Workgroup Report: Drinking-Water Nitrate and Health—Recent Findings and Research Needs

1Humans have altered the nitrogen cycle dramatically over the last half-century, and as a result, nitrate is steadily accumulating in our water resources. Globally, human nitrogen production has increased rapidly since 1950 and currently exceeds nitrogen fixed by natural sources by about 30%. This figure compares with pre-1950 human inputs, which were a small fraction of the input from natural sources. Fertilizer is the largest contributor to anthropogenic nitrogen worldwide; other major sources include animal and human waste, nitrogen oxides from utilities and automobiles, and leguminous crops that fix atmospheric nitrogen. These organic and inorganic sources of nitrogen are transformed to nitrate by mineralization, hydrolysis, and bacterial nitrification. Under reducing conditions, nitrate can be biologically transformed to nitrogen gas through denitrification. Nitrate not taken up by plants or denitrified migrates to streams and groundwater.

2The U.S. Environmental Protection Agency (EPA) maximum contaminant level (MCL) for nitrate in drinking water of 10 mg/L nitrate-nitrogen (nitrate-N) (equivalent to 45 mg/L as nitrate) and the World Health Organization (WHO) guideline of 50 mg/L as nitrate (equivalent to 11 mg/L as nitrate-N) were promulgated to protect against methemoglobinemia, or “blue baby syndrome,” to which infants are especially susceptible. The regulatory level is usually met for public water supplies, which are routinely monitored. Much less is known about private wells, which in the United States are usually required to be tested only when the well is constructed or when the property is sold. Some have suggested recently that the regulatory level for nitrate in drinking water is overly conservative. However, this discussion of the regulatory level has not thoroughly considered studies of other chronic health effects including cancer, adverse reproductive outcomes, and diabetes. Although a causal role for nitrate in these other health outcomes is not conclusive, recent studies that indicate possible adverse effects at nitrate levels below the MCL are of concern.

3In recognition of the widespread contamination of drinking-water sources by nitrate and the potential for health effects in addition to methemoglobinemia, a symposium titled “Drinking Water Nitrate and Health: Recent Findings and Research Needs” took place at the annual meeting of the International Society for Environmental Epidemiology (1–4 August 2004, New York, New York, USA). Invited experts presented results from recent unpublished studies and summarized the state of knowledge on exposure and health effects of drinking-water nitrate, with a focus on cancer and adverse reproductive outcomes. This article summarizes the symposium discussions and recommends promising areas for future research. Specifically, we discuss the epidemiologic evidence for drinking-water nitrate and risk of specific cancers, adverse reproductive outcomes, and other health outcomes in the context of the current regulatory limit for nitrate in drinking water.

4Nitrate is the most common chemical contaminant in the world’s groundwater aquifers. An estimated 42% of the U.S. population uses groundwater as their drinking-water supply. In the United States, total nitrogen in streams and nitrate in groundwater are highest in agricultural areas, followed by urban areas and areas with mixed land use (Figure 1). The most recent data indicate that about 22% of domestic wells in agricultural areas of the United States exceeded the MCL (U.S. Geological Survey, unpublished data). In contrast, 3% of public supply wells in major aquifers (typical sources for public water supplies) exceed the MCL (U.S. Geological Survey, unpublished data).

5The exposure picture is similar in the European Union. Public water supplies are largely below the WHO guideline; however, in some countries, private wells in rural areas have elevated nitrate concentrations reaching 10–15 times the recommended level. Overall, nitrate levels exceeded the guideline in about one-third of the groundwater bodies for which data were available. Several eastern European countries report high levels of nitrate contamination in a large proportion of private wells; for example, in Romania, 20% of 2,000 wells had nitrate levels > 23 mg/L as nitrate-N. Studies from other countries, including China, Botswana, Turkey, Senegal, and Mexico, report private well water levels that exceed the WHO guideline, in some instances at levels > 68 mg/L nitrate-N. Fertilizer is the main contributing factor in agricultural areas; however, nitrogen from human waste appears to be the most important source in urban areas lacking centralized water and sanitation systems. Although systematic information on nitrate levels in groundwater in other parts of the world is more limited, empirical modeling approaches have indicated that users of shallow wells in areas with high nitrogen inputs, well-drained soils, and unconsolidated rocks are most at risk of consuming high-nitrate groundwater.

6Ingested nitrate is reduced to nitrite, which binds to hemoglobin to form methemoglobin (MetHb). Methemoglobinemia occurs when elevated levels of MetHb (exceeding about 10%) interfere with the oxygen-carrying capacity of the blood. Infants are particularly susceptible to developing methemoglobinemia for several reasons, including their increased capacity to convert nitrate to nitrite and their lower levels of the enzyme cytochrome b5 reductase, which converts MetHb back to hemoglobin. Methemoglobinemia in infants fed formula made with well water with high nitrate levels was first reported in 1945 by. The regulatory level for nitrate in drinking-water supplies was determined after a survey of infant methemoglobinemia case reports in the United States indicated that no cases were observed at drinking-water nitrate levels < 10 mg/L nitrate-N. Because an estimated 22% of domestic wells in agricultural regions of the United States exceed the nitrate MCL (U.S. Geological Survey, unpublished data), it is likely that significant numbers of infants are given water containing > 10 mg/L nitrate-N. Nevertheless, few cases of methemoglobinemia have been reported since the MCL was promulgated.

7The risk of methemoglobinemia among infants depends on many factors other than the ingestion of nitrate in drinking water. Some foods and medications contain high levels of nitrate. Enteric infections, potentially caused by fecal bacteria contamination in wells, may lead to the endogenous production of nitrite, as evidenced by numerous published reports of infants with diarrhea and methemoglobinemia but no apparent exposure to exogenous MetHb-forming agents. The consumption of antioxidants such as vitamin C appears to be a protective factor. Finally, polymorphisms in the activity of cytochrome b5 reductase may mediate the effect of ingested nitrate or endogenously produced nitrite.

8Studies that have examined the relationship between nitrate levels in drinking water and MetHb levels in infants have produced mixed results. The few experimental studies are largely negative; however, most of these studies evaluated low levels of drinking-water nitrate and included few infants. Cofactors such as diarrhea and respiratory diseases reportedly increase MetHb levels. An epidemiologic study in South Africa found an increase in MetHb levels in infants fed water with nitrate > 20 mg/L nitrate-N; however, clinical methemoglobinemia was rarely found. A protective effect of vitamin C intake on MetHb was noted. More recently, a retrospective, nested case–control study in Romania found an association between nitrate exposure from drinking water and clinical methemoglobinemia, but also some evidence of an association with diarrheal disease. found cytochrome b5 reductase activities to be higher among those consuming water with high nitrate levels, indicating a level of adaptation to the consumption of high nitrate waters.

9Recently, the role of nitrate exposure alone in causing methemoglobinemia has been questioned. Clearly, we need to better understand the interaction of factors that lead to methemoglobinemia to assess the relative importance of each factor and to identify the conditions under which exposure to nitrate in drinking water poses a risk of methemoglobinemia.

10Nitrate is a precursor in the formation of N-nitroso compounds (NOC), a class of genotoxic compounds, most of which are animal carcinogens. In the human body, nitrate is a stable, inert compound that cannot be metabolized by human enzymes. However, the nitrate-reducing activity of commensal bacteria may convert nitrate into nitrite and other bioactive nitrogen compounds that affect physiological processes and human health. After ingestion, nitrate is readily absorbed from the upper gastrointestinal tract. Up to 25% is actively excreted in saliva, where about 20% is converted to nitrite by bacteria in the mouth. This conversion can occur at other sites including the distal small intestine and the colon.

11Under acidic conditions in the stomach, nitrite is protonated to nitrous acid (HNO2), which in turn spontaneously yields dinitrogen trioxide (N2O3), nitric oxide (NO), and nitrogen dioxide (NO2). NO is a bioactive compound known to play a role in vasodilatation and in defense against periodontal bacteria and other pathogens. N2O3, on the other hand, is a powerful nitrosating agent capable of donating NO+ to secondary and tertiary amines to form potentially carcinogenic N-nitrosamines. Alternatively, HNO2 can be protonated to H2NO2, which reacts with amides to form N-nitrosamides. At neutral pH, nitrite can be reduced by bacterial activity to form NO, which can react with molecular oxygen to form the nitrosating compounds N2O3 and nitrogen tetroxide (N2O4). In addition to the acid-catalyzed and bacterial-catalyzed formation of nitrosating agents, inducible NO synthase activity of inflammatory cells can also produce NO. Together, these three mechanisms of endogenous nitrosation account for an estimated 40–75% of the total human exposure to NOC. Other sources of human exposure include pre-formed NOC found in preserved meats and fish, beer, certain occupational exposures, and tobacco products.

12Several studies support a direct relationship between nitrate intake and endogenous formation of NOC. High intake of drinking-water nitrate (above the MCL) is associated with an increased endogenous capacity to nitrosate proline. In addition, populations with high rates of esophageal and gastric cancer excrete high levels of N-nitrosoproline. Nitrate intake at the acceptable daily intake level (3.67 mg/kg body weight, 0.84 mg/kg as nitrate-N) results in increased urinary excretion of NOC, particularly in combination with increased intake of dietary nitrosatable precursors. However, a Canadian population exposed to nitrate below the acceptable daily intake level showed no relationship between nitrate levels in drinking water and urinary nitrosamines.

13Although intake of high drinking-water nitrate is consistently associated with endogenous nitrosation capacity, intake of dietary nitrate is less likely to increase nitrosation, because of the presence of nitrosation inhibitors in vegetables, the major contributors to dietary nitrate intake. Dietary compounds that inhibit endogenous nitrosation include vitamin C, which has the capacity to reduce HNO2 to NO, and alphatocopherol, which can reduce nitrite to NO. Several epidemiologic studies reported no association or inverse associations between dietary nitrate intake and human cancers, which may be because of the antioxidants and nitrosation inhibitors in nitrate-containing foods. Inhibitory effects on nitrosation have also been described with betel nut extracts, ferulic and caffeic acid, garlic, coffee, and green tea polyphenols. In addition, nondietary factors such as the use of mouthwashes containing chlorhexidine can influence the endogenous nitrosating capacity.

14Apart from the level of nitrosating agents, the level of nitrosatable precursors in the diet, which come predominantly from meat and fish, is a crucial factor in endogenous nitrosation. Dietary intakes of red and processed meat are of particular importance in the formation of fecal NOC. Higher consumption of red meat (600 vs. 60 g/day), but not white meat, resulted in a 3-fold increase in fecal NOC levels. Colon cancer incidence is most consistently associated with consumption of red meat (beef, lamb, and pork), but not with poultry and fish. Dietary supplementation of a diet low in red meat with either heme iron or inorganic iron demonstrated that heme in particular was able to stimulate endogenous nitrosation, thereby providing a possible explanation for the differences in colon cancer risk between red and white meat consumption. Additionally, this linkage may be stronger for processed meat than for fresh meat because of the higher NOC and NOC precursor levels in processed meat.

15Endogenous nitrosation can also be stimulated by inflammatory and other medical conditions. For instance, patients with bilharzia have an increased bladder cancer risk associated with increased urinary levels of nitrite and volatile nitrosamines, most likely generated by the reaction of inflammation-derived NO with amines present in the urine. Inflammatory bowel disease is also related to both increased nitro-sation and cancer risk. During inflammatory bowel disease, increased inducible NO synthase activity can produce excess NO, which is oxidized to nitrogen oxides and nitrite, which in turn react with nitrosatable precursors in colonic contents to produce NOC. Indeed, ulcerative colitis patients showed increased levels of inducible NO synthase in the colonic mucosa and of NO and nitrite in the colonic lumen. Increased levels of fecal NOC have been found in patients with inflammatory bowel disease and in mice with chemically induced colitis.

16NOC are potent animal carcinogens, inducing tumors at multiple organ sites including the esophagus, stomach, colon, bladder, lympatics, and hematopoietic system. NOC cause tumors in every animal species tested, and it is unlikely that humans are unaffected. The number of well-designed epidemiologic studies with individual exposure data and information on nitrosation inhibitors and precursors are few for any single cancer site, limiting the ability to draw conclusions about cancer risk.

17Most studies have been ecologic in design, linking incidence or mortality rates to drinking-water nitrate levels at the town or county level. The early studies focused on stomach cancer mortality, and most used drinking-water nitrate measurements concurrent with the period of cancer mortality. Results were mixed, with some studies showing positive associations, many showing no association, and a few showing inverse associations. Recent ecologic studies of stomach cancer in Slovakia, Spain, and Hungary with historical measurements and exposure levels near or above the MCL have found positive correlations with stomach cancer incidence or mortality. Two studies included other cancer sites. In Slovakia, incidence of non-Hodgkin lymphoma (NHL) and colon cancer was significantly elevated among men and women exposed to public supply nitrate levels of 4.5–11.3 mg/L nitrate-N; there was no association with bladder and kidney cancer incidence. In Spain, there was a positive correlation between nitrate levels in public supplies and prostate cancer mortality, but no relation with bladder and colon cancer.

18In the past decade, several case–control and cohort studies have evaluated historical nitrate levels in public water supplies (largely < 10 mg/L nitrate-N) and risk of several cancers (Table 1). Some studies evaluated factors affecting nitrosation, such as vitamin C intake. A cohort study of older women in Iowa (USA) found a 2.8-fold and 1.8-fold risk of bladder and ovarian cancers, respectively, associated with the highest quartile (> 2.46 mg/L nitrate-N) of the long-term average nitrate levels at the current residence. They observed significant inverse associations for uterine and rectal cancer and no significant associations for NHL, leukemia, colon, rectum, pancreas, kidney, lung, and melanoma. Case–control studies of bladder, brain, colon and rectum, and pancreas cancer in Iowa found no association between cancer risk and average nitrate levels over almost 30 years. Each study evaluated the interaction between nitrosation inhibitors or NOC precursors and nitrate intake from drinking water. For colon cancer, there was a significant positive interaction between 10 or more years of exposure above 5 mg/L nitrate-N and both low vitamin C and high meat intake, factors likely to increase endogenous NOC formation.

19A case–control study of NHL in Nebraska (USA) found a significant positive association between the average nitrate level in public water supplies over about 40 years and risk among men and women. In the highest quartile of nitrate (4.0 mg/L nitrate-N), risk was elevated 2-fold. However, a recent study of NHL in Iowa with similar exposure levels found no association. A case–control study of NHL in Minnesota (USA) with lower levels of nitrate found an inverse association among those with the highest level (> 1.5 mg/L nitrate-N). Case–control studies in Nebraska and Germany found no association with long-term average nitrate levels in public water supplies and adult brain cancer. The Nebraska study found no evidence of an interaction with vitamin C intake. A case–cohort analysis of stomach cancer within a cohort study in the Netherlands found no association with quintiles of water nitrate intake determined from public supply levels.

20Specific NOC are transplacental neurocarcinogens in animal studies. A study of childhood brain cancer measured nitrate levels in water supplies using dipstick measurements, often many years after the pregnancy. Measured levels of nitrate and nitrite were not associated with risk; however, women in western Washington State, one of the three study centers, who used private wells as their drinking-water source during the pregnancy had a significantly increased risk of brain cancer in their offspring.

21In 1961, Schmitz described a possible relationship between high maternal MetHb levels and spontaneous abortion. Since then, at least 10 studies have examined the association between drinking-water nitrate and adverse reproductive outcomes. Table 2 summarizes these studies by location, study design, determination of water nitrate, and key findings. Few studies have been published regarding water nitrate and the outcomes of spontaneous abortions, stillbirths, premature birth, or intrauterine growth retardation. Results of these studies have been inconsistent, possibly indicating no true effect of water nitrate on reproductive outcomes at the levels evaluated in these studies. Alternatively, the inconsistencies may be due to the differing periods over which exposure was assessed, differing levels of water nitrate across studies, or differences in exposure to other cofactors.

22Results of studies evaluating drinking-water nitrate and congenital malformations in offspring are also mixed (Table 2). Four studies found positive associations between drinking-water nitrate and congenital malformations, particularly malformations of the central nervous system, and specifically neural tube defects (NTDs). In each of these studies, water nitrate levels associated with increased risk of these defects were below the MCL, although the 95% confidence intervals (CIs) for some of the risk estimates were consistent with unity and varied by the source of water (groundwater, mixed, or surface). Two of these studies also examined dietary intake of nitrates and nitrates and NTDs and found minimal or no effect on risk. In a study of nitrosatable drug exposure and risk of NTDs, drinking-water nitrates and dietary nitrites/total nitrites substantially modified the risk associated with this drug exposure during the periconceptional period; higher levels of nitrates in food or water significantly increased the risk of NTDs if women were exposed to such drugs.

23Animal studies suggest that nitrate at high doses can competitively inhibit iodine uptake and induce hypertrophic changes in the thyroid. In a human biomonitoring study in the Netherlands, consumption of water with nitrate levels at or above the MCL was associated with thyroid hypertrophy and genotoxic effects. Animal studies provide evidence that NOC can damage the pancreatic beta cells. Three epidemiologic studies that were ecologic in design found a positive correlation between drinking-water nitrate levels below the MCL and the incidence of type I childhood diabetes, although the association observed by van Maanen was not statistically significant. Other studies have found associations between water nitrate exposure and increased blood pressure and acute respiratory tract infections in children.

24Endogenous nitrosation in humans has been demonstrated in relation to drinking-water nitrate ingestion at levels above the MCL. However, further studies are needed to determine the extent of endogenous nitrosation at intermediate drinking-water nitrate levels (5–10 mg/L as nitrate-N) and to clarify the role of nitrate from water versus food sources. Furthermore, the role of precursors and modulators of NOC formation should be more fully investigated. These future studies should be conducted among healthy individuals as well as individuals with medical conditions that increase endogenous nitrosation.

25In view of the complex kinetics of NOC formation and the organ specificity of several of these compounds, more studies are needed to evaluate the relationship between nitrate intake and formation, metabolism, and excretion of NOC. Ideally, a physiologically based pharmacokinetic model should be developed as previously recommended to predict exposure to NOC from all sources of nitrate exposure (exogenous and endogenous), nitrite intake, the transformation of nitrate into nitrite, and antioxidant intake. However, this will require additional data on the formation of individual NOC as well as their respective toxicologic characteristics. The results of these investigations will reveal the value of different markers of NOC exposure in future epidemiologic studies. Future studies linking NOC exposure to early markers of effect or to the actual disease will clarify the role of endogenous nitrosation and NOC exposure as etiologic factors.

26Because many NOC require α -hydroxylation by CYP2E1 for bioactivation and for formation of DNA adducts, it is important to investigate the influence of polymorphisms in the gene encoding for this enzyme. One study found that specific variants in this gene are linked to increased rectum cancer risk, particularly in subjects with high intake of red and processed meat, who are exposed to increased levels of NOC. Moreover, gene expression levels of human CYP2E1 were related to cytotoxicity and DNA damage by nitrosamines in pancreatic beta-cell lines, suggesting that such gene environment interactions are also relevant in type 1 diabetes. These promising lines of research point to a possible interaction between drinking-water nitrate exposure and gene expression of and/or genetic variation in CYP2E1, which may also influence the risk of several adverse health outcomes associated with nitrate exposure.

27Methods must be developed and validated to improve estimates of current and historical exposure to nitrate via food and water, particularly for populations served by private wells, which are less likely to be routinely monitored. Future epidemiologic studies should integrate a) exposure assessment for nitrate intake from drinking water, nitrate and nitrite intake from the diet, and amines and amides from dietary and drug sources, b) endogenous exposure to NOC by analysis of relevant biological media (e.g., saliva, urine, feces), and c) reliable health risk markers (e.g., biomarkers of genotoxicity) or diagnosis of actual disease.

28Future studies should include populations with well-characterized long-term exposures, including those who use private wells, which can have higher nitrate levels than public supplies. With the increasing availability of public water supply monitoring data (many U.S. states have almost 40 years of measurements), further detailed exposure assessment of populations using public supplies is also feasible. Drinking-water contaminants that may occur along with nitrate, such as agricultural pesticides, should also be evaluated. Geographic-based modeling efforts to predict the probability of high nitrate concentrations in groundwater, using information on nitrogen inputs from agricultural and urban sources, is a promising approach for estimating drinking-water nitrate exposure for the population using private wells.

29Additional studies of drinking-water nitrate and cancer are needed to follow up on the suggestive positive findings to date and to evaluate other cancer sites in which endogenously formed NOC may play a role. Studies of reproductive outcomes should address the exposure period most relevant for the specific outcome of interest. Maternal residential mobility between conception and birth may lead to misclassification of exposure if the water source at birth is used in studies of spontaneous abortions and congenital malformations. Studies must be of sufficient size to allow for examination of specific defects rather than groups of defects by system, because combining different defects might mask associations. More research is needed on the relation between water nitrate and the reproductive outcomes of spontaneous abortion, fetal death, premature birth, and intrauterine growth retardation.

30In the design and analysis stage, future epidemiologic studies should consider factors that modulate endogenous nitrosation, as discussed above, to be able to evaluate potential interactions of water nitrate intake with these factors, thus providing stronger evidence for or against an association. In particular, studies of susceptible populations may be fruitful, and epidemiologic studies should be designed with sufficient power to evaluate risk among potentially susceptible subgroups. Such populations include patients with different forms of chronic inflammation (such as inflammatory bowel disease), patients infected with nitrate-reducing bacteria (such as in periodontal disease), those with low intake of vitamins and other known nitrosation inhibitors, or those with a history of high incidence of potentially NOC-related diseases. The people of Linxian County in China, for example, are known for their persistently low intake of several micronutrients and high risk of esophageal cancer. Such populations will likely benefit from preventive measures taken as a result of these investigations.

31Adverse health effects from drinking-water nitrates are most likely the result of a complex interaction of the amount of nitrate ingested, the concomitant ingestion of nitrosating cofactors and precursors, and medical conditions of the host that may increase nitrosation. Furthermore, these effects may be attenuated by inhibitors of endogenous nitrosation such as vitamin C and alphatocopherol. We recommend that future studies take into account such complexities in understanding the relation between drinking-water nitrates and cancer, adverse reproductive outcomes, and other health outcomes.

32Several authors have questioned the importance of nitrate in drinking water as a risk factor for methemoglobinemia and have suggested that the current nitrate standard might be safely raised to 15–20 mg/L nitrate-N with no increase in methemoglobinemia cases. A better understanding of the conditions under which nitrate in drinking water poses a risk of methemoglobinemia is clearly needed, particularly in light of recent cases of methemoglobinemia associated with well water levels between 20 and 30 mg/L nitrate-N. Most importantly, the role of nitrate as a risk factor for cancer and adverse reproductive outcomes must be more thoroughly explored before changes to nitrate water quality standards are considered.