Document XRGNeDz7B91azgxGRoZm0o01d

I URL05470 4m Am Pubiic Health. 1982. 2:85-100 EPIDEMIOLOGIC APPROACHES TO CANCER ETIOLOGY1 Y/l /I PIT.) *iT,I U'Bp/.doo y | iinnTiT oq Xcm p>j.>cjotn erqj, :3DI ION Joseph F Fraumeni, Jr. Environmental Epidemiology Branch, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20205 INTRODUCTION Epidemiology has contributed substantially to knowledge about the causes of human cancer and provides the basis for preventive measures. The ap proach dates from 1700 when the occupational physician Bemadino Ramazrini reported that nuns were at high risk of breast cancer, and 177S when the surgeon Perctvall Pott found that chimney sweeps exposed to soot were prone to scrotal cancer (1). In more recent times the initial leads to epidemiologic investigations have continued to come from astute clinicians who noted an excessive number of patients with the same tumor and traced the "cluster" to a particular cultural, occupational, or iatrogenic exposure. New insights into cancer etiology are provided also by experimental ap proaches to detect carcinogens in laboratory animals or mutagens in short term test assays, and to clarify basic mechanisms of carcinogenesis. In recent years the pace of epidemiologic and experimental research in cancer etiology has accelerated, including efforts to i&entify environmental deter minants, which are generally held responsible for a targe proportion of cancers in the general population (2). 'Tbe US Government has the right to retain a nonexclusive, royalty-free license in and to any oopylifht covering this paper. 65 >86 FRAUMENI DEMOGRAPHIC PATTERNS URL 05471 Cancer is second only to heart disease as a cause of death in the United States, and accounts for 20% of all deaths. It is estimated by the American Cancer Society that in 1981 about 815,000 Americans will develop cancer, excluding in situ carcinomas and nonmelanoma skin cancer, and that about 420,000 will die from this disease (3). The most common cancers arise from the lung in males, the breast in females, and the colon and rectum in both sexes combined. These three sites account for over 40% of cancer rases and deaths in the United States. The Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute has recently published detailed information on the incidence and mortality from cancer during the first five years of operation, 1973-1977 (4). The 11 areas partic ipating in the Program represent over 10% of the US population, and permit analyses according to a variety of demographic variables, including ethnic groups and geographic areas, that provide important clues for epidemiologic investigation. It has been widely reported that 80% or more of all cancer is attributed to environmental influences, particularly those related to life style practices, and that this fraction erf cancer is potentially preventable (5). These esti mates are derived from the substantial international variation in cancer incidence, in which rates for the lowest risk countries are subtracted from the rates prevailing in the United States. The resulting difference is at tributed to extrinsic causes, and the lowest risk is assumed to represent the baseline level for tumors that may be difficult or impossible to avoid. Around the world the reported age-adjusted incidence rates for total cancer vary by a factor of about three, while the range ofrates for certain anatomic sites, particularly the esophagus and liver, is greater than 100-fold (6). The risks for the more common tumors in developed countries differ by factors of about 8 to 40, with substantial differences in rates apparent even when the analyses are confined to European countries (7). Although some of the global variation may be influenced by a genetic component or by reporting practices in different parts of the world, the major contribution ofenviron mental factors is indicated by the patterns of risk among migrant popula tions, such as the Japanese who moved to Hawaii and California (8). Generally, as migrant groups adopt customs of the new land, their risk of various cancers shifts away from the rate prevailing in the country oforigin to approximate that of the host country. The change in incidence for scone cancers, notably the colon, is evident within two to three decades ofmigra tion, whereas the change for other cancers, notably the breast, requires more than one generation. Although variations within countries are not as great as those seen internationally, the mapping of mortality statistics at the CANCER ETIOLOGY 87 r county level in the United States, China, and other countries has revealed geographic peculiarities and clustering that provide starting points for etioiogic studies (9, 10). Variations in cancer incidence and mortality over time may also reflect environmental factors, although some fluctuations can be explained by changing medical practices and reporting procedures (11). Most dramatic has been the increase in lung cancer rates in all countries that have adopted the practice of cigarette smoking. Upward trends have been noted also for thyroid cancer resulting from X-ray exposures to the head and neck during childhood, malignant melanoma from changing clothing habits and recre ational exposures to sunlight, and endometrial cancer from the use of menopausal estrogens (12). The increases in prostatic cancer, multiple myeloma, and certain other neoplasms are at least partly due to improve ments in diagnostic measures. In the black population of the United States, sharp increases over time have been reported for cancers of the lung, eso phagus, prostate, and pancreas, and multiple myeloma, so that these tumors are now more common in blacks than whites. Several cancers have shown little change in whites or blacks, while some have displayed downward trends, notably cancers of the stomach, cervix, and liver. CAUSES OF CANCER Although much remains to be learned about the factors responsible for the geographic and temporal variations of cancer in the general population, several environmental exposures have been identified as carcinogenic in man (Table 1). The evidence is based primarily on case-control studies (comparing the past experience of persons with and without a particular cancer) or cohort studies (following up individuals whose experiences and characteristics are already defined). There is a growing recognition, how ever, that most cancers result from the combined effects of multiple expo sures and susceptibility states. This is consistent with multistage models in which different risk factors accelerate the transition rates at various stages of carcinogenesis (13). Some affect early stages as initiators, others act at late stages as promoters, while still others influence both early and late stages. It is generally thought that cumulative environmental exposures, long latency periods, and multistage processes account for the increasing risk of most cancers with advancing age. < Tobacco l The principal carcinogenic hazard to man is tobacco smoking, which pro duces cancers of the lung, larynx, mouth, pharynx, esophagus, bladder, pancreas, and probably kidney. It is estimated that smoking, especially of i8 F^AUMENI Table 1 Environmental cause* of human cancer Agent Alkylating agents (melphalan, cyclophosphamide, chlorambucil) Androgen-anabolic steroids Alcohol Aromatic amines (benzidine, 2-naphthylamine, 4-aminobiphenyl) Arsenic (inorganic) Asbestos Benzene Bis(chloromethyi)ether Chlonuphazine Chromium compounds Estrogens Synthetic (DES) Conjugated (P/cmarin) Steroid contraceptives Immunosuppressants (azathioprine, cyclosporin) Ionizing radiation Isopropyl alcohol production Mustard gu Nickel dust Phcnacetin-containing analgesics Polycyclic hydrocarbons Tobacco chews and powder - Tobacco smoke Ultraviolet radiation. Vinyl chloride -- Wood dusts Type of exposure Medication Medication --- Drinking Manufacturing of chemicals Mining and smelting of certain ores, pesticide manufacturing and application, medication and contaminated drinking water Manufacturing and application Leather, petroleum, and other industries Manufacturing of ion exchange reams Medication Manufacturing Medication 1 Medication Atomic Masts, treatment and diagnosis, radium dial painting, uranium and metal Manufacturing by strong acid process illiufrill ri*tg Refining Medication Coat carbonization products and some mineral oils Snuff dipping and diewing of tobacco, betel, time Smoking, especially cigarettes Sunlight Manufacturing of polyvinyl chloride Furniture manufacturing Site of cancer leukemia, bladder Liver Mouth, pharynx. esophagus, larynx, liver Bladder Lung, skin, liver (angiosarcoma) Lung, pleura. peritoneum Leukemia Lung Bladder Lung Vagina (adenocardnoma) Endometrium Liver (benign) Lymphoma (histiocy- tic), skin (squamous carcinoma), soft tissue sarcoma Nearly all sites Nasal stiluses Lung, larynx, nasal sinuses Lung, nasal sinuses Renal pelvis Lung, skin (squamous carcinoma) Mouth Lung, larynx, mouth. pharynx. bladder, panacea, kidney Skin, including mela* noma Liver (angiosarcoma) Nasal sinuses \ CANCER ETIOLOGY 89 cigarettes, contributes to about 25 to 35% of all cancer deaths in men and 5 to 10% in women (14). The greatest impact is on lung cancer, with the risk for male smokers of two or more pads per day being about 20 times that ofnonsmokers, based on an evaluationofdie threelargest cohortstudies (15). However, the rates for lung cancer are now rising more sharply in women than in men, reflecting the growing popularity of cigarettes among women in the past 20 to 30 years. Recent studies indicate that smokers of filter-tipped cigarettes with reduced levels of tar and nicotine have lower risks of lung and larynx cancers than do smokers of nonfilter cigarettes, but an even greater reduction ofrisk comes from cessation of smoking (16). The risks from cigar and pipe smoking resemble those of cigarette smoking for cancers of the mouth, pharynx, larynx, and esophagus, but the lung cancer risks are not greatly elevated over those ofnonsmokers, presumably because cigar and pipe smoke is more irritating than cigarette smoke and less conducive to inhalation (17). Smokeless tobacco products are also of con cern, since oral cancer has been linked with snuff dipping, a common practice in rural southern areas of the United States (18). In parts of Asia, oral is very common in people exposed to various tobacco chews, which are often mixed with betel, lime, and other agents that may enhance the risks. Alcohol Consumption of alcoholic beverages has been shown to multiply the car cinogenic effects of tobacco smoking on cancers of the mouth, pharynx, larynx, and esophagus, but no such effect has been shown on the lung (19). For heavy drinkers who do not smoke, the risks appear only slightly ele vated, so that the effects are strongly dependent upon smoking habits. Heavy drinking also increases the risk of liver cancer, particularly among cirrhotic patients. Based on these relationships, it is estimated that alcohol contributes to about 3% of all cancer deaths (20). Although some studies suggest that the carcinogenic effect is greatest following consumption of spirits, the bulk of evidence indicates that the influence of alcohol is inde pendent of the type of beverage. Since pure alcohol is not carcinogenic in laboratory animals, the mechanism by which alcohol promotes carcino genesis is not clear. Under suspicion are nutritional deficiencies associated with heavy drinking, the effects ofcongeners or contaminants (e.g., nitrosamines, hydrocarbons) in alcoholic drinks, and the capacity of alcohol to solubilize carcinogens or enhance their penetration into tissues lining the upper digestive and respiratory tract (21).*Some recent studies have sug gested that beer consumption may be related to cancer of the large bowel, particularly the rectum, although this association has not been confirmed 02). 9p FRAUMENI 1 URL 05473 cancer etiology 91 Sunlight The dominant risk factor for nonmelanoma skin cancer (squamous and basal cell carcinomas) and for malignant melanoma is ultraviolet (UV) radiation from the sun (23). The evidence is based on the tendency for skin cancers to arise on sun-exposed surfaces, the high rates among outdoor workers, the inverse correlation between skin cancer incidence and diatance from the equator, the predisposition of light-skinned and especially faircomplexioned populations who sunburn easily, the resistance of darkskinned populations with protective melanin pigment, the exceptional risks of skin cancer among persons with genetic diseases exacerbated by sunlight (e.g. xeroderma pigmentosum, albinism), and the capacity of UV radiation in repeated doses to induce skin cancer in experimental animals, particu larly in the UV-B spectral range (290-320 nm) that causes delayed ery thema in human skin (24). The intensity of UV-B exposure on the earth's surface is limited by the ozone layer in the stratosphere, but there is concern that this protective barrier may be impaired by certain atmospheric pollu tants, especially the continued release of chlorofluorocarbons used in aerosol propellants, refrigerators, and air conditioners. Based on recent surveys of skin cancer in the United States, it is estimated that the relative impact of ozone depletion would be greater for squamous cell carcinoma, which shows a steeper gradient with UV-B exposure than does basal cell carcinoma (25). The relationship of melanoma to sunlight exposure is less clear cut, but the recent recognition of the dysplastic nevus syndrome as a precursor state has provided a better understanding of host susceptibility to this tumor (26). Radiation Although ionizing radiation probably accounts for less than 3% of all cancer deaths, it appears that virtually no site of the body is spared from its carcinogenic effects (27). It is difficult to measure directly the effects of low doses of sparsely ionizing radiation, such as X or gamma rays, and debate continues about the precise nature of the dose-response relationship at low doses. However, extrapolations are possible by studying populations who have been exposed to high and moderate doses for medical, occupa tional, or military reasons (28). These studies also provide insight into the basic principles of cancer induction. In general, the breast, thyroid, and bone marrow are the most radiosensitive organs (29). Radiogenic shows a wave-like pattern with the excess risks starting about 2-4 years after exposure, peaking at 6-8 years, and dedining to normal within 25 years. In contrast, radiogenic carcinomas of various sites have a minimum latent period of 5 years, and a temporal distribution that closely resembles the natural inddence curve and suggests that age-dependent cofactors influ- cnee tumor expression. Except for dose of radiation, the age at exposure may be the most critical determinant of risk, with the vulnerable age groups varying with die type of cancer. Surveys of medically irradiated populations have revealed excess risks of the following (29): (c) leukemia and other cancers in the radiation field among patients treated for ankylosing spondylitis, metropathia hemorr hagica, and various neoplasms; (b) breast cancer among women treated for postpartum mastitis or who received pneumothorax fluoroscopies for tuber culosis; (c) thyroid cancer among children treated for thymus enlargement, benign head and neck disease, or tinea capitis; and (d) cancers at the deposit sites of radioactive compounds (osteosarcoma with radium-224, leukemia with phosphorus-32, and leukemia and liver angiosarcoma with thorotrast). It is generally felt that significant lowering of the radiation burden to the population can be attained mainly from reduced exposure to X rays in medical and dental practice (27). This will require measures to improve the efficiency ofx-ray equipment, and judicious weighing of the benefits versus the risks of radiation when diagnostic procedures and therapy are under consideration. Occupation Occupational exposures are usually reported to account for about 5% of all deaths (2). However, the percentage varies according to the type of neoplasm and the geographic location, and is likely to change as new hazards are recognized or as protective measures are instituted. Since the present state of knowledge of occupational cancer is so limited, it has been difficult to resolve controversies over the proportion of cancer that may be attributed now or in the future to occupational exposures. The wide range of percentages suggested by various investigators has resulted usually from the kinds of assumptions that are made (e.g. extrapolating from animal . studies, estimating effects of low doses when human data are available only for high exposure levels). In spite ofthis uncertainty, it is clear that no other approach has identified so many human carcinogens as the study ofoccupa tional groups (30). Usually the initial leads have come from clinical and epidemiologic observations, with subsequent confirmation by laboratory studies. Although benzene and inorganic arsenic have not been conclusively shown to be carcinogens in animals, an experimental study of benzene suggests a carcinogenic effect in rats (31), and epidemiologic analyses of lung cancer risk among smelter workers indicate that the effect of arsenic may be cm a late stage in the manner of a promoting agent (32). In the case of mustard gas and vinyl chloride, the risks were detected In man after the substances had been shown to induce tumors in laboratory animals, al though little attention was given to the experimental studies when first ,f 92 FRAUMENI URL reported. The effects ofsome carcinogens, particularly asbestos and radon, are greatly potentiated by cigarette smoking, so that programs to reduce either the workplace exposure or nlriig would substantially lowerbnt not eliminate the occupational risk. It is noteworthy that a number ofmanufac turing industries (e.g. furniture, leather, rubber) are associated with cancer risk, but specific carcinogens remain to be identified (30). Industrial hazards and their detection have important implications beyond the workforce, since most agents are not confined to the plant but ultimately become part of the general environment to which large segments of the population may be inadvertently exposed. Pollution Pollutants in the urban air have long been suspected in the etiology of lung cancer, with fossil fuel combustion products, especially polycyclic hydro carbons, being of special concern (33). In several studies the rates for lung cancer have shown correlations with measurements of benzo(a)pyrene in the ambient air, yet the available evidence suggests that the urban excess oflung.cancer is due mainly to cigarette smoking and partly to occupational exposures. In the large-scale survey of the American Cancer Society, ageand smoking-standardized rates for lung cancer were computed among tw>n not occupationally exposed to dust, fumes, or vapors (34). No major differ ences in mortality were seen between urban and rural areas, or between cities characterized by indices of pollution. Another approach has been to extrapolate from studies of workers heavily exposed to hydrocarbons; the results have suggested only small effects from urban air pollutants, on the order of 10 cases of lung cancer per year per 100,000 men with average smoking habits (35). In some studies the effects of smoking a particular amount were greater in urban than rural areas, suggesting that tobacco smoke may interact with carcinogens in the ambient atmosphere (36). Asbestos bodies and calcified pleural plaques have been reported in huge segments of the urban population, but the effects on lung cancer following nonoccupational exposures Are uncertain. It is clear, however, that meso theliomas may result from neighborhood exposure to asbestos industries and from household contact with asbestos dust, particularly through laun dering of work clothing (37). Another hazard may result from airborne levels of arsenic, since high mortality rates for lung cancer have been reported among male and female residents in communities with arsenicemitting smelters (38). Recently interest has centered on contaminants in drinking water, since several halogenated organic compounds (trihalomethaaes) produced during chlorination are carcinogenic or mutagenic in laboratory tests. Levels of these compounds in drinking water have shown geographic correlations CANCER ETIOLOGY 93 with the rates for cancers of the bladder and large bowel (39). Although it is difficult to evaluate causal relationships, efforts are being made through case-control studies that assess lifelong exposures to drinking waters of varying quality (40). Medications A valuable source of leads to carcinogens comes from studies of patients exposed to medicinal agents, which may presently account for about 1 to 2% of all cancers (41). This approach probably ranks second to occupa tional studies as a means of discovering substances that are carcinogenic in man Although some drugs that cause cancer have been withdrawn from clinical practice, others are retained since risk-benefit considerations may warrant their use in certain conditions. A major impetus to research in this area was the discovery in 1971 that synthetic estrogens given during preg nancy produced adenocarcinomas of the vagina and cervix several years later in daughters exposed in utero (42). This was the first demonstration of transplacental carcinogenesis in humans, and also heightened concern about the hazards ofexogenous estrogens given at any time. Subsequently, a aeries ofstudies firmly linked endometrial cancer to the use of conjugated estrogens for menopausal symptoms and then suggested that breast cancer may also occur excessively in exposed women (43). These findings are consistent with epidemiologic and experimental observations on breast and endometrial cancer indicating that the risks depend upon menopausal and ovarian status, and appear mediated by the metabolism of endogenous estrogens (44, 45). The hazards of oral contraceptives are less clear, but there are relationships to liver adenomas, to endometrial cancer among users of the se<pi*nrial type of contraceptives, and possibly to breast cancer among high-risk women (e.g. with benign breast disease or familial predis position) (41). An risk of acute nonlymphocytic leukemia, and perhaps other cancers, has been seen among patients receiving alkylating agents, espe cially malphfllan, cyclophosphamide, and chlorambucil (46). These risks may be acceptable when treating conditions with a poor prognosis such as nwfrffrtfr cancer, but for conditions with a favorable long-term prognosis the benefits of treatment should be carefully balanced against the risks. These drugs may exert their action in part by breaking chromosomes, since other leukexnogens (radiation, benzene) have a similar effect. Immunosuppressive agents have been assessed primarily by studies of renal transplant recipients, most of whom nave had azathioprine and corti costeroids (47). The risk of histiocytic lymphoma is very high, and first appears within months of transplantation. This explosive onset has sug gested that a latent oncogenic virus may be activated by immunologic >4 fiRAUMENI mechanisms. For all other cancer combined, the excess risk is about two fold and first becomes evident about two years after transplantation. It has not affected all forms of cancer as might be predicted by the hypothesis of "immunologic surveillance,** but increased risks have been noted for can cers of the liver, biliary system, and bladder, and for soft-tissue sarcomas (including Kaposi's sarcoma), adenocarcinoma of the lung, squamous car cinoma of the skin, and malignant melanoma (48). Recently, other groups of patients receiving immunosuppressants have shown an excess of lym phomas, squamous carcinoma of the skin, and soft-tissue sarcomas, but at lower rates than those seen in transplant patients (49). It is noteworthy that the predominance of lymphomas with drug-induced immunosuppression is seen also among patients with heritable immunodeficiency syndromes (50). Infection Viruses have not been linked with certainty to the origins of any human cancer, although several candidate agents are under scrutiny (51). The epidemiologic patterns of cervical cancer have long suggested venereal transmission of an infectious agent, with herpes simplex virus type 2 being a chief suspect The Epstein-Barr virus (EBV) is related to nasopharyngeal cancer and Burkitt's lymphoma, particularly in areas of the world where these tumors are highly prevalent Hepatitis-B infection is related to hepato cellular carcinoma, especially in endemic regions of Africa and Asia. There is little evidence that any form of cancer is transmitted from one case to another, and reports of time-space clustering of leukemia and lymphoma have not been confirmed (52). A viral origin for Hodgkin's disease in young adults, however, is suggested by the association with childhood environ ments, such as small family size, that tend to reduce or delay early life exposures to infections, in a manner resembling paralytic poliomyelitis (S3). In some cases EBV may be involved, since antibody titers appear elevated and an increased risk ofHodgkin's disease has been reported among persons with infectious mononucleosis. Two new developments have expanded the opportunities for work in this area. Kaposi's sarcoma shows a geographic concentration in Africa resem bling that of Burkitt's lymphoma and has been related to infection by cytomegalovirus (54). Recently, in the United States, clusters ofthis usually rare tumor have been identified in homosexual men, often in association with opportunistic infections and immunosuppression of unknown cause (55). Another discovery has been the isolation of a retrovirus from T cell leukemias in the United States and in the southwestern part ofJapan where elevated rates have been reported (56). If viruses are oncogenic in humans, it seems likely that predisposing factors are operating (12). Thus, EBV may interact with certain histocom patibility antigens to produce the high rate of nasopharyngeal cancer in CANCER ETIOLOGY 95 populations, with persistent immunostimulation from malarial in fections to induce African Burkitt's lymphoma, or with a genetic im munodeficiency trait to cause family dusters of lymphoma. Hepatitis-B infection may combine with dietary afiatoxin and other cofactors to produce liver cancer in endemic regions. In animal models the production of im munodeficiency enhances viral carcinogenesis, so that the comparatively narrow range of tumors complicating immunodeficiency states of man sug gests that viruses play only a limited role in human cancer. Parasitic infections affect the risk of cancer in certain developing areas of the world (57). In Africa and Papua New Guinea the geographic patterns of malaria and Burkitt's lymphoma are closely correlated; in the Middle East and north Africa schistosomiasis produces squamous carcinoma of the bladder; and in parts of Asia infestation with liver flukes (clonorchiasis and opisthorchiasis) predispose to cholangtocarcmoma. Nutrition International correlations and migrant studies have suggested that certain features of the affluent Western diet contribute to a large but uncertain proportion of all cancers (58). Various nutritional hypotheses are under study by epidemiologic and experimental investigations, although the mechanisms appear complex and difficult to unravel. The role of dietary fat in colon cancer has been suggested in some studies (59), and may act by increasing the concentration of bile acids in the bowel, which are then metabolized by bacterial flora into carcinogens or co-carcinogens (60). Die tary Gat may also alter the risk of breast cancer (61), perhaps by increasing estrogen production or prolactin release (62). It has been suggested that dietary fat and caloric excess may contribute to endometrial cancer, includ ing its associations with obesity, diabetes, and hypertension (45). Obesity is also infl{ftted with gallbladder cancer, which is consistent with its effect on gallstones, the major risk factor for biliary cancer (63). However, there is evidence that a low intake of certain food classes may predispose to cancer. Several studies indicate that the risk of colon cancer is inversely related to the consumption of fiber, which may protect against intestinal carcinogens or precursors by dilutions! or other effects (64). Micronutrients may also have a protective influence, since cancers of the lung and several other sites have been associated with a low intake of vitamin A and -carotene (65). The risk ofstomach cancer has been related to a deficiency of fruits and vegetables, especially those containing vitamin C, which may act by inhibiting the formation of carcinogenic nitrosamines in the stomach (66). In one study ofcolon cancer, patients ingested compar atively small amounts of cruciferous vegetables (e.g. cabbage, brussels sprouts, cauliflower), which contain indole compounds with the potential to inhibit carcinogenesis in laboratory animals (67). In a recent study of '96 JfRAUMENI i URL 05476 cancer etiology 97 esophageal cancer in the United States, the high rate among black men was attributed to heavy alcohol consumption and a complex nutritional defi ciency state that could not be narrowed down to any particular food class or nutrient (68). A variety of other dietary factors, including additives and contaminants, have fallen under suspicion (58). The consumption of aflatoxin, a carcino genic metabolite of the fungus Aspergillusflaws, correlates closely with the distribution of liver cancer in high-risk areas such as Africa. Coffee intake has been associated with bladder cancer and recently with pancreatic cancer (69), but causal relationships have not been established. The artificial sweet eners, saccharin and cyclamate, are weak bladder carcinogens or co-car cinogens in laboratory animals, but a large-scale study of bladder cancer indicates that the risk in man is very small if present at all (70). Cooking practices may release hydrocarbons and other carcinogens in food, although there are no epidemiologic indications that consumption of various cooked foods is related to gastric or other forms of cancer. Genetic Factors Compared to environmental factors in cancer, genetic determinants are less conspicuous and more difficult to identify by clinical and epidemiologic means (71). Although the racial and ethnic differentials for most cancers appear largely modulated by environmental influences, genetic factors ap pear to contribute to some high rates (e.g. nasopharyngeal cancer among Chinese and gallbladder cancer among American Indians and certain His panic groups) and some low rates (e.g. testicular cancer and Ewing's sar coma among blacks in Africa and the United States). Genetic susceptibility is most evident for skin cancer, since ethnic variations correspond to the degree of protective skin pigmentation. Although only a small percentage of cancer is inherited in a Mendelian fashion, over 200 single-gene disorders have been linked to neoplasia (72). Some tumors seem to arise directly as an inherited trait (e.g. bilateral retinoblastoma, familial polyposis coll), while others occur as a complica tion of inherited precursor states (e.g. neurofibromatosis, xeroderma pig mentosum, chromosome instability, and immunodeficiency syndromes). In certain heritable syndromes, environmental factors contribute to the devel opment ofcancer, and provide valuable insights into genetic-environmental interactions and multistage models ofcarcinogenesis. Neoplasms ofa hered itary nature tend to occur earlier in life than do nonfamilial occurrences of the same tumor, and usually arise from multiple fod within the affected organ. In contrast to the hereditary syndromes, the common human cancers show small familial risks, on the order oftwo- to three-fold (73). However, | j | like the hereditary syndromes, the familial risks for breast and colon cancers may be as high as 20 to 30-fold among subgroups of patients with early onset and bilateral or multifocal origin (74). Familial susceptibility also appears to enhance the effects ofenvironmental exposures, such as smoking in lung cancer, estrogenic compounds in breast cancer, and sunlight expo sure in melanomas derived from dysplastic nevi. In some families there are remarkable aggregations of cancer consistent with an autosomal dominant mode of inheritance. These "cancer families" may display either a single type of cancer or a constellation of multiple cancers, especially adenocar cinomas of the colon and endometrium, or the breast and ovary. Other familifs are prone to diverse ceil types of childhood and adult cancers, particularly soft-tissue and bone sarcomas, breast carcinoma, brain tumors, adrenocortical neoplasms, and leukemia (75). The delineation of genetic and familial syndromes is helpful in applying laboratory probes to clarify the heritable component of carcinogenesis, and in targeting screening and prevention programs designed to protect high-risk individuals. CONCLUSIONS Despite the large gaps in present knowledge of cancer epidemiology and etiology, the wide international variation in cancer incidence has suggested that the bulk ofhuman cancer is related to environmental factors and is not an inevitable consequence ofthe aging process (2). This notion has provided a great stimulus to etiotogic research by indicating that cancer is in principle a preventable disease. Although the established risk factors still appear to account for a minority of cancer cases, leads to nutritional factors seem especially promising in resolving key etiologic questions for several com mon cancers (76). Most cancers appear to result from the cumulative effects of multiple factors, including susceptibility states, and initiating and pro moting agents acting at various stages ofcarcinogenesis. Even when specific causes are related to a particular tumor, further work should be encouraged to identify other factors that contribute to the multistage process that eventuates in cancer. The overall impact of genetic factors is difficult to gauge at present, but should not be downplayed as a residual category after subtracting the environmental component. Genetic-environmental interactions may ac count fox a proportion of cancer, sd that the delineation of suscepti bility mechanisms is likely to have far-reaching implications to the ultimate prevention of cancer. Although experimental research is the approach re quired for a full understanding of mechanisms, epidemiologic observations may provide valuable guidance in clarifying, for example, the role of ixnmunosurvetllance in cancer etiology. Jk. t- 45-557 ia3utn no* -dt3*j )u|d$(rejt-[tj tii joouejjo qstg 746! Jf J 7 `noinnuj `g `joaooh '4t \t-UV-Ltl paw f 7Sug w -jooiro rousAO -ouiOJj JOtunx Jo sjxdfr ypoff -dints m -oojj aj 'omsodxo otoo&iv proop -uancoo oj pspdda ttnooSoonnojo jo atp aSajsnptin aqi jo *oopv9fldtBi t -pips lioqoo jo tttXpnn atp joj potpatn jo Xauoqi ixi?fa-8iniiXqi* joyo araiA J*3o y mi o "3J *110 `O O *Q*tua Tf -not ojnoy '4461 O "H `9uoa "Jf 'd'7 `luourmuj `J3AOOJJ ` -g g 'jaajp'H -9^ if-tSE :0i ojotxrj -poff uo|ti|S{iiiiflp fuo dd 6881 ql ioq Xq ciu foiAa*9BS>>dS ob *Xnscqq -JBH 8uuds PlOO ^JOA mon 79*455 auoi-StKri :uszdm) jo tpaja oneSoab . dd `doisuim V T '*/% 'Q 7 `n*!H jfd otp jo uouuirooaxap })ttuiind I `H H P9 vwuotunfj Jo nttSuo iwtj 6461 O `ojroroos *Pn[*H If nj 'jwits puiotnopao oj mrapj JXHtDJ fapods t|jtM stsonofoupjuo wuinq ui 0Z l l *wm|OA cqdKjSoooat jaip jo 9foi oqx -446! g luojjsauy St :*iep [aunro pat araraq no ponq 60l*t4:! 'y tomapfds jomno jsMjq trrainq jo XSoforcDopido sqi jo mwaoj y '6461 ~1 7 `Xpg tt ZZ-LlSVLi Juw jreoiq jo qsu pro osn ttosojtso f*md -oow 1861 Jf d 7 `roanmwj 'yi `opfZS "N a `joaooh "V 1 *uo|opB ft sroamq 01 jwtoo qjM paproosst emu -cud pruisnpin pro i|MUMp jo 0091 -iqBA# tiy 0861 **0 DWl -0E 06-I80I:4t-wnD noi|!puni9!paiB 8tnonoj jooroo 1861 a f `**>8 tOZl~L6U-60Z xmops tp ftpu 8amoof jo esop of tooj; sqsu 18-848 :t8Z T&N 'f 72ug -fj uotuou 3imoX tn ootnueodde jomni qiM Adcjotp jojjs -oqftts pnuoism jo nopapossy atnSeA aqj jo SFajomojaaouo'pv -|46l O Q `jozroqsod ` H `JapPJlil "1 V 'JSRJ3H Tt 08-1401 If jsautr) -jooroo poonpui-Stua 1861 Jf d 7 `ttwimmij `-a 'joaooh It sjd u] 7Jaduaj HH3H vojfdug *.)/& 8uptnup ui sjiue8jo potmuopp jo Xjpraoloinareo jo oauopi -as fWHSoionuoptdg T86T "d "31 `J0jro3 0t 8-646-79 7nif JOaUDQ 7{Off / JOJBM Suiquup tn soucqiouiofaq qjw* Xji|*moui iaotito jo mopEtoosry 8461 7 1 *3CltD0W '{ A `uostjv ` H `-33AOOH "d "X 'iojiiB3 6f n-ziiz ixwrj jootno 8un| pm uoijn[jod ji* |9itu&iy 746! Jf d f 'nistunajd ` f `m *ioig -gf J8-08t:61 T3H -^O 'f 'SJOtunj [aipqiowui jo ^8o|oq)ed on{du8oo8 oqx LL6\ ~1 W 4wnoqoN LZ 9i~H-ZZ Psduaj tfjfoaff vathtug jooueo 8un| oj aouo jojoj [ctoads q|i uoiinpod Jiejo sjaajp 11 '8/161 d `JO{J0pao `i 'tioquj 9c If-CVZZ yi/Diff -uaiu ug uootreo 8unj puv nopnnod ouaqdsouny g6l d `HOQ `?C 1I-90T-6 TH WS P*!"n 3tp nj JKUK! puf uonnuod Jtt [CJ9U0> 0861 1 `Plu^O "D -a `puouranrH tf SSOlIJ qid -AlQf) :JOOtTtta tWUBJ jMroo foroamtg *0861 `H O *Porl *8f 93*6ir6 7*W S3IBIS paipifi atp tn Xiqnima jootos o) aonvipu ftnznx* jo oopnquinoo u. 0861 in ~o f `na "s ,>nf lz dd 60S tiopwis v 9qtud rqJOA **N '99 -6EI dd *op8apsjq *f 7q *naipfoo I "1 ' Jf WD H M V muovoiw luouitpifl uownfj tq 'vtnoropm iron -Sqvai jo iububa Xi^tperoq oqx '6461 jf j f `nnmrojj n 71 *000010 **511 ECtr-t8 (HIN) ^ 1<W M3HO "isui jootod iin :-pw **P*n -og soms P*n*n *f> V J93*BD "PtS DUiovofawuo/if Jo XMUJH5VJ *|86I 'Jf J f Ttcitunajj `-g x tJWJ "f `Ottooe `SZ 94-KI 'dd SI FH S -oopwpo iPS T861 Jf J f 'raatnnBJj `-g 'X <'J * f "opoos >1 0C-4JZ;6 7jy 'ahj -rotn tn jootoo npp pro nopmpaj lOfOtAUim -0861 'i ft f9-tSt J J33UIT) f 7UJ -SJOqjOM XjOAOiq qstroci 3uotro tpop JO tno pro Xltptqiom jootno 6461 *W O `oswof XL zs-is8r6f J9outo j ii9 pntpj-ioqoop ui sopnmjioddo (*9 -iSopmnoptda '6^61 'if *j 7 `potnrojx -\z 64-t4I:6 Wf ,ASJd uopdoms -nos loqo[ 01 a|(|K)nq|Jn jootne jo aoniodojd sqx '0861 7 TI `treramoa xfi nuotfjjy *6X61 *nninn<>d |eiuaamaRA -ogjo cioojp potSoiotg pro potpoH uo 0S-6fI " *9 pa S pqoafV Ji6I 7 7 *nnnpoa 61 33Tjnmuo3--tpncO p*N *ff ssoid u] otyO Itmtuputo *1861 Sl~Zl `un 6t-St4 :tof th r 7m -wis ppn tmipnos oqi tn ooidoa 8ootro joa roo juo pro 8tnddipjpmg *1861 *Jf'd 7 'roonrauj -y *opo|OX -A -j *on9ij "W O *7 M ion '-jq 'a *!A *81 9900S-64 (SHd) h mi M3Ha 'ASH wa sn Oa otpSaniWM XK-fI-<-1 -inO-motmS' jMby p- nffjmff pm fyp/outs I 'Jn ooroqoi jo kujoj JqJO '6461 loflwo JOJ -lMO '41 44-l4t79 vujjoomj 1toff 7* -Xptps topnoo-ono V :qni aoroo nuCn| pro nnt| no o8t -*n oponto joqq auoi-Soot jo pndiuj '6461 fl -S *uuqp>S *n *3 *91 dd e/|j -uoparos wiqdpp^iqa T8-t9S <" "Jf nptonwj -j 7 *ppj -aouoips a T 'yefttmojj pm Oop> -tu&pfdg jnmj sj vnojd pro tinri T86I 7 A *>ia "Jf d 7 *n*BMJd 'SI' C4"69!:6 W WIS peipn ^ m J09TO0 pro tnp|omg *0861 *H wnnpPS "D *3 *ptioimwH >1 68"446:t9 W row FW / Jww jo nottooAOM AJttmja pro tppom o8i njinw *0861 O O "3 N *^a tt dd 4g> uoAa ^JOA *N *f9"IS dd `Mvqs -a -3 *aguo 'D'Vp tmj -ay Jo tutspmfaapf pm uoj/Datfpuapf .Twawma/erp tq -jooroo p toipiiis |*9 -tSofontspidg 6461 Jf d 7 `ronmnia Zl 14-SKX39 Wfjaamo 7J*w / t4-sc6i ptun n m sprop Xiqvuom pro aooopptn joooiq 146! *X *Q *>ouoAns "S *8 *a tt *Ef-4riH)I jofutapKfa f -juf -qoraocu pan (OJinoo jooma joj pm oy :towp jo oqqndoa s,o|ddoa oig ai XiqTjjbai jooroo jo soj -tv *1861 as **a "a-*x `""S "r"z *WD "A 'O tl "6- S *nn "A-7 Y| t)l dd f/7 nwtttnQ v otnuo TJOA N '44-59 '** yg -po nuawy Jo ams P*tl*/J atp pm twpojb atjqnda^ t.ajdoaj tip ui ipjoasa-g jaatno 01 ems pajran otp tn jooroo jo sajopod 9tqdj8ooo 0861 'Jf d 7 1wwnJd *7 76 **>18 '6 89-CtW W/jxmo vp/i 7 loms pnnifi oqj tn osorodn/ noon sosnosip jomo pro mats mag Xiqipojq x *s)rojxnn osoavdif jo soi *PtS *8961 'If ***,,tiJB3 *`M *pooH 1 otiBiuJupH -uopaox '9S"ftI 'dd 'joprj i n tioiStRaiXs 1 -po "Oofoamofitnoftpyunoj atfpua -PS svowafddns tiofoauo "I tproosoj poi80{oniiopido tn sjpspan oouojjmoo J09tx jo swn oqx p86l 3 N `^a `4 *dd amupBOy 50C-C6? <" "Jf raonmuj *d 7 *P* jewoQ pm tSopttg jaawQ Of ipeojddy ayuxrmQjo ysry H9fH ** fuocjatf tq wotrundod qsu-qfpj tq oofl 1UW lootT*tuojtii *5461 -g p *jmp| -9 dd 8I0t *Pmi * *n infqdpptffqd *90f-ltl ** m PJd *3 WH -d 7 po `mpipajf jaaavj tq AJojonnopjdg 746! *S O 'JPH `7 IkwuiHih 7 T80!~I:4S J*wW w/ jaaaoa jipff LL-U61 U*P XjqnKra pro 09aoppm :srpoj poo pro 'X3o[o -itnopido WonpAjns '1861 'JP* "f y *ojny -q -3 S3ii ` jf t f *8onoA t *dd |f *900 J09TO3 uiy :qA J**M mi PTM JWtf jaautr) 0861 '^opos joou3 roouowy 7 IOC I -I6IPI9 vajjaaapj jtojf 7 Xopoj *0}S poipn tn J09TO9JO npu o(q -pNMjo sopranro 0Atmnron{> uooro? jo SOSTO9 oqx 786! 71 Wj "a TPCI T *dd t4S -wmom :ni *PP9 -8mjds 74-09 -dd `ppjrojioips a -po `aoffaauaj pm Gofofuiapwj JtaviQ tq XSofonoaptdo jostoo 01 rfmaptitq [couoistq otuog 5461 a W `"PI'unS 7 paitj ajntojatrj uoiScj83Atn AioiBJoqBi pan nsidoyoiuispido jo sjjog? oqi Saimqajoo saqomidcfe oaiibaouiit siinboi k;m jnqi surerireqoatn 3[jqns ptre vo^duioo q8nojqj job Xbui pue `XSojoiaiopids jo S3iS -ojEjjs jsnsn oqi opnp Xbui sjoiobj osoqt jo omo$ 'Xjrpqpdoosns otjouo? ptre `siuBinijod iBiuouraojiAtio *sosiuia OfuoSoouo `sjojobj oqoqBjoto ptre fenoq uinu ?uipn{3in `spjtrzeq orao3otnoj90 jo joqiuira b oiBppnjo 01 Supdraoj -jb at sitraq Jfqt paqoBOj XjJBoa 0Aq Xbui XSo|opnop;da jo soqoBOidds IBootiBAiosqo fgpopTpBJi oqx tiny 9qi 01 pottsjnd ptre poiBtqBAO ore spraj osoqi jBtp smisito oy uo^oBioiat Xnrat|dpstp;itnui oSunootio 01 iireuodun si II sistSojoraopido jo `sreqBitioniiJodxo `suBtonitp yoje 8atpnjout 'soojnos inojojip Xubui ujojj out00 Xbui uotiBSnBO jooubo oj speo^ pspnn sqx 66 AOCHOI13 H3DNVD iiwo ion iNHwav^d is %00 *. `fRAUMENI 48. Hoover, R. 1977. Effects of drugs: Im munosuppression. See Ref. 45, pp. 369-79 49. Kinlen, L Jn Shed, A.OJL, Pete, J., Dofl, R. 1979. Collaborative United Kingdom--Australasian study of can cer in patients treated with immunosup pressive drugs. Br. Med J. 2:1461-66 30. Ftlipovich, A H., Spector, B. D,, Ker sey, J. 1980. Immunodeficiency in hu mans as a riskfactor h the development of malignancy. Pm. Med. 9:252-59 51. Evans, A. S. 1982. Viruses. See Ref. IS, pp. 364-90 52. Smith, P. G. 1982. Spatial and temporal clustering. See Ref. 15, pp. 391-407 53. Gutensohn, N., Cole, P. 1981. Child hood social environment and Hodg kin's disease, N. EngL J. Med 304: 133-40 54. Giraldo, G.. Beth, E., Huang. E.-S. 1980. Kaposi's sarcoma and its relation ship to cytomegalovirus (CMV). HI. CMV DNA and CMV early in Kapoti's sarcoma. Ini J. Cancer 26t 23-29 55. Hymea, K Cheng, T., Greene, J. B., Prose, N. S, Mama, A_ Ballard, R, William, D. C, Laubcaatcm, L. J. 1981. Kaposi's sarcoma in homosexual men: A report of eight cases. Lancet 2:598-600 56. Poiesz, B. J., Ruscetti, F. W.. Gazdar, A F., Buna, P. A., Minna, J. >., Gallo, R. C. 1980. Detection and isolation of type C retrovirus particles from fresh and cultured lymphocytes of a patient with cutaneous T-ceU lymphoma. Proc. Nad. Acad ScL USA 77:7415-19 57. Buxton, O. J. 1982. Parasites. See Ref. 15, pp. 408-18 58. Armstrong, B. K-, McMkhsd, A. J., MacJ-ennan, R. 1982. Diet. See Ref. IS, pp. 419-33 59. Jam, M., Cook, G. M, Davis, F. G., Grace, M. G,, Howe, G. R., Miller, A B. 1980. A case-control study of diet and colo-rectal canoer. Ini J. Cancer 26:757-68 60. Reddy, B. S. 1976. Dietary factors and cancer ofthelarge bowel Semin. OacoL 3351-59 61. Milkr, A a, Kdly, A. Choi. N. W, Matthews, V., Morgan, R. W,, Munaa, L., Burch, J. D, Feather, J., Howe, G. K, Iain, M. 1978. A study of diet and breast cancer. Am. J. EpUemfoi 107:499-509 62. Wynder, B. L, Hffl, P. 1977. Prolactin, oestrogen, and lipids in bread fluid. Lancet 2:840-41 63. Lew, E. A^ Garfinkd, L. 1979. Varia tions fat mortality by weight among 750,000 men and women. /. Chrun. Die. 32463-76 64. InLAcency Res. Cancer Intestinal Mi- croecology Group. 1977. Dietary fibre, transit-time, faecal bacteria, steroids, and colon canoer in two Scandinavian populations. Lancet 2407-11 65. Peto, R., Doll. R.. Buckley, J. D., Spora, M. B. 1981. Can dietary betacarotene materially reduce hTM" cancer rates? Nature 290401-8 66. Weisburgcr, J. H., Marquardt, H., Mower, H. F., Hirota, N., Mori, H., Williams, G. 1980. Inhibition of car cinogenesis: Vitamin C and the preven tion of gastric cancer. Pm Med 9:352-61 67. Graham, S., Dayal, R, Swanson, M., Mittdman, A, Wilkinson, G. 1978. Diet in tike epidemiology of cancer of the colon and rectum. 7. Nad Cancer Inst 61:709-14 68. Pottem. L M* Morris, L. Biot, W. If Ziegler, R. G., Prsnnniu, J. F. Jr. 1981. Esophageal canoer among black men in Washington. D. G: 1. Alcohol, tobacco, and other risk factors. J. Nad. Cancer hut 67:777-83 69. MacMahon, H, Yea, S-, Trichopoulos, D., Warren, K. Naidi, G. 1981. Coffee and cancer of the pancreas. N. EngL J. MU 304.-630-33 7a Hoover, R. N, Strasser, P. H. 198a Artificial sweeteners and human Mad der cancer: Pretiminarr results. Ferret 1:837-40 *. riwwcw, j. r. jt. tvu. ueneoc xac- ltaonrdt.,IEn.CParndcemrM.ePdhiciliandeeelpdh. iJa.;FL.eHaoAl Febiger. 2nd ed. In press 72. MuraMn, J. J. 1977. Genetic repertory of human neoplasia, In Genetics ofHu man Cancer, ed. J. J. Mulvihill, R. W. Miller, J. F. Fraomenj Jr- pp. 137-43. New York; Raven. 519pp. 73. Fraumeni, J. F. Jr. 1977. pat tens offamilial cancer. See Ref. 72. pp. 223-33 74. Anderson, D. E. 1982. Familial predis position. See Ref. 15, pp. 483-9$ 75. Biattncr, W. A,, McGnfre, D. B., Mul vihill, J. J, lampkin, B. C, HtMnim, J., Fraumeni, J. F. Jr. 1979. Genealogy otcuxtstnAhsDdy.J.AmMUAsoK. 241459-61 76. Miller, A B. 198a Nutrftfan and can cer. Pm MU 9:189-96 I URL 05418 Ann. Rex Public Health. 1961 3:101-39 CONTROL OF CIGARETTE SMOKING FROM A PSYCHOLOGICAL PERSPECTIVE1 Robert C Benfari and Judith K. Ockene Harvard School of Public Health, Department of Behavioral Sciences, Boston, Massachusetts 02115 Kevin M. McIntyre Harvard Medical School, Boston, Massachusetts 02115 sad Veterans Administration Hospital, West Roxbury, Massachusetts 02132 Cigarette smoking has sociological, physiological, and psychological deter minants related to the initiation and maintenance of the behavior as well as important medical and economic implications. The interrelationship of all of these must be understood in order to control the behavior. Epidemiological studies have proven repeatedly that acute myocardial infarction, sudden death, lung cancer, and other disease processes are strongly associated with cigarette smoking (1-4). Nevertheless, 30 billion packages of cigarettes are sold in the United States each year. Helping people to stop smoking cigarettes presents one of the greatest challenges facing public health and preventive medicine in the United States and other industrialized countries today. The strongest case for a causal relationship between tobacco and any specific disease is for lung cancer. Before 1930, lung cancer was a rare disease, ten times less common than stomach cancer; five times less common than colon and rectal cancers (5). Since then, the age-adjusted death rates for lung cancer have increased almost 20-fbld in males. An encouraging epidemiologic finding his been that former smokers have coronary heart disease (CHD) mortality rates well below the rates of cur- The US Government has the right to retain a nonexclusive, royalty-free license in and to any copyright covering this paper. 101