Cancer Today: Origins, Prevention, and Treatment (1984)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

5 The Epidemiology of Diet and Cancer The causes of cancer are extraordinarily difficult to discern. A small number of cancers are clearly of familial, or genetic, origin. Yet the majority seem to be caused either by spontaneous meta- bolic events or by specific external factors—substances encoun- tered in the course of normal living. Many of the clues to the origins of cancer have come from epidemiologists, who search for the causes of disease among the broad characteristics of a population. Like detectives, epidemiol- ogists work backwards from the crime: they study the incidence and distribution of a disease and then try to deduce the factors that caused it. They look for any telltale differences among people—such as their race or age, the foods they eat, or their exposure to viruses or industrial pollutants—that might explain why some are afflicted with a disease and others are spared. This chapter is based on the presentation given by Anthony B. Miller, National Cancer Institute of Canada, University of Toronto, at the 1983 annual meeting of the Institute of Medicine. 49

CANCER TODAY As such, epidemiology is descriptive and suggestive. It uncovers associations, points to populations at risk, provides clues about factors to blame, and raises hypotheses for further study. Over the past few decades, epidemiologists have been able to map the global patterns of cancer occurrence. They have found striking variations in both the incidence and types of cancers that occur in different countries, which suggests that external factors, including lifestyle, may be responsible. These variations are not due to ethnic origin, for when people migrate from one country to another, they tend to acquire the pattern of cancer typical of their new home. Epidemiologists now believe that 80 to 90 percent of all cancers are caused by external, or environmental, factors. Identifying the culprits will be an arduous task, for the long list of suspects includes smoking, diet, industrial chemicals, substances in air and water, the regions where a person lives or works, viruses and other in- fectious agents, and other personal habits. Nonetheless, many sci- entists see the role of external factors as the most encouraging finding to emerge from cancer research: it means that cancer may be largely preventable. The Search Narrows Some of the environmental causes of cancer are fairly obvious. As early as the eighteenth century, occupational exposure to certain toxic chemicals was implicated by epidemiological studies when an English surgeon, Percival Pott, noted the abnormally high in- cidence of scrotal cancer among chimney sweeps. Later epide- miological studies revealed that workers who handled aniline dye faced an increased risk of bladder cancer. More recently, the in- cidence of lung cancer in coastal regions of the United States has been linked to exposure to asbestos in the shipbuilding industry during World War II. About 30 years ago, epidemiologists noticed an association be- tween cigarette smoking and lung cancer. Many years elapsed before their suspicions were borne out in laboratory experiments, but now tobacco is thought to account for one-third of all cancer deaths in the United States alone. Lung cancer is the most prevalent 50

EPIDEMIOLOGY OF DIET AND CANCER form of cancer in the Western nations, and its incidence has been rising throughout this century as an increasing number of men, and later women, started smoking. Cigarette smoking also con- tributes to cancers at other sites, including the bladder and prob- ably the pancreas and kidney. Recently, suspicion has turned to both the intended and unin- tended products of industrialization, such as industrial chemicals or pollution from oil and gas consumption. So far, the fear that the environment has become increasingly carcinogenic has not been confirmed by epidemiological studies. Although the number of people who develop and die of cancer has risen, much of the increase is related to smoking. The rest can be largely explained by increasing life expectancy: more people are living to the age when cancer usually takes its toll. When adjusted for age, the incidence of those cancers unrelated to smoking has actually de- clined since World War II. Nevertheless, industrial products are constantly changing; new chemicals are being synthesized. It is not known whether these new products will increase the cancer risk. Meanwhile, it is clear that a major cause of today's cancers remains to be discovered. Given the fairly constant cancer rate of the past 30 years, it would seem that some longstanding aspects of lifestyle, such as diet, are responsible. Diet—the foods and portions people eat and the method of preparation—varies greatly from society to society and could account for many of the observed differences in international cancer incidence. But it has been only in the past two decades that cancer researchers have begun to suspect that certain features of an abundant, apparently normal diet could be related to cancer. An Imprecise Tool Of all the possible external causes of cancer, diet is perhaps the most difficult to evaluate. It is far easier to determine whether and how much people smoke than to find out what and how much they eat, or particularly what they ate a decade or two ago. The problem is not simply the vagaries of memory or the variation in individual eating habits, but a general paucity of knowledge about 51

CANCER TODAY the contemporary food supply. Foods are complex substances, the exact composition of which is often not known. The increasing reliance on processed foods compounds the difficulty. To determine dietary intake, epidemiologists often rely on in- direct data, some of which have been collected for entirely different purposes. One measure is national per capita food intake data, also known as food disappearance data, which takes into account the amount of food produced in a country as well as the amount exported, imported, fed to livestock, lost in storage, etc. Another approach is the household food inventory, in which individuals record purchases or menus. Using these techniques, epidemiologists have in the past two decades uncovered specific associations between types of cancer and various diets. They have documented a correlation between the diet of affluent societies, for instance, and cancers of certain organs—primarily breast, colon, and uterus. They have identified some foods that seem to increase the risk of cancer and others that seem to decrease it. These studies, however, reveal only associa- tions; in none of them has diet been proved to cause cancer. Nonetheless, when numerous epidemiological studies repeat- edly turn up the same associations, the weight of evidence becomes difficult to ignore. That is what happened with smoking and cancer some 20 years ago. And that is what is happening now with diet and cancer. In addition, laboratory studies have tended to support the epidemiological data. The evidence is now strong enough for a number of leading scientists, including a recent committee of the National Research Council (NRC),1 to conclude that diet is associated with many common cancers. (At this preliminary stage, however, the NRC committee could not determine exactly how much cancer can be attributed to diet.) A more difficult task will be to determine precisely which dietary components are either protective or to blame and the mechanisms by which they influence the development of cancer. 1 The Committee on Diet, Nutrition, and Cancer of the National Research Council. The studies described in this chapter are discussed in greater detail in its report, Diet, Nutrition, and Cancer. Washington, D.C.: National Academy Press, 1982. 52

EPIDEMIOLOGY OF DIET AND CANCER Migrant Studies Some of the first evidence that diet might be related to cancer came from studies of migrants. In the 1960s and 1970s, a number of epidemiologists determined the cancer incidence among certain populations that had migrated from one country to another. A consistent pattern emerged: for first-generation migrants, the pat- tern of certain cancers remained similar to that in the country of origin, or perhaps intermediate between the old and new country. Yet after one or more generations, migrants acquired the cancers characteristic of the host country. For example, among the Japanese who migrated to the western United States in the early 1900s, the first generation was found to have a far lower mortality rate for breast cancer than the Caucasian population. (In the United States and other affluent nations, breast and colon and rectal, or colorectal, cancer are particularly preva- lent.) By the second generation, the mortality from breast cancer had increased dramatically, and by the third generation was almost identical to that of the native population. This occurred without any outbreeding in the Japanese community—instead, it was due to acculturation. Similar changes have been observed among migrants to Israel, who come both from affluent and less developed nations. The Israel Cancer Registry provides a wealth of information for epi- demiologists, for it regularly publishes data on the incidence of cancer according to the birthplace of its citizens. First-generation migrants show a marked increase in colorectal cancer: one study showed that for middle-aged women who had migrated from Asia and Africa, the incidence of colorectal cancer increased by 39 per- cent over a 10-year period. For women born in Israel, the incidence increased by 102 percent during that same period. Other migrants to Israel who came from Western European countries, the United States, and Canada—where colorectal cancer is much more prev- alent—showed very little increase. A similar pattern occurs for breast cancer—Asian and African migrants and those born in Israel show a marked increase, while the incidence for migrants from Western nations remains stable. Thus, the biggest changes in co- lorectal and breast cancer incidence seem to be among migrants 53

CANCER TODAY TABLE 5-1 Comparison of Rates of Colorectal Cancer Among Native and Migrant Jewish Females Ages 35-64, in Israel Increase in Incidence (percent) Native Country 1960-1966 1967-1971 1972-1976 1966-1976 Israel 17.0 26.6 34.3 102 Africa/Asia 11.9 13.9 16.5 39 Europe/N. America 35.2 35.7 39.4 12 (SOURCE: Anthony B. Miller, National Cancer Institute of Canada.) TABLE 5—2 Comparison of Rates of Breast Cancer Among Native and Migrant Jewish Females Ages 35-64, in Israel Native Country . Increase in Incidence (percent) 1960-1966 1967-1971 1972-1976 1966-1976 Israel 85.9 54.0 132.4 125.2 132.3 60.5 133.6 133.6 136.2 54 49 Africa/Asia Europe/N. America 3 (SOURCE: Anthony B. Miller, National Cancer Institute of Canada.) from Asia and Africa who are adopting Israel's more affluent life- style (see Tables 5-1 and 5-2). What is particularly interesting is that the change in the incidence of breast cancer occurs rapidly—the incidence is as high for the foreign-born migrants from Asia and Africa as it is for natives of Israel. This is in direct contrast to the pattern observed in Japanese migrants. In that group, the incidence of colon cancer increased rapidly, while the changes in rates of breast cancer did not occur for one or often two generations. Epidemiologists suspect that this may reflect different rates of acculturation. Some migrant groups retain many of their cultural habits, including diet, for several generations. This has been shown for some Japanese populations in Canada, who after 35 to 50 years in their new country continued to eat more fish and rice and less beef, potatoes, bread, and milk than the Canadian-born population. 54

EPIDEMIOLOGY OF DIET AND CANCER Religious Groups Certain religious groups have proved to be incredibly fruitful subjects for epidemiologists, as they follow distinct diets that differ from that of the population at large. Most of these studies of the association between diet and cancer have focused on cancers of the gastrointestinal tract and the breast and other tissues subject to hormonal influence, as nutrition is known to affect hormone levels. The Seventh Day Adventists abstain from tobacco and alcohol, and roughly half of them are lacto-ovovegetarians—that is, they eat dairy products but no meat. In several studies, Roland Phillips and his colleagues at Loma Linda University in California have found that this diet seems to protect against colorectal cancer: Seventh Day Adventists had lower rates of both this diet-related cancer and smoking-related cancers than did a control group of comparable age, sex, and social status. The incidence of breast cancer was also reduced, but not significantly. The Mormons abstain from alcohol, tobacco, coffee, and tea. Many follow a diet rich in grains, fruits, and vegetables, with a moderate consumption of meat. The incidence of colorectal cancer is also lower among Mormons, according to a study by Joseph L. Lyons of the University of Utah in Salt Lake City. In India, the incidence of cancer varies dramatically among dif- ferent religious groups. The Parsis, who eat a Western-style diet, have a much higher incidence of breast, colon, and rectal cancer than do the Hindus, who generally follow a strict vegetarian diet. Which Dietary Components? These broad, descriptive surveys point to several dietary com- ponents, such as fat and animal protein, that might influence can- cer. Other studies have examined the association between specific dietary components and other variables with cancer incidence and mortality. In 1975, Bruce Armstrong and Sir Richard Doll of Oxford University reported the results of their massive study, which in- cluded over 30 nations and correlated the incidence and mortality 55

CANCER TODAY rates of cancer at nearly 30 sites or organs with a range of dietary and other variables. The strongest correlations were for meat and fat intake with cancers of the colorectum, breast, uterus, and ovary. They also found associations between total fat intake and testicular and kidney cancer. Armstrong and Doll also detected other direct correlations: in- testinal cancer with sugar; pancreatic cancer with eggs, animal protein, and fat; ovary and bladder cancer with fats and oils. Some inverse relationships also came to light: gastric cancer with meat, animal protein, and fat; cervical cancer with protein and fruit. Kenneth Carroll of the University of Western Ontario calculated that there was a strong correlation worldwide between per capita total fat intake and age-adjusted mortality from breast cancer. The highest mortality rates are found in those nations with the highest fat intake—the United States, the United Kingdom, and Canada. The lowest rate is found in Japan and other Eastern countries, where people consume far less fat. The correlation was strongest for total fat, almost as strong for animal fat, but nearly absent for vegetable fat. In Armstrong and Doll's study, many of the dietary variables, especially animal protein and animal fat, were correlated with each other and with gross national product. Other studies have also revealed a correlation between particular cancers and socioeco- nomic status, which is a major indicator of diet and lifestyle. In the United States, breast cancer—the leading cancer among women—is correlated with higher socioeconomic status among postmenopausal women. Cancer of the colon and rectum in men is also linked with higher socioeconomic status. For both sexes, renal cancer is more prevalent at higher socioeconomic levels. By contrast, stomach and lip cancer appear to be predominately found among less affluent groups. Similar findings have been observed in other countries. When the data are adjusted for occupation, strong correlations still remain, suggesting that social class is a major indicator of cancer risk. Other variables, which may in fact be nutrition-related, have also been found to be associated along with diet in the incidence of or mortality from cancer. Gregory Gray and his colleagues at the University of California at Los Angeles found that in addition 56

EPIDEMIOLOGY OF DIET AND CANCER to intake of animal protein and fat, a woman's height and weight are positively associated with breast cancer. Other studies have shown that the risk of dying of breast cancer increases propor- tionately with body weight. Epidemiologists can begin to move from the detection of as- sociations to inferences about causality when they collect data from individuals rather than rely on broad estimates of a population's diet. By interviewing individuals, epidemiologists can get a far more accurate assessment of actual dietary intake, although the technique is not foolproof. For one, epidemiologists must trust in an individual's recall and accuracy. In addition, the underlying assumption is that the current diet reflects the individual's diet at the time the cancer was initiated, one or perhaps two decades ago. One approach is the case-control study, in which investigators collect detailed information on dietary intake from individuals who have a particular type of cancer. This is then compared with dietary intake of a selected control population whose members do not have cancer. This difficult and time-consuming procedure has been used most frequently to study breast and colorectal cancer, ex- amining in greater detail the associations brought to light by the descriptive and correlation studies. Breast Cancer A major study on breast cancer was completed in Canada in 1978 by Anthony B. Miller of the National Cancer Institute of Canada and the University of Toronto. He and his colleagues collected dietary intake data on six nutrients from 400 breast cancer cases and 400 controls. For both pre- and postmenopausal women, the strongest correlation was with total fat consumption. For pre- menopausal women, there were also weaker associations with sat- urated fat and cholesterol. Of major importance in these investi- gations is evidence for dose-response—does the risk increase proportionately with dose? For total fat, Miller's group initially found no evidence of a dose-response ratio, but their subsequent, more detailed analyses suggest that for saturated fat, the risk may be greater with increasing consumption. Miller and his colleagues have estimated that a larger share of 57

CANCER TODAY breast cancer can be attributed to high fat intake than it can to the other variables known to be involved in the etiology of the disease, such as age at first pregnancy, family history of breast cancer, and weight. Another study reveals a similar correlation. Jay Lubin and his colleagues at the National Cancer Institute and the Cross Cancer Institute in Edmonton, Alberta, have found that risk of breast cancer increases significantly with frequent consumption of beef and other red meat, pork, and desserts—all of which reflect fat consumption. Colorectal Cancer A greater number of case-control studies have explored the re- lation of diet and colorectal cancer. Diverse subgroups have been examined—blacks in California, Japanese in Hawaii, and other groups in the Caribbean countries, Scandinavia, Norway, Aus- tralia, New York, and Canada. Some of the results have been conflicting—perhaps reflecting differences in accuracy of the var- ious methodologies—but in all, they point to a strong association between total fat intake and colorectal cancer. The association appears to be particularly strong for saturated fat. Several early, descriptive studies have turned up what seems to be a protective role for dietary fiber from grains and vegetables. As mentioned earlier, the Seventh Day Adventists have a low incidence of colon cancer. Similarly, the Punjabis of northern In- dia, who eat a diet rich in cellulose, roughage, and vegetable fiber, have almost no colon cancer. Yet efforts to document these as- sociations in case-control studies have yielded inconsistent results. In several of these studies, epidemiologists have found that vic- tims of colon cancer do consume less fiber than do controls. But one difficulty with these studies is that the amount of fiber in the diet is usually not measured; instead, it is estimated from the intake of certain vegetables, fruits, and whole grains. In one study that did quantify fiber intake, the investigators did not find a protective effect for fiber. They did observe, however, that incidence of colon cancer was inversely related to intake of one fiber component, the pentosan fraction, which is found in whole wheat products and 58

EPIDEMIOLOGY OF DIET AND CANCER other foods. That may help to explain the different results of these case-control studies: it might not be fiber per se but rather a par- ticular component or components that protect against colon can- cer. People who eat a lot of vegetables also tend to fare better in terms of colon and rectal cancer. A number of studies suggest a protective role for cruciferous vegetables—cabbage, broccoli, cau- liflower, and brussels sprouts. For instance, Saxon Graham of the State University of New York at Buffalo found that individuals who rarely eat these vegetables—once a month or less fre- quently—face the highest risk of colon cancer. The lowest risk was for those who consume cruciferous vegetables weekly. He also found a protective effect against rectal cancer, though it was far weaker. Anthony Miller and others have also documented a weak protective effect for these vegetables. At this stage, it is still unclear which component of the vegetables is responsible. Accumulating Evidence Epidemiological studies are only one-half of the evidence im- plicating diet in the development or prevention of cancer. Labo- ratory studies are now under way to test the various hypotheses. The results to date have tended to confirm the epidemiological data. For instance, a number of nutritive and non-nutritive com- pounds present in cruciferous vegetables have been shown to in- hibit carcinogenesis in cells in culture. Yet even in laboratory experiments, it is still not clear which components exert this ben- eficial effect. Similarly, experiments on animals have consistently shown that dietary fats promote tumorigenesis (the formation of tumors), especially in the breast and colon. When fat intake is increased from 10 to 40 percent of total calories in the diet of laboratory animals, tumor incidence in various tissues also rises. On the positive side, the consumption of some ingredients in high-fiber foods, such as cellulose and bran, has been shown to inhibit the induction of colon cancer by certain chemical carcin- ogens, but not consistently. Although laboratory studies support the epidemiological fmd- 59

CANCER TODAY ings, it is still too soon to say with certainty whether the specific associations between dietary components and different cancers are causal. More basic research is necessary to determine the mecha- nisms by which diet might trigger the development of cancer. Some clues about possible mechanisms have come from study- ing animals in which tumors have been induced by known chem- ical carcinogens. In most cases, carcinogenesis seems to be a mul- tistep process. It begins when the animal is exposed to an initiating agent, presumably a mutagen that damages DNA. Often initiation does not appear to be sufficient to give rise to a tumor; the action of other agents, known as promoters, is necessary to complete the process. Promoters may not be mutagens; instead, they seem to modify or accelerate the normal processes of cell differentiation and promote cell proliferation. Dietary components could influence either the early or late stages of carcinogenesis—both stages are poorly understood. Many car- cinogenic initiators seem to be created by metabolic interactions within the cell—that is, one chemical is converted to another, carcinogenic form within the cell. Some dietary components may actually be initiating substances; others may modify different sub- stances within the cell, converting them to initiators. Even less is known about the late stages of carcinogenesis, al- though preliminary evidence suggests that diet may exert its great- est influence there. Some dietary components have been shown to have a promoting or inhibiting effect; they can raise or lower the incidence of a cancer that has already been induced by exposure to an initiator. Dietary Guidelines Despite the uncertainties about the role of diet in cancer, many scientists believe that sufficient evidence exists to recommend pre- liminary dietary guidelines for reducing the risk of cancer. After a two-year study of diet and cancer, a committee of the National Research Council (National Academy of Sciences) proposed such guidelines in its 1982 report Diet, Nutrition, and Cancer. In early 1984, the American Cancer Society began a campaign to urge the public to follow this diet. 60

EPIDEMIOLOGY OF DIET AND CANCER The NRG committee recommended that the population reduce its consumption of both saturated and unsaturated fat from the present level of approximately 40 percent to 30 percent of total calories; eat fruits, vegetables, and high-fiber grains regularly; re- duce consumption of cured, pickled, or smoked foods, which have been linked to cancer of the stomach and esophagus; and drink alcohol only in moderation. These guidelines are, by necessity, preliminary. Given the cur- rent gaps in our understanding, there is no precise formula for preventing diet-related cancers. The situation is reminiscent of the uncertainty surrounding the role of smoking and cancer 20 years ago. If the population had been persuaded to stop smoking when the association with lung cancer was first reported, approximately 30 percent of today's cancer deaths could have been prevented. Many scientists, including the NRG committee, now believe that it may eventually be possible to define a diet that will significantly reduce the incidence of cancer in the United States. The next and by no means trivial step will be to persuade people to adopt that diet. 61

Dietary Carcinogens and Anticarcinogens As epidemiological studies revealed numerous links between diet and cancer, efforts were increased to determine whether chem- ical mutagens or carcinogens occur naturally in foods. Epide- miological studies have focused on foods containing specific nu- trients, such as protein, fats, and dietary fiber. Yet nutrients are only a small proportion of the chemicals that constitute various foods. Foods also contain many thousands of non-nutritive sub- stances, as well as additives and natural and synthetic contami- nants. In recent history, food additives and contaminants such as chem- icals used in food processing or residues of pesticides have engen- dered the greatest concern as possible health hazards. In the past two decades, many food additives have been tested for toxicity— although only a small number have so far been tested for carci- nogenicity. Several laws and regulations are designed to minimize This chapter is based on the presentation given by Bruce N. Ames, University of California-Berkeley, at the 1983 annual meeting of the Institute of Medicine. 63

CANCER TODAY human exposure to known or suspected toxins and carcinogens in food. The Delaney Clause of the Federal Food, Drug, and Cosmetic Act, for instance, prohibits the use of any food additive that has been shown to cause cancer in laboratory animals or human beings. By contrast, the natural constituents of food have received much less attention. In the past 10 years or so, scientists have begun screening some of these substances, as well as some naturally oc- curring contaminants like mold toxins, to determine whether they, too, pose a cancer risk. These biochemical assays have uncovered an extraordinary va- riety of naturally occurring mutagens and animal carcinogens in foods. In addition, some common cooking methods have been shown to generate mutagens. These studies reveal that mutagens and potential carcinogens are ubiquitous in the human diet—they can be found in celery, peanuts, hamburgers, and toast, for in- stance. Yet the magnitude of the risk they pose is not at all clear. Part of the problem is the difficulty in testing. Because cancer is thought to arise from damage to DNA, substances that cause mutations are considered to be potential carcinogens. The first step, then, in testing a chemical is a mutagenicity assay. Numerous studies have shown that a chemical that causes mutations in any living cells, including bacteria, should be suspected as a human carcinogen and studied further. Lengthy and expensive tests in whole animals are necessary to prove that a substance is a carcin- ogen—even in that animal. Few mutagens in foods have under- gone such tests. Identifying a chemical as a mutagen or a carcinogen is only the start. Before its risks can be assessed, its potency must be deter- mined—a difficult task when trying to extrapolate from labora- tory animals to humans—as well as its prevalence in the diet and route of exposure.1 Such work is just beginning on these natural mutagens and carcinogens. At this point, there is no evidence that any one of them is a major contributor to total cancer risk in the United States, yet a possible hazard cannot be ruled out. 1 For a discussion of risk assessment, see Risk Assessment in the Federal Govern- ment: Managing the Process. Washington, D.C.: National Academy Press, 1983. 64

DIETARY CARCINOGENS AND ANTICARCINOGENS The picture is not as bleak as it may sound, however. A variety of natural foods have also been found to contain anticarcinogens— substances that inhibit carcinogenesis in laboratory experiments. The mechanisms by which they act are not known, nor is it clear if these chemicals have the same inhibitory effect in the living human body as they do in the laboratory. Nonetheless, the lab- oratory findings are intriguing in light of several epidemiological studies suggesting that certain foods protect against cancer. In numerous laboratories around the world, efforts are under way to identify carcinogens and anticarcinogens in the human diet and to elucidate their mode of action. There is little conclusive evidence about how either might act to promote or inhibit car- cinogenesis. One provocative—if quite speculative—hypothesis is that some optimum balance of carcinogens and anticarcinogens in the diet might act to keep cancer at bay, and thus by increasing consumption of anticarcinogens, it may be possible to prevent certain cancers. Naturally Occurring Carcinogens It is well known that certain molds produce highly mutagenic or carcinogenic toxins. The aflatoxins and sterigmatocystin, for instance, are among the most potent carcinogens known. Afla- toxin-producing molds are ubiquitous, but the toxin is especially prevalent in tropical climates, where conditions favor the molds' invasion of stored foods. In the United States, aflatoxins can enter the diet through crops that are invaded by these molds before harvest—usually peanuts, corn, and cottonseed, and less fre- quently, tree nuts such as almonds, walnuts, pecans, and pista- chios. Other mutagens and carcinogens can be present in mold- contaminated corn, grains, peanut butter, bread, cheese, fruit, and apple juice. Nitrate and nitrite are widely dispersed in the human diet. Ni- trate is present in numerous vegetables, including lettuce, beets, celery, spinach, radishes, and rhubarb, in varying concentrations depending on the species, culture conditions, time of harvest, and other factors. In the body, nitrate can be converted to nitrite, which in turn can react with amines to form nitrosamines and other N- 65

CANCER TODAY nitroso compounds, which are potent carcinogens. Nitrite is also ingested directly, predominantly in cured meats. Tobacco also contains nitrosamines—the average smoker receives 10 to 100 times more nitrosamines from cigarettes than from dietary sources. Nature's Pesticides Plants synthesize a variety of toxic chemicals, which are used to ward off bacteria, fungi, insects, and other predators. For the past 100 years, organic chemists have been isolating and charac- terizing these plant chemicals: tens of thousands are already known, and others are still being discovered. Bruce Ames, the University of California biochemist who developed the widely used Ames mutagenicity assay, estimates that in general these natural pesti- cides constitute 2 to 10 percent of a plant's weight. Despite the prevalence of these natural plant chemicals in the human diet, little is known of their toxicology. Only in the past few years have they been tested for mutagenicity or carcinogen- icity. Many of them have been found to be mutagens, carcinogens, or teratogens—substances that cause birth defects. According to Ames, these plant toxins can enter the diet in several ways, either by the direct ingestion of plants containing them, or indirectly, through the milk of animals that have eaten toxin-producing plants, for instance. The following examples, drawn from the work of Bruce Ames, illustrate the diverse sources of these naturally occurring mutagens and carcinogens. Edible mushrooms contain a group of chemicals known as hy- drazines, many of which are mutagenic and some of which are carcinogenic in laboratory animals. The most common commer- cial mushroom, Agaricus bisporus, contains several of these carcin- ogenic hydrazines or their breakdown products. The widely eaten false morel (Gyromitra esculenta) contains 11 hydrazines, 3 of which are known animal carcinogens. One of these is especially potent, inducing lung tumors in mice at a very low dose.2 2 For more details on extrapolating doses from rodents to human beings, see Identifying and Estimating the Genetic Impact of Chemical Mutagens. Washington, D.C.: National Academy Press, 1983. 66

DIETARY CARCINOGENS AND ANTICARCINOGENS Pyrrolizidine alkaloids are present in thousands of plants, gen- erally in nonedible species. In some plants, they are present in trace amounts; in others they constitute up to 5 percent of the plant's dry weight. Many of these pyrrolizidine alkaloids are extremely potent carcinogens, causing lung and other tumors in laboratory animals. Plants containing pyrrolizidine alkaloids sometimes con- taminate forage crops and food grains, resulting in acute or chronic poisoning in livestock. They can also enter the human diet directly through the ingestion of herbal teas, comfrey leaves, and occa- sionally, honey. Black pepper and oil of sassafras, used in sarsaparilla root beer and in other foods as a flavoring agent, contain safrole, an allylic benzene derivative and known animal carcinogen. Black pepper also contains a closely related chemical, piperine, in large amounts— nearly 10 percent of dry weight. There is limited evidence that extracts of black pepper cause tumors in mice. Tarragon and anise contain another carcinogen, estragole, which is also an allylic ben- zene derivative. Celery, parsnips, parsley, and other members of the Umbrel- liferae family commonly contain furocoumarins, such as psomlen derivatives, which are potent carcinogens. One hundred grams of celery, for instance, typically contains 100 micrograms of psoralen derivatives. Moreover, the level can increase 100-fold if the celery is stressed or diseased. Celery pickers often develop skin rashes from handling diseased celery. Because these plant chemicals are carcinogenic in laboratory animals, they represent a potential human cancer risk. The mag- nitude of that risk, however, has yet to be assessed. Numerous other plant toxins cause mutations in bacterial cells, which raises suspicions that they may be carcinogens as well. Some of the most widespread of the known naturally occurring mutagens arejlavonoids, which are contained in the edible portions of many plants. In the United States, the average daily intake of flavonoids is estimated at 1 gram. Roughly 25 percent of this intake comes from tea, coffee, cocoa, fruit jams, red wine, beer, and vinegar. One of the flavonoids, quercetin, is particularly prevalent in the human diet. Some data suggest that it is a carcinogen, others do not. 67

CANCER TODAY Glycoalkaloids, the chemicals that potatoes produce to ward off predators, are quite toxic. Indeed, two of these, solanine and cha- conine, can be lethal to human beings at high concentrations. Dif- ferent varieties contain varying concentrations of the toxic gly- coalkaloids; one variety bred to withstand insects had to be withdrawn from use because of its toxicity. As is the case with the psoralen derivatives in celery, the concentration of glycoal- kaloids can increase when the potato is diseased, bruised, or ex- posed to light. Some evidence suggests that solanine and chaconine may be teratogens. Mutagenic or carcinogenic plant toxins have also been identified in cocoa, tea, coffee, mustard, horseradish, rhubarb, fava beans, and other foods or flavorings. In addition, some alcoholic beverages, including Scotch and North American whiskeys, have been shown to be mutagenic in laboratory tests, but investigators have not yet identified the re- sponsible component or components. There is also some incon- clusive epidemiological evidence linking alcohol and coffee with specific cancers: alcohol with cancer of the mouth, esophagus, pharynx, and larynx; coffee with cancer of the bladder, ovary, pancreas, and large bowel. Browned and Burned Foods Two decades ago, several investigators reported that beef grilled over a charcoal or gas fire contained a variety of mutagenic chem- icals known as polynuclear aromatic hydrocarbons (PAHs). That was one of the first indications that methods of cooking can gen- erate mutagens. In that case, the mutagenic PAHs were derived from the smoke created when fat dripped from the meat onto the hot coals. If the meat was grilled so that it was not exposed to the smoke, PAH contamination was eliminated or reduced. PAHs have since been detected in a variety of cooked or smoked foods, including coffee. In the past few years, it has been learned that many burned or browned foods are rife with mutagens and potential carcinogens. Much of the research was performed in Japan by Takashi Sugimura of the National Cancer Center Research Institute, and others. Sev- 68

DIETARY CARCINOGENS AND ANTICARCINOGENS eral years ago Sugimura tested the charred surface of the fish his wife was grilling on their hibachi and found it was highly mu- tagenic. He subsequently discovered that the mutagens are created by the heating of protein—and the process can also occur during the cooking of other protein-containing foods, such as dairy prod- ucts. Other researchers have found that mutagens can be formed in meats that are cooked at a normal, not a high temperature. Sugimura has identified nearly a dozen mutagens that are produced by the heating of protein or amino acids. According to Ames, roughly 7 percent of these substances have been shown to be carcinogenic as well. The browning of starchy foods also creates mutagens, appar- ently through the reaction of amino acids and sugars. In the past few years, reseachers have found that french fries, toast, cara- melized sugar, the crust of French bread, and other browned foods all contain highly mutagenic—but as yet unidentified—chemicals. Anticarcinogens Epidemiological studies suggest that some foods, including cab- bage and other cruciferous vegetables and some components of dietary fiber, protect against cancer. The particular constituents responsible for these protective effects are not known with cer- tainty. In laboratory studies, certain substances have been found to inhibit the process of chemically induced carcinogenesis. Some of these anticarcinogens, as they are called, are described in the following paragraphs. Several vitamins are known or suspected anticarcinogens. In epidemiological studies, food rich in vitamin A or its precursors, carotenoids, are associated with a reduced risk of cancer of the lung, bladder, and larynx. (Vitamin A is present in liver and carot- enoids are present in green and yellow vegetables.) These studies provide only estimates of vitamin A intake, however, and it is not clear whether the protective effects are due to vitamin A, the carotenoids, or other substances. Conversely, laboratory experi- ments have revealed that vitamin A deficiency tends to increase susceptibility to chemically induced cancers. Some epidemiological studies show that consumption of foods 69

CANCER TODAY containing vitamin C (ascorbic acid) is associated with a low risk of cancers of the stomach and esophagus, although it is not clear which constituent of the foods is responsible. In laboratory tests, ascorbic acid can inhibit the formation of carcinogenic N-nitroso compounds, which suggests one possible mechanism for the ob- served effect. There is much less evidence to suggest that vitamin C can affect the action of already-formed carcinogens. Vitamin E (tocopherol) also inhibits the formation of N-nitroso compounds, and there is some evidence that it can inhibit chem- ically induced cancers in the laboratory. Epidemiological studies are rare, however, because vitamin E is so prevalent in common foods, such as vegetable oil, whole grain cereals, and eggs, and thus it is difficult to distinguish its effect from those of other compounds in the foods. The mineral selenium, which is present in seafood, organ meats, and grains, as well as in drinking water, has been shown to protect against cancer in laboratory studies and to a limited extent in epidemiological studies. In animal experiments, it inhibits the chemical induction of skin, liver, colon, and mammary tumors. It also inhibits the induction of mammary tumors by viruses. It is not known if or how readily this protective effect can be achieved at normal dietary concentrations; most of these studies were con- ducted using selenium concentrations far in excess of the normal intake, often at levels bordering on toxic. Oxidative Damage The mechanisms by which dietary anticarcinogens might pro- tect against cancer are not known. Ames speculates that many antimutagens and anticarcinogens may act by protecting against cell damage caused by mutagenic forms of oxygen, such as su- peroxide, hydrogen peroxide, and the hydroxyl radical. Ames believes that these forms of oxygen may be a major contributor to the destructive processes associated with aging and cancer. It is thought, for instance, that ionizing radiation acts to damage a cell by generating the hydroxyl radical. Specifically, Ames suspects that they both initiate and promote carcinogenesis. These active forms of oxygen are created as the by-products of 70

DIETARY CARCINOGENS AND ANTICARCINOGENS normal oxidative metabolism, as well as by other ways. Because of their reactivity, these forms of oxygen cause abundant damage within the cell. They damage DNA, cross link DNA to DNA, or DNA to protein, and generally disrupt cell organization. They can also trigger the destructive chain reaction of rancidity, known as lipid peroxidation, within the cell. This in turn leads to the generation of additional mutagens and carcinogens, many of which are oxidants. (Oxidants increase the oxygen content of a com- pound by removing electrons and thereby increasing the com- pound's valence, or ability to react with other compounds.) The body has many defense mechanisms to protect against these mutagenic forms of oxygen. Several enzymes, for instance, protect cells from oxidative damage. In addition, several molecules con- sumed in foods are antioxidants, substances that prevent aberrant oxidation and the generation of rancidity mutagens. Many of the dietary substances identified as antimutagens or anticarcinogens, such as vitamin C, vitamin E, selenium, beta-carotene, glutathi- one, and uric acid, are in fact antioxidants. If active oxygen forms do prove to be involved in carcinogen- esis, as Ames suspects, then one approach to cancer prevention may be to adjust the dietary intake of antioxidants, he says. At present, the optimum level of dietary antioxidants is not known, and excessive consumption of several of them, including vitamin A and selenium, can be toxic. Assessing the Risk As shown by the examples in this chapter, naturally occurring mutagens and possible carcinogens are widely distributed in the human diet. However, with a few exceptions, such as the afla- toxins and A/-nitroso compounds, the significance of these sub- stances for human health is not known. In Ames' view, this recent—and continuing—detection of nat- ural carcinogens in the diet will eventually require a reappraisal of health hazards and sources of carcinogens. The presence of pes- ticide residues in foods has engendered considerable concern and is aggressively regulated; by contrast, little attention has been paid to the intake of nature's pesticides, the plant toxins, says Ames. 71

CANCER TODAY He suspects that they may turn out to pose a greater risk than synthetic pesticides. Additional research, including epidemiological studies, and long- term cancer studies in laboratory animals, will be necessary to evaluate the magnitude of risk posed by dietary carcinogens and mutagens. Before society responds to these potential threats, other factors will also need to be evaluated, such as the trade-offs in- herent in removing these substances from the diet. Cooking fish or meat, for instance, does create mutagens. Yet cooking also destroys parasites and pathogenic microorganisms. Similarly, some plants containing mutagenic flavonoids are highly nutritious. There are other possible implications of the association between cancer and dietary constituents. For example, in an attempt to reduce dependence on synthetic pesticides, some breeders are de- veloping plants that contain higher levels of the natural toxins that confer resistance to insects. It may be prudent to determine which pesticides—natural or synthetic—pose the greatest health hazard, Ames suggests. Questions of relative risk aside, it is far easier to minimize the occurrence of intentional additives and industrial contaminants in foods than it is to control the natural risk factors in diet. Histor- ically, food safety regulation has tended to focus on such "foreign" sources of risk. Yet if it is determined that natural substances in the diet—major nutrients and perhaps these naturally occurring carcinogens—are the greater danger, then consumers may have to assume a greater share of the responsibility for protecting them- selves. To explore many of the questions raised in this and the preceding chapter, the National Cancer Institute has embarked on an ambitious research program. Only when it is known which substances in foods either promote or inhibit carcinogenesis will it be possible to devise prudent strategies for preventing diet- related cancer. 72

7 Diet and Cancer: New Policy Questions Since the passage of the 1906 Food and Drug Act, the task of ensuring the safety of foods has fallen primarily to the Food and Drug Administration (FDA). During this century, regulatory pol- icy has grown incrementally with the passage of statutory amend- ments addressing one potential risk or another, such as food ad- ditives or the residues of drugs used in food animals. In the past 25 years, FDA has specifically sought to eliminate possible cancer-causing substances from foods. The focus has been on exogenous chemicals—food additives, coloring agents, indus- trial contaminants, natural contaminants, and the like. In many ways, says Richard Merrill, dean of the University of Virginia School of Law and former chief counsel for FDA, the agency's strategy has resembled a "search and destroy" mission, in which FDA has attempted to identify and when possible remove from This chapter is based on the presentation given by Richard A. Merrill, University of Virginia School of Law, at the 1983 annual meeting of the Institute of Medicine. 73

CANCER TODAY the diet those few agents thought to pose a cancer or other health risk. In regulating potential carcinogens, the agency has been tough- est on intentional food additives—e.g., the 1958 Delaney Clause prohibits the use of any additive known to cause cancer in human beings or other animals. Working with the Environmental Pro- tection Agency, FDA has also set permissible limits for some industrial contaminants, such as pesticide residues. (However, the mechanisms for detecting whether a hazardous substance has in- advertently entered the food supply are weak, says Merrill, who cites the recent ethylene dibromide [EDB] episode.) In addition, FDA has set permissible limits for naturally occurring contami- nants known to be carcinogenic, such as aflatoxins and other mold toxins. Throughout this process, the natural constituents of foods have received comparatively less attention. This partially reflects a longstanding assumption that foods that are grown and harvested are generally safe as long as they are free of additives and contam- inants. It also reflects the feasibility of regulation: It is far easier to monitor and control the level of intentional additives in foods than it is their natural constituents. Everyday Risks The accumulating evidence about the relation between diet and cancer described in the preceding two chapters challenges some of the assumptions that underlie existing regulatory policy for food safety. As outlined in Chapter 6, recent studies suggest that a typical, everyday diet may pose a greater cancer risk than do food additives and contaminants, at least at current levels. In the past 20 years, epidemiologists have found that the foods people eat strongly influence the probability of their developing certain types of cancer. As described in Chapter 6, the high rate of breast and colon cancer in the United States and other affluent nations has been linked to a high intake of fats and fatty meats. Furthermore, frequent consumption of grains, vegetables, and fruits is associated with a lower incidence of colon, esophageal, and stomach cancers. Some scientists now believe that dietary factors may be responsible for anywhere from 10 to 70 percent of all cancers in the United 74

DIET AND CANCER: NEW POLICY QUESTIONS John Branch (San Antonio Express-News) States. By contrast, food additives are typically thought to cause at most 5 percent of all cancers, and environmental pollutants no more than 2 percent. These are preliminary estimates at best. In addition, it has recently been found that natural mutagens and possible carcinogens are pervasive in a traditional diet. These substances have been detected in a variety of foods—including celery, potatoes, peanuts, coffee, cocoa, tea, and meats. Some of these foods—and vegetables in particular—also contain anticarcin- ogens, substances that seem to protect against cancer (see Chapter 6). Self-Protection Existing regulatory structures were not designed to address these new concerns, says Merrill, who thinks that as the scientific evi- 75

CANCER TODAY dence strengthens, both Congress and FDA may need to rethink government strategy for preventing diet-related cancer. Diet now appears to be a far greater factor in cancer than was generally thought 80 years ago when the food safety law was written. More- over, the problem is not specific, individual contaminants, as pre- viously believed. Nor is it specific foods per se, as some of the dietary components found to promote cancer, such as fat, are essential nutrients. Rather, the probability of developing cancer seems to be influenced by the quantity and proportion of certain dietary components, and perhaps even by methods of cooking. In short, the problem is one of total dietary choice. Indeed, two dietary cancer risks—too much fat and perhaps too little fiber—are associated with those choices over which the con- sumer has the readiest control. Yet past regulatory efforts have relied little on consumer self-protection; when additives and con- taminants were considered the major hazards, the responsibility for prevention was thought to rest not with the consumer but with the government, which had the power to ban them. This "pollution control" approach to prevention—the attempt to eliminate offending substances from the diet—may be out- moded, Merrill says. FDA will undoubtedly continue to regulate food additives and other exogenous chemicals in an effort to min- imize exposure to carcinogenic substances. But given the plethora of natural mutagens and carcinogens, such regulations cannot en- sure a risk-free diet. For foods that cannot practically be eliminated from the diet, the government may be forced to rely on labels warning of po- tentially harmful constituents, as Congress has done for saccharin. Yet there are practical limitations to this approach. Fruits and vegetables are hard to label, for instance, and their characteristics vary both seasonally and regionally. Instead, FDA may be per- suaded to resume regulating the composition of processed foods. In the 1940s and 1950s, FDA set standards for the content of cheeses, jams, mayonnaise, and numerous other foods. Such con- trols could be revived and modified to reflect new understanding of dietary risk factors, suggests Merrill. For example, FDA could set a limit on the amount of saturated fat allowed in processed foods. 76

DIET AND CANCER: NEW POLICY QUESTIONS However, this and other traditional approaches to control the characteristics of marketed foods may not be sufficient, according to Merrill. An additional and perhaps more effective strategy for preventing cancer may simply be to provide information that will help consumers make intelligent choices. And as the government's focus broadens from the control of the composition of foods to education, it may turn out that FDA is no longer the most ap- propriate agency for the task, Merrill concedes. Several educational programs are already under way. Both the American Cancer Society and the federal government embarked on new efforts in 1984 to alert the public to the risks and benefits of certain dietary patterns. The federal cancer prevention effort, led by the Department of Health and Human Services, will focus on diet, smoking, and to a lesser extent, occupational safety. In the first year, nearly $700,000 will be spent on television com- mercials, pamphlets, and other materials. In addition, groups that have a commercial stake in what Americans eat will probably launch their own educational campaigns. Personal Autonomy Implementing such a policy will be a formidable task. Although it is relatively easy to eliminate contaminants in foods, it is much more difficult to influence dietary choice, Merrill suspects. The toughest obstacle will probably not be technical feasibility or cost, although a massive educational campaign could be quite expensive. Rather, it may be personal autonomy. People have an obstinate attachment to eating foods they enjoy, and they do not always welcome advice on how to change their diet. Moreover, even those individuals who choose to change their eating habits often have an extremely hard time doing so, as shown by the vast numbers of Americans who try perpetually, but unsuccessfully, to lose weight. There are practical impediments to such a program as well. One is the complexity of the subject and the difficulty of translating the key information into accessible language. In addition, the po- tential for "information overload" is high, as FDA noted in 1979 when it was criticized for failing to require labeling of all foods 77

CANCER TODAY containing possible carcinogens. Their response, published in the Federal Register, was blunt: "A requirement for warnings on all foods that may contain inherent carcinogenic ingredients or car- cinogenic contaminants would apply to many, perhaps most of the foods in the supermarket. Such warnings would be so nu- merous they would only confuse the public. It would not promote informed consumer decision-making. It would not advance the public health." The continuing scientific uncertainty also complicates the issue. While the weight of evidence implicating diet in cancer is strong, the associations between certain nutrients and cancer have not been proved definitively, nor have the biological mechanisms been worked out. Thus, consumers would be asked to alter their behavior on the basis of uncertain information. Moreover, because the data on diet and cancer are still open to different interpretations, the public may be bombarded with conflicting information as it was 20 years ago on the issue of smoking and cancer. Determining the appropriate means of communication is an- other problem. Restaurant sales account for roughly 40 percent of all food consumed in the United States, yet FDA has no regulatory authority to require labeling or other information disclosures within restaurants. Similarly, large segments of the population, such as members of the military services or persons in various institutions, have limited choice about their diet. Changing Habits Perhaps the major impediment to an effective educational pro- gram is the paltry understanding of consumer behavior generally and dietary change specifically. Surprisingly little is known about the quantity and types of foods people eat, or the exact compo- sition of those foods. Even less is known about why people select certain foods, or why they change their eating habits, as they often do spontaneously. At this stage, there are no proven methods for eliciting long-term dietary change. This is not for lack of trying. Numerous studies have been conducted to assess the effects of nutrition information programs on dietary habits. The effect of saccharin warning labels on diet 78

DIET AND CANCER: NEW POLICY QUESTIONS soft drink sales, for example, has recently been studied. Those investigators found that sales did drop at the height of publicity surrounding this suspected human carcinogen, but advertising also dropped at the same time. When advertising increased to its pre- publicity level in 1980, sales reached an all-time high. Weight loss programs generally have a dismal rate of success. Very few people who participate in one of these programs are able to maintain their weight loss. There is some evidence, however, suggesting that a substantial number of people who fail repeatedly to reduce their weight can eventually succeed. Perhaps most relevant to the diet and cancer issue is our ex- perience with programs designed to prevent heart disease. In the recent Multiple Risk Factor Intervention Trial, a 7-year study con- ducted by the National Heart, Lung, and Blood Institute involving 13,000 men at high risk for heart attack, half of the group was given "special intervention" to help them reduce the factors that contribute to the risk of heart attack. Specifically, they were en- couraged to stop smoking, reduce their blood pressure, and mod- ify their diets to lower serum cholesterol. They did have reduced mortality from heart attacks during the study period. But the control group also modified these risk factors and had a lower mortality rate from heart disease, as did the entire U.S. population. It is difficult to evaluate how much of the change can be attributed to active intervention, how much to "spontaneous" change. For over two decades, the American Heart Association has con- ducted an educational campaign to persuade the public to adopt a diet and lifestyle that would minimize the risk of heart disease. As the study described above and others suggest, many people in the United States have changed their diet and lifestyle in an effort to stay healthier. Moreover, mortality from heart disease has declined substantially in the past 15 years. It is not known, however, whether these lifestyle changes are responsible for the mortality decline, or in turn whether the lifestyle changes were the result of specific educational programs. Although the underlying reasons for these changes in dietary habits are not known, they are encouraging. It may be that the public is responding to the growing scientific and medical con- sensus, communicated through many formal and informal chan- 79

CANCER TODAY nels, that diet is related to a wide spectrum of diseases, including cancer. Additional research may help in the design of more effec- tive educational programs. Such programs will always be imper- fect tools for preventing cancer, but as the accumulating evidence about diet and cancer makes clear, any effort toward prevention will have to emphasize consumer self-protection. 80