans (Preston-Martin et al., 1990). Consideration of the positive and negative influence of chemicals on all factors relevant to cancer development may be necessary to elucidate the role of dietary chemicals in human cancer. Defining the term 'carcinogen'
A complicating factor when discussing carcinogenic chemicals is that whereas our understanding of the carcinogenic process and its modulating factors has improved enormously, the evolution of appropriate terminology to describe the process has been less impressive. Thus, the classic definition of a carcinogen (i.e. any substance that can cause an increase in the incidence or rate of onset of malignant tumour formation) takes no account of factors such as potency, route of exposure, or other confounding factors such as diet, stress and general health. If the evidence for carcinogenicity comes from animal studies, species differences must also be considered. The assumption behind this definition of a carcinogen is that a chemical that is labelled as a 'carcinogen' on the basis of epidemiological evidence or animal bioassay is hazardous to human health given appropriate exposure levels and circumstances. Unfortunately, this has been misinterpreted by some as meaning all exposure levels and all circumstances. Use of animal bioassays to identify carcinogens began initially with the assumption that carcinogenic chemicals could be clearly separated from noncarcinogenic chemicals and thus eliminated from the human environment, or at least controlled. Although this has been effective for the more potent carcinogens, it has become clear that many chemicals are capable of inducing turnouts in animals under particular experimental conditions that may not be relevant to human exposure. This has led to a reappraisal of how such results should be interpreted. Whether or not one should classify such chemicals as potential human carcinogens is a matter of judgement, taking into account the conditions under which tumours were induced in the animal studies and the
Carcinogenic chemicals in food dose levels used, as well as the expected route and level of exposure in humans (Abbott, 1990). Use of the term 'carcinogen' without qualification can no longer be justified, particularly in relation to the low-level chemicals in food. Many more factors need to be recognized as influencing potential human carcinogenicity, and the conditions under which a chemical may be justifiably suspected of human carcinogenic activity need to be clearly identified.
Natural and synthetic carcinogens in food Regulation of human exposure to chemicals in food has been, and continues to be, almost exclusively directed towards synthetic chemicals. This is a result of limited toxicological information on chemicals that occur naturally in food and difficulties in controlling their levels, together with the public perception that chemicals that are natural must be harmless. Although the acutely toxic components of many plants have been identified, subtle aspects of the toxicity of these and other components are not as well understood, particularly their potential for longterm effects, including cancer (Cheeke, 1989). Because of these limited data on naturally occurring chemicals, it is difficult to make a fair comparison with synthetic chemicals. Pesticides, for example, are extensively tested in animals before use and the levels permissible in crops or food products is controlled. Many of the naturally occuring chemicals in plants act as natural pesticides (Ames et al., 1990a; Beier, 1990): these include coumarins, glycoalkaloids, isoflavonoids, stilbenes and terpenoids, and their level can vary with different plant varieties and at different phases of growth. Selection for disease-resistant plants can, in some cases, be directly related to the level of naturally occurring pesticides. The advent of biotechnology may allow the possibility of altering the levels of naturally occurring plant constituents, either increasing the levels of toxic components to provide enhanced insect resistance, or decreasing these levels to reduce toxicity (IFBC, 1990). Hence the need for more information on the toxicology of these natural components. The limited animal testing that has been performed has confirmed the potential carcinogenicity of a number of naturally occurring compounds, such as caffeic acid, benzylacetate, safrole and allyl isothiocyanate (Ames et al., 1990a; Rosenkranz and Klopman, 1990) and has led to some questioning of the relatively greater emphasis placed on the toxicity testing of synthetic chemicals (Ames et al., 1990b).
Natural defence systems Animals, including humans, have evolved effective defence systems against potentially harmful dietary chemicals at the whole-animal, cellular and molecular levels. These systems operate equally well for both naturally occurring and synthetic chemicals. For FCT 30/4---E
329
those chemicals with high acute toxicity, the most important and effective barriers are the more general mechanisms of emesis and shedding of surface cells in the oesophagus, stomach and gastro-intestinal tract. For the majority of chemicals in food, however, the major defence mechanism is the inducible pathways of the liver. The oxidation reactions, which are catalysed by the numerous cytochrome P-450 enzymes, can operate on a wide range of substrates. For the majority of chemicals, oxidation reactions catalysed by P-450 enzymes result in detoxification and excretion. For some chemicals, however, oxidation reactions can result in a more highly reactive molecule capable of covalent binding to DNA to form pro-mutagenic lesions. Some chemicals may also enhance DNA damage through the formation of reactive oxygen species such as superoxide, hydroxyl, peroxyl and alkoxyl radicals (Ames, 1983; Epe et al., 1990) although the extent to which DNA is damaged in this way is unknown. Cells contain enzymes such as superoxide dismutase and glutathione peroxidase, which protect against oxidative damage. Dietary antioxidants such as r-carotene, selenium, ascorbic acid and glutathione have also been shown to reduce cell damage and, under certain circumstances, to reduce tumour incidence (Ames, 1989). DNA damage, regardless of the origin, can be repaired by a variety of DNA-repair enzymes that recognize small or bulky lesions, DNA-strand breaks or gaps. The importance of D N A repair in reducing mutation rate and maintaining cell integrity can be observed in humans who have a genetic deficiency associated with repair of damage caused by ultraviolet light (Xeroderma pigmentosum patients); such individuals have a significantly increased incidence of skin cancer. Further understanding of D N A repair mechanisms may enable the identification of individuals with increased cancer risk. DNA damage, if not repaired, may result in a mutational event, but although there is substantial evidence that mutation is an essential step for tumour induction, it is only one of many steps and is likely to be a relatively common event, compared with the occurrence of tumours. In comparing the toxicity of natural versus synthetic chemicals, a common argument is that, in terms of detoxification systems, humans are biologically better equipped to handle naturally occurring chemicals in food. This argument is not particularly convincing when we consider, first, the incredibly wide variety of chemical components in food and the enormous metabolic diversity of the liver, and secondly, that many of the foods we now consume are relatively new in evolutionary terms and for which we could never have evolved specific detoxification systems. A third point is that humans consume very different regional diets, which clearly influence cancer incidence. There is some evidence that naturally occurring dietary components may be linked with this variation (Lu et al., 1986).
330
P.J. ABBOTT
Current testing for carcinogenic chemicals Identifying those chemicals that may cause an increase in cancer incidence in humans has proved to be a difficult task for a variety of reasons. Most chemicals classified by the International Agency for Research on Cancer (IARC) as proven human carcinogens have been identified from epidemiological data and subsequently supported by experiments in laboratory animals. Indeed, the close correlation between the response of animals and humans to the potent carcinogens that were first identified has been the driving force to use animals to test other chemicals for carcinogenicity. This extensive use of animals for testing, although providing a large quantity of data, has raised new questions regarding interpretation of the results, particularly the difficulty of making species and dose extrapolations without reference to the mechanism of carcinogenesis and the route and level of human exposure (Clayson, 1987). This uncertainty is reflected in the large number of chemicals that have been classified by IARC as 'probable' or 'possible' human carcinogens. The major difficulty in using animal data has been extrapolation of dose. Humans are exposed only to low or very low dose levels whereas the same chemicals are tested in animals at relatively high dose levels. Although the high dose levels improve the statistical power of the assay, they also lead to complications that may influence the outcome. Secondary factors such as tissue damage and regenerative hyperplasia, hormone imbalance, mineral deposition and saturation of metabolic pathways can alter both the rate of tumour development and possibly the initiation of new tumours. Consideration of these factors is necessary when interpreting the results of animal bioassays. In many cases, carcinogenic mechanisms have been identified in animal studies that have no relevance at lower dose levels or in the human species. The induction of ~-2-microglobulin in male rat kidney is an example of a carcinogenic mechanism that is irrelevant to humans (Alden, 1986).
A threshold dose level for carcinogens? The question of a threshold dose for carcinogens has been a controversial one and stems from the close correlation between carcinogenicity, mutagenicity and D N A damage for those chemicals first identified as human carcinogens. The potential for mutation from a single D N A lesion, together with the clear mutagenicity of most of the potent human carcinogens, led to the idea that no safe dose existed for carcinogens. Despite advances in our understanding of carcinogenesis and the significant flaws in this argument, this theory underlies much of the fear regarding carcinogenic chemicals. The relationship between mutagenicity and carcinogenicity has undergone close examination as the number of chemicals undergoing extensive rodent carcinogenicity testing has risen. Whereas those
chemicals that are clear mutagens have generally continued to be shown to be carcinogens, the reverse relationship has not proved so reliable. Approximately half the chemicals that have given rise to tumours in either mice or rats in studies conducted by the US National Toxicology Program (NTP) were negative in a range of mutagenicity assays (Ashby and Tennant, 1988). Although there may have been some selection bias by the NTP towards non-mutagenie chemicals, the results indicate that other than mutagenic routes to cancer initiation can exist. These so-called 'non-genotoxic' carcinogens were generally positive in animal bioassays only at the higher dose levels of the range used. The mechanism of tumour formation may vary for individual chemicals but appears to involve a sustained disruption to tissue homoeostasis (Clayson, 1989). At dose levels below which such disruption occurs, no tumour formation is evident. Although D N A mutation may still occur at a later phase as part of the carcinogenic process induced by these chemicals, it does not represent the initial step in the process, and a clear threshold for the induction of cancer can be demonstrated. Examples of such chemicals that can occur in food include contaminants such as the pesticide amitraz, oestrogenic agents such as diethylstilboestrol, goitrogenic agents such as amitrole, peroxisome proliferators such as phthalates, and the artificial sweetener saccharin. For the majority of such chemicals, human exposure occurs at levels far below the demonstrated threshold for tumour induction. For chemicals with demonstrated genetic activity in the somatic cells of animals, a cautious approach is to assume a propensity to induce cancer in humans. This is based on the premise that genotoxic carcinogens have a stochastic mode of action, that is, the probability of occurrence of cancer depends only on the dose, with no threshold dose level. Given the considerable barriers associated with physical absorption, distribution, metabolism and D N A repair, it seems more likely that a threshold dose level for mutation will occur in the target organ. Establishing such a threshold, however, may be difficult, although recent advances in measuring low-level mutagenic lesions in D N A may provide some progress in determining the dose-response relationship at the low dose levels to which humans are normally exposed (Swenberg et aL, 1990). In time, it may be possible to determine a virtually safe dose for mutagenic chemicals, based on the level and rate of accumulation of D N A lesions in target organs. Evidence for the existence of thresholds for either genotoxic or nongenotoxic carcinogenic chemicals may be able to provide a more scientific basis for establishing safe exposure levels (Clayson and Arnold, 1991).
Identifying carcinogenic risk factors Diet is one of the factors that can influence human cancer levels, and individual components can have
Carcinogenic chemicals in food either a positive or negative influence. Of the potential dietary risk factors, considerable public attention to date has focused on synthetic chemicals such as additives and pesticide residues; the natural constitutents, both micro-components, including antioxidants and mutagens, and macro-components, such as fat, fibre, salt and carbohydrate intake, have received less attention. In a consideration of carcinogenic risk factors, the micro-components (both synthetic and natural) may be divided into either mutagenic or non-mutagenic chemicals. The non-mutagenic chemicals in food are unlikely to be consumed at a level capable of causing a disruption to tissue homoeostasis (as discussed previously for 'non-genotoxic' carcinogens) and thus to initiate cancer by this secondary route. Exceptions to this generalization include some of the natural fungal toxins such as the ochratoxins, the levels of which must be carefully monitored. The mutagenic chemicals in food, which include polycyclic aromatic hydrocarbons, nitrosamines and aflatoxins, are considered to be a carcinogenic risk factor, although at present it is generally not possible to establish their activity in target organs. Their potential to cause mutation must also be considered in relation to the spontaneous mutation rate. Given that spontaneous mutation may also lead to cancer initiation, it may be difficult to establish the contribution of a particular chemical to the mutagenic burden. A better understanding of spontaneous D N A damage and mutation in relation to cancer incidence may improve this situation. Progress in establishing the real risks associated with low levels of mutagenic chemicals in the diet may come from a better understanding of the capacity of the natural defence systems and also from the development of better techniques for measuring the potential D N A or protein lesions in target organs. The current dependence on long-term animal testing of individual chemicals seems unlikely to provide meaningful answers to assist risk estimation. Individual chemicals may also need to be considered in the context of the entire diet, as well as taking into account other cellular controls on cancer development. Despite much speculation, there is little evidence that the synthetic chemicals found in food are linked with human cancer. This is perhaps not surprising given that, for any chemical suspected of being carcinogenic to humans, the risk is minimized by strict measures to control exposure. On the other hand, the levels of human cancer remain high despite these controls, perhaps because other major influences have not been recognized. Indeed, it seems likely that the greatest potential for any further reduction in human cancer levels lies in a better understanding of the influence of the macrocomponents of the diet. Epidemiological evidence (Weisburger, 1987) is beginning to support the view that dietary imbalance is a risk factor for human
331
cancer. However, given the number of confounding variables, the mechanism by which such components modulate cancer incidence may take some time to elucidate. Future directions
The next decade is likely to see a greater appreciation of the multitude of complex factors that influence cancer incidence in both animals and humans. This could lead to some changes in current procedures used to identify those dietary chemicals suspected of carcinogenic activity in humans. It is to be hoped that emphasis will shift further from a study of the carcinogenicity of individual chemicals to a consideration of the interaction between dietary components in relation to their modulation of cancer incidence. Other changes that are beginning to occur include closer attention to mechanistic studies, which may identify some of the early changes in carcinogenesis. Further research on biological monitoring of exposure at the tissue and D N A level may also allow for better risk assessment through the identification of biological thresholds. Finally, improvement in communicating the scientific basis for establishing risk factors associated with human cancer will, it is hoped, lead to a better appreciation of the relative carcinogenic risk of both the natural and the synthetic components of the diet. REFERENCES
Abbott P. J. (1990) Identifying human carcinogens: the need for a multidirectional approach. Australian Cancer Society, Cancer Forum. 14, 37-40. Alden C. L. (1988) A review of unique male rat hydrocarbon nephropathy. Toxicologic Pathology 14, 109-111. Ames B. N. (1983) Dietary carcinogens and anticarcinogens: oxygen radicals and degenerative diseases. Science, New York 221, 1256-1264. Ames B. N. (1989) Endogenous DNA damage as related to cancer and aging. Mutation Research 214, 41-46. Ames B. N., profet M. and Gold L. S. (1990a) Dietary pesticides (99.99% all natural). Proceedings of the National Academy of Sciences of the U.S.A. 87, 7777-7781. Ames B. N., Profet M. and Gold L. S. (1990b) Nature's chemicals and synthetic chemicals: comparative toxicology. Proceedings of the National Academy of Sciences of the U.S.A. 87, 7782-7786. Ashby J. and Tennant R. W. (1988) Chemical structure, Salmonella mutagenicity, and extent of carcinogenicity as indices of genotoxic carcinogens among 222 chemicals tested in rodents by the US NCI/NTP. Mutation Research 204, 17-115. Beier R. C. (1990) Natural pesticides and bioactive components in foods. Reviews of Environmental Contamination and Toxicology 113, 47-137. Cheeke P. R. (Editor) (1989) Toxicants of Plant Origin. Vols 1-4. CRC Press Inc., Boca Raton, FL. Clayson D. B. (1987) The need for biological risk assessment in reaching decisions about carcinogens. Mutation Research 185, 243-269. Clayson D. B. (1989) Can a mechanistic rationale be provided for non-genotoxic carcinogens identified in rodent bioassays? Mutation Research 221, 53~7.
332
P. J. AeeoTT
Clayson D. B. and Arnold D. L. (1991) The classification of carcinogens identified in the rodent bioassay as potential risks to humans: what type of substance should be tested next? Mutation Research 257, 91-106. Cohen S. M. and Ellwein L. B. (1990) Cell proliferation in carcinogenesis. Science, New York 249, 1007-1011. Davis D. L. (1989) Natural anticarcinogens, carcinogens, and changing patterns in cancer: some speculation. Environmental Research 50, 322-340. Doll R. (1988) Epidemiology and the prevention of cancer: some recent developments. Journal of Cancer Research and Clinical Oncology 114, 447-458. Doll R. and Peto R. (1981) The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today. Journal of the National Cancer Institute 66, 1191-1308. Epe B., Hegler J. and Wild D. (1990) Identification of ultimate DNA damaging oxygen species. Environmental Health Perspectives 88, 11 I-115. Farber E. (1984) The multistep nature of cancer development. Cancer Research 44, 4217-4223. IARC (1987) IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Supplement 7. Overall evaluation of carcinogenicity: an updating of IARC Monographs volumes 1 to 42. International Agency for Research on Cancer. International Food Biotechnology Council (1990) Biotechnologies and food: assuring the Safety of foods produced by genetic modification. Regulatory Pharmacology and Toxicology 12, 1-129. Ito N., Fukushima S. and Hirose M. (1987) Modification of the carcinogenic response by anti-oxidants. In Toxicological Aspects of Food. Edited by K. Miller. pp. 253-293. Elsevier, London. Larsen G. and Poulsen E. (1987) Mutagens and carcinogens in heat-processed food. In Toxicological Aspects of Food. Edited by K. Miller. pp. 205-252. Elsevier, London.
Le Marchand L., Yoshizawa C. N., Kolonel L. N., Hankin J. H. and Goodman M. T. (1989) Vegetable consumption and long cancer risk: a population-based case control study in Hawaii. Journal of the National Cancer Institute 81, 1158-1164. Lippman S. M. and Meyskens F. L. (1988) Retinoids as anticancer agents. Progress in Clinical and Biological Research 259, 229-244. Lu S. H., Montesano R., Zhang M. S., Feng L., Luo F., Chui S. X., Umbenhauer D., Saffhill R. and Rajewsky M. F. (1988) Relevance of N-nitrosamines to eosophageal cancer in China. Journal of Cellular Physiology, Suppl. 4, 51-58. Mullaart E., Lohman P. H., Berends F. and Vijg J. (1990) DNA damage metabolism and aging. Mutation Research 237, 189-210. Overvik E. and Gustafsson J.-A. (1990) Cooked food mutagens: current knowledge of formation and biological significance. Mutagenesis 5, 437-446. Preston-Martin S., Pike M. C., Ross R. K., Jones P. A. and Henderson B. E. (1990) Increased cell division as a cause of human cancer. Cancer Research 50, 7415-7421. Rosenkranz H. S. and Klopman G. (1990) Natural pesticides present on edible plants are predicted to be carcinogenic. Carcinogenesis 11, 349-353. Swenberg J. A., Fedtke N., Fennell T. R. and Water V. E. (1990) Relationship between carcinogen exposure, DNA adducts and carcinogenesis. In Progress in Predictive Toxicology. Edited by D. B. Clayson, I. R. Munro, P. Shubik and J. A. Swenberg. pp. 161-184. Elsevier, London. 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. Weisburger J. H. (1987) On the mechanisms relevant to nutritional carcinogenesis. Preventive Medicine 16, 586-591.
