As early as the 1930s, experiments on laboratory animals revealed that diet can considerably influence the process of cancer causation and development (carcinogenesis) (Tannenbaum 1942a, 1942b; Tannenbaum and Silverstone 1953). It was several decades later, however, that the first epidemiological studies appeared to indicate that diet could play a role in human cancer. A key conference held in 1975, entitled "Nutrition in the Causation of Cancer," summarized the existing knowledge and hypotheses (Wynder, Peters, and Vivona 1975). From that moment, research in experimental systems, including animal models and epidemiological studies, increased rapidly, providing extensive information on the impact of nutritional traditions and specific macro- and micronutrients on several types of cancer. Considerable progress had already been made in several underlying sciences. For example, advances had been achieved in understanding the mechanisms of action of nutrients, the process of carcinogenesis, and the classification of carcinogens according to their mode of action (Kroes 1979; Weisburger and Williams 1991).

In particular, epidemiological studies on the international variations in incidence rates for certain cancers pointed to the existence of one or more exogenous factors that could be controlled. Observational studies had been conducted with migrants from countries with lower incidence rates to countries with higher incidence rates. A rapid increase from the lower to the higher incidence in those migrants supported the suggestion that environmental causes, and especially prevailing dietary habits, may influence the development of a number of neoplasms.

For example, in the Western world, high incidences of cancers in the lung, colon and rectum, breast, prostate, pancreas, endometrium, and ovary were observed. But these diseases were, until recently, quite rare in Japan and other Asian countries, where cancers of the stomach, esophagus, and liver were the major diseases. Yet, in migrant populations the risk of cancer shifted from the risk prevalent in their native country to that seen in the new host country, sometimes within the same generation (colon), sometimes in the next generation (breast) (Parkin 1993).

The results of experimental and epidemiological studies have led to hypotheses concerning factors involved in cancer causation. In fact, E. L. Wynder and G. B. Gori (1977) and R. Doll and R. Peto (1981) estimated that 35 percent of cancer mortality may be attributable to diet, thereby placing diet at the same "risk factor" level as smoking, which was calculated to cause about 30 percent of cancer mortality. In fact, current views suggest that dietary habits may account for 55 to 60 percent of cancer mortality. The interpretation of such multidisciplinary investigations has shed light on underlying mechanisms of cancer causation and has provided an understanding of the role of specific dietary risk factors.

In this chapter, the focus is the role of diet in cancer causation. First, attention is drawn to aspects of mechanisms in carcinogenesis. The major strengths and weaknesses of experiments in animal models and epidemiological studies are discussed, and then the major diet-related causes are examined and their suggested mechanisms and possible preventive measures described. The safest (but in the public’s perception the most dangerous) intentional additives in food are evaluated for their cancer risk, as are nonintentional contaminants, along with substances that occur naturally in food. Finally, the key role that diet may play in an integrated approach to (chronic) disease prevention is presented. Health promotion through research-based nutrition not only is important for the individual but, on a global level, may also appreciably lower the cost of medical care. A research-based healthy lifestyle is designed to allow people to "die young, as late in life as possible," as Dr. Ernst L. Wynder, president of the American Health Foundation, has said.

Tumor Models: Strengths and Weaknesses

Animal tumor models are often used in experimental designs to study the mechanisms of cancer causation, to examine the effects of modulating factors on the genesis and development of cancer, to assess therapeutic modalities, and to explore possible adverse effects.

Tumor models are specifically used to investigate etiologic and physiopathological properties or processes, especially those which, for obvious practical and ethical reasons, cannot be studied in humans. The ideal animal tumor is histologically similar to the human neoplasm of concern, and latency period, growth, and tendency to metastasize should both be predictable and resemble those of the human neoplasm. The animals should be cost effective, be easily available, and have a genetic uniformity. Variables should be controllable, thereby making it possible to investigate the influence of isolated factors (Davidson, Davis, and Lindsey 1987; Galloway 1989; Weisburger and Kroes 1994).

For research into matters of nutrition and cancer, chemically induced tumors in animals, predominantly those in rats and mice, are the most important models because they usually best mimic existing types of human cancer (Kroes et al. 1986). For most, if not all, nutrition-related human cancers, tumor model systems are available. They provide an ideal research tool for investigating the influence of individual factors, either in the initiation phase or in the promotion—
progression phase of tumor development, or in the overall process.

Possible leads from epidemiological studies can be further investigated in models, thus providing more detailed information, especially regarding risk factors and mechanisms that can be the basis for new epidemiological studies to test a presumed hypothesis. However, animal tumors are only approximations that are rarely identical to human disease. For example, relatively high dosages of genotoxic carcinogens are used to induce the tumor, which tends to distort circumstances when modulating factors are investigated.

Tumor metabolism, growth, potential for and pattern of metastasis, and clinical features can also differ from the human disease. Multiple tumors often occur in models, but seldom in humans. Moreover, the nutrition patterns of experimental animals can be quite distinct from those of humans. Nonetheless, diets can be designed in terms of macro- and micronutrient intake that mimic specific human nutritional traditions.

Animal studies are best understood as providing support for epidemiological studies. It is the integration of results from epidemiological and animal studies that provides the best insight into the etiology and growth of cancer, as well as its treatment, and brings us nearer to the ultimate goal of such research, cancer prevention.

Epidemiological Studies Concerning Nutrition and Cancer

Epidemiological research on cancer and diet seeks to associate exposure to certain dietary factors with the occurrence of cancer in selected population groups. Studies can be descriptive, reporting the occurrence of cancers in populations, in subgroups of a given population, or in a certain population over time. Observed patterns may be related to particular variables, such as diet, but although the results of such investigations are suggestive, they are certainly not definitive. Their use is especially valuable, however, in identifying populations at risk.

Correlation studies investigate the possible relationships between more-or-less crude exposure data and cancer incidence data in different populations or nations in order to generate new hypotheses. But they are generally of limited value, because national per-capita food intake and cancer incidence data are only approximations and differ from place to place in terminological definition and accuracy. Results should be used mainly as an indication of trends or relationships.

Another type of investigation is the case control study, in which the investigator is able to collect data from individuals instead of groups, and confounding variables can be controlled to a certain extent. In studies of a specific type of cancer, food intake data are collected and are compared to data similarly obtained in matched controls. When known biases and chance can be excluded, associations can be made between exposure and disease. Case control studies are relatively cheap and are of short duration but never prove causal relationships.

Much more expensive, time consuming, and elaborate are prospective cohort studies that focus on individuals in a given (large) population and establish exposure or nutrition data before the occurrence of the disease. Thus, at the time of disease occurrence, exposure data of the patients are compared to data from people not having the disease, thereby providing evidence for a possible causal relationship. Drawbacks in such studies are that the exposure assessment at any given time may not be representative of the individual’s whole life, and this is especially likely in the case of diet.

Finally, intervention studies provide the investigator with the possibility of a random assignment of subjects of a given (often high-risk) population into groups that are treated or fed differently under controlled conditions. In this type of trial, causal relationships can be established.

In investigations focused on the relationship between nutritional factors and cancer, the methods used for determining dietary intake are crucial and difficult. All dietary intake measures share certain limitations, because people vary in their abilities to estimate the amount of something they have eaten. Indeed, sometimes they may even fail to notice or report consumption of certain foods, and they usually possess insufficient knowledge about the ingredients in the foods they consume. A further handicap is that accuracy in dietary recall deteriorates over time. Finally, in the case of cancer, it is very difficult to relate varying diets to the disease because the latter has a long course of development before becoming clinically manifest (10 to 40 years).

The major strength of epidemiological studies, however, is their focus on human populations, thus avoiding the need to extrapolate from other species. They tend to afford an opportunity to examine different effects at different exposure levels, and they are always realistic, in contrast to the high exposure levels usually employed with animals (National Research Council 1982; International Agency for Research on Cancer 1990; Weisburger and Kroes 1994). In this connection, however, it should be noted that in specific investigations into the role of nutrients such as fat, the experimental design in laboratory animals usually faithfully mimics the situation of human populations at high or low risk, thus providing relatively reliable comparative data. But at the same time, as noted, it should also be recognized that the normal food patterns of laboratory animals differ from those of humans.

Mechanisms of Carcinogenesis

The concept that chemicals can induce cancer through a variety of modes of action is derived from a greater understanding of the complex processes of carcinogenesis (Williams and Weisburger 1991). Cancer causation and development involves a series of essential steps (Figure IV.F.7.1). In the first step, a reactive form of carcinogen (often produced metabolically from a procarcinogen) binds to DNA, or DNA is altered by the effective generation of hydroxy radicals. This reaction, in turn, leads to translocation and amplification of specific genes, proto-oncogenes, or a mutation in tumor suppressor genes, that translate to a distinct expression of the properties of the cells bearing such altered genes (Williams and Weisburger 1991; Ronai 1992; Miller 1994).

The property to bind to DNA is the basis for the development of specific rapid, efficient, and economical in vitro bioassays, such as a mutation assay in prokaryotic or eukaryotic cell systems. Advantage can also be taken of the presence of enzyme systems performing DNA repair; these can provide effective complementary test systems to outline the possible DNA-reactivity of chemicals (Weisburger and Williams 1991; Weisburger 1994). Moreover, the 32P-postlabeling procedure of K. Randerath and colleagues (1989) yields information about the presence of reactive carcinogen—DNA adducts.

When a chemical displays properties of reacting with DNA and inducing mutations and DNA repair in a number of cell systems, it can be considered DNA-reactive or genotoxic. Most human carcinogens are genotoxic. Other agents, such as the hormonoïd diethylstilbestrol (DES) or the hormone estradiol, may give rise to genotoxic products (Liehr 1990). The fact that a given product is genotoxic may signal that, with sufficient dosage and chronicity of exposure, it poses a cancer risk to humans.

Reactive oxygen, hydroxy radicals, or hydrogen peroxide can be generated in metabolic processes. On the other hand, there are endogenous defense mechanisms, such as catalase-destroying hydrogen peroxide or superoxide dismutase or glutathione peroxidase (in turn, involving glutathione [GSH] or sulfhydryl amino acids), that neutralize reactive oxygen. Exogenous antioxidants (such as those in vegetables, tomatoes, and tea), carotene, vitamins C and E, selenium, and sources of GSH aid in the disposition of reactive species (Harris 1991; Olson and Kobayashi 1992; Packer 1992; Sai et al. 1992).

The consequences of DNA—carcinogen interactions are beginning to be explored in terms of proto-oncogene — oncogene codon translation and amplification that can be measured by highly specific and sensitive polymerase chain reaction techniques (Brugge et al. 1991). Errors may be introduced into DNA during biosynthesis (Echols and Goodman 1991). Mutational events in tumor-suppressor genes also yield abnormal DNA, which is of growing interest (Brugge et al. 1991). The rate of cell duplication is important in generating abnormal DNA. A rapid rate decreases the chances of successful repair and is a reason that growing organisms or proliferating tissues are often more sensitive to carcinogens (Cohen and Ellwein 1992). For example, radiation exposure during the Hiroshima atom-bomb explosion caused a fourfold higher breast-cancer incidence in the 10- to 15-year-old age group (with mammary gland cells in rapid DNA synthesis and mitosis) than in younger or older groups (Land et al. 1980).

Reactive carcinogens modify not only DNA but also proteins. Both types of interactions can serve as sensitive markers for qualitative and quantitative analysis, especially the readily measured hemoglobin adducts (Brugge et al. 1991).

Nongenotoxic carcinogens and promoters cannot cause cancer without an antecedent mutational event and cell change. For example, mice with mammary tumor virus (MTV) develop mammary tumors proportional to the level of estrogen administered, but those without MTV do not, no matter what dose of estrogen is used (Highman, Norvell, and Shellenberg 1977). The action of promoters requires their presence at relatively high levels for a long time, and that action is often tissue specific. For example, bile acids are promoters of colon cancer, and very high dosages of sodium saccharin act as a promoter for cancer of the urinary bladder. The interruption of gap junctions and intercellular communication plays a key role in promotion, and promoters can be detected through this characteristic (Yamasaki 1990; Trosko and Goodman 1994).

As promoters are nongenotoxic substances, linear extrapolation for health-risk assessment seems unrealistic and, actually, scientifically improper (Kroes 1987; Williams and Weisburger 1991; Weisburger 1994). In order to better assess risk using appropriate epidemiological and biostatistical approaches, new procedures to define the mode of action of epigenetic (nongenotoxic) agents are being developed. Most likely, dose-response studies will yield a typical pharmacological S-shape response, with a definite no-effect level. This is especially important when fundamental insights into the properties of carcinogens and promoters are applied to the area of nutritional mechanisms in cancer causation.

Thus, chemical carcinogens can be classified into two main groups: (1) DNA-reactive substances that are genotoxic in appropriate test systems and (2) epigenetic (nongenotoxic) agents operating by producing some other specific biological effect as the basis for their carcinogenicity. Genotoxic carcinogens alter DNA, are mutagenic, and lead to transformed cells with neoplastic attributes. Nongenotoxic carcinogens involve other mechanisms, such as cytotoxicity, chronic tissue injury, hormonal imbalances, immunologic effects, or promotional activity.

Diet-Related Cancers

Several extensive reviews addressing diet-related neoplasms have been published in the past decades. In the Western world, cancers associated with nutrition account for a substantial percentage — about 35 to 45 percent — of premature deaths. In this section, characteristics of the different types of cancer, their established or suggested relationships with dietary factors, and their presumed mechanisms of action are described. In addition, possible measures to prevent or decrease the risk of developing the disease are discussed.1

Oral Cavity, Pharynx, and Esophagus

Cancers of the oral cavity and pharynx account for approximately 400,000 new cancer cases each year in the world. High incidences are noted in France, Switzerland, northern Italy, Central and South America, parts of Pakistan, and India. These cancers occur much more frequently in males than females, and differences between high- and low-incidence areas may be as much as 20-fold.

High-incidence rates of cancer of the esophagus are found in the so-called Asian esophageal cancer belt, which extends from eastern Iran, along the Caspian Sea, through Turkmenistan, Tajikistan, Uzbekistan, and Kyrgyzstan, and into parts of China. Except for the high-incidence areas, where the sex ratio almost equals 1:1, males show predominantly higher incidences. World incidence rates differ more than 100-fold, and globally more than 300,000 new cases occur each year.

Cancer of the esophagus is especially common among individuals who chew or smoke tobacco and drink alcoholic beverages. Consumption of alcohol alone — especially hard liquor — seems to be a risk factor as well (Seitz and Simanowski 1991; Castelleto et al. 1994). Smoking and alcohol consumption have a synergistic effect on carcinogenesis in the upper alimentary tract. In Asian and African populations, dietary deficiencies of zinc, riboflavin, vitamins A and C, manganese, and molybdenum may play a role, as well as mycotoxins, bracken fern, opium pyrolysates, and betel quids. The consumption of salted fish is an established risk factor in southern Chinese populations, probably because of the formation of specific nitrosamines (Craddock 1992; Zeng et al. 1993). Consumption of very hot beverages, along with the use of substances that irritate the oral cavity, pharynx, and esophagus, all of which lead to increased cell proliferation, may enhance the incidence of neoplasia.

However, substantial differences in incidence between high- and low-risk areas indicate that there exists considerable potential for prevention. Frequent consumption of fresh fruits and vegetables, as well as tea, appears to be associated with a lower risk for these types of cancer. The potential reduction has been estimated to be around 75 percent (Negri et al. 1993). Preventive measures involve the avoidance of tobacco and very hot beverages, along with moderate alcohol use and a well-balanced diet that includes a sizable increase in the regular consumption of vegetables and fruits (Block, Patterson, and Subar 1992).

Stomach

In the 1980s, gastric cancer was still considered to be the most common cancer in the world. Indeed, with almost 700,000 new cases per year, it represented approximately 10 percent of all cancers. Differences between high- and low-incidence areas vary by 40-fold. However, a large decrease in rates has occurred in most populations during the last four to five decades, indicating a reduction in exposure to tissue-specific carcinogens and/or the introduction of a protective agent.

Males suffer approximately twice the incidence and mortality of females, although the sex ratio is not constant by age group. The sex ratio equals 1:1 in people under the age of 30, but the disease is rare in this group. High-risk populations usually consume considerable quantities of pickled vegetables, dried salted fish, smoked fish, and other smoked, salted, and dried foods. Consumption of certain salted and pickled fish has yielded high levels of mutagenicity and evidence of carcinogenicity. One of the mutagens present has been identified as 2-chloro-4-methylthiobutanoic acid. This finding was totally unexpected, because in the past, nitroso compounds were associated with stomach cancer (Chen et al. 1995).

By contrast, a negative association has been established between mutagenicity and the regular intake of green leafy vegetables and citrus fruits. Laboratory experiments show that vitamins C and E block the formation of mutagens when fish is treated with nitrite, mimicking pickling (Weisburger 1991).

Infection with Helicobacter pylori and associated conditions, such as atrophic gastritis, ulceration, partial gastrectomy, bile acid reflux, and pernicious anemia, are additional risk factors. Several of these increase cell-duplication rates, rendering the gastric cells more sensitive to genotoxic carcinogens.

A high level of consumption of salted, pickled, or smoked food was once customary in the Western world. However, better access to home refrigeration, improved and cheaper transport — and, therefore, increased availability — of fresh fruit and vegetables seems to correlate well with the decline of this type of cancer (Howson, Hiyama, and Wynder 1986; Weisburger 1991). The relevant mechanism begins with the development of atrophic gastritis due to the cytotoxic activity of salt and vitamin deficiencies. The consequent decrease in gastric acidity permits uninhibited bacterial growth. Bacterial growth then converts dietary nitrates to nitrites, which are further metabolized into carcinogenic nitroso derivatives or reactive carcinogens (Correa 1992; Chen et al. 1995). Because vitamins C and E are known to be effective inhibitors of nitrosation, it is plausible that an increased intake of these vitamins, or foods containing them, should reduce the risk of gastric cancer by inhibiting nitrosation.

Preventive measures are the introduction of food refrigeration, the reduction of salt and pickled food intake, and an increased consumption of fruits and vegetables. Especially in areas with high prevailing environmental nitrate levels, vitamin C and vitamin E supplementation may be useful for preventing formation of nitrite-derived reactive carcinogens and reducing nitrite produced by conversion of nitrate in the mouth. The preventive potential has been estimated to be about 50 percent but may well be much higher.

Colon and Rectum

Approximately 600,000 new cases of colorectal cancer are diagnosed worldwide each year. It is particularly a disease of the developed countries, which to some extent reflects increasing life expectancy. Differences in incidence may be 60-fold. The lowest incidence rates are found in Africa and Asia, although incidences are rising, especially in areas where the risk was formerly low, as in Japan. Colon cancer affects the sexes equally. The distribution for rectal cancer is similar to that for colon cancer. Incidences are usually lower, and there is a male—female ratio of 1.5:2.0, especially in high-incidence areas.

Epidemiological evidence suggests that (Western) lifestyle is an important determinant of risk for colorectal cancer: Migrants to Western countries acquire a higher risk for the disease in the first generation, and Mormons and Seventh Day Adventists enjoy a low risk. Familial polyposis, ulcerative colitis, and Crohn’s disease are identified risk factors for colon cancer, but these are uncommon conditions.

Diets high in fats and low in fiber and vegetables are associated with increased risk for colon cancer. A fat—fiber interaction has been suggested, and the type of fiber is important as well (Kroes, Beems, and Bosland 1986; Weisburger 1992). It is interesting to note that some polyunsaturated fats found in fish and some vegetable seeds inhibit colon cancer formation. Moreover, olive oil intake, as in the Mediterranean countries, does not increase the risk of the nutritionally linked cancers or heart disease, a fact also documented in animal models (Reddy 1992). An inverse relationship has also been found for the consumption of fruits and vegetables, as well as for calcium intake and regular exercise; the same is true of coffee and tea for colon and rectal cancer, respectively (Baron, Gerhardson de Verdier, and Ekbom 1994).

Experimental and epidemiological research has revealed that bile acids promote cancer formation (Reddy 1992). The case is similar with alcohol, especially for rectal cancer, perhaps accounting for the higher male-to-female ratio (Seitz and Simanowski 1991). Recent surveys also indicate that intake of heavily fried or grilled meat and gravies is positively related to colorectal cancer, suggesting that chemicals produced during the frying or grilling of meats (heterocyclic aromatic amines) may be the initiating carcinogens, particularly for breast, colon, and, perhaps, prostate and pancreatic cancer (Adamson et al. 1995).

Suggested mechanisms in colon cancer development are the increased bile acid concentrations in individuals consuming high levels of many types of dietary fat. The higher concentrations of bile acids may lead to increased turnover of the epithelial cells of the intestines, reflecting increased risk of carcinogen—DNA adducts to cause translocation and amplification of abnormal genes or mutated tumor-suppressor genes. This phenomenon is inhibited by increased dietary calcium. The toxicity of bile acids is also reduced at a lower luminal pH. Alcohol may act as a promoter at the level of the rectum, and its metabolite acetaldehyde, which occurs at higher concentration in the rectum, may induce cytotoxicity, thus leading to increased cell proliferation and turnover.

Dietary fibers at adequate concentrations are thought to dilute and particularly bind the genotoxic agents present in the gut, decreasing fecal mutagenic activity. Fibers also modify the metabolic activity of the gut flora and lower luminal pH. Wheat bran increases the bulk of the gut contents, thus diluting bile acids and decreasing their adverse effect on the mucosal lining of the bowel (Reddy 1992). Regular physical exercise also lowers transit time of the luminal contents and appears to decrease risk of colorectal cancer. Fruits and vegetables generally reduce cancer risk through several mechanisms (Block et al. 1992). They provide fibers and antioxidants that can detoxify active genotoxins and also contain a number of chemopreventive agents, such as indole derivatives, that are anticarcinogenic. Tea antagonizes the effect of heterocyclic amines present in fried or broiled meats, which are thought to be carcinogens for the colon.

Potential reduction of colorectal cancers through prevention has been estimated at 35 percent, mainly of distal colon cancer and rectal cancer. The risk factors for proximal colon cancer are not well known, although general recommendations for lower risk may apply to colorectal cancer overall. This would involve a low fat intake (20 to 25 percent of calories), use of monounsaturated fats such as olive oil, an adequate fiber intake (25 to 30 g/day), moderate alcohol consumption (an average of 2 drinks/day), fish 2 to 3 times a week, an increased calcium (lowfat milk or yoghurt) intake (1,200 to 1,500 mg), increased consumption of vegetables and fruits (ideally 5 to 9 servings/day), tea (4 to 5 cups per day), and regular exercise.

Breast

Breast cancer is the third most common cancer in the world; every year about 600,000 new cases are detected, which is about 9 percent of the global cancer burden. It is important to distinguish between premenopausal breast cancer, in which diet plays a minor role (except for some protection afforded by consumption of vegetables and fruits, including soy products), and peri- and postmenopausal disease, in which diet may exert important controlling effects. Breast cancer occurs almost exclusively in women, and in high-risk areas (North America and western Europe), the incidence is about 4 to 30 times higher than in low-risk areas like China, Japan, and Sri Lanka, although there has been an appreciable increase in Japan during the last decade — the result of a westernization of dietary customs.

Important risk factors for breast cancer are a family history of the disease, a low number of offspring, avoidance of breast feeding of infants, a late age at first pregnancy, an early menarche, a late age at menopause, and high consumption of fats (about 30 to 40 percent of calories) and, possibly, alcohol. During the last decade, increasing evidence has been adduced indicating that there is an inverse relationship between breast cancer and increased intake of vegetables and fruits. Food antioxidants (such as selenium, retinoids, and polyphenols), as well as bran cereal fibers, have been suggested as inhibiting factors. Obviously, endocrine factors are important in breast cancer development. Fat may increase breast cancer risk by its control of hormonal regulation. In addition, high fat and high energy intakes, coupled with lack of exercise, lead to obesity, a possible contributory factor in breast cancer in postmenopausal women.

Obesity, however, seems inversely related to the risk of breast cancer in premenopausal women. F. De Waard has developed a unifying concept on the etiology of breast cancer, which focuses on the events that occur during adolescence and early reproductive ages (see Weisburger and Kroes 1994). He has suggested that preneoplastic lesions develop at early ages, from 15 years onward. Several factors, such as nutritional status, high fat intake, low consumption of protective vegetables, fruit, and fibers, along with reproductive life, interact in inducing a long period of cell proliferation without sufficient differentiation in the breast.

On the other hand, early pregnancy and long-term lactation will raise the differentiation of cells, thus limiting the proliferation of less differentiated cells, the latter being more vulnerable to genotoxic attack. Fat may also influence the immune system, increase prostaglandin synthesis, and increase membrane fluidity, all phenomena bearing on the promotion and growth of neoplastic cells. Therefore, the appropriate dietary preventive measures are avoidance of heavily fried or broiled meats, a limited fat intake (possibly as low as 20 to 25 percent of total calories), preference for monounsaturated fats such as olive oil, an increased intake of vegetables, fruits, tea, and insoluble bran cereal fiber, and an energy intake that balances energy need with the avoidance of obesity.
In this latter connection, an increase of exercise
has been shown to lower risk and assist in weight control.

Endometrium and Ovary

Endometrial cancer strikes approximately 150,000 women in the world each year, with tenfold differences in incidence, depending on location. High incidences are found in Argentina, the United States, Canada, and western Europe, whereas a low incidence has been noted in Asian populations. Identified risk factors are, in particular, endogenous estrogen and higher amounts of exogenous hormones employed for the management of menopausal and postmenopausal symptoms. Obesity and fat consumption are also associated with increased risk. Estrogen therapy, as practiced for postmenopausal symptoms between 1960 and 1975, has been documented as a causal element for endometrial cancer, most probably because it was given in relatively large dosages and was not balanced by progesterone. If the action of limited amounts of estrogens is balanced by progesterone, cancer risk is decreased.

The role of obesity or high fat consumption in endometrial cancer may be explained by the fact that fat cells produce estrogen, which itself is a key effector in neoplastic development through its specific effect on endometrial tissue and on overall endocrine balances. As dietary factors may be responsible for an appreciable percentage of cases, limited fat intake and avoidance of excessive energy intake are suggested preventive measures. Regular exercise, likewise, constitutes a protective element.

Ovarian cancer is common in western Europe and North America, whereas it has a low frequency in Indian, Japanese, and other Asian populations. Unlike that of many other types of cancer, the incidence of ovarian cancer in the Western world has remained rather constant over time. The risk factors for ovarian cancer are the same as those for breast and uterine cancer, meaning a positive association with endocrine factors and dietary fat intake and a negative association with parity and elements that suppress ovulation. Thus, oral contraceptives may substantially reduce the risk of ovarian cancer. Limited fat intake (perhaps 20 to 25 percent of calories or less) and consumption of vegetables and fruits are suggested as preventive measures.

Pancreas

Pancreatic cancer occurs more frequently in developed countries, comprising approximately 3 percent of the worldwide cancer burden. The disease, however, is increasing in incidence over time and has a very high mortality rate because of late diagnosis and, thus, has low success in therapy. Every year, approximately 140,000 new cases are diagnosed. In the last 40 years, pancreatic cancer incidence has doubled in western Europe and quadrupled in Japan (Hirayama 1989).

Tobacco smoking has been implicated as a major risk factor, which can explain the increasing incidence, especially in those countries where the pancreatic cancer incidence is still relatively low. Convincing evidence also exists from experimental animal research that carcinogens from tobacco and a high fat intake are positively related, whereas caloric restriction, selenium, and retinoids are inversely related. Of interest is the role shown by trypsin inhibitors in pancreatic carcinogenesis in experimental animals. These trypsin inhibitors do reduce trypsin levels in the gut, stimulating the secretion of cholecystokinin (CCK) as a feedback phenomenon. CCK stimulates pancreatic growth, thus promoting pancreatic carcinogenesis. Trypsin inhibitors, present in soy proteins, are heat labile. Soy proteins are high-quality foods, but they should be incorporated in foods and cooked (Watanapa and Williamson 1993).

Epidemiological research reveals a positive relationship for dietary fat, fried or grilled meats, and, possibly, alcohol or cholesterol, whereas an inverse relationship has been observed for caloric restriction, omega-3 fatty acids (fish and some seeds like flax seed), and fresh fruits and vegetables (Bueno de Mesquita 1992). Preventive potential has been estimated to be 70 percent. Cessation of tobacco smoking, moderate alcohol use, low fat consumption, and increased intake of vegetables and fruits are the main measures for prevention. This is particularly important because of its grim prognosis. Thus, control is optimal through prevention by lifestyle adjustment.

Prostate

Prostate cancer is the fifth most common cancer among males, and especially predominant in older males. Approximately 240,000 new cases of clinical invasive prostate cancer occur each year, and high-incidence areas are northwestern Europe and North America; in the latter, African-Americans have a particularly high incidence. Low rates are found in India, China, and Japan. There exists a 50-fold difference between populations with the highest rates of prostate cancer (blacks in Detroit, Michigan) and populations with the lowest incidence (Asians in Shanghai, China) (Nomura and Kolonel 1991). Endocrine factors may play a role in prostate carcinogenesis, but geographic pathology indicates that dietary factors are probably also important. Populations with a tradition of high fat and high protein intake have a high risk. The diet controls the endocrine balance.

Negative associations have been suggested for vitamin A, beta-carotene, vegetables, fruits, selenium, fish, and fiber. Sugar and egg consumption are weakly positive (Bosland 1986). Genetic, sexual, and dietary factors seem to play a role in prostate carcinogenesis, indicating a multifactorial process. As is true for other endocrine-controlled neoplasms, a dietary regime low in fat and rich in vegetables and fruits, coupled with regular exercise, may contribute to lower risk irrespective of sexual and genetic elements (Wynder, Rose, and Cohen 1994).

Lung

It is surprising to note that more and more data have become available to indicate that lung cancer is influenced by dietary factors. Clearly, the disease is associated with cigarette smoking, but since E. Bjelke (1975) and G. Kvale, Bjelke, and J. J. Gart (1983) found in metabolic epidemiological studies that smokers with a higher level of vitamin A in plasma had a lower risk of lung cancer, more attention has been given to dietary factors (Ziegler et al. 1992; Le Marchand et al. 1993). Also, for humans, an inverse relationship between lung cancer development and fruit and vegetable intake has been observed, whereas other data suggest a positive relationship between dietary fat intake and lung cancer (Wynder, Taioli, and Fujita 1992). In addition, the antioxidants in tea may provide a protective effect. Currently, there are more smokers in Japan than in the United States or the United Kingdom, but the incidence of lung cancer is lower in Japan. It has been suggested that the Japanese have a lower risk because of a lower total fat intake and more frequent intake of fish, soy foods, and tea.

Therefore, although the first recommendation should be to quit smoking — or, in fact, never to start — an increased intake of fruits and vegetables (especially those containing retinoids) and also of fish, soy-derived foods, and tea, coupled with a lowered fat consumption, may serve as preventive measures and could be particularly appropriate for ex-smokers.

Food Additives, Contaminants, and Natural Toxins

For decades, the possibility of cancer risks from food additives and contaminants has been widely publicized, especially in the developed countries, where there has been an increase in the addition of various substances to food for preservative and commercial purposes. Thus, food additives and contaminants are viewed by many as a major threat to human health — and one that may cause cancer. Scientific information, however, shows exactly the opposite: Food additives are safer than everyday traditional nutrients, and the same is true for most contaminants (Miller 1992; Weisburger 1994; Weisburger and Kroes 1994).

Such opposite perceptions may be explained by the misinterpretation of epidemiological reports in the late sixties, when the term "environmental" (as in "environmental factors") was used to account for major causes of cancer. In fact, what was meant was as lifestyle factors, but the general public (and especially the news media) misinterpreted this to mean synthetic chemicals, including food additives and contaminants.

In addition, several episodes have enhanced this misconception, as, for example, when certain food additives (that is, some food dyes in Western countries and the preservative AF-2 in Japan) were first permitted and later correctly withdrawn because of their demonstrated carcinogenicity in animals (Sugimura 1992). Regulatory action, especially in the United States, aimed at such chemicals as sodium saccharin and cyclamate, further deepened public suspicion. Yet the latter substances are now considered safe, at least at the normal intake levels that humans experience. In fact, certain substances with antioxidant properties, which are used as food additives, are even believed to reduce cancer risk. Thus, Wynder and Gori (1977), as well as Doll and Peto (1981), have estimated that cancer mortality from food additives ranges from —5 to +2 percent, the negative score specifically addressing the beneficial aspects of antioxidants used in foods.

Additives are used to improve the stability and storability of foods, as well as their flavor, consistency, appearance, texture, and nutritional quality. In certain cases, they are a necessity, such as in the case of preservatives that prevent food-borne microbial infections. And in any event, the risk of disease from food additives today is minimal, because efficient and effective control practices are available and applied to ensure safety.

Contaminants of human-made origin are, like food additives, extensively tested in animals before use, and the levels permissible in crops are well controlled internationally. Thus, the margin of safety for pesticide residues in food runs usually between 1,000 and several millions, whereas for several macro- and micronutrients, the margin of safety is as small as 2 to 10 (Kroes in press). In fact, B. N. Ames and colleagues (1990, 1992) have listed a number of naturally occurring substances in food that, because of uncontrolled exposure, provide much more concern for cancer risk than synthetic chemicals. About half of such natural chemicals that have undergone standard high-dose animal cancer tests proved to be animal carcinogens, such as the mold-generated hepatic carcinogen, aflatoxin (International Agency for Research on Cancer 1993b). In addition, as noted, powerful carcinogens are formed during the cooking of meats and during the salting and pickling of some fish and meats.

The natural defenses of humans, however, may make them capable of detoxifying low doses of most toxins, whether synthetic or natural. For example, despite a continuing low-level presence of aflatoxin B1 in some foods, the incidence of primary liver cancer in the United States and Europe is not significant. Yet it is quite high in parts of Africa and China, where the dietary contamination is appreciable and where more people carry the hepatitis B antigen, potentiating the action.

Certainly, in light of the foregoing, it seems relevant to invest more research capacity in the identification of possible risks and benefits of naturally occurring substances. This is especially true because many are also known to possess anticarcinogenic properties — properties that are believed to be the reason for the inverse relationship between several cancers (and heart diseases) and the regular intake of vegetables, fruits, and tea.

Food preparation has entailed cancer risk in the past and will continue to do so in the future. Preservation methods, for example, such as the use of salt or pickling solutions, are associated with a high risk of stomach cancer and in some areas, such as China, with cancer of the esophagus. Salted fish causes nasopharyngeal cancer, and salt and high nitrate (saltpeter) concentrations in several meat products can lead to the formation of carcinogenic nitroso compounds, or of the chloro analog of methionine, either in the food itself or in the stomach. Salt is cytotoxic to the gastric mucosa, translated by increased cell duplication rates and, in turn, to more efficient carcinogenesis. Some salted, pickled foods contain direct-acting mutagens thought to be gastric carcinogens (Weisburger 1992; Chen et al. 1995). Salt, not balanced by potassium from vegetables, and calcium from dairy products is also a cause of hypertension and stroke. In Japan (Sugimura 1992) and in Belgium (Joossens, Hill, and Geboers 1985), formal plans were introduced to lower salt intake by people.

Charcoal-broiled meats or fish have at their surface polycyclic aromatic hydrocarbons that are established animal carcinogens. But it is important to note that the ordinary cooking (broiling, frying) of meats or fish can produce powerful mutagens, consisting of about 19 heterocyclic amines (also established animal carcinogens) for specific target organs. They are believed to be the key carcinogens causing increased incidence of several human cancers, such as those in the breast, prostate, colon, and pancreas. Certainly it has been shown that those who generally eat well-done meat increase their risk of colon cancer. The formation of heterocyclic amines during the heating of meats can be reduced by preliminary brief microwave cooking (removing essential creatinine) or by the addition of antioxidants, soy protein, or the indole amino acids tryptophan and proline, which all compete with creatinine in the so-called Maillard reaction, forming heterocyclic amines (Weisburger and Kroes 1994).

Prevention: An Integrated Approach

A substantial amount of solid epidemiological and experimental evidence indicates that the majority of human cancers, and indeed many other chronic diseases, such as heart disease, hypertension, and adult-onset diabetes, are largely preventable. Complex causes have been, or are being, identified, and the underlying mechanisms elucidated. Control of many major diseases of humankind in the past, such as scurvy, pellagra, rickets, polio, smallpox, rabies, and tuberculosis, has been achieved by prevention strategies. Therefore, a clear and balanced prevention approach to the effective control of human cancers (and other chronic illnesses, such as cardiovascular diseases) is likely to be successful as well.

Experience in the past two decades with other chronic ailments, such as cardiovascular diseases, indicates that the application of sometimes even simple measures can have a considerable impact on their outcome (Meyskens 1992). In fact, cancer prevention runs a decade behind the scientific understanding of the disease, as a number of lifestyle-associated factors contributing significantly to cancer risk are well known. Tobacco and nutritional traditions, in particular, and — to a lesser extent — radiation, some chemicals, and certain viruses, are documented, avoidable risk factors.

Wynder and Gori (1977), Doll and Peto (1981), and J. H. Weisburger (1992) have contributed substantially to the evidence of preventive potential for many types of cancer through their listing and documentation of avoidable risks of cancer. Unfortunately, not only are such concrete factors as tobacco, diet, lifestyle, and radiation contributing to the cancer burden, but poverty does as well, because the need for essential lifestyle changes has been difficult to communicate effectively to the lower socioeconomic groups (Tomatis 1992).

Cancer prevention programs should be based on reliable epidemiological and laboratory evidence, and on ethical and moral responsibility, and ought to specify clearly achievable outcomes in mortality and morbidity reduction. Moreover, they should be integrated with other chronic-disease-prevention programs. The multifactorial elements, such as nutritional traditions, sedentary habits, and tobacco use, which represent a risk for diseases like coronary heart disease, hypertension, stroke, obesity, and many neoplastic diseases, have to do with lifestyle. In fact, these factors are the major causes and modulators of these diseases.

Table IV.F.2.1 depicts several realistic actions to take in order to lower the risk for certain common diseases. In the Western world, nutritional traditions with a relatively high fat intake (38 to 46 percent of energy intake), low cereal fiber, and low vegetable and fruit consumption, along with a lack of regular physical exercise, are associated with high incidence and high cost of the management of chronic diseases, such as cardiovascular diseases, diabetes, obesity, and specific types of cancer. In the Far East and in Central and South America, prevailing illnesses seem to stem from other nutritional traditions, such as the use of highly salted and pickled foods and a limited variety in diet. Currently, the changing nutritional habits in Japan to a Western style parallel an increase in heart disease and the kind of cancers common in the Western world. This provides strong additional support for the thesis that dietary customs and specific chronic diseases are related.

Today’s knowledge enables us to recommend a healthy dietary regime in which fats should be replaced by complex carbohydrates (starches) that should provide around 70 to 75 percent of the calories needed for energy. Furthermore, a protein intake of between 10 to 15 percent from animal and vegetable sources (more in growing young children, less in older individuals) is recommended. Fats should consist of a fair proportion of monounsaturated oils, such as olive oil or canola oil, and omega-3 fatty acids as found in fish and some seeds, like flaxseed.

Excessive salting, pickling, and smoking of food as a mode of preservation should, ideally, be abandoned, and foods should be preserved by refrigeration or freezing, or eaten fresh. Caloric intake should be equal to energy need as an effective means to avoid obesity. The intake of fruits and vegetables should be increased considerably — ideally to more than 5 servings per day. Bran cereal fibers, or breads baked with high-fiber flour, increase stool bulk, avoid constipation, and lower the risk of colon and breast cancer and perhaps other diseases. A low intake of total salt, 3g/day or less, and adequate calcium (1,000 to 1,500 mg) and magnesium (300 mg) are beneficial; less is needed on a low-protein diet. Moderate, but regular, physical exercise is also part of a healthy lifestyle. Alcohol consumption ought to be moderate, but adequate fluid intake (2 to 2.5 liters daily for adults) is essential for maintenance of physiologic functions. Tea, an extract of the plant Camellia sinensis, is, after water, the second-most-used beverage in the world. Because it is made with boiling water, it is sterile even if the water source is not pure. Tea is rich in fluoride, potassium, and especially in antioxidants that lower the risk of coronary heart disease and many types of cancer (Weisburger 1996; Weisburger and Comer this volume).

These recommendations will contribute to better health by lowering the risk for major chronic diseases. In education and in medical practice, emphasis must be placed on the importance and the efficacy of available methods for chronic disease prevention. One task of practicing nutritionists and home economists ought to be that of devising practical recipes for cooks to incorporate the essence of a new, health-promoting lifestyle that the public will find attractive. Indeed, it is essential to devise appealing dishes and drinks for the public that are also designed for chronic disease prevention, and when such preventive approaches are successful, health-care costs should decrease. Good health into old age is not only a desirable goal for the individual but may have major economic savings for the population at large. To repeat our earlier quotation, the ultimate goal, as expressed by Ernst Wynder, should be "to die young, as late in life as possible" (Wynder et al. 1994).

Robert Kroes
J. H. Weisburger

Research in Weisburger’s laboratory is supported by USPHS-NIH grants and contracts from the National Cancer Institute, grants from the Tea Trade Health Research Association and the American Cancer Society, and gifts from Texaco, Inc., and the Friends Against Cancer Team.

Note

Bibliography

Adamson, R. H., J. A. Gustafsson, N. Ito, et al. 1995. Heterocyclic amines in cooked foods. Proceedings of the 23rd International Symposium of the Princess Takamatsu Cancer Research Fund. Princeton, N.J.

Ames, B. N., and L. S. Gold. 1992. Animal cancer tests and cancer prevention. Journal of the National Cancer Institute Monographs, no. 12: 125—32.

Ames, B. N., M. Profet, and L. S. Gold. 1990. Nature’s chemicals and synthetic chemicals: Comparative toxicology. Proceedings of the National Academy of Sciences 87: 7782—6.

Baron, J. A., M. Gerhardson de Verdier, and A. Ekbom. 1994. Coffee, tea, tobacco, and cancer of the large bowel. Cancer Epidemiology, Biomarkers, Prevention 3: 565—70.

Bjelke, E. 1975. Dietary vitamin A and human lung cancer. International Journal of Cancer 15: 561—5.

Block, G., B. Patterson, and A. Subar. 1992. Fruit, vegetables and cancer prevention: A review of the epidemiological evidence. Nutrition and Cancer 18: 1—29.

Bosland, M. C. 1986. Diet and cancer of the prostate: Epidemiologic and experimental evidence. In Diet, nutrition, and cancer: A critical evaluation, Vol. 1, ed. B. S. Reddy and L. A. Cohen, 125—49. Boca Raton, Fla.

Brugge, J., T. Cuppan, E. Harlow, and F. McCormick. 1991. Origins of human cancer; a comprehensive review. Plainview, N.Y.

Bueno de Mesquita, H. B. 1992. On the causation of cancer of the exocrine pancreas; a population based case-control study in the Netherlands. Thesis, University of Utrecht.

Castelleto, R., X. Castellsague, M. Muñoz, et al. 1994. Alcohol, tobacco, mate drinking and esophageal cancer in Argentina. Cancer Epidemiology, Biomarkers, Prevention 3: 357—64.

Chen, W., J. H. Weisburger, E. S. Fiala, et al. 1995. 2-Chloro-4- Methylthiobutanoic acid, a newly identified mutagen in salt, nitrite-treated fish. Nature 374: 599.

Cohen, S. M., and L. B. Ellwein. 1992. Risk assessment based on high dose animal exposure experiments. Chemical Research in Toxicology 5: 742—8.

Committee on Diet and Health, Food and Nutrition Board. 1989. Diet and health: Implication for reducing chronic disease risk. Washington, D.C.

Correa, P. 1992. Human gastric carcinogenesis: A multistep and multifactorial process. First American Cancer Society Award Lecture on Cancer Epidemiology and Prevention. Cancer Research 52: 6735—40.

Craddock, V. M. 1992. Aetiology of oesophageal cancer: Some operative factors. European Journal of Cancer Prevention 1: 89—103.

Davidson, M. K., I. K. Davis, and J. R. Lindsey. 1987. Requirement and selection of an animal model. International Journal of Medical Sciences 23: 551—5.

Doll R., and R. Peto. 1981. The causes of cancer: Quantitative estimate of avoidable risks of cancer in the United States today. Journal of The National Cancer Institute 66: 1191—308.

Dutch Nutrition Council. 1986. Factoren in de voeding en het ontstaan van kanker (Dietary factors in relation to the causation of cancer). Dutch Nutrition Council Report. The Hague.

Echols, M., and M. F. Goodman. 1991. Fidelity mechanisms in DNA replication. Annual Review of Biochemistry 60: 477—511.

Galloway, D. J. 1989. Animal models in the study of colorectal cancer. Cancer Surveys 8: 170—88.

Harris, E. D. 1991. Regulation of anti-oxidant enzymes. Federation of American Societies for Experimental Biology Journal 6: 2675—83.

Highman, B., M. J. Norvell, and T. E. Shellenberg. 1977. Pathological changes in female C3H mice continuously fed diets containing diethylstilbestrol or 17-beta-estradiol. Journal of Environmental Pathology and Toxicology 1: 1—30.

Hirayama, T. 1989. Epidemiology of pancreatic cancer in Japan. Japanese Journal of Clinical Oncology 19: 208—15.

Howson, C. P., T. Hiyama, and E. L. Wynder. 1986. Decline of gastric cancer: Epidemiology of an unplanned triumph. Epidemiology Reviews 8: 1—27.

International Agency for Research on Cancer. 1990. Cancer: Causes, occurrence and control. International Agency for Research on Cancer Scientific Publication no. 100. Lyon.

1993. Some naturally occurring substances, food items and constituents, heterocyclic aromatic amines and mycotoxins. Monographs on the evaluation of carcinogenic risk to humans, no. 56. Lyon.

Joossens, J. V., M. J. Hill, and J. Geboers, eds. 1985. Diet and human carcinogenesis. Amsterdam.

Kroes, R. 1979. Animal data, interpretation and consequences. In Environmental carcinogenesis, ed. P. Emmelot and E. Kriek, 287—302. Amsterdam.

1986. Diet in relation to cancer. In New concepts and developments in toxicology, ed. P. L. Chambers, P. Gehring, and F. Sakai, 91—105. New York.

1987. Contribution of toxicology towards risk assessment of carcinogens. Archives of Toxicology 60: 224—8.

In press. Risk assessment: Past, present and future. Proceedings of the 2d Princess Chulabhorn Science Congress. Thailand.

Kroes, R., R. B. Beems, M. C. Bosland, et al. 1986. Nutritional factors in lung, colon and prostate carcinogenesis in animal models. Federation Proceedings 45: 136—41.

Kvale, G., E. Bjelke, and J. J. Gart. 1983. Dietary habits and lung cancer risk. International Journal of Cancer 31: 397—405.

Land, C. E., J. D. Boice, C. E. Shore, et al. 1980. Breast cancer risk from low dose exposure to ionizing radiation: Results of parallel analysis of three exposed populations of women. Journal of the National Cancer Institute 65: 353—76.

Le Marchand, L., J. H. Hankin, L. N. Kolonel, et al. 1993. Intake of specific carotenoids and lung cancer risk. Cancer Epidemiology, Biomarkers, Prevention 2: 183—7.

Levi, F., F. Lucchini, and C. La Vecchia. 1994. Worldwide patterns of cancer mortality, 1985—1989. European Journal of Cancer Prevention 3: 109—43.

Liehr, J. G. 1990. Genotoxic effects of estrogens. Mutation Research 238: 269—76.

Meyskens, F. L. 1992. Strategies for prevention of cancer in humans. Oncology 6 (supplement): 15—24.

Miller, S. A. 1992. Food additives and contaminants. In Toxicology: The basic science of poisons. Fourth edition, ed. L. J. Casarett and J. Doull, 819—53. New York.

1994. A brief history of chemical carcinogenesis. Cancer Letters 83: 9—14.

National Research Council. 1982. Diet, nutrition, and cancer. Washington, D.C.

Negri, E., C. La Vecchia, S. Franceschi, and A. Tavani. 1993. Attributable risk for oral cancer in northern Italy. Cancer Epidemiology, Biomarkers, Prevention 2: 189—93.

Nomura, A. M. Y., and L. N. Kolonel. 1991. Prostate cancer: A current perspective. American Journal of Epidemiology 13: 200—27.

Olson, J. A., and S. Kobayashi. 1992. Antioxidants in health and disease: An overview. Proceedings of the Society for Experimental Biology and Medicine 200: 245—7.

Packer, L. 1992. Interactions among antioxidants in health and disease: Vitamin E and its redox cycle (43433). Proceedings of the Society for Experimental Biology and Medicine 200: 271—6.

Parkin, D. M. 1993. Studies of cancer in migrant populations. In Cancer in Italian migrant populations, eds. M. Geddels, D. M. Parkin, and M. Khlat et al. Lyon: International Agency for Research on Cancer, Scientific Publication no. 123, 1—10.

Reddy, B. S. 1992. Animal experimental evidence on macronutrients and cancer. In Macronutrients: Investigating their role in cancer, ed. M. S. Micozzi and T. M. Moon, 33—54. New York.

Reddy, B. S., and L. A. Cohen. 1986. Diet, nutrition, and cancer: A critical evaluation. In Macronutrients and cancer, Vol. 1. Boca Raton, Fla.

Ronai, Z. 1992. Ras oncogene detection in preneoplastic lesions: Possible applications for diagnosis and prevention (Review). Oncology Research 4: 45—8.

Sai, K., T. Umemura, A. Tagaji, et al. 1992. The protective role of glutathione, cysteine and vitamin C against oxidative DNA damage induced in rat kidney by potassium bromate. Japanese Journal of Cancer Research 83: 45—51.

Seitz, H., and U. A. Simanowski. 1991. Alcohol and cancer: A critical review. In Alcoholism: A molecular perspective, ed. T. N. Palmer, 275—96. New York.

Sugimura, T. 1992. Multistep carcinogenesis: A 1992 perspective. Science 258: 603—7.

Tannenbaum, A. 1942a. The genesis and growth of tumors: II. Effects of caloric restriction per se. Cancer Research 2: 460—7.

1942b. The genesis and growth of tumors: III. Effect of a high fat diet. Cancer Research 2: 468—74.

Tannenbaum, A., and H. Silverstone. 1953. Nutrition in relation to cancer. Advances in Cancer Research 1: 453.

Tomatis, L. 1992. Poverty and cancer. Cancer Epidemiology, Biomarkers, Prevention 1: 167—75.

Trosko, J. E., and J. I. Goodman. 1994. Intercellular communication may facilitate apoptosis: Implications for tumor promotion (Review). Molecular Carcinogenesis 11: 8—12.

U.S. Public Health Service. 1988. The Surgeon General’s Report on Nutrition and Health, Department of Health and Human Services (Public Health Service) Publication no. 88—50210. Washington, D.C.

Watanapa, P., and R. C. N. Williamson. 1993. Experimental pancreatic hyperplasia and neoplasia: Effects of dietary and surgical manipulation. British Journal of Cancer 67: 877—84.

Weisburger, J. H. 1991. Nutritional approach to cancer prevention with emphasis on vitamins, antioxidants and carcinogens. American Journal of Clinical Nutrition 53: 226—37.

1992. Mechanisms of macronutrient carcinogenesis. In Macronutrients: Investigating their role in cancer, ed. M. Dekker, M. S. Micozzi, and T. E. Moon, 3—31. New York.

1994. Commentary. Does the Delaney Clause of the U.S. Food and Drug Laws prevent human cancers? Fundamental and Applied Toxicology 22: 483—93.

1996. Tea antioxidants and health. In Handbook of Antioxidants, eds. Enrique Cajenas and Lester Packer. New York. 469—86.

Weisburger, J. H., and R. Kroes. 1994. Mechanisms in nutrition and cancer. Meeting report. European Journal of Cancer Prevention 3: 293—8.

Weisburger, J. H., and G. Williams. 1991. Critical effective methods to detect genotoxic carcinogens and neoplasm promoting agents. Environmental Health Perspectives 90: 121—6.

Williams, G. M., and J. H. Weisburger. 1991. Chemical carcinogens. In Casarett and Doull’s toxicology. Fourth edition, ed. M. V. Amdur, J. Doull, and C. D. Klaassen, 127—200. New York.

World Cancer Research Fund. 1997. Food, nutrition, and the prevention of cancer: A global perspective. Washington, D.C.

Wynder, E. L., and G. B. Gori. 1977. Contribution of the environment to cancer incidence: An epidemiologic exercise. Journal of the National Cancer Institute 58: 825—32.

Wynder, E. L., J. A. Peters, and S. Vivona, eds. 1975. Nutrition in the causation of cancer, a symposium. Cancer Research 35: 3231—35.

Wynder, E. L., O. P. Rose, and L. A. Cohen. 1994. Nutrition and prostate cancer: A proposal for dietary intervention. Nutrition and Cancer 22: 1—10.

Wynder, E. L., E. Taioli, and Y. Fujita. 1992. Ecologic study of lung cancer risk factors in the U.S. and Japan with special reference to smoking and diet. Japanese Cancer Research 83: 418—23.

Yamasaki, H. 1990. Gap junctional intercellular communication and carcinogenesis. Carcinogenesis 11: 1051—8.

Zeng, Y., H. Oshima, G. Bouvier, et al. 1993. Urinary excretion of nitrosamino acids and nitrite by inhabitants of high and low risk areas for nasopharyngeal carcinoma in southern China. Cancer Epidemiology, Biomarkers, Prevention 2: 195—200.

Ziegler, R. G., A. F. Subar, N. E. Craft, et al. 1992. Does beta-carotene explain why reduced cancer risk is associated with vegetable and fruit intake? Cancer Research 52: 2060—6.