In this chapter, the dose-response relationships for the health effects of nitrates and nitrites are characterized so that the health effect of primary concern can be identified.
Very few of the reports of studies performed on nitrate-induced methemoglobinemia in humans that include dose information involve cases occurring at nitrate concentrations below 50 mg/L, and most cases occurred when bacterial contamination of water supplies was present as well. In some cases, important dietary sources of nitrate (such as spinach) were also identified.
The current MCLG for nitrate is based on the study of Walton (1951). The American Public Health Association sent questionnaires to all 48 states investigating the morbidity and mortality among infants due to methemoglobinemia induced by nitrate-contaminated water. The survey identified 278 cases and 39 deaths that could be “definitely associated with nitrate-contami-
nated water.” Nitrate exposures were known for 214 cases, and all of them exceeded 50 mg/L; of the 214 cases, 81% occurred above 220 mg/L, 17% at 90-220 mg/L, and only 2% at 50-90 mg/L. The presence of nitrite, of bacteriologic contamination, and of gastrointestinal disease and methemoglobin concentrations were not reported.
In a similar survey in Germany, 745 cases of methemoglobinemia among infants were identified (Simon et al. 1964); data on exposure were available for 249 of the cases. Nitrate concentration in water exceeded 100 mg/L in 84%, was 50-100 mg/L in 12%, and was less than 50 mg/L in only 4%. The only three cases that occurred at concentrations below 20 mg/L were associated with nitrite and substantial dietary nitrate exposure as well. Of the 306 cases for which additional information was available, 98% occurred in infants aged 3 months old or younger, and 53% of the infants had diarrhea, an indicator of bacterial contamination and a factor associated with endogenous nitrate formation.
Dose-response relationships for nitrate exposure and methemoglobin concentrations have been reported in several studies of infants. For example, normal methemoglobin concentrations (less than 3%) were observed in infants fed water that contained nitrate at up to 50 mg/L, with mean methemoglobin concentrations increasing with nitrate intake up to 6.6% in those consuming over 100 mg/L (Würkert 1978; Toussaint and Würkert 1982). Similar dose-response relationships have not been observed in children or adults, in whom increasing nitrate exposure has little or no effect on methemoglobin concentration (Craun et al. 1981). In adults, methemoglobinemia has been reported only in cases of accidental ingestion of large amounts of nitrite. However, the concentration of methemoglobin that constitutes an adverse health effect has not been established definitively. Other factors, such as infantile diarrhea, can influence methemoglobin concentrations in the absence of increased concentrations of nitrate in food or water.
Epidemiologic data do not support a straightforward association between exogenous nitrate or nitrite exposure and human carcinogenesis. A discussion of dose-response relationships between human carcinogenesis and nitrate or nitrite exposure is not appropriate without supporting epidemiologic data and a physiologically based pharmacokinetic model that would permit analysis of the complex relationships between exogenous and endogenously formed nitrate, nitrite, and N-nitrosamines. In addition, because there is no evidence that either nitrate or nitrite alone is carcinogenic in animals, a discussion of dose-response relationships between carcinogenesis in animals and nitrate or nitrite exposure is not possible. The only evidence of a role of nitrite in carcinogenesis comes from studies in which nitrite was administered simultaneously with a nitrosatable amine; in these cases, carcinogenesis can be attributed to the endogenous formation of carcinogenic nitrosamines.
Cancer risk associated with endogenous nitrosamine formation is a function of four variables: the amount of nitrite ingested or formed from nitrate, the amount of nitrosatable substances ingested, the rate of in vivo nitrosation, and the carcinogenic potency of the resulting nitrosamine. Establishing human dose-response relationships for a phenomenon that has so many variables is not straightforward. The carcinogenic potencies of nitrosamines based on rodent bioassays vary by at least a factor of 1,000 (Shephard et al. 1987). Daily dietary intakes of nitrosatable amines, amides, guanidines, and ureas have been estimated to range from less than 1 mg to hundreds or even thousands of milligrams (Shephard et al. 1987). Dietary nitrate intake is estimated to vary from about 75 to 270 mg (NRC 1981), and the extent to which nitrate is reduced to nitrite endogenously depends on gastric acidity and the nature
and number of bacteria present; dietary nitrite intakes are much lower. It is possible that the development of physiologically based pharmacokinetic models for nitrate and nitrite metabolism and nitrosamine formation would allow each of those variables to be evaluated and the bounds on the dose-response relationships to be estimated (see Appendix A).
REPRODUCTIVE AND DEVELOPMENTAL TOXICITY
Studies in humans are inadequate to support an association between nitrate or nitrite exposure and reproductive or developmental effects. Therefore, developing dose-response relationships based on human data is not possible. Developmental effects and fetal toxicity have been reported among rats and mice receiving both nitrite and nitrosatable amines, but (as discussed above for cancer) developing dose-response relationships for humans on the basis of these data and estimating rates of nitrosamine formation and potency are so uncertain as to be meaningless.
Nitrate has not been reported to produce reproductive effects in animal bioassays. One study reported that nitrate produced alterations in rat neurobehavioral development at 7.5 mg/kg-day (Markel et al. 1989); although the study has several drawbacks, this is the lowest dosage at which any developmental effects have been reported. This dosage can be converted to a human adult dosage as follows:
(7.5 mg/kg-day)¾(70 kg) = 317 mg/day,
where the exponent ¾ is used to account for the difference in surface area between rats and humans (Federal Register 1992) and
the average human is assumed to weigh 70 kg. Assuming that the average human infant 0-3 months old weighs about 5 kg, the adult dosage is equivalent to an infant's dosage of about 23 mg/day.
Reproductive effects attributable to nitrite exposure have been reported in animal bioassays at dosages that might have been associated with maternal methemoglobinemia. Developmental effects of nitrite that have been reported at lower dosages in rodents appear to result from exposure after birth and not in utero (Roth et al. 1987). The effects included anemia and reduced weight gain. The lowest dosage at which the effects were reported was 275 mg/kg-day in rats (Roth et al. 1987). That dosage can be converted to a human infant dosage as follows:
(275 mg/kg-day)¾(5 kg) = 338 mg/day,
where the same assumptions were used as for nitrate.
In most mammals, including humans, the vasodilator effects of sodium nitrite overlap the dosage ranges that cause methemoglobinemia (Sollman 1957). The discussion of dose-response relationships for methemoglobinemia is thus applicable to the vasodilator effects as well. Although the stagnant hypoxia that results from prominent vasodilation might contribute to the anemic hypoxia resulting from methemoglobinemia, it is clear that methemoglobinemia is the primary cause of death. Methylene blue can reverse nitrite-induced methemoglobinemia and protect against death. Maintenance of normal blood pressure has never been shown to protect against nitrite lethality (Smith and Wilcox 1994).