Although dietary contaminants and alcohol in particular contribute significantly to the genesis of the cancer burden in the human population throughout the world, a factor that is probably more important in human carcinogenesis is the effect of dietary nutrients, both macro and micro. We have already noted in Chapter 8 that such dietary constituents demonstrably play major roles in both the spontaneous and induced development of neoplasia in lower animals. In one of the ear- lier, extensive studies of the effect of human diet on cancer development, Hoffman (1937) con- cluded that “overnutrition is common in the case of cancer patients to a remarkable and exceptional degree, and . . . overabundant food consumption unquestionably is the underlying cause of the root condition of cancer in modern life.” Doll and Peto (1981) proposed that as many as one-third or more of human cancers were causatively associated with dietary nutrient factors. As seen from Table 11.2, the IARC has concluded—on the basis of epidemiological and experimental evidence—that dietary factors are a sufficient cause of human neoplasia, specifi- cally those neoplasms listed in the table and potentially some more. There is ample experimental evidence to support this classification, as we saw from Chapter 8. While it is not possible in most instances to study the long-term incidence of cancer in individuals on balanced, restricted diets, inferences may be made from studies of relatively common human conditions such as obe- sity, weight gain, and body stature, where diet appears to play a significant role. In particular, obesity and a high body mass index (BMI), defined as the weight in kilograms of an individual divided by the square of his or her height in meters, are associated with increased mortality in men and women (Garfinkel, 1986; Manson et al., 1995). In middle-aged obese women who had never smoked, the relative risk of death from cancer and cardiovascular disease was 2.1 and 4.1 respectively (Manson et al., 1995). Similar trends had been seen earlier (Garfinkel, 1985). Obe- sity has been associated with increased risks of cancer of the breast, colon, endometrium, esoph- agus, liver, pancreas, and prostate (Osler, 1987; Ingram et al., 1989; Garfinkel, 1985; Møller et al., 1994b). Albanes et al. (1988) have also reported that short stature in both men and women is associated with a reduced risk of cancer, suggesting a role for nutrition early in life in human carcinogenesis. While these data on obesity and stature do not prove that dietary excesses and restriction are related respectively to increases and decreases in human cancer incidence, they are quite suggestive and clearly in line with data from experimental studies (Chapter 8).
Although it is difficult to study the overall effects of caloric restriction and excess on the incidence of human cancer, epidemiological and nutritional investigations directed toward spe- cific nutrients, both macro and micro, have indicated a variety of associations, both positive and negative, with the development of specific human cancers. Table 11.3 gives a tabulation of many of the major nutritional dietary factors involved in site-specific human carcinogenesis. The inci- dence of bladder cancer is not affected by a variety of dietary factors, although Risch et al.
(1988) reported that the daily intake of cholesterol was positively associated with a mild increase in risk of developing this neoplasm. On the other hand, consumption of green and yellow vegeta- bles (G&YV) or increased vitamin C in women was associated with a significantly decreased risk of bladder cancer development (La Vecchia et al., 1989; Nomura et al., 1991). Although animal studies have not supported the cholesterol finding, investigations in animals have demon- strated effects of retinoids and vitamin C in preventing the development of experimental neopla- sia (Sporn et al., 1977; cf. Block, 1991). Serum levels of selenium have also been shown to be inversely related to the incidence of bladder cancer in the human (Helzlsouer et al., 1989).
The evidence for diet as a major role in the incidence of gastrointestinal cancer in humans comes from studies of the lifestyle and dietary habits of particular societal groups. The Seventh Day Adventist population abstains from smoking and drinking, and about 50% of these individ- uals eat a largely vegetarian diet, avoiding the use of coffee, tea, hot condiments, and spices. General cancer mortality in this group is 30% to 50% lower than that in the general population for most cancer sites that are unrelated to smoking and drinking. In particular, cancers of the lower gastrointestinal tract are between 30% and 35% lower than in the controls. Another impor- tant source of evidence for the importance of diet in cancer incidence is the study of cancer inci- dence in migrants moving from one part of the world, where the incidence of a specific cancer is high or low, to another area, where the opposite is true. One of the most striking examples of this is seen in the dramatic decrease within only two generations in the incidence of stomach cancer in Japanese migrating to the United States, and the simultaneous marked increase in the inci- dence of colon cancer in this population on migration (cf. Liu et al., 1993). Willett (1990) has also presented evidence that human cancer risks are inversely correlated with the level of blood retinol and with the dietary content of β-carotene, both derivatives of vitamin A.
A number of studies both in animals (Hawrylewicz et al., 1982) and humans (Toniolo et al., 1994; Vatten et al., 1990) have indicated that diets high in protein or in “red meat” are asso- ciated with increased incidences of mammary neoplasia, but there has been considerable contro- versy on the role of dietary fat in enhancing mammary neoplasia, especially in humans. While a number of studies, including Howe et al. (1991) and Wynder et al. (1997), indicate a significant role for high fat diets in promoting the development of human breast cancer, a number of other investigations have not supported such a role. Rather, evidence suggests that the association is linked to total energy consumption rather than to levels of dietary fat per se (Willett, 1997). Fig- ure 11.3 shows the now classic relationship between fat consumption and breast cancer deaths in a variety of countries (Carroll and Khor, 1975), on which are superimposed the more recent data of Willett et al. (1987). As can be noted, both fat consumption and calories from fat consumed are related directly to the death rate due to breast cancer.
Unfortunately, as pointed out by Schatzkin et al. (1989) and noted in the figure, the rela- tive risk of breast cancer for women in the highest compared with the lowest fat-consuming groups would be only 1.5. Since animal studies support both viewpoints (Freedman et al., 1990; Rose, 1997), it is most likely that both the total energy content of the diet and dietary lipid itself play roles in the development of mammary cancer both in humans and in animals. In general, studies on the role of specific protective factors in mammary cancer development in humans have been conflicting. Freudenheim et al. (1996) described a study in humans that indicated that intake of fruits and vegetables decreased the risk of breast cancer development, and Rose (1990) reported that dietary fiber may have a protective effect because of its influence on estrogen me- tabolism and excretion or by other mechanisms. A protective effect of dietary fiber in N-meth- ylnitrosourea–induced mammary cancer in rats has been published (Cohen et al., 1996), and several studies (Zile et al., 1986; Rao et al., 1990) have indicated an inhibitory effect of retinoids and antioxidant micronutrients in the development of experimental mammary neoplasia.
Figure 11.3 Relation between age-adjusted breast cancer mortality rates, per capita fat consumption, and percentage calories in the diet from fat (after Schatzkin et al., 1989, with permission of the authors).
In the case of colon carcinogenesis, the most convincing data in humans are the associa- tion of red meat in the diet with an increased risk of the development of colon cancer (Giovan- nucci and Goldin, 1997). As discussed earlier in this chapter and in the literature by Potter (1993) and others, this association may be directly related to the production of mutagens and heterocyclic amines (Figure 11.2) during the cooking process. A more controversial association is that of dietary fat with an increased incidence of colon cancer. One of the most detailed stud- ies was that of Willett et al. (1990) among a large cohort of middle-aged and older women. This study demonstrated a positive association of the risk of colon cancer development with increased animal fat intake. Other studies (cf. Statland, 1992) have tended to confirm this finding, but an- other view is that neither case-control nor cohort studies have found that total fat composition of the diet increased the risk of colon cancer (Giovannucci and Goldin, 1997). Interestingly, at the experimental level, Nutter et al. (1983) reported that the feeding of beef protein to mice did not significantly affect the response of the intestines to 1,2-dimethylhydrazine–induced carcinogen- esis. Unfortunately, the study did not indicate the actual treatment of the beef protein. On the other hand, dietary fat enhances experimental intestinal carcinogenesis, both chemically induced (Zhao et al., 1991) and genetically based (Wasan et al., 1997).
In four out of five prospective studies, an association has been noted between low serum cholesterol and an increased risk for colon cancer (cf. Burnstein, 1993). On the other hand, feed- ing of high-cholesterol diets to rodents during chemical induction of colon carcinogenesis sig- nificantly enhanced the number of colon neoplasms per animal (Hiramatsu et al., 1983; Makino et al., 1989). While the reasons for this association are not understood, it is likely that changes in bile acid metabolism may be the important factor. A number of epidemiological studies have revealed that fecal bile acid concentrations are elevated in populations having a high incidence of colon cancer, notably those consuming a high-fat, western-style diet (cf. Cheah, 1990). Bile acids have also been found to act as promoting agents in the experimental induction of colon carcinogenesis in animals (cf. Cheah, 1990). Recently, however, Narisawa et al. (1998) found that one bile acid, ursodeoxycholic acid, prevented N-methylnitrosourea–induced colon carcino- genesis in rats. Gupta and associates (1983) isolated from human feces and determined the struc- ture of a number of hydrocarbon ethers, which they designated as fecapentaenes. These materials are evidently synthesized by bacteria within the colon and exhibit strong mutagenic activity (Kingston et al., 1990). However, their effectiveness as initiators or complete carcino- gens in rats and mice has not been repeatedly verified in animals (Weisburger et al., 1990; Ward et al., 1988; Shamsuddin et al., 1991), and as yet the role of these unique mutagens in the induc- tion of human colon cancer is not clear.
In a recent review, Weinberg (1994) reported that studies in animal models and relatively large groups of humans indicate a positive association of excessive iron with colorectal carcino- genesis. In contrast, administration of excessive calcium salts decreases the risk and develop- ment of colon cancer in both humans (Newmark and Lipkin, 1992) and animals (Pence et al.,1995). In the United Kingdom, a recent study indicated that patients selecting diets containing significantly less fruit and vegetables than were present in the diets of age-matched controls had an increased risk of developing colorectal cancer (Matthew et al., 1997). In experimental studies, Narisawa and associates (1996) had shown that a variety of natural carotenoids found in fruits and vegetables inhibited the development of preneoplastic aberrant crypt foci in the colons of rats receiving N-methylnitrosourea. There is now developing a substantial body of epidemiolog- ical data indicating a relationship between the status of the vitamin folic acid and the risk of developing colorectal neoplasia (Mason, 1994). These studies indicate that folate deficiency is related to an increased risk of colorectal neoplasia—a finding that is clearly supported by animal investigations (Mason, 1994; Cravo et al., 1992). Dietary fiber content has also been associated with a decreased risk of cancers of the colon and rectum in a number of epidemiological studies (Howe et al., 1992). This has been confirmed more recently (Le Marchand et al., 1997), where study of a cohort of a variety of ethnic groups indicated a strong, dose-dependent, inverse associ- ation in both sexes with fiber intake measured as crude fiber, dietary fiber, or nonstarch poly- saccharides. Experimental evidence supports these findings (Jacobs, 1986), although experimental studies indicated that certain fibers were much more effective than others in such preventive effects. Matthew et al. (1997) reported that a group of patients at increased risk of colorectal cancer selected diets containing significantly less fruit and vegetables than those not at risk. This further supports the more general finding that diets high in fruits and vegetables tend to decrease the risk of a number of cancers (Hirayama, 1994).
Studies on the association of dietary factors with endometrial cancer, like cervical cancer, have not been extensive. In at least two studies (Zheng et al., 1995a; Levi et al., 1993), evidence was presented that consumption of various types of meats was associated with an increased risk of this disease. However, other investigations found either the reverse (Barbone et al., 1993) or no significant effect (Tzonou et al., 1996). The latter authors, studying a cohort in Greece, did present evidence that increased intake of monounsaturated fat, mostly olive oil, was somewhat protective. Few if any animal studies in this area have surfaced.
Carcinoma of the esophagus has long been known to be associated with smoking and al- cohol abuse (Gao et al., 1994a; Launoy et al., 1997; Gammon et al., 1997). Dietary factors are also directly as well as indirectly related to the development of esophageal cancer in the human. At least two studies have demonstrated an increased risk of this lesion in populations consuming corn or maize as a predominant portion of their diet (Sammon and Alderson, 1998; van Rensburg,1981). Sammon and Alderson (1998) have postulated that one possible mechanism concerns the presence of high concentrations of linoleic acid, a precursor of gastric prostaglandin synthesis, in maize. Together with other dietary abnormalities, this may lead to an increased concentration of prostaglandin E2 in the gastric mucosa, with subsequent pyloric relaxation and related events leading to a greater reflux of gastric juices into the esophagus. It is this mechanism that is felt to be important in the development of a preneoplastic condition termed Barrett esophagus, the inci-dence of which has increased dramatically in North America and western Europe in recent years (Cameron, 1994). Approximately 10% of individuals with some degree of acid reflux in the esophagus suffer from this condition, the overall incidence of which may be in the neighborhood of 1 in 80 adults in the general population in these areas (cf. Cameron, 1994). Although esoph- ageal cancer is a relatively uncommon cause of death in patients with Barrett esophagus (van der Burgh et al., 1996), patients with this condition have a 30- to 125-fold greater risk of developing esophageal cancer than the general population (cf. Cameron, 1994). Smoking and alcohol do not appear to be related to the predisposition of the changes seen in Barrett esophagus, but they are strongly associated with the development of adenocarcinoma in patients with established Barrett esophagus (Gray et al., 1993). Thus, diet, which in turn affects gastric acidity and reflux of gas- tric contents into the esophagus, may be related to an increased incidence of esophageal carci- noma. Increased intake of animal fat and cholesterol have also been related to an increased incidence of esophageal cancer (Launoy et al., 1998; Kaul et al., 1986).
However, no significant associations were reported with total calories and esophageal cancer incidence (Brown et al.,1995), but salty and fried food intake was associated with increased risk (Gao et al., 1994b). The protective action of antioxidant vitamins such as A, C, and E, as well as fruits and vegetables, has been reported in several studies (Zheng et al., 1995b; Gao et al., 1994b; Brown et al., 1995; Launoy et al., 1998). At least one study in animals has shown a protective effect of vitamin E on nitrosamine-induced esophageal neoplasia in mice (Odeleye et al., 1992). Although one study involving dietary supplementation in a cohort in the People’s Republic of China showed no spe- cific effect of antioxidant vitamins and zinc on the prevalence of premalignant lesions of the esophagus, those individuals with large increases in these components in the blood were more likely to have a histologically normal esophagus at the end of the trial (Wahrendorf et al., 1988). In contrast, zinc deficiency in rats caused the development of a significantly higher incidence of esophageal and forestomach neoplasia in animals administered precursors of the carcinogen N- nitroso-N-benzylmethylamine (Fong et al., 1984). Thus, while diet may not directly be a major cause of esophageal cancer in the human, indirect effects leading to esophageal reflux and the significant potential for prevention indicate that diet plays an important role in the ultimate de- velopment of this disease in the human.
Cancer of the gallbladder is not uncommon in the human but is unusual in most animal species, even in experimental studies. In the human, studies have indicated that both a high- carbohydrate diet as well as one containing high levels of fats and oils and a high total energy intake are associated with an increased incidence of this disease (Moerman et al., 1993; Zatonski et al., 1997). These factors may also be related to the close association of biliary tract cancer and the presence of gallstones and obesity (cf. Tominaga and Kuroishi, 1994). In one Japanese study, intake of animal proteins and fats, as well as ingestion of fruits and vegetables, decreased the risk of gallbladder cancer (Kato et al., 1989). In contrast, our knowledge of dietary factors in the development of liver cancer, primarily hepatocellular carcinoma, comes from a knowledge of infectious agents. including the hepatitis B and C viruses (Chapter 12) and dietary contaminants such as aflatoxin (see above).
Alcohol also plays a significant role in the development of cancer of this organ, but specific major dietary factors have not been studied extensively in the human. In one study (Yu et al., 1995), a low consumption of dark green vegetables was associated with an increased risk of hepatocellular carcinoma, as was a low serum retinol level. In animals a number of studies have shown the effects of specific dietary constituents on the development of liver neoplasia (Chapter 8). He et al. (1997) demonstrated that dietary extracts of carrots, toma- toes, and orange juice decreased the number of preneoplastic foci during initiation and also in- hibited the growth of focal lesions during the stage of promotion. In general, administration of retinoids during hepatocarcinogenesis inhibits the process (Mack et al., 1990; Moreno et al.,1995; Gradelet et al., 1998). However, retinol itself reportedly enhanced carcinogenesis by 3′- methyl-4-dimethylaminoazobenzene (Ohkawa et al., 1991). Some studies actually indicated that β-carotene and related compounds were more effective than vitamin A and its derivatives in in- hibiting the development of preneoplastic lesions in experimental hepatocarcinogenesis (Sarkar et al., 1995; Moreno et al., 1995). Similarly, vitamin E reportedly prevented the induction of preneoplastic enzyme-altered foci in rats (Ura et al., 1987), but it actually enhanced the growth of hepatic focal lesions in mice (Kolaja and Klaunig, 1997).
Dietary factors as causative of lung cancer have not been extensively investigated, since the primary cause of this disease is tobacco abuse. High dietary fat may increase the risk of lung cancer (Byers et al., 1987), and cholesterol intake was associated with significant increased risk in one investigation by Jain et al. (1990). Most studies of the relationship of dietary factors to lung cancer risk have been concerned with antioxidant vitamins and related dietary constituents both in the pure form and within fruits and vegetables. Le Marchand et al. (1989) demonstrated a strong inverse association between lung cancer risk and the intake of all vegetables, while an- other investigation (Yong et al., 1997) indicated a protective effect of both fruits and vegetables, especially those containing carotenoids. Ingestion of crude retinoids and carotenoids, especially β-carotene and vitamin E from fruits and vegetables as measured by serum levels of these com- ponents, indicated a protective action in decreasing the risk of lung cancer (Yong et al., 1997; Menkes et al., 1986; cf. Fontham, 1990). In contrast, two studies on vitamin E and β-carotene supplements to smokers demonstrated the absence of any preventive effect and found that β- carotene supplementation may slightly increase lung cancer incidence in cigarette smokers (Albanes et al., 1996; Omenn et al., 1996). An earlier study by Blot et al. (1994) in a different population suggested that supplementation with β-carotene, vitamin E, and selenium decreased the risk of lung cancer in smokers. In part, these discrepancies may be related to the fact that several reports indicate that vitamin E and carotenoids may exert their effects in patients who are relatively light smokers compared with those smoking heavily (Knekt, 1993; Yong et al., 1997). A preventive effect on the incidence of mesothelioma was also related to the intake of vegeta- bles, especially cruciferous vegetables, whose consumption was inversely related to risk of this lesion in the lung induced by a different carcinogen—asbestos (Schiffman et al., 1988). Simi- larly, the prevalence of bronchial metaplasia, a preneoplastic condition, in asbestos-exposed workers was inversely related to the dietary intake of vitamin A (Mayne et al., 1998). Thus, sim- ilar dietary relationships have been noted for two entirely different histological types of neo- plasms within the pulmonary cavity in the human. In nonsmokers, both men and women, dietary β-carotene, raw fruits and vegetables, and vitamin E supplements reduced the risk of lung cancer (Mayne et al., 1994).
While there are few if any data on the effect of diet on leukemias, myeloma, and sarcomas, at least one study (Chiu et al., 1996) did demonstrate that excessive dietary fat and animal protein increased the risk of non-Hodgkin lymphomas in older women. However, considerable data have accumulated, as noted earlier (Chapter 6), on the effect of retinoids, especially trans- retinoic acid (Tallman et al., 1997), in inhibiting the growth of acute promyelocytic leukemia. In this instance, administration of high levels of these retinoids induced differentiation with the potential for increased apoptosis of leukemic cells, the mechanism of which was discussed earlier (Figure 6.6). In a Japanese cohort, daily meat consumption was significantly associated with the occurrence of ovarian cancer (Mori and Miyake, 1988). The consumption of carrots and presumably β-carotene was found to be inversely associated with the risk of this neoplasm (Engle et al., 1991).
Several studies have indicated dietary factors as causative in the development of pancre- atic cancer in the human. Increased intake of dietary meat as well as increased caloric intake reportedly enhance the risk of developing this lesion (Farrow and Davis, 1990; Ohba et al., 1996; Silverman et al., 1998). At least one study of a cohort from Poland indicated a strong positive association with the intake of cholesterol (Zatonski et al., 1991) as well as some positive associ- ation with dietary protein. However, Farrow and Davis (1990), studying a cohort in western Washington in the United States, did not find any effect of dietary cholesterol. In rodents, espe- cially rats, dietary fat enhances or promotes the development of preneoplastic acinar lesions in the pancreas (cf. Roebuck, 1992). This is primarily true of dietary unsaturated fats of the omega-6 but not the omega-3 series (Chapter 8). Thus, promotion of these lesions by dietary fat appears only partly related to the high caloric content of the fat itself. Frequent consumption of fruits, vegetables (particularly carrots), and plant foods/protein lowered the risk of pancreatic cancer (Norell et al., 1986; Ohba et al., 1996). While no exactly comparable animal studies have been carried out, several investigations have demonstrated the inhibition of the development of preneoplastic foci in the pancreas of azaserine-treated rats by the administration of synthetic retinoids (Roebuck et al., 1984), β-carotene, and selenium (Woutersen and van Garderen- Hoetmer, 1988).
Just as with cancer of the breast, a major dietary factor in the development of prostatic carcinoma is dietary fat (Giovannucci, 1998; Whittemore et al., 1995), which finding is also re- flected in the relationship of obesity to the risk of fatal prostate cancer (Snowdon et al., 1984). As in mammary cancer, there is also evidence that total energy intake is related to the develop- ment of preclinical prostate cancer (Meyer et al., 1997). However, not all investigations have supported these findings (e.g., Key et al., 1997). In a recent review, Giovannucci (1998) argued that the strongest correlation for a positive association of dietary factors with prostate cancer exists with dairy products, especially milk and high-fat milk. Just as in Figure 11.3, studies have demonstrated a quasilinear correlation between age-adjusted prostate cancer mortality and di- etary animal fat (Rose et al., 1986). These workers have thus put the two relationships together and demonstrated an interesting quasilinear relationship between the age-adjusted mortality of prostate cancer and that of breast cancer in populations of a number of countries throughout the world (Figure 11.4). Dietary protein/meat intake has also been positively associated with an in- creased risk of prostate cancer in several studies (Heshmat et al., 1985; Talamini et al., 1992; Vlajinac et al., 1997). As with a large number of neoplastic cell types, fruits and vegetables and their constituents, especially carotenoids, have generally been found to play a protective role in the development of prostate cancer (Ohno et al., 1988; Mettlin et al., 1989; Mills et al., 1989). The single vegetable found most influential in several studies is the tomato, and within this vegetable the presence of the carotenoid lycopene appears to be a consistently effective inhibitor of the development of this lesion in humans (Giovannucci et al., 1995; Clinton et al., 1996). Although β-carotene and related carotenoids have been shown in some studies to be negatively associated with the risk of prostate cancer, other investigations have argued the opposite (Kolonel et al., 1988). Later investigations by this group in Hawaii (Le Marchand et al., 1991) have tempered their earlier findings, but several other studies have also shown a positive or no effect of vitamin A and related retinoids on the risk of prostate cancer (cf. Giovannucci, 1998). Interestingly, an inverse relationship between serum retinol and the risk of prostate cancer has been reported by at least two groups (Hsing et al., 1990; Hayes et al., 1988). In animal investiga- tions, Wynder and associates (1994), in a review, have pointed out that the preponderance of studies on the effect of high fat intake in experimental prostate carcinogenesis have not reported any effect, although caloric restriction may cause a significant inhibition of tumor incidence. On
Figure 11.4 Relationship between breast and prostate cancer mortality rates for 30 countries for the years 1978–79. The axes indicate the age-adjusted mortality rate (MR) per 100,000 individuals. (Adapted from Rose et al., 1986, with permission of the authors and publishers.)
the other hand, synthetic retinoids fed in the diet can inhibit chemically (Pollard et al., 1991) and genetically induced prostate cancer in rodents (Slawin et al., 1993). In contrast to the extensive studies on the effect of diet on prostate cancer risk, relatively few investigations have been car- ried out on the effect of diet on the development of renal carcinoma. Two studies (Mellemgaard et al., 1996; Wolk et al., 1996) have indicated the importance of total dietary energy intake in the risk of renal cell carcinoma, but they found discrepant results on the protective effect of fruits and vegetables. These investigations included populations from Australia, Denmark, Sweden, and the United States.
Unlike many other neoplasms of the gastrointestinal tract and their related components, a major dietary factor associated with increased risk of gastric cancer is the high salt or sodium chloride content of ingested foods. This finding has been demonstrated in numerous investiga- tions in the human, several of which are listed in the table and confirmed with experimental studies in rodents (Takahashi et al., 1994). In animals, the promoting or enhancing effect of so- dium chloride on the development of gastric cancer can be related to a concentration-dependent tissue damage and subsequent cell proliferation in the stomach mucosa of the rodent (Charnley and Tannenbaum, 1985; Furihata et al., 1996). As a possible factor in these toxic effects of the salt, Takahashi et al. (1991) demonstrated that administration of sodium chloride is associated with enhanced lipid peroxidation in the gastric mucosa. In addition to gastric carcinogenesis, the development of nasopharyngeal carcinoma, while primarily related to infection with the Epstein-Barr virus (see below) has also been positively associated with a high-salt diet (Yu et al.,
1986; Zou et al., 1994). The ingestion of sodium nitrate as a potential precursor of nitrite and subsequent nitrosation of secondary amines within the stomach (see above) have been argued to be potentially significant factors in the dietary causation of stomach cancer in the human (Hart- man, 1983). However, other studies do not appear to bear this out (Pobel et al., 1995), and stud- ies on the carcinogenicity of sodium nitrate in rats were negative (Maekawa et al., 1982). Pickled vegetables, which may also contain high levels of salt as well as possible carcinogenic nitro- samines (see above), have also been associated with an increased risk of gastric cancer (Ahn,
1997; Ramón et al., 1993). The relatively high incidence of gastric cancer in Iceland may be related to the high level of smoked fish consumed in that country. But as yet the dietary factors involved in the high morbidity and mortality rates of stomach cancer in Japan have not been clearly identified. The suggestion has been made that one reason for the decrease in gastric can- cer seen in the United States over the past four decades is the addition of antioxidants to dairy products and other foods containing fat; however, this proposal has not been verified. As with other cancers of the gastrointestinal tract, consumption of fruits and vegetables is inversely asso- ciated with the incidence of stomach cancer in the human (Coggon et al., 1989; Graham et al.,1990; Tuyns et al., 1992). In mainland China, it appears that ingestion of Chinese cabbage, a staple food for lower income groups, may have an important role in reducing the risk of the development of stomach cancer (Hu et al., 1988). The components of fruits and vegetables that have been most closely associated with their preventive effect are ascorbic acid and β-carotene (Hansson et al., 1994; Kromhout and Bueno-de-Mesquita, 1997). The most important factor ap- pears to be the concentration of vitamin C within the gastric juice rather than serum levels of ascorbate (Correa et al., 1998). In some animal studies (cf. Krinsky, 1994), administration of carotenoids may inhibit chemically induced development of gastric carcinoma.