ARSENIC IN DRINKING WATER: Is the Federal Maximum Contaminant Level Safe?

Introduction

Tap water: we drink it, we wash with it, we prepare food with it, we play in it, and we water our gardens with it. Public drinking water systems (PDWS) provide potable water to a significant portion of United States residences; thus, tap water presents itself as one of the greatest potential sources of exposure to environmental contaminants (WHO, 2001). The quality of our drinking water supplies is therefore of utmost importance in terms of public health.

Contaminants present in our drinking water supply may be from natural and/or anthropogenic sources. The federal government regulates the presence of contaminants in PDWS through the Safe Drinking Water Act (SDWA) of 1974 (Public Law 93-523), which gives the United States Environmental Protection Agency (EPA) the authority to set drinking water standards through the National Primary Drinking Water Regulations (NPDWR; Part 141 of Title 40, Code of Federal Regulations). These standards are only part of a more comprehensive approach to ensuring high quality drinking water. Other components include the protection of drinking water sources, regulation of drinking water treatment and distribution systems, and presentation of water quality data from each PDWS to the public. The NPDWR are legally enforceable standards that apply to all public water supply systems, and protect public health by limiting the levels of contaminants present in drinking water. In accordance with the SDWA, the NPDWR specifies Maximum Contaminant Levels (MCLs), which are "the maximum permissible level of a contaminant in water which is delivered to any user of a public water system." (40 CFR 141.2). MCLs may be created or revised through the 1996 Amendments to SDWA.

Of recent interest is the promulgation of the revised arsenic MCL in 2001, from 50 micrograms per liter (m g/l) to 10 m g/l. Through 1996 amendments to the SDWA, the EPA was charged with revising the 50 m g/l MCL for arsenic. In 1999, the National Academy of Sciences (NAS) determined that the former MCL, which was set in 1975 and based on a Public Health Service Standard established in 1942, was not adequately protective of public health (NAS, 1999). Subsequently, EPA evaluated several potential MCLs, ranging from 3 m g/l to 20 m g/l. After approximately two years of evaluation, which included input from the NAS and other public and private entities, the EPA promulgated a final MCL of 10 m g/l in January 2001 (USEPA, 2001).

What does this new standard mean for United States residents who depend on public water supplies? There are numerous cases of arsenic toxicity associated with drinking water exposures; particularly striking examples are epidemiological studies conducted in India, Bangladesh and South America, in which a large portion of village residents exhibit overt signs of arsenic-associated toxicity. Although there is strong evidence indicating that high concentrations of arsenic in drinking water may be associated with both carcinogenic and noncarcinogenic effects, chronic, low level exposures must also be taken into account. In light of recent toxicological and epidemiological data available, is the current MCL protective of public health in the United States? In this paper, we seek to answer this question relative to both arsenic toxicity and the potential for exposure in the United States population.

Arsenic is present in the earth’s crust as the 20^th most abundant element; it is typically associated with igneous and sedimentary rocks in an inorganic form, such as arsenopyrite. Although most of the arsenic is bound up in either rock or soil, some arsenic may be released to the environment through volcanic emissions or wind-blown dust (ATSDR, 2000).

Man-made releases of arsenic to the environment greatly exceed releases from natural sources (ATSDR, 2000). Anthropogenic sources of arsenic include industrial processes (in particular, mining and smelting), ash residues from power plants, pesticide and fertilizer applications, wood preservatives, industrial and municipal waste, and sewage sludge. Inorganic arsenical pesticides and herbicides were widely used in the US since the mid 19^th century; however, health concerns as well as occupational risks related to their use in agriculture ultimately caused inorganic arsenicals to be banned for agricultural use. Today inorganic arsenic is mainly used only in certain types of insect control (such as ant baits and flea dips) as well as in wood preservatives, such as chromated copper arsenate (CCA) (ATSDR, 2000). However, organic arsenicals as well as purer forms of arsenic are commonly used in agriculture as well as industry. Arsenic has been and presently is also used as a therapeutic agent to treat various ailments, ranging from parasitic disease to leukemia (ATSDR, 2000; NRC, 1999).

Arsenic is typically released to the environment in an inorganic form, and it tends to adsorb strongly to soils. Leaching into subsurface soils is generally not significant, except under reducing conditions (ATSDR, 2000). Physical soil characteristics, such as pH, organic carbon content, cation exchange capacity (CEC), and iron oxide content, tend to govern the leaching potential. Soluble forms of arsenic in soil may either run off into surface water bodies or leach into shallow groundwater. Arsenic may also be introduced into aqueous systems through natural weathering of soil and rock. The transport and partitioning of arsenic in water depends on not only the form of arsenic, but on interaction with other material (i.e., adsorption onto suspended particulates in the water column and onto sediments). Arsenate (AsO₃) is the predominant form in groundwater, although arsenite (AsO₂) may be present in significant proportions depending on local geology and water characteristics.

Concentrations of arsenic in both soil and groundwater vary considerably throughout the world. In the earth’s crust, arsenic is present at an average concentration of 2-5 milligrams per kilogram (mg/kg) (NAS, 1977; Tamaki and Frankenberger, 1992); however, in the conterminous United States, soil arsenic concentrations range from less than 0.1 mg/kg to 97 mg/kg (Shacklette and Boerngen, 1984). Regional patterns of arsenic concentrations may be attributable to local geology and climate (Welch et al., 2000). Many areas of New England have elevated arsenic concentrations (although not at the same magnitude as the southwestern states), due to inorganic arsenic complexes present in the overburden and bedrock, and the preferential accumulation of alluvial sediments in glacial valleys. Central Massachusetts (i.e., the Worcester arsenic belt), for example, has relatively high concentrations of arsenic in soil (up to 90 mg/kg) relative to other parts of the state (Doherty, 2002).

In groundwater, arsenic concentrations may range from parts per billion to parts per thousand. The southwestern portion of the United States, in particular, has found concentrations as high as 3,400 m g/l in certain areas, with extreme values up to 48,000 m g/l in mining areas. Across the United States, groundwater levels in public groundwater supplies are relatively low, with median concentrations at or less than 1 m g/l (Focazio et al., 2000). On a regional scale, the highest groundwater arsenic concentrations are most frequently observed in the western United States, whereas natural arsenic concentrations are lowest in the Appalachian Highlands and the Atlantic Plains (Welch et al., 2000). Concentrations of arsenic in surface water bodies (i.e., rivers, streams, ponds and lakes) are generally lower than those found in groundwater, with most levels approximately less than 10 m g/l (EPA, 1992; Smith et al., 1987), and less than 1 m g/l in US surface water supplies (EPA, 2000). However, arsenic concentrations in both surface water and groundwater tend to increase with proximity to mining areas and other industrial sites (ATSDR, 2000).

Arsenic has long been classified as a poison; it is an odorless and tasteless substance that, when ingested at high concentrations, may cause death. Lower levels of exposure to arsenic can cause a number of detrimental results such as cancer and disease. The toxicity of arsenic depends on its chemical nature. Inorganic versus organic state, valence number, solubility, physical state and purity, and rates of absorption and elimination all contribute to the relative toxicity of arsenic (ATSDR, 2000). Although arsenic exists at different oxidation states (-3, 0, +3, +5), toxicologists are most interested in the inorganic forms, trivalent arsenite (As(III)) and pentavalent arsenate (As(V)) (Hughes, 2002). In humans, 60-90% of ingested soluble arsenic is absorbed by the gastrointestinal tract (ATSDR, 2000). Acute arsenic poisoning results in vomiting, esophageal and abdominal pain, bloody urine, diarrhea, anuria, shock, convulsions, coma, and death (Hughes, 2002). The lethal range of inorganic arsenic in humans is a dose of 1-3 mg As/kg (Ellenhorn, 1997). Symptoms associated with chronic exposure include detrimental effects on the skin, cardiovascular system, nervous system, the liver, bone marrow, endocrine system and kidneys (Hughes, 2002). The results of long term exposure due to elevated levels of inorganic arsenic in drinking water supplies include skin, lung, liver and bladder cancers (Kitchin, 2001) as well as malignant neoplasms, diabetes (Tseng et al., 2002) and vascular diseases (Chen et al., 1995). There has also been compelling evidence linking arsenic in drinking water with pregnancy irregularities including stillborns and spontaneous abortions (Yang et al., 2002). It has been shown in animal studies that arsenic crosses the placenta, resulting in detrimental effects of offspring including a variety of tumors (Waalkes, 2003).

Once ingested, arsenic accumulates in the liver, spleen, kidneys, lungs and gastrointestinal tract. Mammals metabolize inorganic arsenic through methylation to methlyarsonic acid (MMA) and dimethylarsinic acid (DMA). The steps of metobolism include reducing the pentavalent arsenic to the trivalent state and the addition of a methyl group from donor s-adenosylmethionine. MMA and DMA are less reactive with the tissues exposed and are more readily excreted with urine (Vahter, 2002). The exposed tissues rapidly clear the toxicant, but two to four weeks after exposure, amounts may be found in keratin-containing tissues such as hair, nails and skin (ATSDR, 2000). The biological half-life of arsenic is 10 h (Casarett and Doull(C&D), 819). Trivalent arsenic is known to be more toxic than pentavalent arsenic at acute levels. It was also found that derivatives of the intermediate substances in the metabolizing pathway are more reactive with tissues than arsenic itself. Trivalent forms of MMA and DMA were shown to be more toxic to hepatocytes, epidermal keratinocytes and bronchial epithelial cells than trivalent arsenic. These intermediates were also found to be more genotoxic in their ability in damaging DNA (Vahter 2002). In experimental animals, DMA was recently found to promote the carcinogenic effects of chemical genotoxins on the kidney, liver, thyroid gland and respiratory tract (Yamamoto et al 1995). The LD₅₀ of various forms of arsenic are shown in Table 1. This table demonstrates the degree of lethality of trivalent MMA and arsenite in comparison to other forms of arsenic.

Modified from Hughes 2002

Other examples of arsenic’s replacement of phosphate in biological processes include the sodium pump and anion exchange (Hughes, 2002). Interruption of such cellular processes has the potential to wreak havoc on the regulation of cellular functions. The cell relies on the tight regulation of ion concentrations for such processes as signal transduction, intracellular transport, and enzyme activation and deactivation. Altering the ion balance of the cell also disrupts homeostasis and could lead to cell lysis. Trivalent forms of arsenic directly react with molecules containing thiol, such as growth stimulating hormone. This binding may inhibit the protein’s biological function and lead to toxicity. On the other hand, if the arsenic binds to a non-essential part of the protein, the result may be detoxification because the arsenic would be sequestered (Aposhian, 1989). This mechanism has the potential to block the action of a number of various enzymes posing a serious threat to cellular function.

Arsenic has a number of modes of action in its capacity as a carcinogen. Administration of arsenic in hamster embryo cells resulted in chromosomal aberrations, DNA-protein crosslinks, and sister chromatid exchanges (Kochhar, 1996). In vitro studies on other cell lines, including human lymphocytes and fibroblasts gave similar results. These aberrations were characterized by chromosomal gaps, breaks and endoreduplications (Hughes, 2002). Kessel et al. (2002) found that arsenic induced oxyradicals are also potent mutagens in mammalian cells. Arsenicals have also been found to induce tetraploidy and arrest the cell cycle in mitosis. Mitotic arrest was induced by exposure to MMA, DMA, and trimethylarsine oxide (TMAO), all metabolites of arsenic. Tetraploidy was induced by DMA and TMAO (Eguchi et al., 1997). All of the above chromosomal aberrations can result in cancerous cell induction. The presence of tumors in kidneys, liver, lungs, gastrointestinal tract and skin could all result from such insults.

Arsenic exposure has also been linked to the onset of diabetes mellitus. Numerous studies in Taiwan, where well water is contaminated with arsenic in particular villages, showed a higher incidence of diabetes when compared to the general population. Studies in these areas determined the dose-response relationship as a function of cumulative arsenic exposure (CAE). The resulting odds ratios were 6.61 and 10.05 for CAEs of 0.1-15 and ? 15.1 mg/l years. The biological plausibility for the incidences of diabetes and chronic exposure to arsenic was demonstrated by a study showing higher levels of glycosylated hemoglobin in individuals chronically exposed to arsenic. This suggests that arsenic somehow interferes with the metabolism of glucose. This dose-response relationship between arsenic exposure and the onset of diabetes was also supported by studies in Bangladesh and Sweden (Tseng et al., 2002). There is a known relationship between long-term arsenic exposure and vascular diseases including diabetes, peripheral artery disease, hypertension, ischemic heart disease, and cerebrovascular disease. Studies looking at accelerated arteriosclerosis showed that even in the absence of the traditional coronary risk factors, hardening of the arteries still occurred (Wang et al., 2002).

Arsenic has been shown to readily cross the placenta, thus adding another route of exposure for its toxicity. Animal studies investigating transplacental exposure to arsenic have shown substantial levels of arsenic in all fetal tissues at any period of gestation (NRC, 1999). Waalkes et al. (2003) performed studies with mice that indicated inorganic arsenic as a transplacental carcinogen resulting in tumors in the liver, adrenal, lung and ovary in offspring of treated adults. The levels of arsenic that the animals were treated with for these experiments were 0, 45.2 and 85 ppm (mg/L). This is much higher than the MCL, but this study looked at a brief exposure and long-term ingestion of low levels of arsenic could potentially put the fetus at risk. These studies showed that brief maternal exposure to arsenic- contaminated drinking water results in malignant, benign, and preneoplastic lesions in adult offspring long after arsenic exposure occurs. This suggests that the insult of arsenic toxicity is irreversible (Waalkes et al., 2003). Yang et al. (2003) studied the effect of arsenic contaminated drinking water on the risk of adverse pregnancy outcomes, specifically spontaneous abortions, low birth weight and preterm delivery. This study reported a much higher occurrence of these outcomes in exposed areas where the arsenic level is greater than 0.1 mg/L, as compared with the control area where levels are less than 0.02 mg/L.

Arsenic exposure affects the ubiquitin (Ub)-proteosome pathway, which the cell uses to degrade proteins. This pathway is important in many cell-signaling pathways and in regulatory processes. Misfolded or damaged proteins are tagged for degradation by the modification of cysteine residues with the addition of poly(Ub) chains. Specific enzymes respond to the Ub chain and induce the degradation of the protein. Many highly regulated proteins are also degraded through this pathway. These include the cell cycle cyclins, p53, I? B, apoptosis, stress response and transcription, to name a few (NRC, 2001). Interruption of the function of these proteins would seriously disrupt cellular homeostasis. Kirkpatrick et al. (2003) used in vitro studies to show that low levels of arsenic also perturb the ubiquitin pathway. These studies showed that at levels as low as 0.1 ? M cause an increase in high molecular weight Ub-conjugated proteins and a decrease in proteosome activity. This study also noted a decrease in the ubiquitination in certain histone proteins. Both occurrences could potentially have negative effects on cell signaling and regulatory processes.

The potential damage that arsenic can have on the human body is substantial. The level of arsenic exposure directly affects the extent of this damage. High doses of arsenic can cause damage to skin, liver, lungs, kidneys and gastrointestinal tract. As the evidence presented shows, arsenic also can disrupt cellular function. Whether lower levels of arsenic can subtly cause damage is unknown. It can be inferred through knowledge of the action of arsenic on cellular function and on DNA that the mere presence of the toxin may be detrimental in cells with functions essential to human health.

In the United States, public water supplies are regulated such that water must be treated, if necessary, to meet the MCL. In some areas of the world where treatment of drinking water is not routine, elevated concentrations of arsenic (as well as other contaminants) may present a significant public health threat. Of particular relevance are epidemiological studies that have related high arsenic concentrations in drinking water to commensurate increases in skin cancer, internal cancers, and numerous noncancer effects in rural populations of India and Bangladesh.

Numerous case studies of arsenic contamination in groundwater have been documented in West Bengal, India and Bangladesh. Sources of groundwater contamination include the following (Harun-ur-Rashid and Mirdha, 1998):

• Natural weathering of geological deposits located at shallow depths (40-150 feet),
• Industrial discharges, pesticides, and other anthropogenic sources. In Bangladesh, these sources may include wooden electric poles that are treated with chromated copper arsenate (CCA) , amoniacal copper arsenate (ACA) and amoniacal copper zinc arsenic (ACZA); and
• The use of arsenic-based fertilizers and pesticides contribute to arsenic contamination in soil and shallow groundwater.

Figure 3: Correlation of Arsenic with Iron.

The authors present a few arsenic removal methods. See Mitigations for these methods of removal.

Each method has its own advantages and disadvantages. Disposal of arsenic sludge presents the biggest problem. Also, costs of chemical methods are a big concern since the regions involved are very poor. Simple filtration methods are the best way to go since they are most simple and cost effective.

Epidemiological case studies in India and Bangladesh

In the article "Arsenic Mitigation in Bangladesh" by Nadia S. Halim, appearing in The Scientist 14[5]: 14, Mar. 6, 2000 is an example of arsenic contamination in Bangladesh. The article discusses the fact that researchers estimate that "as many as half of the four million tube wells in Bangladesh are pumping out groundwater contaminated with naturally occurring arsenic". These tube wells were used as alternatives to well water starting in the 1970’s, which may have contained bacterial contaminants. The levels of arsenic in many of these wells exceeded 500 ppb, 50 times the MCL. A British Geological Survey, completed in 1998, estimates that 20-25 million people are at risk from groundwater arsenic exposure. Arsenic attacks proteins that have sulfur-sulfur bonds, such as keratins found in hair and skin, thereby resulting in skin lesions and hair loss. It also interferes with ATP generation during cellular respiration, causing weakening in exposed patients. Other symptoms may include coughing, fever, numbing of legs and arms, liver and kidney damage; long term effects can include cancer. The article’s author speaks to Tony Fletcher of the London School of Hygiene and Tropical Medicine. He estimates that 1.5 million people in Bangladesh with show hyperpigmentation, 715,000 suffer keratosis, 900,000 will develop skin cancer, and 20,000-250,000 will develop internal cancers of various types as a result of arsenic-contaminated drinking water.

The article discusses alternative water sources. One example includes the harvesting of rainwater, since the rainy season (monsoon) lasts several months. Storing the water in large reservoirs can do this, but this can be costly and needs a lot of space, something overpopulated countries do not have. There is also a risk of bacterial contamination. Another possibility is arsenic removal from the water. This can include precipitating the water with iron fillings or alum, an aluminum compound. The precipitate is then filtered off; this technique readily works in a laboratory setting but may not work in an actual rural setting. Nikalaos Nikolaidis, an associate professor of civil and environmental engineering at University of Connecticut, has developed a system that would remove arsenic using iron fillings under anoxic conditions. It has been successfully tested on a landfill in Maine. This system would filter all the water pumped from the ground, but 75% of it would be used for irrigation, so it may not be the most efficient method and disposal of solid arsenic waste is also problematic. The article basically closes by stating the need for more money for arsenic mitigation programs.

The article "Groundwater Arsenic Contamination in Bangladesh and West Bengal, India: from Environmental Health Perspectives (Volume 108, number 5, May 2000) is going to be discussed next. In this case study, the authors investigate nine districts in West Bengal, India and 42 districts in Bangladesh and found that they had arsenic levels above the WHO maximal permissible level of 50 m g/l. The investigators started their investigation in 1989 in West Bengal. At that time they identified only 22 affected villages in 12 police stations/blocks of five districts. As the years progressed they discovered more and more affected villages. By 10 years they discovered 985 affected villages in 69 police stations/blocks of 9 districts affected by arsenic. Nine districts of West Bengal were found to have arsenic levels >50 m g/l in the groundwater. Seven of these nine districts had people suffering from arsenical skin lesions. They, however, expected all nine districts to have arsenic patients.

In their survey of 64 districts of Bangladesh, 52 districts contained arsenic levels of over the WHO guidelines of 10 m g/l and 42 districts had arsenic levels over the maximum level of 50 m g/l. Of these 42 districts, only 25 districts contained patients with skin lesions. In two districts they found critical arsenic patients that were suffering from such conditions as keratosis, hyperkeratosis, gangrene, and melanosis. Of the nine districts in West Bengal that contained arsenic at levels >50 m g/l, 45% of tube wells were found to contain water that was safe to drink. Higher arsenic contamination was found in Bangladesh than in West Bengal. Groundwater samples containing >1,000 m g/l are much more abundant in Bangladesh. Two hundred and thirty-three samples out of 10,991 samples from 42 districts had such a high concentration of arsenic. In West Bengal only 38 of 58,166 samples contained arsenic levels >1,000 m g/l. The existing arsenic species in the water were arsenate and arsenite, indicating that monomethylarsonic acid and dimethylarsinic acids were not present in the groundwater.

The investigators analyzed some 3,332 hair samples, 3,321 nail samples, 1,043 urine samples, and 165 skin-scale samples from West Bengal. Twenty percent of these samples, excluding skin-scale, were from patients with arsenical skin lesions. All skin scales had elevated levels of arsenic. From Bangladesh they analyzed 7,381 nail samples, 7,135 hair samples, 9,795 urine samples, and 165 skin-scale samples. Sixty percent of these samples were from patients suffering from skin lesions. Elevated arsenic levels were found in 81% of hair samples, 94% of nail samples, and 95% of urine samples collected from West Bengal. For Bangladesh, 57% of hair, 83% of nail, and 89% of urine samples had elevated arsenic levels. While many villagers weren’t suffering from skin lesions, many still had elevated levels of arsenic in their hair and nails. Many more villagers were found to be subclinically affected; arsenical neuropathy was present in 37.2% of 413 arsenicosis patients treated in West Bengal.

The authors of the paper talk about symptoms of arsenical toxicity that are discussed in the other parts of our paper in more detail. The symptoms generally show up after 6 months to 2 years, varying with the amount of water intake and arsenic concentration in the water. Some symptoms may include darkening of the skin (diffuse melanosis) in the whole body or the parts of the hands. This is an early-onset symptom, and people suffering from arsenical toxicity will not necessary show this symptom. Spotted melanosis is also another early symptom of arsenic poisoning. Usually melanosis is present in the back, chest or limbs. Keratosis is a late symptom of arsenical toxicity.

The investigators also gave some suggestions to combat the arsenic problem. They include monitoring arsenic levels in safe tube wells every 3-6 months. Additionally, they recommend labeling safe tube wells with the color green and the unsafe ones with red, so villagers may use the green tube wells for drinking and cooking and red tube wells for bathing, washing, toilet, etc. Epidemiological research for arsenic-affected areas is needed to document long-term exposure. Educating people on arsenic contamination in the regions affected is also an important strategy, as many of these people are uneducated.

As with the previous paper, the authors state that arsenic contamination is far worse in Bangladesh than in West Bengal (which is next to Bangladesh). Urine samples collected from populations in Bangladesh contained up to 235 times normal arsenic levels, nail samples up to 35 times greater, and hair samples up to 80 times greater than samples from West Bengal. Skin was shown to be the best test for visual detection of arsenic poisoning.

The paper discusses chronic arsenic exposure as having three stages in which changes in skin pigments, hyperkeratosis, development of skin ulcerations, and risks of cancer to bladder, skin, liver, and kidney will occur. These three steps are called the initial stage, second stage, and final stage. In the initial stage, dermatitis, hepatitis, conjunctivitis, bronchitis, and gastroenteritis occur. In the second stage, peripheral neuropathy, melanosis, depigmentation, and hyperkeratosis occur. Gangrene of the limbs, malignant neoplasm and cancer may occur in the final stage.

Arsenic contamination of drinking water supplies is widespread in certain areas of India and Bangladesh. To mitigate this problem, the following strategies are suggested:

• Eliminate the source of contaminated drinking water.

• Educate people about the risks of drinking and cooking with contaminated water.
• Provide alternative water sources like treated water.
• Treatment of contaminated water with copper sulfate or sedimentation to remove arsenic.
• Installation and use of deeper tube-wells or pond water, with slow sand filtration.
• Simple decantation and filtration method. Iron at high concentrations can also be toxic, but they both coexist in groundwater. Its concentration is higher than that of arsenic and can therefore be used to remove arsenic. About 75% of arsenic and 85 % of iron can be removed using this method.
• Filtration and application of activated alum.
• Ion exchange and membrane processes.

Drinking water exposures are one of the greatest sources of exposure to arsenic from a public health perspective (WHO, 2001; ATSDR, 2000). From the India and Bangladesh epidemiological studies presented in the previous section, it is obvious that high-level exposures via drinking water can cause significant public health issues. However, in the United States, such elevated exposures to arsenic in public drinking water supplies are not common; over 99% of PDWS systems have arsenic concentrations below the 1974 MCL of 50 m g/l (EPA, 1984a). Exposures to concentrations at this level are unlikely to cause acute health effects; however, numerous studies have found that chronic ingestion of elevated arsenic levels in drinking water may cause a multitude of both cancer and noncancer effects. Increased risks of lung and bladder cancer, as well as skin lesions, have been associated with drinking water exposures of less than 50 m g/l (WHO, 2001). The National Academy of Sciences concluded in a 1999 report that the "current standard [of 50 m g/l] does not achieve EPA’s goal of protecting public health" (NRC, 1999).

The MCL must consider quantifiable as well as nonquantifiable benefits from a reduction in health risks; costs of compliance; any increases in health risks resulting from compliance; incremental costs and benefits of alternative MCLs; and susceptibility by sensitive subpopulations (EPA, 2001-fr). The EPA, as a result of the 1996 Amendments to the SDWA, and in response to the NAS findings, prepared a comprehensive evaluation of the proposed arsenic MCL, which included a human health risk characterization, cost-benefit analysis, and technical feasibility analysis. Along with the MCL, the EPA is also required to provide a maximum contaminant level goal (MCLG), which is the "level of a contaminant in drinking water at which no known or anticipated adverse effect on the health of persons would occur, and which allows an adequate margin of safety" (40CFR141.2). Thus the MCLG, although nonenforceable under the SDWA, nevertheless provides a public health-based guideline independent of technical feasibility. The SDWA specifies that the MCL must be set as close to the MCLG as is technically feasible, unless the cost of treatment required to meet this level would either increase the risk from other co-occurring contaminants or if the benefits from this level would not justify the costs (Section 1412(b)(4-5)).

To conduct the arsenic risk assessment to support the MCLG, the EPA considered chronic cancer and noncancer effects resulting from arsenic exposures in their evaluation of the MCLG. Acute effects were not evaluated, as exposures equivalent to or less than the 1974 MCL of 50 m g/l are not typically associated with these concentrations (EPA, 2001). Noncancer endpoints included vascular effects, based on epidemiological evidence linking drinking water exposures to vascular diseases such as Blackfoot Disease. Although other noncancer effects such as neuropathy, diabetes, and reproductive/developmental effects were qualitatively considered, the EPA did not include these as endpoints due to either lack of sufficient data, weak correlation, inconsistency among studies, or confounding factors. Cancers that have been associated with ingestion of inorganic arsenic in drinking water include those of the skin, bladder, liver, lung, and kidney. Accordingly, EPA evaluated epidemiological studies for these types of cancers from the United States, Asia, South America, and Mexico. Ultimately, EPA chose lung and bladder cancer as their final endpoint, as the studies upon which effects and exposure were based had the strongest and most consistent conclusions among those evaluated.

The EPA used the Chen et al. (1988, 1992) and Wu et al. (1989) Taiwanese studies to derive a dose-response relationship for arsenic. As there is no consensus on the definitive mode of action for arsenic as a carcinogen, the EPA assumed linearity when extrapolating from higher doses (presented in the studies) to lower doses, in accordance with draft EPA cancer risk assessment guidelines (EPA, 1996). Based on their evaluation, the EPA presented a MCLG of zero, relying on the linear model of carcinogenicity. In other words, the linear model assumes that any level of exposure has an associated risk. Therefore, the most protective concentration would be that which causes no excess cancer risk, or zero.

As previously mentioned, the SWDA directs the EPA to set the MCL as close to the MCLG as is technically feasible. Laboratory measurements are only as good as the available technology from which they are generated; consequently, the EPA derived a practical quantitation limit (PQL) of 3 m g/l, which the majority of laboratories would be expected to meet on a routine basis (EPA, 2001) and which "can be reliably measured within specified limits of precision and accuracy during routine laboratory operating conditions" (EPA, 1985: 50 FR 46906). This PQL was determined to be technologically feasible for most water treatment systems in the United States, given the current technology.

The EPA evaluated four different arsenic concentrations to derive the final MCL: 3 m g/l (which is the lowest technically feasible concentration), 5 m g/l, 10 m g/l, and 20 m g/l. Each of these concentrations was evaluated with respect to arsenic occurrence distributions for seven regions of the United States, as well as potential health benefits vs. monetary costs associated with each proposed MCL. Health benefits included a quantitative analysis of the number of reduced mortality/morbidity/cancer cases resulting from the different levels of exposure and a qualitative analysis of benefits associated with the decreased MCL. Some of the qualitative health benefits evaluated included a reduction in skin, kidney, liver, nasal and prostate cancers; cardiovascular and pulmonary effects; and neurological, immunological, and endocrine effects. Both quantitative and qualitative health benefits were weighed against estimated costs of implementing arsenic treatment systems to achieve the proposed MCLs. Costs included the dollar amount (the total national, state, and household costs) to bring treatment systems into compliance for each proposed MCL. As a result of this analysis, the EPA determined a final MCL of 10 m g/l, which "best maximizes health risk reduction benefits at a cost that is justified by the benefits" (EPA, 2001: fr).

Most public water supplies in the United States will likely have arsenic concentrations lower than the revised MCL; however, approximately 8% of PDWS systems are estimated to have arsenic concentrations exceeding 10 m g/l (Focazio et al., 2000), with the majority of those exceedances occurring in small PDWS systems (i.e., serving less than 10,000 residents). Water treatments may effectively reduce arsenic concentrations to drinking water standards. However, there are numerous areas of the country in which arsenic is present in water supplies at concentrations less than 10 m g/l, as shown in Figure 4 below: 

(modified from Focazio et al., 2000: http://webserver.cr.usgs.gov/trace/pubs/geo_v46n11/fig1.html)

The revised MCL is obviously more protective than the 1974 value of 50 m g/l; nevertheless, the MCL is not set at the health-based MCLG of zero, and there are numerous studies indicating that effects may occur at low-dose exposure levels. Although few would argue that arsenic causes cancer, the EPA acknowledges various uncertainties in the studies they used to form the basis of the drinking water standard, which could potentially over- or underestimate risks for the American population. Some of the greatest uncertainties include the following:

• Extrapolation of dose-response data from the Taiwan study (Tseng et al., 1968) to American populations. Taiwanese typically have a much higher exposure to arsenic from food (especially rice and sweet potatoes) and through cooking with contaminated water than do most Americans.
• Nutritional status of exposed populations. Both the rural populations evaluated in the Taiwan and South American studies were generally undernourished, with deficits in essential nutrients such as selenium. Malnourishment may present a predisposition to cancer and other ailments.
• Background incidence of cancer. Cancer risks are expressed as the excess risk (over a lifetime of exposure) of developing cancer relative to background cancer rates. The United States cancer rate for lung cancer is approximately 2-3 times lower than that of Taiwan. If risks of bladder and lung cancer are normalized to US cancer rates, rather than Taiwan rates, risk estimates for the US population actually increase (NRC, 2001).
• Assumption of linearity in the dose-response relationship. The linear model of carcinogenicity was chosen by the EPA due to the lack of adequate information suggesting otherwise. However, there are suggestions that the relationship may be better described as sublinear (with a possible threshold dose), as the mode of arsenic toxicity may be indirect (NRC, 1999, 2001; Abernathy et al., 1996; Slayton et al., 1996).

Overall, the uncertainties provided above suggest that the current MCL may be conservative for the United States population. Studies conducted in United States populations (such as the Utah study by Lewis et al., 1999) indicated a much lower rate of cancer and other arsenic-related effects than those noted in the Taiwan and South American studies (ATSDR, 2000). As the dose-response relationship has yet to be elucidated, there may even be beneficial effects associated with low-level exposure. Although arsenic is regulated by the EPA and other agencies as a Known Human Carcinogen (Class A) it is typically present in the diet and even exhibits some evidence of essentiality (Teaf, 1999). Arsenic has been used as a known therapeutic agent since Thomas Fowler in the late 18^th century introduced Liquor Arsenicalis (or Fowler’s solution) to cure agues, fevers, and headaches (NRC, 1999). Despite the incidence of gastrointestinal and skin side effects, Fowler’s solution and other inorganic arsenic preparations were used extensively throughout the 19^th and early to mid 20^th centuries to treat a broad spectrum of diseases and ailments, ranging from fevers, asthma, and pain to eczema and other skin disorders. More recently, arsenic trioxide has been used in treating acute promyelocytic leukemia (CenterWatch, 2002 ). Indeed, there is growing evidence of the hormetic effects of numerous compounds, including arsenic (Hively, 2002).

However, given the health status of the rural Taiwanese population relative to that of many residents of the United States, the Taiwan study may possibly be indicative of sensitive subpopulations, which were not specifically evaluated in the EPA MCL. Furthermore, there is increasing evidence that low-level, chronic exposure to arsenic may result in increased risk of noncancer effects, such as diabetes and hypertension (NRC, 2001). The uncertainties in the existing database of arsenic studies highlight the need for additional research in this area; currently, "the magnitude of possible risk that exists at low levels is non-quantifiable" (NRC, 2001).

The decrease in the arsenic MCL is a step in the right direction, in terms of public health policy in the United States. Numerous cancer and noncancer adverse effects have been associated with groundwater exposures of arsenic at concentrations at or greater than the previous MCL of 50 m g/l, as indicated by various Asian, South American, and Indian studies. However, uncertainty remains regarding whether the current MCL of 10 m g/l is adequately protective of all subpopulations. The current MCL is not solely health-based, and there is limited, although growing, evidence that long-term exposure to low levels of arsenic may continue to pose health problems, which points to the fact that more information is necessary to refine our public health goals. In addition to health issues, the best available technology as well as monetary concerns will ultimately limit federal control of arsenic contamination.

Thus, for most of the United States population, the MCL appears to be adequately protective, since arsenic concentrations in US drinking water supplies are relatively low, and some conservatism is built into the data upon which the MCL is based. Hopefully the present uncertainty surrounding the MCL will be captured through future research on low-dose toxicity.

Below we have listed some links pertinent to arsenic toxicity, drinking water standards, groundwater exposure, and other items of interest.

United States Geological Survey (USGS): Arsenic in drinking water. http://webserver.cr.usgs.gov/trace/arsenic/

World Health Organization (WHO): Arsenic Fact Sheet. http://www.who.int/inf-fs/en/fact210.html

US EPA. Arsenic in Drinking Water: http://www.epa.gov/safewater/arsenic.html

Low-level exposures: http://www.belleonline.com

MIKE--AUTHORS? 2000. Groundwater Arsenic Contamination in Bangladesh and West Bengal, India. Environmental Health Perspectives 108: pages???

Agency for Toxic Substances and Disease Registry (ATSDR). 2000. Toxicological Profile for Arsenic. September, 2000. Prepared for the US Department of Health and Human Services by Syracuse Research Corporation.

Abernathy, C.O., Chappell, W.R., Meek, M.E. 1996. Is ingested inorganic arsenic a "threshold" carcinogen? Fund. Appl. Toxicol. 29(2): 168-175.

Aposhian, H. V., Biochemical toxicity of arsenic. 1989. In: Hodgson, E., Bend, J.R., Philpot, R.M.(Eds), Reviews in Biochemical Toxicity. Elsevier, New York, pp.265-299.

Slayton, T.M., Beck, B.D., Reynolds, K.A. 1996. Issues in arsenic cancer risk assessment. Env. Health Perspect. 104(10): 1012-1018.

CenterWatch Clinical Trials Listing Service. Newly Approved Drug Therapies Listing: Trisenox (arsenic trioxide). January 15, 2002. http://www.centerwatch.com/patients/drugs/dru643.html.

Chen, C.J., Wu, M., Lee, S.S., Wang, J.D., Cheng, S.H., and Wu, H.Y. 1988. Atherogenicity and carcinogenicity of high-arsenic artesion well water. Multiple risk factors and related malignant neoplasms of blackfoot disease. Arteriosclerosis. 8:452-460.

Halim, N.S. 2000. Arsenic Mitigation in Bangladesh. The Scientist 14[5]: 14, Mar. 6, 2000.

Hughes, Michael F., Arsenic toxicity and potential mechanisms of action. 2002. Toxicol Lett, 133, 1-16.

Kessel, Maris, Liu, Su Xian, Xu, An, Santella, Regina, and Hei, Tom K., Arsenic induces oxidative DNA damage in mammalian cells. 2002. Molec Cell Biochem, 234-235, 301

Hively, Will. 2002. Is Radiation Good for You? Or dioxin? Or arsenic? Discover, Vol. 23 No. 12

Wang, Chih-Hao, Jeng, Jiann-Shing, Yip, Ping-Keung, Chen, Chi-Ling, Hsu, Lin-I, Hsueh, Yu-Mei, Chiou, Hung-Yi, Wu, Meei-Mann, and Chen, Chien-Jen. 2002. Biological gradient between long-term arsenic exposure and carotid atherosclerosis. Circulation, 105. 1804-1809.

World Health Organization (WHO). 2001. Fact Sheets: Arsenic in Drinking Water. Fact Sheet No. 210. Revised May 2001. http://www.who.int/inf-fs/en/fact210.html.

Wu, M.M., Kuo, T.L., Hwang, Y.H., and Chen, C.J. 1989. Dose-Response relation between arsenic concentration in well water and mortality from cancers and vascular diseases. American Journal of Epidemiology 130: 1123-1132.

Yamamoto, S., Konishi, Y., Matsuda, T., Murai, T., Shibata, M.A., Matsui-Yuasa, I., Otani, S., Kuroda, K., Endo, G., and Fukushima, S. Cancer induction by an organic arsenic compound, dimethylarsinic acid (cacodylic acid), in F344/DuCrj rats after pretreatment with five carcinogens. 1995. Cancer Res., 55, 1271-1276.

Yang, Chun-Yuh, Chang, Chih-Ching, Tsai, Shang-Shyue, Chuang, Hung-Yi, Ho, Chi-Kung, and Wu, Trong-Neng. Aresenic in drinking water and adverse pregnancy in an arseniasis-endemic area in Northeastern Tawain. 2003. Environ Res, 91, 29-34.

Welch, A.H., Westjohn, D.B., Helsel, D.R., and Wanty, R.B., 2000, Arsenic in ground water of the United States-- occurrence and geochemistry: Ground Water, Vol. 38, No. 4, p. 589.

Teaf, C.M. 1999. Low Level Exposures: Implications for Regulatory Agencies. Biological Effects of Low-Level Exposures (BELLE), Vol. 8, No. 1. http://www.belleonline.com/n6v81.html