The Safe Drinking Water Act (SDWA) directs the U.S. Environmental Protection Agency (EPA) to establish national standards for contaminants in public drinking-water supplies. Enforceable standards are to be set at concentrations at which no adverse health effects in humans are expected to occur and for which there are adequate margins of safety. Enforceable standards are standards that can be achieved with the use of the best technology available.
Arsenic is a naturally occurring element present in the environment in both inorganic and organic forms. Inorganic arsenic is considered to be the most toxic form of the element and is found in groundwater and surface water, as well as in many foods. A wide variety of adverse health effects, including skin and internal cancers and cardiovascular and neurological effects, have been attributed to chronic arsenic exposure, primarily from drinking water. EPA's interim maximum contaminant level (MCL) for arsenic in drinking water is 50 micrograms per liter (µg/L). Under the 1996 SDWA amendments, EPA is required to propose a standard (an MCL) for arsenic in drinking water by January 2000 and finalize it by January 2001.
The Charge to the Subcommittee
In 1996, EPA's Office of Water requested that the National Research Council (NRC) independently review the arsenic toxicity data base and evaluate the scientific validity of EPA's 1988 risk assessment for arsenic in drinking water. The NRC assigned this project to the Committee on Toxicology (COT), which convened the Subcommittee on Arsenic in Drinking Water, whose membership includes experts in toxicology, pharmacology, pathology, chemistry, nutrition, medicine, epidemiology, risk assessment, and biostatistics. The subcommittee was charged with the following tasks: (1) review
EPA's characterization of human health risks from ingestion of arsenic compounds found in food and drinking water and the uncertainties associated with that characterization; (2) review available data on cancer and noncancer health effects from exposure to arsenic compounds in drinking water and the implications of these effects on the assessment of the human health risks from arsenic exposure; (3) review data on the toxicokinetics, metabolism, and mechanism or mode of action of arsenic and ascertain how these data could assist in assessing human health risks from drinking-water exposures; and (4) identify research priorities to fill data gaps. EPA did not request, nor did the subcommittee endeavor to provide, a formal risk assessment for arsenic in drinking water.
The Subcommittee's Approach To Its Charge
The subcommittee evaluated data relating to key elements of the risk-assessment processhazard identification, dose response, and risk characterizationthat addresses the protective nature of the current MCL. Specifically, the subcommittee reviewed information on the health effects of arsenic exposure and data on the disposition and the mechanism or mode of action of arsenic. The subcommittee also evaluated other information that could affect the risk assessment, such as variations in human susceptibility, and current capabilities to measure arsenic in various media, including biological tissues. The major conclusions and recommendations of the subcommittee in each of those areas are discussed in the remainder of this summary. The implications of these findings on the assessment of human health risk is provided below in the section on risk characterization.
The Subcommittee's Evaluation
The subcommittee concludes that there is sufficient evidence from human epidemiological studies in Taiwan, Chile, and Argentina that chronic ingestion of inorganic arsenic causes bladder and lung cancer, as well as skin cancer. With minor exceptions, epidemiological studies for cancer are based on populations exposed to arsenic concentrations in drinking water of at least several hundred micrograms per liter. Few data address the degree of cancer risk at lower concentrations of ingested arsenic. Noncancer effects resulting from chronic ingestion of inorganic arsenic have been detected at doses of 0.01 milligram per kilogram (mg/kg) and higher per day. Of the noncancer
effects, cutaneous manifestations of exposure have been studied most widely. Developmental and reproductive effects resulting from chronic ingestion of inorganic arsenic have not been demonstrated in humans, although arsenic is known to pass through the placenta. Parenteral administration of inorganic and organic forms of arsenic are known to be teratogenic in a number of mammalian species, and oral administration affects fetal growth and prenatal viability. Arsenic has not been tested for essentiality in humans, nor has it been found to be required for any essential biochemical processes. Arsenic supplementation at very high concentrations (e.g., 350-4,500 nanograms per gram (ng/g)) in the diet has been shown to affect growth and reproduction in minipigs, chicks, goats, and rats.
Additional epidemiological evaluations are needed to characterize the dose-response relationship for arsenic-associated cancer and noncancer end points, especially at low doses. Such studies are of critical importance for improving the scientific validity of risk assessment. With respect to cancer, studies are recommended to refine the dose-response relationship between arsenic ingestion and cancer of the skin, bladder, and lung, and to investigate the effect of arsenic on cancer at other sites. With respect to noncancer effects, particular emphasis should be placed on epidemiological study of arsenic-associated cutaneous effects, cardiovascular and cerebrovascular disease, diabetes mellitus, and adverse reproductive outcomes.
Future studies on the beneficial effects of arsenic in experimental animals should carefully monitor the amount and speciation of arsenic in diets and water, use biomarkers to assess arsenic exposure and bioavailability, and use techniques that assess the toxicity and benefits of arsenic in a more specific manner than is possible through measurement of growth and reproductive success. In humans, the concentration of arsenic in total parenteral nutrition (TPN) should be determined by validated analytical methods and related to the health status of patients on long-term TPN.
Disposition (Absorption, Distribution, Metabolism, and Excretion)
In humans, inorganic arsenic is readily absorbed from the gastrointestinal tract and is primarily transported in the blood bound to sulfhydryl groups in proteins and low-molecular-weight compounds, such as amino acids and peptides. The half-life of arsenic in the body is about 4 days, and it is primarily excreted in the urine. Humans and some animals methylate inorganic
arsenic to forms that are less acutely toxic and more readily excreted. However, the methylation process varies among animal species, making most animal models less suitable for studying the disposition of arsenic in humans. The methylation of ingested arsenic is not inhibited or overloaded, unless acute toxic doses are ingested. Substantial variations in the fractions of methylated forms of arsenic in urine are also known to occur among different populations and individuals within the same exposed population. Such variations might be indicative of genetic differences in the enzymes responsible for the methylation of arsenic. Methylation of arsenic might also be influenced by such factors as the arsenic species absorbed, high acute doses, nutrition, and disease. The extent to which variation in arsenic methylation affects its toxicity, including carcinogenicity, is not known.
Because of interspecies differences in the disposition of arsenic, more human studies are needed, including research using human tissues. Factors influencing the methylation, tissue retention, and excretion of arsenic in humans also need to be investigated.
Mechanism or Mode of Action
The mechanism or mode of action by which inorganic arsenic causes toxicity, including cancer, is not well established. In vivo studies in rats and mice to determine the ability of inorganic arsenic to act as a cocarcinogen or as a promoter have produced conflicting results. Studies on the arsenic metabolite, dimethylarsinate (DMA), suggest that it is not an initiator but might act as a promoter. However, those studies used very high doses, making interpretation of the results difficult, especially if DMA is formed in situ following the administration of inorganic arsenic.
The most accepted explanation for the mode of action for arsenic carcinogenicity is that it induces chromosomal abnormalities without interacting directly with DNA. These markers of tumor response would lead to a dose-response curve that exhibits sublinear characteristics at some undetermined region in the low-dose range, although linearity cannot be ruled out.
The mechanism of action by which arsenic induces noncancer effects is centered on its inhibitory effects on cellular respiration at the level of the mitochondrion. Hepatotoxicity is a major health effect related to decreased cellular respiration. Oxidative stress might also have an important role in both cancer and noncancer effects.
Identification of proximate markers of arsenic-induced cancers and their application in carefully designed epidemiological studies might better define the cancer dose-response curves at low concentrations. Molecular and cellular characterization of neoplasms from arsenic exposed populations and appropriate controls might aid in identifying the mechanism by which arsenic induces tumors. Chronic low-dose studies in a suitable animal model (mouse, hamster, or rabbit) might increase our understanding of the mode of action of arsenic carcinogenicity, particularly the potential role of chromosomal alterations.
A greater understanding is needed of the inter-relationships between arsenic's effects on cellular respiration and its effects on biochemical processes, including methylation, formation of reactive oxygen species, oxidative stress, and protein stress response.
Variation in Human Sensitivity
Human sensitivity to the toxic effects of inorganic arsenic exposure is likely to vary based on genetics, metabolism, diet, health status, sex, and other possible factors. These factors can have important implications in the assessment of risk from exposure to arsenic. A wider margin of safety might be needed when conducting risk assessments of arsenic because of variations in metabolism and sensitivity among individuals or groups. For example, people with reduced ability to methylate arsenic retain more arsenic in their bodies and may be more at risk for toxic effects. One study suggests that children have a lower arsenic-methylation efficiency than adults. Similarly, poor nutritional status might decrease the ability of an individual to methylate arsenic, resulting in increased arsenic concentrations in tissues and the development of toxic effects. There is some evidence from animal studies that low concentrations of S-adenosylmethionine, choline, or protein decrease arsenic methylation.
Factors that influence sensitivity to or expression of arsenic-associated cancer and noncancer effects need to be better characterized. Particular attention should be given to the extent of human variability and the reasons for it with respect to arsenic metabolism, tissue accumulation, and excretion (including total and relative amounts of urinary arsenic metabolites) under
various conditions of exposure. Gene products responsible for metabolism, diet, and other environmental factors that might influence the susceptibility to or expression of arsenic-associated toxicity also need to be characterized in human studies and in suitable animal models. Potential differences between young children and adults in arsenic-methylation efficiency need to be validated and considered in any risk assessment of arsenic. Finally, quality-control data are needed to ensure that reported variations are not due to the analytical methods or procedures used. Standard reference materials are needed to analyze arsenic species in urine.
Assessment of arsenic exposure via drinking water is often based on the measurements of arsenic concentrations in drinking water and assumptions regarding the amount of water consumed. Such data are estimates, the uncertainty of which will depend on the method used. The subcommittee evaluated various biomarkers (e.g., arsenic in urine, blood, hair, and nails) to measure the absorbed dose of inorganic arsenic and concluded that blood, hair, and nails are much less sensitive than urine as biomarkers of exposure. Specifically, the subcommittee concluded that the total concentration of inorganic arsenic and its metabolites in urine is a useful biomarker for both recent (previous day) and ongoing exposure. The concentration of urinary inorganic arsenic and its metabolites is less influenced by the consumption of seafood than is the total concentration of urinary arsenic. The concentration of arsenic in blood is a less-useful biomarker of continuous exposure because the half-life of arsenic in blood is short (approximately 1 hr), the concentration might be markedly affected by recent consumption of seafood, and it is difficult to speciate arsenic in blood. Measurements of arsenic in hair and nails have little use as biomarkers of absorbed dose, largely because of the difficulty in distinguishing between arsenic absorbed from ingestion and arsenic uptake in hair and nails from washing with contaminated water.
At present, the practical quantitation limit (PQL) for arsenic in water in most commercial and water utility laboratories is 4 µg/L. Measurement of total concentration of arsenic in drinking water is adequate for regulatory purposes.
More data are needed that tie biomarkers of absorbed arsenic dose (especially urinary concentrations of arsenic metabolites) to arsenic exposure
concentrations, tissue concentrations, and the clinical evidence of arsenic toxicity. Data are particularly lacking for people living in different parts of the United States. Possible relationships between arsenic concentrations in urine, blood, hair, and nails need to be evaluated. In particular, the degree of external binding of arsenic to hair and nails should be examined.
There is a need for further development of analytical techniques to determine the chemical species of arsenic in various mediawater, food, urine, and biological tissues. Quality-control data and certified standards for arsenic speciation are also needed.
In the context of its task, the subcommittee was asked to consider whether cancer or noncancer effects are likely to occur at the current MCL. No human studies of sufficient statistical power or scope have examined whether consumption of arsenic in drinking water at the current MCL results in an increased incidence of cancer or noncancer effects. Therefore, the subcommittee's characterization of risks at the current MCL is based on observed epidemiological findings, experimental data on the mode of action of arsenic, and available information on the variations in human susceptibility.
In the absence of a well-designed and well-conducted epidemiological study that includes individual exposure assessments, the subcommittee concluded that ecological studies from the arsenic endemic area of Taiwan provide the best available empirical human data for assessing the risks of arsenic-induced cancer. The cultural homogeneity of this region reduces concern about unmeasured confounders, although the potential for bias still exists due to considerable uncertainty about the exposure concentrations assigned to each village. Ecological studies in Chile and Argentina have observed risks of lung and bladder cancer of the same magnitude as those reported in the studies in Taiwan at comparable levels of exposure.
Information on the mode of action of arsenic and other available data that can help to determine the shape of the dose-response curve in the range of extrapolation are inconclusive and do not meet EPA's 1996 stated criteria for departure from the default assumption of linearity. Of the several modes of action that are considered most plausible, a sublinear dose-response curve in the low-dose range is predicted, although linearity cannot be ruled out. In vitro studies of the genotoxic effects of arsenic indicate that changes in cellular function related to plausible modes of carcinogenesis can occur at arsenic concentrations similar to the current MCL. However, the subcommittee believes that those data and the confidence with which they can be linked to
arsenic-induced neoplasia are insufficient to determine the shape of the dose-response curve in the low-dose range (point of departure). The subcommittee also finds that existing scientific knowledge regarding the pattern of arsenic metabolism and disposition across this dose range does not establish the mechanisms that mitigate neoplastic effects.
Human susceptibility to adverse effects resulting from chronic exposure to inorganic arsenic is likely to vary based on genetics, nutrition, sex, and other possible factors. Some factors, such as poor nutrition and arsenic intake from food might affect assessment of risk in Taiwan or extrapolation of results in the United States.
The subcommittee also concludes that the choice of model for statistical analysis can have a major impact on estimated cancer risks at low-dose exposures, especially when the model accounts for age as well as concentration. Applying different statistical models to the Taiwanese male bladder-cancer data revealed that a more stable and reliable fit is provided by Poisson regression models that characterized the log relative risk as a linear function of exposure. The estimation of risk at low doses using those models is substantially higher than that using the multistage Weibull model. As an alternative to model-based estimates of risk, the subcommittee finds that the point-of-departure methods discussed in the 1996 draft EPA guidelines for cancer risk assessment give much more consistent low-dose estimates across a wide range of dose-response models. For male bladder cancer, a straight-line extrapolation from the 1% point of departure yielded a risk at the MCL of 1 to 1.5 per 1,000. Because some studies have shown that excess lung cancer deaths attributed to arsenic are 2-5 fold greater than the excess bladder cancer deaths, a similar approach for all cancers could easily result in a combined cancer risk on the order of 1 in 100.1 It is also instructive to note that daily arsenic ingestion at the MCL provides a margin of exposure less than 10 from the point of departure for bladder cancer alone. The public health significance of daily ingestion of a given amount of arsenic in drinking water will be influenced by the background levels of arsenic consumed in food
On the basis of its review of epidemiological findings, experimental data on the mode of action of arsenic, and available information on the variations in human susceptibility, it is the subcommittee's consensus that the current
1Two of the 16 members of the subcommittee did not agree with the 1 in 100 estimate pending further analysis of the risk of lung cancer, as done for bladder cancer in Chapter 10.
EPA MCL for arsenic in drinking water of 50 µg/L does not achieve EPA's goal for public-health protection and, therefore, requires downward revision as promptly as possible.
To improve future risk characterizations, the following are recommended: Sensitivity analyses should be conducted to determine whether the results, including the way exposure concentrations are grouped together, are sensitive to the choice of model. The potential effect of measurement error and confounding on the dose-response curve and associated confidence limits should be further addressed.
To assist in the application of cancer data observed in different populations to cancer risks predicted for the United States, information on nutritional factors in study populations that pertains to susceptibility to arsenic-induced cancer should be investigated.
Modeling of epidemiological data should not be limited to the multistage Weibull model. Other models, including those which incorporate information from an appropriate control population, should be considered. The final risk value should be supported by a range of analyses over a broad range of feasible