This chapter summarizes the qualitative aspects of the health effects known to be associated with nitrate and nitrite exposure.
The primary adverse health effect associated with human exposure to nitrate or nitrite is methemoglobinemia. Nitrite converts hemoglobin to methemoglobin by oxidizing the Fe2+in heme to Fe3+, which cannot transport oxygen. Low concentrations of methemoglobin (0.5-3.0%) occur in normal people, although concentrations up to 10% can occur without clinical signs (EPA 1990a; Walton 1951). Concentrations above 10% can cause cyanosis, characterized by bluish skin and lips, and concentrations above 25% are associated with hypotension, rapid pulse, and rapid breathing, as a result of the vasodilator effects of nitrite. Concentrations above 50% can be fatal (EPA 1990a).
To cause methemoglobinemia, nitrate must be converted to nitrite. The conversion is performed by bacteria in the mouth and stomach. The extent of toxicity of nitrate depends on the extent to which it is converted to nitrite, which depends on the concentration and type of bacteria in the mouth and stomach. In adults, the acidity of the stomach is usually great enough that bacterial growth and the consequent conversion of nitrate to nitrite are negligible; probably about 5% of a dose of nitrate is reduced to nitrite, on average (ECETOC 1988), and methemoglobinemia is rare. In infants, however, the low gastric acidity is thought to favor the growth of nitrate-reducing bacteria, and infants are the group most susceptible to methemoglobinemia. Most infant victims of methemoglobinemia have reportedly been fed infant formula mixed with well water that contained high concentrations of nitrate, but cases have also been associated with the consumption of spinach and carrots, which are high in nitrate and nitrite (Bruning-Fann and Kaneene 1993). However, several authors have reported that achlorhydria and gastric nonsterility are rare in infants, even infants with methemoglobinemia (Bodo 1955; Simon et al. 1962; Agunod et al. 1969); that observation suggests a role of pathways other than gastrointestinal nitrite synthesis. Other investigators have reported that infant gastric acidity is low enough to support bacterial growth (EPA 1990a). Hegesh and Shiloah (1982) studied newborns hospitalized for acute diarrhea and found no correlation between ingestion of food or water containing high concentrations of nitrate or nitrite and methemoglobinemia; they concluded that endogenous synthesis of nitrite resulting from diarrhea was the principal cause of infantile methemoglobinemia. Thus, diarrhea apparently can be a major cause of infant methemoglobinemia unrelated to the nitrate content of food and water (endogenous synthesis is discussed in Chapter 4).
A 1990 EPA publication (EPA 1990a) provides a thorough review of the literature available on the occurrence of methemoglo-
binemia in infants, children, and adults, including case reports, reports of surveys, clinical studies, and epidemiologic studies. The subcommittee could find no studies of nitrate-induced methemoglobinemia reported since the 1990 EPA publication. The absence of reports might in part be due to the lack of requirements for reporting cases of methemoglobinemia. In addition, the studies reviewed by EPA are of uncertain quality largely because of the lack of controls for the presence of confounding factors.
Nitrate and nitrite have been tested for carcinogenicity in laboratory animals, and epidemiologic studies of human cancer rates among populations with high nitrate or nitrite exposure concentrations have been performed. In general, nitrate and nitrite are not carcinogenic in laboratory animals when administered in the absence of nitrosatable amines. When nitrite and nitrosatable amines are administered together, however, carcinogenic nitrosamines can be formed in the stomach and lead to various tumors. Similar results have not been reported for simultaneous administration of nitrate and nitrosatable amines. Nitrosamine formation is inhibited by antioxidants, such as vitamin C and vitamin E. Results of epidemiologic studies have not supported an association between high nitrate or nitrite exposure from drinking water in the United States and increased cancer rates in humans. Both the animal and human studies are reviewed in detail in publications of EPA (EPA 1990a) and the European Chemical Industry Ecology and Toxicology Centre (ECETOC 1988). Human studies reported since the EPA review are also included here. The subcommittee could find no animal-carcinogenicity studies reported since the 1990 EPA publication.
Sodium nitrate and sodium nitrite have been administered in the drinking water and diet of male and female rats, mice, hamsters, and guinea pigs to assess their carcinogenicity. Several studies of nitrate and numerous studies of nitrite have shown that when administered alone at doses up to 1,755 mg/kg-day and 550 mg/kg-day (as nitrate or nitrite), respectively, these substances yielded no evidence of carcinogenicity (EPA 1990a). When sodium nitrite is administered simultaneously with secondary or tertiary amines, however, tumor incidence increases at a number of sites. The increases are thought to result from the formation of nitrosamines in the stomach. Nitrosamine formation from nitrite and nitrosatable amines is of concern because nitrosamines are potent mutagens and carcinogens at various sites and in various species of laboratory animals (Peto et al. 1984; ECETOC 1988).
For example, Greenblatt and Mirvish (1972) fed a diet containing piperazine at 0 or 6,250 ppm to male strain A mice for 20 weeks in addition to supplying drinking water containing NaNO2 at 0, 50, 250, 500, 1,000, or 2,000 mg/L. There was an increased rate of lung-tumor formation at all dosages except 50 mg/L, with a clear dose-response relationship. No increase in tumor rate was observed among animals receiving NaNO2or piperazine alone or in other groups of animals receiving piperazine and NaNO3. Taylor and Lijinsky (1975) fed drinking water containing 0.2% heptamethyleneimine (133 mg/kg-day) and 0.2% NaNO2(90 mg/kg-day) to male and female Sprague-Dawley rats. That treatment elicited oropharyngeal, tongue, esophageal, and forestomach tumors. No tumors were observed among rats receiving either component alone. Thamavit et al. (1988) administered drinking water containing aminopyrine at 0 or 1,000 mg/L and NaNO2at 0 or 1,000 mg/L NaNO2(nitrite at 0 or 65 mg/kg-day) to male
Syrian golden hamsters. An increased incidence of liver tumors was observed among animals receiving both NaNO2 and aminopyrine, whereas no liver tumors were observed among untreated controls or animals receiving only one treatment. Mokhtar et al. (1988) found that the hepatotumorigenic effect of simultaneous administration of NaNO2 and dibutylamine in mice was diminished by concurrent feeding with ascorbic acid.
None of the investigations described above determined actual nitrosamine formation. Yamamoto et al. (1989), however, detected N-nitrosobis(2-hydroxypropyl)amine in the urine of rats receiving bis(2-hydroxypropyl)amine in the diet and NaNO2 in the drinking water. Treated animals exhibited increased rates of lung, esophageal, and nasal-cavity tumors. Untreated animals and those receiving NaNO2 or the amine alone neither developed tumors at those sites nor excreted the nitrosamine. Other investigators have reported detecting in vivo formation of nitrosamines in the gastro-intestinal tracts of rats and hamsters receiving oral doses of nitrite and nitrosatable amines in short-term experiments (Inui et al. 1980; Massey et al. 1988). Kamm et al. (1975) reported hepatotoxicity associated with increased plasma nitrosodiethylamine in rats administered aminopyrine and nitrite. Nitrosamine formation from nitrite and nitrosatable amines is discussed further in Chapter 4.
There are several reasons why nitrosamine formation and consequent cancer risk among laboratory animals receiving nitrite and nitrosatable amines might not be relevant to similar human exposures. Nitrosamine formation in the stomach is favored by low pH. Rodents have higher gastric pHs than most humans (about 5.0 versus about 1.5), so rodents are less likely to produce nitrosamines than humans. Humans transport nitrate from blood to saliva, whereas rats do not. In addition, the dosages used in the experiments reported here are often much higher than the concentrations encountered by humans in drinking water or in the diet. For these
reasons, pharmacokinetic characteristics must be considered if extrapolations are made. Exposure is discussed in greater detail in Chapter 4.
No studies in animals have shown that cancer risk is increased by the simultaneous administration of nitrate and nitrosatable amines.
Several cohort, case-control, and geographic-correlation studies have attempted to evaluate the effects of nitrate and nitrite exposure on human cancer risk. Those studies have been reviewed in detail by EPA (1990a) and ECETOC (1988) and provide inadequate evidence of an association between nitrate or nitrite exposure and increased cancer risk.
Two cohort studies—Fraser et al. (1982) and Al-Dabbagh et al. (1986)—examined cancer mortality among male fertilizer workers exposed to nitrate-containing dust in England and Wales. The latter study found that nitrate concentrations in saliva were higher in exposed workers than in controls. In both studies, however, neither overall cancer mortality nor mortality for any particular cancer site was statistically significantly higher among the exposed populations than in controls. The occupational exposures to nitrate experienced by the study populations were likely to have been much greater than the nitrate exposures expected from drinking water.
One case-control study evaluated the association between nitrate exposure from drinking water and gastric-cancer risk (Rademacher et al. 1992). Persons who had died of gastric cancer in Wisconsin in 1982-1985 were matched with controls who had died from other causes, and information on their drinking-water nitrate concentrations was compared. Nitrate concentrations rarely exceeded 50
mg/L, and no association between nitrate exposure and gastric-cancer risk was found.
Several case-control studies have evaluated dietary risk factors among patients with stomach cancer or chronic atrophic gastritis, a condition that can be a precursor of stomach cancer. For example, Risch et al. (1985) calculated intakes of specific dietary constituents on the basis of diet questionnaires and U.S. Department of Agriculture data on food composition. Associations were reported between stomach cancer and intake of nitrite, smoked meat, unsaturated fats, grains, chocolate, eggs, cream desserts, or non-refrigerated food, whereas intake of nitrate, fiber, citrus fruits, or vitamin C was associated with a reduction in stomach-cancer risk. The contributions of waterborne nitrate or nitrite were not considered in the analysis. Although the diet questionnaires evaluated current dietary practices, gastric-cancer risk was likely to be associated with dietary practices 20 or 30 years previously.
A case-control study in Germany evaluated the relationship between several dietary factors and confirmed cases of glioma and meningioma (Boeing et al. 1993). The study focused on nitroso compounds because several nitrosoureas had been found to induce tumors of the nervous system in animal experiments (Preussmann and Stewart 1984). On the basis of a food-frequency questionnaire and questions on food preparation and food supply, the authors concluded that dietary intake of nitrate and nitrite was not associated with an increase in risk of glioma or meningioma in the population examined. Bunin et al. (1993, 1994) performed case-control studies to evaluate the relationship between maternal dietary exposure to nitrosamine precursors and the risk of astrocytoma and primitive neuroectodermal tumors of the brain in children in the United States and Canada. Increased nitrate intake was associated with a decrease in the risk of primitive neuroectodermal tumors, whereas there was no association between nitrate or nitrite intake and astrocytoma risk.
A case-control study on diet and gastric cancer conducted in
Spain found a positive correlation between gastric cancer and high consumption of foods containing nitrite but an inverse association with consumption of foods containing nitrate (González et al. 1994). Buiatti et al. (1990) also reported a negative trend in the association between risk of gastric cancer and nitrate intake as estimated with a food-frequency questionnaire in Italy, whereas Hansson et al. (1994) reported no association in a Swedish study.
Fontham et al. (1986) actually measured nitrate and nitrite in the gastric juice of gastritis patients and found no differences from measurements of controls. Gastritis patients reportedly had consumed less fruit, vegetable, and vitamin C than controls, however. Those results were confirmed by Sobala et al. (1991), who determined concentrations of nitrate and nitrite in the gastric juice of patients with and without precancerous conditions of the stomach and found no differences related to gastric pathology. Knight et al. (1991) performed a similar study and found an inverse relationship between gastric nitrate concentration and the severity of gastric disease, but no association with gastric nitrite concentration.
The case-control studies of nitrate exposure and cancer risk thus show either no association or an inverse correlation. Negative associations are likely to be a result of the fact that vegetables are the primary dietary source of nitrate; diets rich in vegetables have consistently been shown to be associated with lower cancer risk (NRC 1989).
Most epidemiologic studies of nitrate, nitrite, and cancer are geographic-correlation studies. They attempt to make associations between average cancer incidences in a geographic area and nitrate or nitrite concentrations in the food or water of that area. Such studies are useful for developing hypotheses, but they cannot demonstrate causality, because exposure to nitrate or nitrite is not established and because they seldom take other risk factors into account. In addition, correlations are generally sought between current nitrate or nitrite concentrations and current cancer inci-
dence or mortality, whereas cancer risk is a function of past exposure.
Geographic studies of nitrate and nitrite concentrations in drinking water and cancer rates have not found consistent correlations. For example, Davies (1980) evaluated death rates from stomach cancer in a town in England where the water contained nitrate at 90 mg/L and found no differences from rates in towns where nitrate concentrations were not high. Juhasz et al. (1980) found that in one county in Hungary stomach-cancer incidence correlated with drinking-water nitrate concentrations in some locations but not in others. A study of esophageal-cancer rates in Iran found no correlation with drinking-water nitrate or nitrite concentrations (Joint Iran-IARC Study Group 1977). Gilli et al. (1984) reported a positive association between gastric-cancer incidence and drinking-water nitrate concentrations in Italy, but this study did not control for other important risk factors for gastric cancer, such as age, diet, alcohol consumption, and occupation. Anquela et al. (1989) reported a positive association between drinking-water nitrate concentration and gastric-cancer mortality but also did not control for other risk factors. In a similarly uncontrolled study, Xu et al. (1992) examined a population at high risk for gastric cancer in China and reported a positive correlation between nitrate intake via drinking water and gastric disease, including cancer, but no relationship with nitrite intake. Chen et al. (1993) found no relationship between dietary nitrate and nitrite exposure and esophageal-cancer mortality when different communities in China were compared but reported a positive correlation between the urinary concentrations of N-nitrosoamino acids and nitrate that was associated with the consumption of nitrate-rich vegetables. Cuello et al. (1976) found that people in Colombia at high risk for stomach cancer lived in areas where nitrate concentrations in drinking-water wells were higher than in lower-risk areas, drank well water more frequently, and had higher urinary nitrite concentrations. In contrast, both Beresford (1985) and Forman et al.
(1985) found that stomach-cancer mortality rates were lower in parts of the United Kingdom where nitrate intake was higher.
In the context of a multistep model for gastric carcinogenesis, Correa (1992) has proposed that the overall epidemiologic and pathologic evidence suggests that ingestion of ascorbic acid and nitrate (determinants of intragastric nitrosation) is associated with effects on the intermediate stages of gastric carcinogenesis, namely, intestinal metaplasia and dysplasia. A positive correlation between urinary nitrate and nitrosamine excretion has been reported in persons with intestinal metaplasia and dysplasia in a population at high risk for gastric cancer in Colombia (Stillwell et al. 1991). A positive association between the proportion of detectable nitrite in gastric juice and the risk of gastric precancerous lesions was found in the same population (Chen et al. 1990).
For various reasons, it is difficult to show a relationship between nitrate and nitrite intake from drinking water and cancer incidence or mortality in humans. First, of course, it is possible that there is no such relationship or that long latency periods or threshold effects make it difficult to establish a relationship. Second, humans are exposed to nitrate and nitrite from many sources other than drinking water and also make nitrate endogenously, so individual exposures vary widely. Third, many dietary factors inhibit nitrosamine formation from nitrite, such as antioxidants, and individual exposure to these varies widely. Fourth, intake of nitrosatable amines varies widely. Finally, the epidemiologic studies that have been conducted to date suffer from a variety of limitations, such as lack of historic exposure measurements, small sample size, and confounding by concomitant exposures. It is likely that considering nitrate or nitrite exposure without also considering exposure to nitrosatable amines is not adequate to determine cancer risk. Thus, although nitrate reduction to nitrite and later nitrosamine formation are possible in humans, many variables affect this process, and exposure to drinking water that contains
nitrate or nitrite is unlikely to be the rate-limiting factor. However, people with a low intake of fruits and vegetables, who are at increased risk for many epithelial tumors (including stomach cancer) will receive a relatively greater proportion of nitrate from drinking water than those with high intakes of fruits and vegetables (Steinmetz and Potter, 1991). People with low fruit and vegetable intakes will also have a relatively lower intake of ascorbic acid and other vegetable-derived antioxidants, the proportion of Americans consuming the recommended five servings of fruits and vegetables per day is less than 20% (Lanza et al., 1987), so some overlap between low dietary intake and high water-nitrate concentration can occur. Diets low in fruits and vegetables might or might not be high in nitrosatable amines, but this possibility contributes some uncertainty to the conclusion that exposure to nitrate or nitrite from drinking water is unlikely to be associated with increased human cancer incidence.
REPRODUCTIVE AND DEVELOPMENTAL TOXICITY
Numerous studies have been conducted in laboratory animals to evaluate the reproductive and developmental toxicity of nitrate and nitrite. In general, little evidence of toxicity has been found except at relatively high doses, which also can produce maternal methemoglobinemia. A single study in rats has reported developmental effects of exposures encountered by humans (Markel et al. 1989). But the few studies that have been conducted in humans have yielded no evidence of any reproductive or developmental effects of nitrate or nitrite. EPA (1990a) has reviewed the studies of the reproductive and developmental toxicity of nitrate and nitrite. Several studies reported since the EPA publication are also described here.
In a large study of both reproductive and developmental effects, nitrate was administered by gavage during pregnancy to rats, mice, hamsters, and rabbits. Mice and hamsters received nitrate at up to about 300 mg/kg-day and rats and rabbits up to 180 mg/kg-day. Similar studies were performed with nitrite at up to 15 mg/kg-day in mice and rabbits, 6.5 mg/kg-day in rats, and 17 mg/kg-day in hamsters. No fetal toxicity, malformations, or effects on maternal reproductive characteristics were observed (FDA 1972a,b,c,d). Kammerer (1993) and Kammerer and Siliart (1993) studied the reproductive and developmental effects of nitrate in female rabbits and reported no effects on reproductive performance, fertility, litter size, or weight at birth or at weaning in association with nitrate at 250 or 500 mg/L of drinking water (about 9 or 18 mg/kg-day).
Studies of nitrate or nitrite mutagenicity in mammalian germ cells for the purpose of determining genotoxic activity are few. Increases in sperm-head abnormalities were reported in mice that were sacrificed 5 weeks after receiving nitrate at 870 mg/kg-day or nitrite at 40 or 80 mg/kg-day by gavage for 14 days (Alavantic et al. 1988a). The pregnancy rate was reduced among females caged with males 10 days after males were treated with the highest dose of nitrite. The basis for this effect is not known. Similar effects were not seen among the mice receiving lower doses of nitrite or in the nitrate-treated groups. Sperm-head abnormalities were also increased in males sacrificed 11 or 17 days after treatment with nitrite at 80 mg/kg-day for 3 days but not in males that received nitrate at 875 mg/kg-day or nitrite at 40 mg/kg-day (Alavantic et al. 1988b). None of the 3-day treatments induced unscheduled DNA synthesis in treated spermatids.
In guinea pigs receiving nitrate at 5,000 mg/kg-day in their drinking water for 143-204 days, no fetotoxicity or effects on fertility were observed (Sleight and Atallah 1968). Effects on
reproductive performance were observed only at the highest dosage, which was associated with a decrease in number of live births that might have been attributable to maternal methemoglobinemia. In a similar study with nitrite, reproductive performance was unaffected up to a dosage of 600 mg/kg-day but higher dosages severely reduced the numbers of live births (Sleight and Atallah 1968).
Effects of nitrite on body and organ growth were reported in a three-generation study on rats (Hugot et al. 1980). Chronic administration of nitrite at about 300 or 525 mg/kg-day in the diet of females had no effects on reproduction or fertility. The rate of weight gain by pups was lower than that in untreated controls, however, and the weights of several organs were abnormal. Similar results were obtained by Roth et al. (1987), who noted anemia and reduced weight gain among the pups of dams supplied with drinking water containing nitrite at 145 or 200 mg/kg-day during gestation and 275 or 340 mg/kg-day during lactation. Those effects were not observed in pups exposed only during gestation, and it was concluded that postnatal exposure was more important.
Markel et al. (1989) reported alterations in neurobehavioral development among the offspring of rats given nitrate at 7.5 or 15 mg/kg-day in their drinking water during gestation and lactation. The offspring were also provided with the nitrate-containing water on weaning. Nitrate-exposed offspring developed hearing-startle reaction and mature righting and cliff-avoidance reflexes earlier than untreated controls. Hyperactivity was reported soon after birth, but hypoactivity was observed at day 20. As adults, the treated male rats showed deficits in rewarded-discriminative-learning tests and in establishing active-avoidance response. No dose-response relationship was observed, however, so it is not possible to eliminate indirect behavorial effects. It is likely that the motor changes reported when the animals were young affected their learning behavior; no additional studies were performed to see
whether the neurobehavioral effects reported were unrelated to the motor effects. Isaacson and Fahey (1987) and Viveiros and Tondat (1978) reported behavioral deficits in rats resulting from sodium nitrite exposure, but their studies were subject to the same limitations as the study of Markel et al. (1989). Additional drawbacks of the Isaacson and Fahey (1987) study include the use of a single dose administered intraperitoneally and failure to rule out a number of confounding factors. Taken together, these studies provide scant evidence that nitrate or nitrite produces neurobehavioral effects, although they provide suggestive evidence of effects that might be clarified by further research. It is not known whether the observations in rats can be extrapolated to humans.
Developmental and fetal toxicity has been reported among rats and mice treated simultaneously with nitrite and nitrosatable amines (Ivankovic et al. 1973; Teramoto et al. 1980). Similar tests with nitrate have not been reported. Of the offspring of rats receiving ethylurea at 100 mg/kg and nitrite at 33 mg/kg, 60% developed hydrocephalus and died within 8 weeks of birth; this effect was not observed when ascorbic acid at 250 mg/kg was also administered. Decreases in fetal survival and increases in fetal malformations were observed when pregnant mice were given simultaneous doses of ethylenethiourea at 400 mg/kg and nitrite at 67 or 134 mg/kg. These effects were not seen when nitrite was administered 2 hours after ethylenethiourea or when either chemical was administered alone. Results of both studies indicate that the toxic effects observed were due to nitrosamine formation. Extrapolating the results to humans must be approached with caution for the same reasons that the relevance of the carcinogenicity bioassays discussed must be questioned: the gastric acidity of humans favors nitrosamine formation but that of rodents does not, and the doses used were often much higher than those encountered in the environment. Therefore, pharmacokinetic characteristics must be considered if extrapolations are made.
Several case-control studies have been performed to evaluate the potential relationship between nitrate exposure from drinking water and reproductive or developmental effects in humans. One study compared people with CNS defects to controls without such effects in New Brunswick, Canada, and found no significant association between birth defects and nitrate exposure (Arbuckle et al. 1988). A study in Boston found that exposure to nitrate in water was associated with a decrease in the risk of spontaneous abortion but that exposure to nitrite had no effect (Aschengrau et al. 1989). An increase in the risk of birth defects was reported in South Australia among women consuming groundwater containing nitrate at 5-15 mg/L, compared with women consuming rainwater containing nitrate at less than 5 mg/L (Scragg et al. 1982; Dorsch et al. 1984). The authors could not attribute their results to nitrate, however, because actual exposure to nitrate was not determined and because other contaminants of groundwater were probably present as a result of the runoff of agricultural chemicals and chemicals used in the local wood-processing industries.
Inorganic nitrite, but not inorganic nitrate, can produce hypotension in humans as a result of its action as a smooth muscle relaxer, especially in the vascular system. This effect is similar to that of some organic nitrates and is thought to result from the production of S-nitrosothiols or nitric oxide; both stimulate guanylate cyclase activity (Knowles et al. 1989). Hypotension as a result of sodium nitrite administration has been recognized for well over a century. It occurs in most, if not all, mammals, including humans,