Aflatoxins

A very extensive effort has gone into the study of this group of mycotoxins, especially to examine its most potent member, aflatoxin B₁. Much more is known about the occurrence and toxicity of the aflatoxins than about any other mycotoxin and, probably, most other natural contaminants.

The scattered data pertaining to worldwide occurrence of aflatoxins in food were compiled for a conference on mycotoxins, which was sponsored by the Food and Agriculture Organization, the United Nations Environment Program, and the World Health Organization (1977). More recently, Stoloff (in press) compiled data on the occurrence of aflatoxins in the United States.

The aflatoxin-producing molds Aspergillus flavus and A. parasiticus are ubiquitous. They are frequently encountered as outgrowths on stored commodities under conditions prevailing in many tropical areas. In the United States, aflatoxin contamination is generally restricted to those crops invaded by the aflatoxin-producing molds before harvest: most frequently peanuts, corn, and cottonseed, and to a much lesser extent tree nuts, including almonds, walnuts, pecans, and pistachios. The extent of contamination is greater in the southeastern United States.

In the United States, humans are exposed to aflatoxin mostly from corn and peanuts (U.S. Food and Drug Administration, 1979). Other direct dietary sources, such as tree nuts, are of minor significance, either because contamination is infrequent or because only small quantities are consumed.

It is unlikely that secondary exposures result from the ingestion of aflatoxin residues in tissues of animals fed aflatoxin-contaminated feed (Stoloff, 1979), except for aflatoxin M₁, a metabolite that appears in the milk of lactating mammals exposed to aflatoxins. But, although large amounts of milk are consumed, this exposure is negligible compared to the direct exposure from peanuts and corn.

Aflatoxins are classified as unavoidable contaminants. In the United States, the maximum allowable limit of total aflatoxins in consumer peanut products is currently 20 μg/kg (U.S. Food and Drug Administration, 1980b).

Epidemiological Evidence. Oettlé (1965) was the first investigator to draw serious attention to the hypothesis that aflatoxin ingestion might cause liver cancer. He suggested that the geographic distribution of liver cancer in Africa could be explained by differing levels of exposure to aflatoxin in the diet. Keen and Martin (1971) reported an apparent association between the consumption of groundnuts contaminated with aflatoxin and the occurrence of liver cancer in different areas of Swaziland. Alpert et al. (1971) made a similar correlation of contaminated foodstuffs and incidence of hepatoma by tribe and by province or district in Uganda. In a later study in Swaziland, Peers et al. (1976) analyzed aflatoxin levels in foods consumed by a representative sample of the population in 11 geographic areas. He reported a significant correlation between aflatoxin contamination and incidence of primary liver cancer among adult males. A similar study in the Murang'a district of Kenya (Peers and Linsell, 1973) indicated that there was a correlation between aflatoxin levels in dietary staples of three district subdivisions and the incidence of liver cancer. Mozambique has particularly high rates of liver cancer, perhaps the highest in the world, and studies of aflatoxin contamination of foods indicated that the estimated daily intake of aflatoxin in that country was higher than that reported for any other country (van Rensburg et al., 1974). One problem recognized by the researchers in all of these studies is the inadequacy of the data on liver cancer incidence, since cancer registration is not well established in these areas.

Detailed studies of aflatoxin contamination of ingested foodstuffs have also been conducted in Thailand, where there was an overall correlation between estimated aflatoxin intakes in two regions and liver cancer incidence (Shank et al., 1972a, b; Wogan, 1975). The frequency with which aflatoxin was detected in foods has also been correlated with liver cancer mortality in Guangxi province in China (Armstrong, 1980). In Taiwan, where liver cancer mortality rates are high, Tung and Ling (1968) reported that dietary staples (e.g., peanuts and peanut oil, which is widely used in cooking) are frequently contaminated with aflatoxin.

Linsell and Peers (1977) observed a strong correlation between estimated levels of aflatoxin ingested and liver cancer incidence from various studies conducted in Africa and Asia. They further noted that there were no areas where high levels of aflatoxin ingestion have been associated with low rates of liver cancer.

Although the studies described above suggest that aflatoxin causes primary hepatocellular carcinoma (PHC), numerous other reports have also documented a high correlation between PHC and exposure to hepatitis B virus (Chien et al., 1981; Prince et al., 1975; Simons et al., 1972; Tong et al., 1971; Vogel et al., 1970). These studies do not indicate whether present or past exposure to this virus is more closely associated with the development of PHC. However, Kew et al. (1979) reported that active hepatitis B viral infection is present in approximately 80% to 90% of the patients with PHC. Approximately 5% to 10% of the victims of hepatitis B infection actually develop chronic active hepatitis with persistent liver damage. The liver cells of these individuals are believed to regenerate more rapidly, thereby increasing the likelihood that a biochemical lesion that initiates neoplasia will become fixed in the genes of the subsequent cell population.

The worldwide occurrence of hepatitis B viral infection is similar to that of primary hepatocellular carcinoma. However, it is possible that the influences of aflatoxin and hepatitis B virus on the risk for PHC are not completely independent. Van Rensburg (1977) reviewed the evidence for both risk factors and concluded that preexisting viral infection is probably a prerequisite for malignant transformation by aflatoxin.

The possibility that aflatoxin may also be involved in the etiology of esophageal cancer is suggested by the correlation between mortality from esophageal cancer and the consumption of large amounts of pickled vegetables and other fermented or moldy food in Linxian county of Henan province in northern China (Yang, 1980). Although Aspergillus flavus has been isolated from some products, it is difficult to determine the role of aflatoxin in the etiology of this disease because these foods also contain other fungal species, mutagens, and carcinogens, including N-nitroso compounds.

Epidemiological studies have not been undertaken in Western countries, but there have been reports indicating the presence of aflatoxin B₁ in autopsy samples from liver cancer patients in Czechoslovakia (Dvořačková et al., 1977), New Zealand (Becroft and Webster, 1972), and the United States (Siraj et al., 1981). Siraj et al. (1981) detected aflatoxin B₁ in four of the six liver samples obtained from patients with PHC in the United States. The significance of these findings is not yet known.

Experimental Evidence: Carcinogenicity. Aflatoxin B₁ is the most potent hepatocarcinogen known, being about 1,000 times more powerful than butter yellow (p-dimethylaminoazobenzene) in rats. The carcinogenicity of aflatoxins has been examined in several studies in a variety of species and strains of laboratory animals, including mice, marmosets, tree shrews, trout, ducks, rhesus monkeys, hamsters, and several strains of rats (Wogan, 1973). Of the various species tested, the male Fischer 344 rat was the most sensitive to aflatoxin-induced carcinogenesis (Wogan, 1973).

Aflatoxin B₁ induced mainly hepatocellular carcinomas in rats. However, other studies in rats have indicated that it may also induce a very low incidence of carcinomas of the glandular stomach (Butler and Barnes, 1966), cancers of the colon (Newberne and Rogers, 1973; Wogan and Newberne, 1967), renal epithelial neoplasia (Epstein et al., 1969), and lung adenomas (Newberne et al., 1967). Within a susceptible species and strain, males are much more susceptible than females to challenge with aflatoxin (Wogan and Newberne, 1967).

Mice are resistant to aflatoxin-induced carcinogenesis under conditions that result in 100% tumor incidence in Fischer rats. However, hepatomas were induced in 82 of 105 inbred (C57BL X C3H)F₁ mice injected intraperitoneally during the first 7 days after birth with doses of aflatoxin B₁ as low as 1.25 μg/g body weight (bw) and killed 82 weeks later (Vesselinovitch et al., 1972).

In comparison to Fischer rats, nonhuman primates (170 animals in 12 different investigations) were relatively resistant to aflatoxin-induced carcinogenesis (Stoloff and Friedman, 1976). Liver tumors do not occur spontaneously in monkeys (O'Gara and Adamson, 1972), but a female rhesus monkey developed a primary liver carcinoma after ingesting approximately 500 mg of aflatoxin B₁ over a 6-year period (Adamson et al., 1973). In another study, one of nine marmosets developed liver tumors after 50 weeks on a diet (5 days a week) containing aflatoxin B₁ at 2 μg/g (Lin et al., 1974). However, the authors also observed liver cirrhosis,which is not a symptom of aflatoxicosis in rats. Reddy et al. (1976) reported that 9 of 18 tree shrews intermittently fed aflatoxin B₁ at 2 μg/g diet developed liver cancers after 74 to 172 weeks of treatment.

Experimental Evidence: Mutagenicity. Aflatoxin B₁ was shown to be mutagenic to Salmonella typhimurium strains TA98 and TA100 with and without S9 fraction (Ueno et al., 1978). It was positive in the Bacillus subtilis rec assay (Ueno and Kubota, 1976). In FM3A mouse cells, aflatoxin inducd 8-azaguanine-resistant mutants as well as chromosome aberrations (Umeda et al., 1977). Aflatoxin M ₁, the metabolite of aflatoxin B₁, was mutagenic in the Ames test (Wong and Hsieh, 1976), but inactive in B. subtilis rec assay (Ueno and Kubota, 1976).

Other Mycotoxins

Table 12-1 summarizes the data on the occurrence, carcinogenicity, and mutagenicity of mycotoxins other than aflatoxins that may be found in food. Although most of these mycotoxins are mutagenic in bacterial systems and other short-term tests and/or are carcinogenic in laboratory animals, there are no epidemiological studies pertaining to their role in neoplasia in humans.

TABLE 12-1

Occurrence, Carcinogenicity, and Mutagenicity of Mycotoxins Other than Aflatoxins.

Summary and Conclusions: Aflatoxins and Other Mycotoxins

A consistent body of evidence, all based on correlational data, associates the contamination of foods by aflatoxin with a high incidence of liver cancer in parts of Africa and Asia, but there is no epidemiological evidence that aflatoxin contamination of foodstuffs is related to cancer risk in the United States. Epidemiological studies have also indicated a high correlation between primary hepatocellular carcinoma and exposure to hepatitis B viral infection. Aflatoxin is carcinogenic in several species of animals, including rats, mice, trout, ducks, monkeys, and marmosets, and there is evidence of dose response. It induces mainly tumors of the liver and, to a lesser extent, tumors in the kidney, lung, stomach, and colon, more readily in males and in the young. The carcinogenicity of aflatoxin is paralleled by its mutagenicity in various systems.

There is no reliable information about the role of other mycotoxins in carcinogenesis in humans.

Agaricus bisporus

Agaricus bisporus is a commonly eaten cultivated mushroom in Europe, North America, and other parts of the world. The exact consumption figures for Agaricus bisporus are unknown, but the U.S. Department of Agriculture (1981) has estimated that approximately 213 million kilograms of this mushroom were available for consumption (production and imports) in the United States during 1980.

Agaricus bisporus contains agaritine -- β-N-[γ-L(+)-glutamyl]-4-hydroxymethylphenylhydrazine (Toth et al., 1978) -- and 4-(hydroxymethyl)benzenediazonium ion (Levenberg, 1962). 4-Hydroxymethylphenylhydrazine and 4-methylphenylhydrazine, which are breakdown products of agaritine, have also been found in A. bisporus (Levenberg, 1964).

Experimental Evidence: Carcinogenicity. N'-Acetyl-4-(hydroxymethyl)phenylhydrazine as a 0.0625% solution in drinking water administered continuously to Swiss mice from 6 weeks of age to the end of their lives induced lung and blood vessel tumors (Toth et al., 1978).

4-(Hydroxymethyl)benzenediazonium tetrafluoroborate administered to Swiss mice in 26 weekly subcutaneous injections at 50 μg/g bw resulted in an increased incidence of tumors of the subcutis and skin (Toth et al., 1981).

4-Methylphenylhydrazine hydrochloride administered to Swiss mice in 7 weekly intragastric instillations of 250 μg/g bw induced lung and blood vessel tumors (Toth et al., 1977).

Experimental Evidence: Mutagenicity. N'-Acetyl-4-(hydroxymethyl)phenylhydrazine was most mutagenic in S. typhimurium TA1537 without metabolic activation, and it exhibited marginal DNA-modifying activity only when the S9 fraction was included (Rogan et al., in press).

4-(Hydroxymethyl)benzenediazonium tetrafluoroborate was weakly mutagenic in TA1535 and strongly mutagenic in TA1537, exhibiting toxicity in both strains (Rogan et al., in press).

Agaritine produced equivocal results in both in vitro assays. There was a slight enhancement of mutagenicity in S. typhimurium TA1537 without metabolic activation, and marginal DNA-modifying activity in the presence of S9 fraction (Rogan et al., in press).

4-Methylphenylhydrazine hydrochloride was also found to be mutagenic with and without S9 fraction in S. typhimurium TA98 and TA100 (Shimizu et al., 1978).

Gyromitra esculenta

Each year, approximately 1 million people throughout the world eat the mushroom Gyromitra esculenta (Simons, 1971); 100,000 of these people reside in the United States (S. Miller, personal communication). The literature contains more than 500 reports of poisonings resulting from the ingestion of this mushroom. Some of these incidents were fatal (Franke et al., 1967).

Experimental Evidence: Carcinogenicity. Eleven hydrazines and hydrazones have been identified in G. esculenta. Studies have been conducted to determine the carcinogenicity of many of these compounds.

Continuous administration of 0.0078% N-methyl-N-formylhydrazine (MFH) in drinking water to 6-week-old outbred Swiss mice for life produced tumors of the liver, lung, gallbladder, and bile duct. A higher dose (0.0156% MFH) given under identical conditions had no tumorigenic effect, since it proved too toxic for the animals (Toth and Nagel, 1978). Subsequently, the carcinogenicity of MFH was confirmed in mice (Toth and Patil, 1980, 1981) and in Syrian hamsters (Toth and Patil, 1979).

Acetaldehyde methylformylhydrazone, the main ingredient of G. esculenta, was administered to Swiss mice in propylene glycol in 52 weekly intragastric instillations at 100 μg/g bw (Toth et al., 1981). The treatment induced tumors of the lungs, preputial glands, forestomach, and clitoral glands.

Drinking water solutions of 0.001% hydrazine, 0.01% methylhydrazine, and 0.001% methylhydrazine sulfate were administered continuously to 5-and 6-week-old randomly bred Swiss mice for their lifetimes. Hydrazine and methylhydrazine sulfate significantly increased the incidence of lung tumors in Swiss mice, whereas methylhydrazine enhanced the development of this neoplasm by shortening its latent period (Toth, 1972).

A 0.01% solution of methylhydrazine was administered daily in the drinking water of 6-week-old randomly bred Syrian golden hamsters for the remainder of their lifetimes. The treatment produced malignant histiocytomas of the liver and tumors of the cecum (Toth and Shimizu 1973).

Experimental Evidence: Mutagenicity. N-Methyl-N-formylhydrazine, which is present in G. esculenta, was mutagenic only in S. typhimurium TA1537 without activation and had no DNA-modifying activity (Rogan et al., in press).

Methylhydrazine was mutagenic in S. typhimurium TA1535 and TA1537. The addition of S9 fraction activating system enhanced the mutagenicity in both strains (Rogan et al., in press). The DNA-modifying activity was observed earlier by von Wright et al. (1977).

Summary and Conclusions: Hydrazines

Studies have shown that some chemical constituents of the Agaricus bisporus mushroom are carcinogenic in mice and mutagenic in bacterial systems. One constituent has also been shown to be carcinogenic in hamsters. But the findings of these studies are not sufficient for conclusions to be drawn concerning the risk to humans.

Some derivatives of hydrazines in the fungus Gyromitra esculenta have proven carcinogenic in a number of organs and tissues of mice and hamsters. Two of them were mutagenic in bacterial systems. There are no epidemiological studies concerning the carcinogenicity of these mushrooms in humans.

Pyrrolizidine Alkaloids

Pyrrolizidine alkaloids occur in many nonedible plant species, including the genera Senecio (ragworts), Crotalaria (rattleboxes), and Heliotropium (heliotropes), in amounts ranging from trace amounts to as much as 5% of the dry weight. In general, members of this group that contain a nuclear double bond alpha to an esterified carbinol are very potent toxins in the liver and lung of rodents and certain farm livestock (Hirono, 1981; Hirono et al., 1979; International Agency for Research on Cancer, 1976).

Experimental Evidence: Carcinogenicity. Monocrotaline, retrorsine, lasiocarpine, heliotrine, senkirkine, symphytine, and petasitenine, all of which are α, β-unsaturated esters, are carcinogenic when administered to rats orally or parenterally under conditions that permit long-term survivals. Most frequently, tumor induction has involved multiple doses of the alkaloids at moderate levels (e.g., a 0.01% solution of petasitenine in drinking water for 480 days) (Hirono et al., 1979), but low incidences of tumors after long latent periods have apparently resulted from only one or a few doses. Tumors have also been induced in rats after the administration of plants, such as coltsfoot (Tussilago farfara) or comfrey (Symphytum sp.), which contain high levels of pyrrolizidine alkaloids. he tumors occur most frequently in the liver, but some have developed in other tissues, including the skin and lungs.

Plants containing the pyrrolizidine alkaloids may contaminate forages and food grains. Such contamination has resulted in acute and chronic poisoning of livestock in some parts of the world (Schoental, 1976). Humans may also be exposed by consuming such alkaloid-containing plants as drugs or foods. For example, one species of comfrey (Symphytum officinale) is consumed as a green vegetable in Japan (Hirono et al., 1979). The carcinogenic potency of some pyrrolizidine alkaloids and their widespread occurrence have led to the suggestion that these α, β-unsaturated esters may play a role in the induction of hepatic cancer in humans in some parts of the world; however, there are no reliable data to support this hypothesis.

Experimental Evidence: Mutagenicity. Retrorsine, lasiocarpine, heliotrine, senkirkine, symphytine, and petasitenine, but not monocrotaline, have been shown to be mutagenic in the Salmonella/microsome assay (Hirono et al., 1979; Wehner et al., 1979; Yamanaka et al., 1979).

Allylic and Propenylic Benzene Derivatives

Numerous allylic and propenylic benzene derivatives are present in the essential oils of a wide variety of plants (Guenther, 1948-1952; Guenther and Althausen, 1949), and some of these plants or their extracts are used as flavoring agents for human foods or as medicines consumed by humans. Of the known naturally occurring allylic benzene derivatives, safrole (l-allyl-3, 4-methylenedioxybenzene), which is a major component of oil of sassafras, and estragole (l-allyl-4-methoxybenzene), which is present in tarragon and anise, have been the most comprehensively studied.

Experimental Evidence: Carcinogenicity. Safrole has induced low-to-moderate incidences of hepatic tumors in adult rats fed at levels of 0.5% or more of the diet for as long as 2 years (International Agency for Research on Cancer, 1976). Both safrole and estragole induced hepatic tumors and subcutaneous angiosarcomas within 18 months after they were fed to adult female CD-1 mice at levels of 0.25%-0.5% for approximately 1 year (Miller et al., 1979). Administration of less than 1 mg of either compound or of methyl eugenol to CD-1 or (C57BL/6 x C3H/He)F₁ male mice prior to weaning resulted in a high incidence of hepatomas by the age of 12 months (Miller et al., 1979).

Experimental Evidence: Mutagenicity. Safrole was mutagenic in vitro and in the host-mediated assay (Green and Savage, 1978). However, McCann et al. (1975), Swanson et al. (1979), and Wislocki et al. (1977) reported that it was not mutagenic in the Ames test. It was positive in Bacillus subtilis rec assay (Rosenkranz and Poirier, 1979) and in Saccharomyces cerevisiae D3 (Simmon, 1979).

Estragole was mutagenic to S. typhimurium TA100 (Swanson et al., 1979). Eugenol was not mutagenic to Ames Salmonella strains in vitro and in the host-mediated assay (Green and Savage, 1978; Swanson et al., 1979).

Bracken Fern Toxin(s)

Bracken fern (Pteridium aquilinum) occurs widely in nature and is consumed by humans in several parts of the world, especially in Japan (Hirono, 1981). For at least 30 years, it has been known that consumption of this plant causes damage to the bone marrow and intestinal mucosa of cattle, but the precise compound(s) responsible for these toxic effects have not been identified.

Epidemiological Evidence: Carcinogenicity. In a prospective cohort study in Japan, Hirayama (1979) found a significantly higher risk of esophageal carcinoma associated with the daily intake of hot gruel or bracken fern every day, especially in people who ate both foods daily. However, Howe et al. (1980) found no association between bladder cancer and consumption of fiddlehead greens (related to bracken fern) in a case-control study in Canada.

Experimental Evidence: Carcinogenicity. The carcinogenicity of bracken fern was first suspected by Pamukcu in 1960, who found polyps in the urinary bladder mucosa of cattle fed large amounts of bracken fern for long periods (Pamukcu and Bryan, 1979). Since that time, ingestion of high levels of bracken fern (25% to 40% of the diet) has been found to result in the formation of urinary bladder carcinomas in cattle, urinary bladder carcinomas and intestinal adenocarcinomas in rats, urinary bladder tumors in guinea pigs, pulmonary adenomas in mice, and intestinal adenocarcinomas in Japanese quail (Evans, 1976).

Hirono (1981) reported that the greatest concentration of the toxin(s) is present in young plants before the fronds have uncurled, and the carcinogenic activity of the rhizome is greater than that of the stalk or fronds. The toxicity of the fern is reduced, but not eliminated, by cooking.

A number of studies have been conducted to identify the carcinogenic agent(s) in bracken fern (Evans, 1976; Hirono, 1981; Pamukcu and Bryan, 1979). Quercetin (3, 3', 4', 5, 7-pentahydroxyflavone) occurs as a conjugate in bracken fern and in numerous other plants. In culture, this compound has induced morphological transformation of cryopreserved golden hamster embryo cells (Umezawa et al., 1977) and mutations in S. typhimurium (Bartholomew and Ryan, 1980), but its carcinogenicity in rats continues to be disputed. In one study, administration of 0.1% quercetin in the diet of rats for as long as 1 year resulted in an 80% incidence of intestinal tumors and a 20% incidence of urinary bladder tumors (Pamukcu et al., 1980). However, in another laboratory, administration of quercetin as 1% or 5% of the diet for 540 days or as 10% of the diet for 850 days did not result in a significant incidence of tumors in ACI rats (Hirono et al., 1981).

Interest in the possibility that bracken fern might play a role in the induction of cancers stems from the knowledge that it is used by humans as food in several parts of the world (Hirono, 1981). Indirect evidence for its carcinogenicity is derived from observations that milk from cows fed high levels of bracken fern contained compounds that were shown to be carcinogenic in rats. Carcinomas of the intestine, urinary bladder, and kidney pelvis were observed in rats fed high levels of fresh or powdered milk from cows that had consumed 1 g of bracken fern per kilogram of body weight daily for approximately 2 years, but not in rats fed milk from control cows (Pamukcu et al., 1978).

Estrogenic Compounds

The plant estrogens include estrone (from palm kernels), genistein (from soybean and clover), coumestrol (from alfalfa and other forage crops), and mirestrol (from certain legumes) (Schoental, 1976; Stob, 1973). Zearalenone, a product of Fusarium molds that sometimes infect grains, also possesses estrogenic activity.

Plant estrogens are very weak estrogens compared to the hormones from animals; however, they can occur in relatively large amounts. For example, fat-free soybeans may contain as much as 0.1% of genistein (Verdeal et al., 1980).

Experimental Evidence: Carcinogenicity. Other than one report on zearalenone (discussed earlier in this chapter), there are no data pertaining to the carcinogenicity of plant estrogens. Some nonsteroidal phytoestrogens that are natural components of some foods compete for estrogen receptors in rat uterine cytosol in tissue sections from 7, 12-dimethylbenz[a]anthracene-induced mammary tumors, and in mammary tumor tissue from humans (Verdeal et al., 1980). The significance of these findings in the etiology of neoplasia in humans is not known.

Experimental Evidence: Mutagenicity. Genistein and coumestrol were not mutagenic in the Salmonella microsome assay (Bartholomew and Ryan, 1980).

Coffee

Epidemiological Evidence: Carcinogenicity: Coffee drinking has been associated with elevated risk for bladder cancer in several case-control studies (Bross and Tidings, 1973; Cole, 1971; Fraumeni et al., 1971; Howe et al., 1980; Miller et al., 1978; Simon et al., 1975; Wynder and Goldsmith, 1977). However, with only two possible exceptions in males (Bross and Tidings, 1973; Wynder and Goldsmith, 1977), there has been no evidence of a dose-response relationship, and it appears that the association is not causal.

A direct association of coffee consumption with risk of pancreatic cancer based on case-control data was reported by MacMahon et al. (1981). They provided evidence for a dose-response relationship. In another report, Lin and Kessler (1981) noted an association between pancreatic cancer and the use of decaffeinated coffee specifically. In an earlier geographical correlation of per capita food intake and mortality from cancer, Stocks (1970) observed a significant association between coffee drinking and pancreatic cancer.

Other reported associations of coffee drinking with cancer have been scattered and inconsistent. Martinez (1969) found an association between oral and esophageal cancers combined and consumption of hot beverages, mostly coffee, whereas Stocks (1970) did not find a significant correlation between coffee consumption and esophageal cancer. Shennan (1973) reported a direct correlation between per capita coffee intake and mortality from renal carcinoma (r=0.8), and the association, though less strong, appeared also in the correlational data of Armstrong and Doll (1975). On the other hand, case-control studies of renal cancer (Armstrong et al., 1976; Wynder et al., 1974) have not confirmed this association. Stocks (1970) also found a direct correlation of prostate cancer mortality with per capita coffee intake. This finding did not appear in a similar analysis by Armstrong and Doll (1975), who reported an association with per capita fat intake and a high correlation between these two dietary factors.

Experimental Evidence: Carcinogenicity. Sprague-Dawley rats were fed a diet containing 5% instant coffee for 2 years. No bladder tumors were noted in rats fed diets containing the equivalent of up to 85 cups of coffee per day (Zeitlin, 1972).

Maximum tolerated doses of regular and decaffeinated instant coffees (6% of the diet) fed to Sprague-Dawley rats for 2 years produced no evidence of carcinogenesis (Würzner et al., 1977). The authors also reported that high levels of caffeine led to a lower incidence of tumors. However, Challis and Bartlett (1975) reported that readily oxidized phenolic compounds--which are constituents of coffee--catalyze nitrosamine formation from nitrite and secondary amines at gastric pH. For example, these experiments showed that 4-methylcatechol and the phenolic component of chlorogenic acid (approximately 13% of the dry weight of the soluble constituents of coffee), exerted catalytic effects on nitrosamine formation. This finding implies that several foodstuffs and beverages, including coffee, may have cocarcinogenic properties.

Experimental Evidence: Mutagenicity. Coffee is mutagenic to Salmonella typhimurium strain TA100, whether it is brewed, instant, or decaffeinated (Aeschbacher and Würzner, 1980; Aeschbacher et al., 1980; Nagao et al., 1979). Although caffeine has been reported to be mutagenic to bacteria (Clarke and Wade, 1975; Demerec et al., 1948, 1951; Gezelius and Fries, 1952; Glass and Novick, 1959; Johnson and Bach, 1965; Kubitschek and Bendigkeit, 1958, 1964; Novick, 1956), it could not have been responsible for the mutagenicity of coffee observed in these reports, since decaffeinated coffee was as mutagenic as regular coffee and caffeine itself was not detected as a mutagen under the test conditions used (Aeschbacher et al., 1980; Nagao et al., 1979).

Methylxanthines

Experimental Evidence: Carcinogenicity. Another widely consumed class of compounds are the methylxanthines, which include caffeine. There appear to be no published studies on the carcinogenicity of caffeine in laboratory animals following chronic oral administration. In one as yet unpublished study (Takayama, personal communication), Wistar rats were divided into three dose groups each containing 50 males and 50 females. The first two groups were given 0.2% and 0.1% caffeine in their drinking water for 18 months beginning at the age of 8 weeks. Then, normal water without caffeine was given to the surviving animals for an additional 6 months. The third group, which served as the control group, was given normal water throughout the experiment. All remaining animals were sacrificed 24 months after the caffeine treatment had begun. The investigators concluded that there was no significant increase in the incidence of any type of tumors in caffeine-treated animals, as compared to control animals.

Experimental Evidence: Mutagenicity. The methylxanthines--a class of compounds that are present in tea and coffee--are mutagenic in at least some test systems. Three of the compounds--caffeine, theophylline, and theobromine--have been reported to be mutagenic to bacteria and to cause abnormalities in the chromosomes of plant cells (see reviews by Kihlman, 1977, and Timson, 1975). However, the mutagenic effects of these compounds in mammals have not been clearly demonstrated in vivo. Caffeine can enhance the genetic effects of other chemicals, even in vivo (Frei and Venitt, 1975; Jenssen and Ramel, 1978). This activity is presumably due to the ability of caffeine to inhibit repair of DNA damage caused by chemical mutagens.

Cycasin

Cycasin (methylazoxymethanol-β-glucoside) is one of the most potent carcinogens found in plants (International Agency for Research on Cancer, 1976; Magee et al., 1976). This compound and at least one related glucoside (macrozamin) are present in the palmlike cycad trees of the family Cycadaceae. These trees have provided food for natives and their livestock in tropical and subtropical regions. The sliced nuts are generally extracted with water prior to use, but acute poisonings have been reported.

In Guam and Okinawa, which have high rates of liver cancer, the ingestion of cycasin in cycad nuts has been proposed as an etiologic factor. However, in a descriptive study conducted in the Miyako Islands of Okinawa, investigators found no correlation between mortality from hepatoma and the ingestion of cycad nuts (Hirono et al., 1970). Therefore, there is no evidence for the carcinogenicity of cycasin in humans.

Experimental Evidence: Carcinogenicity. When administered orally, cycasin is highly carcinogenic in the liver, kidney, and colon of rats, and also induces tumors in other species (Laquer and Spatz, 1968). The tissues of rats contain low levels of β-glucosidase, which hydrolyzes cycasin. However, the hydrolysis generally depends on the action of intestinal bacteria (Matsumoto et al., 1972). The product, methylazoxymethanol (MAM), decomposes at neutral pH to an electrophilic intermediate that methylates nucleic acids and proteins both in vitro and in vivo (Matsumoto and Higa, 1966). These findings and the carcinogenic activity of MAM (Laquer and Spatz, 1968) have implicated MAM as a proximate carcinogenic metabolite of cycasin. The methylating species formed from MAM and cycasin appears to be similar or identical to that formed during the metabolic activation of the synthetic carcinogen nitrosodimethylamine, which has carcinogenic properties similar to those of cycasin (Magee et al., 1976).

Experimental Evidence: Mutagenicity. Cycasin was not mutagenic in the standard Ames test (Ames et al., 1975), but it became mutagenic when preincubated with almond β-glucosidase (Matsushima et al., 1979).

Thiourea

Thiourea occurs naturally in laburnum shrubs and in certain fungi (e.g., Verticillium albo-atrum and Bortrylio cinerea).

Experimental Evidence: Carcinogenicity. Thiourea has been shown to cause thyroid tumors, hepatic adenomas, and epidermoid carcinomas of Zymbal's gland when administered to rats as 0.2% of the drinking water or diet for as long as 2 years (International Agency for Research on Cancer, 1974).

Experimental Evidence: Mutagenicity. Thiourea was negative in the standard Ames Salmonella/microsome assay (Simmon, 1979), but positive in the host-mediated assay (Simmon et al., 1979). It also induced transformations in hamster embryo cells (Pienta, 1981).

Tannic Acid and Tannins

Tannins are contained in many plants. These compounds are divided into two groups--the nonhydrolyzable condensed tannins and the hydrolyzable tannins, which are subdivided into ellagitannins or gallotannins. Commercially, the term tannic acid generally applies to hydrolyzable gallotannins, including taratannic acid. Tannins are widely distributed in plants, and are present naturally in small amounts in coffee and tea. Tannic acid has also been used by U.S. food processors as a clarifying agent in the brewing and wine industries and as a flavoring agent in such products as butter, caramel, fruit, brandy, maple, syrup, and nuts (National Academy of Sciences, 1965).

Experimental Evidence: Carcinogenicity. The investigations of Korpássy showed that subcutaneous administration of tannic acid in doses of 150 to 200 mg/kg bw produced skin necrosis, ulcers, and hepatic tumors in rats (Korpássy, 1959, 1961; Korpássy and Mosonyi, 1950, 1951). No adequate studies have been conducted to test the carcinogenicity of orally administered tannins.

In mice, repeated subcutaneous injections of three condensed non-hydrolyzable tannins produced liver tumors and sarcomas (Kirby, 1960).

Experimental Evidence: Mutagenicity. Tannic acid was found not to be genotoxic or mutagenic to Saccharomyces cerevisiae D4 and Ames S. typhimurium strains with and without metabolic activation (Litton Bionetics, Inc., 1975). Tannins from various sources such as apple juice, grape juice, wine, and betel nuts were found to be strongly clastogenic for Chinese hamster ovary cells, but they lacked the capacity to induce mutations in the Ames test (Stich and Powrie, in press).

Coumarin

Coumarin is present in a number of plants, including tonka beans, cassia, and woodruff, and in their essential oils (International Agency for Research on Cancer, 1976).

Experimental Evidence: Carcinogenicity. Coumarin (o-hydroxycinnamic acid-δ-lactone) has induced bile duct carcinomas in rats fed 0.35% to 0.5% of the compound in the diet for approximately 18 months.

Experimental Evidence: Mutagenicity. Coumarin was negative in the E. coli pol A assay (Rosenkranz and Leifer, 1981). It interferes with excision repair processes in ultraviolet-damaged DNA and with host cell reactivation of ultraviolet-irradiated phage T1 in E. coli WP2 (Grigg, 1972).

Parasorbic Acid

Parasorbic acid occurs in concentrations ranging from 0.2 to 2 μg/g in the ripe berries of the Moravian mountain ash Sorbus aucuparia var. edulis. It has not been found in a number of common fruits (pears, apples, lemons, cranberries, grapes, oranges, or tomatoes) (International Agency for Research on Cancer, 1976).

Experimental Evidence: Carcinogenicity. Sarcomas resulted within 2 years in rats that had received repeated subcutaneous injections of parasorbic acid in total doses of either 13 or 128 mg per animal (International Agency for Research on Cancer, 1976).

Experimental Evidence: Mutagenicity. No studies concerning the mutagenicity of parasorbic acid could be identified.

Tryptophan and Its Metabolites

Experimental Evidence: Carcinogenicity. Dogs fed high levels of tryptophan (7 g/day, i.e., 7 times the amount fed to controls) for long periods developed hyperplasia of the urinary bladder (Radomski et al., 1971). When tryptophan was given to rats as 2% of the diet after subcarcinogenic doses of a nitrofuran, tryptophan exerted a promoting effect on the formation of tumors in the urinary bladder (Cohen et al., 1979). In other studies, four metabolites of tryptophan (3-hydroxykynurenine, 3-hydroxyanthranilic acid, 2-amino-3-hydroxyacetophenone, and xanthurenic acid-8-methyl ether) each induced bladder tumors when implanted as pellets in the urinary bladders of mice (Clayson and Garner, 1976). However, attempts to relate the development of tumors in the urinary bladder of humans to abnormalities in the metabolism of tryptophan have not been definitive (Clayson and Garner, 1976).

Experimental Evidence: Mutagenicity. Tryptophan and its metabolites were not mutagenic in the Salmonella/microsome assay (Bowden et al., 1976).

Hormones

Experimental Evidence: Carcinogenicity. A number of endogenous peptide and steroid hormones facilitate the development of tumors of the endocrine glands of laboratory animals (Clifton and Sridharan, 1975; Furth, 1975). However, because humans consume only very small amounts of hormones from the tissues of animals and because there is no indication that hormones from food sources are significant factors in the development of cancer in humans, they are not considered in this report. One exception is diethylstilbestrol (DES), which is discussed in Chapter 14.

Ethyl Carbamate (Urethan)

Ethyl carbamate, or urethan, is a fermentation product. The detection of low levels of ethyl carbamate in wines treated with the synthetic sterilant diethyl pyrocarbonate (Ehrenberg et al., 1976) led to investigations into the natural occurrence of ethyl carbamate. These studies have demonstrated that naturally fermented foods and beverages (e.g., wines, bread, beers, and yogurt) contain detectable, but very low levels of ethyl carbamate, usually less than 5 μg/kg. The ethyl carbamate probably results from the reaction of ethanol and carbamoyl phosphate--both normal metabolic products in the yeast (Ough, 1976).

Experimental Evidence: Carcinogenicity. For many years, ethyl carbamate has been studied as a synthetic carcinogen in the rat, mouse, and hamster. Its ability to induce tumors has been demonstrated by administering the compound during the prenatal and preweanling periods as well as to adult animals. Ethyl carbamate is active when administered orally, by inhalation, or by subcutaneous or intraperitoneal injection. The susceptible tissues include the lungs, lymphoid tissue, skin, liver, mammary gland, and Zymbal's gland. Most frequently, tumors are induced with doses ranging from 0.5 to 3 mg/g bw (International Agency for Research on Cancer, 1974; Mirvish, 1968). However, lung adenomas were induced in mice with a single dose of 0.01 mg/g bw (Nomura, 1975).

The significance of naturally occurring ethyl carbamate in foods in the development of human cancer is unknown, but the levels are very low in comparison to those used to induce tumors in laboratory animals (i.e., the consumption of 5 μg/day by a 70-kg person would provide an annual intake of approximately 0.05 μg/g bw).

Experimental Evidence: Mutagenicity. Urethan was not mutagenic to Ames Salmonella strains (Simmon, 1979) or in the host-mediated assay (Simmon et al., 1979). However, it did induce transformations in hamster embryo cells (Pienta, 1981).

Epidemiological Evidence

Studies conducted in Colombia, Chile, Japan, Iran, China, England, and the United States (Hawaii) have indicated that there is an association between increased incidence of cancers of the stomach and the esophagus and exposures to high levels of nitrate or nitrite in the diet or drinking water. (See, for example, Armijo and Coulson, 1975; Armijo et al., 1981; Correa et al., 1975; Cuello et al., 1976; Haenszel et al., 1972; Higginson, 1966; Meinsma, 1964).

Exposures to nitrate, nitrite, or N-nitroso compounds were not directly measured in these epidemiological studies. The associations with cancer were based either on correlations of high risk population groups with corresponding exposures in food and water supplies, or on comparisons of the frequency of consumption of foods containing these substances (plus secondary amines) by gastric cancer patients and by controls.

Bladder cancer has been correlated with nitrate in the water supply or with urinary tract infections in some epidemiological studies (Howe et al., 1980; Wynder et al., 1963). However, Howe et al. (1980) reported that there was no difference between cases and controls in consumption of nitrite-preserved meats, such as hams and sausages. Nevertheless, it is of interest that nitrosamines, which are presumably formed from dietary precursors, have been found in the urine of patients with urinary tract infections and could presumably be carcinogenic in the bladder (Hicks et al., 1977; Radomski et al., 1978).

The Committee on Nitrite and Alternative Curing Agents in Food concluded that these reports do not provide conclusive evidence of a causal relationship, and that alternative explanations for the findings have not been ruled out (National Academy of Sciences, 1981).

Studies of occupational exposure have not contributed significant information on possible associations between N-nitroso compounds and cancer risk.

Experimental Evidence: Carcinogenicity

The data on the carcinogenicity of nitrate and nitrite in animals are not definitive. The few experiments conducted in animals have provided no evidence that nitrate is carcinogenic (Greenblatt and Mirvish, 1973; Lijinsky et al., 1973; Sugiyama et al., 1979).

There have been very few adequate studies to test the carcinogenicity of nitrite. Most of the information is derived from data on tumor incidence in experiments that were designed primarily to study nitrosation in animals given nitrite and amine simultaneously. These data were compared with data on control animals given nitrite alone (usually in drinking water). Because the control animals were usually sacrificed after a few months, there may not have been sufficient time for tumors to develop (Aoyagi et al., 1980; Inai et al., 1979; Mirvish et al., 1980; Shank and Newberne, 1976).

In a larger lifetime study conducted for the U.S. Food and Drug Administration, Newberne (1978, 1979) fed various doses of nitrite to groups of approximately 68 male and 68 female Sprague-Dawley rats under a variety of conditions. In comparison to the controls, the treated rats had not only a higher incidence of malignant tumors of the lymphatic system, but also a higher incidence of alterations (immunoblastic cell proliferation) in the spleen and, occasionally, in the lymph nodes of the treated groups (Newberne, 1979). These results were interpreted by the author to indicate that nitrite may be an enhancer or promoter of carcinogenesis in rats. However, a Joint Committee of Experts, which was established to review the study, diagnosed fewer lymphomas than those reported by Newberne (U.S. Food and Drug Administration, 1980a). The discrepancy between the two diagnoses involved the differentiation of lymphomas from extramedullary hematopoiesis, plasmacytosis, or histiocytic sarcoma. Furthermore, the Joint Committee was unable to confirm the diagnosis of immunoblastic hyperplasia.

In addition, the Committee on Nitrite and Alternative Curing Agents in Food reviewed 21 reports in which the carcinogenicity of nitrite was examined (National Academy of Sciences, 1981). The committee concluded that three of the 21 reports were too brief to evaluate adequately. Of the remaining 18 studies, 9 were conducted in rats, 8 in mice, and 1 in guinea pigs. The experimental design was inadequate in many cases, and varied greatly with regard to the end points for carcinogenicity. However, none of the remaining 18 studies provided sufficient evidence that nitrite was carcinogenic (see, for example, Greenblatt et al., 1973; Lijinsky et al., 1980).

The absence of evidence that nitrite is a direct carcinogen does not diminish the possibility that it can interact with specific components of diets consumed by humans and animals or with endogenous metabolites to produce N-nitroso compounds that induce cancer.

N-Nitroso compounds have been studied extensively to determine their carcinogenic effects. Druckrey and his colleagues (1967) reported tests in rats exposed to 65 N-nitroso compounds, most of which were potent carcinogens. Lijinsky and Reuber (1981) examined many other N-nitroso compounds for their carcinogenic potential, mainly in rats. Approximately 300 different N-nitroso compounds have been tested, and a majority of them have been shown to induce cancer in various tissues of one or more species of laboratory animals when administered by any of several routes (Preussmann and Steward, personal communication, 1981). In addition to both nitrosodimethylamine and nitrosodiethylamine, a number of other N-nitroso compounds detected in the environment are carcinogenic in animals (see, for example, Druckrey et al., 1967; Preussmann et al., 1981).

The carcinogenic action of several N-nitroso compounds can be inhibited in systems where the formation of N-nitroso compounds has been prevented (Mirvish, 1981). Nitrosation is inhibited when ascorbic acid and a variety of other agents compete with the nitrosatable agent for the available nitrite in the acidic conditions of the stomach. A number of other agents that interact readily with nitrite can inhibit nitrosation. Among these are other isomers of ascorbic acid, sorbic acid, some phenols, and α-tocopherol. Most of these interactions have been observed at the chemical rather than at the biological level.

Formation of N-nitroso compounds can also be enhanced since a variety of ions, especially thiocyanate and iodide, may catalyze the nitrosation reaction in the stomach (Mirvish et al., 1975). Since these ions are present in foodstuffs, these catalysts could be important in determining the outcome.

The additive or synergistic effects of N-nitroso compounds on other carcinogens with similar organotropy has been emphasized by Schmähl (1980).

Experimental Evidence: Mutagenicity

Nitrate does not appear to be directly mutagenic (Konetzka, 1974). In microbial systems, nitrite may be mutagenic by three different mechanisms (Zimmerman, 1977): deamination of DNA bases in single-stranded DNA; formation of 2-nitroinosine, intrastrand, or interstrand lesions leading to helix distortions in double-stranded DNA; and formation of mutagenic N-nitroso compounds by combination with nitrosatable substrates. Except for the results of one study in which a high dose of nitrite was used, there is no evidence that nitrite is mutagenic in mammalian systems.

Many N-nitroso compounds have been found to be mutagenic in a variety of test systems, including bacterial tests and Drosophila, and under a variety of conditions (Montesano and Bartsch, 1976).

Summary and Conclusions: Nitrate, Nitrite, and N-Nitroso Compounds

Epidemiological evidence suggesting that nitrate, nitrite, and N -nitroso compounds play a role in the development of cancer in humans is largely circumstantial. However, the findings from several epidemiological studies of certain geographical/nationality groups are consistent with the hypothesis that exposure of humans to high levels of nitrate and/or nitrite may be associated with an increased incidence of cancers of the stomach and esophagus. In these studies, the level, duration, and time of exposure were not studied in relation to cancer incidence, and exposure to other known or suspected carcinogens was not excluded.

In animals, nitrate has not been shown to be carcinogenic or mutagenic per se. The data on nitrite, which has been tested more extensively than nitrate, indicate that nitrite is probably not carcinogenic but that it is mutagenic, at least in microbial systems.

As a group, the N-nitroso compounds are clearly carcinogenic in numerous species of animals in which they have been tested. Positive results have been obtained for nearly all of the approximately 300 N-nitroso compounds tested for carcinogenicity in one or more species. Many of these compounds are also mutagenic.

The Committee on Nitrite and Alternative Curing Agents in Food recommended that exposure to nitrate, nitrite, and N-nitroso compounds should be reduced (National Academy of Sciences, 1981).