Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
4 Variability and Uncertainty VARIABILITY VERSUS UNCERTAINTY Quantitative health risk assessments rely on mathematical models to combine data and assumptions with the goal of providing useful information to the decision-maker about health risks. In the context of developing a dose-re- sponse relationship, risk assessors evaluate the weight of the scientific evi- dence of an effect. Given sufficient evidence, they combine data from multi- ple studies in the same species and from epidemiological and animal studies to estimate the health risks. These risks are estimated in the dose range for which observations have been made and are then extrapolated from these dose ranges to Tower doses that might be of concern for regulation. Science and Judgment in RiskAssessment MARC 1994) emphasized the important distinc- tions between variability and uncertainty in risks, highlighting their very different ramifications for risk management in the following statement: Uncertainty forces decision-makers to judge how probable it is that risks will be overestimated or underestimated for every member ofthe exposed population, whereas variability forces them to cope with the certainty that different individuals will be subjected to risks both above and below any reference point one chooses (NRC 1994, p. 237~. 133
134 ARSENICINDR1NKING WATER. 2001 UPDATE Uncertainty refers to a lack of knowledge in the underlying science. Variabil- ity in a risk assessment means that some individuals in a population have more or less risk than others because of differences in exposure, dose-response relationship, or both. Arsenic in Drinking Water MARC ~999) discussed the variability and uncertainty associated with an arsenic risk assessment. This chapter focuses on data that have become available since that report. These data provide further insight into the variability and uncertainty in the risk assessment and allow a better characterization of how they affect the overall characterization of risk. This chapter is divided into two sections, one that primarily discusses variability and one that primarily discusses uncertainty. It is important, however, to recognize that uncertainty might exist about vari- able quantities and that variability and uncertainty are best defined and under- stood within a decision-making context ARC ~ 994; Thompson and Graham 1996). VARIABILITY AND UNCERTAINTIES DISCUSSED IN TTIE 1999 REPORT In its discussion of variability in an arsenic risk assessment, the 1999 Subcommittee on Arsenic in Drinking Water concluded that characteristics of different individuals could contribute to variability in risk estimates. Human susceptibility to inorganic arsenic could vary because of differences in genet- ics, metabolism, diet, health status, and sex. In the 1999 NRC report, the subcommittee presented risk estimates that were calculated using different statistical models, which provide some indica- tion ofthe impact ofthe model on the estimates, and concluded that the statis- tical model used to analyze the data can have a major impact on the estimated cancer risks at Tow-dose exposures. The ~ 999 NRC report quantified some of the impacts of different dose-response mode! choices, but did not quantify the impacts of variability in the population of the risk estimates. In most epidemiological studies of arsenic, biomarkers of individual ar- senic exposure have not been used; therefore, assumptions must be made about the amount and source of wafer consumed to estimate exposures. Those assumptions add to the uncertainty in the risk assessment. There is the poten- tial for bias in the results of the Taiwanese studies (Chen et al. 1985, 1988, 1992; Wu et al. 1989) because ofthe uncertainty in exposure estimates. How- ever, similar associations between arsenic exposure and cancers (lung and bladder) were seen in studies in Chile (Smith et al. 1998) and Argentina (Hopenhayn-Rich et al. 1998~; therefore, the uncertainty in the Taiwanese
VARIABILITYAND UNCERTAINTY 135 data, and the subsequent risk assessment using those data, is decreased ARC 1999~. Additional uncertainty comes from the lack of knowledge of the pre- cise intake of arsenic from food in both the U.S. and Taiwanese populations. Differences in the intake between the populations could affect the assessment of risk from exposure to arsenic in Taiwan and the extrapolation of that risk to the U.S. population. The lack of knowledge regarding population differ- ences adds to the uncertainty in the risk estimates. SOURCES OF VARIABILITY Exposure Different individuals or subpopulations are exposed to different amounts of arsenic, leading to variability in the risk estimates that could be calculated for a given population. Many factors can affect the amount of arsenic to which different individuals are exposed through drinking water, including differences in the amount of arsenic present in drinking water, the amount of water that individuals consume, and physical factors, such as sex, age, and body weight. Exposure to arsenic through sources other than drinking water can also lead to variability in risk estimates. Environmental arsenic concentra- tions affect exposures that do not come from drinking water, and those con- centrations vary among locations because of different natural background levels and local anthropogenic sources. In addition, different foods contain varying concentrations of arsenic (IOM 2001~; therefore, an individuals food preferences will affect arsenic intake. Although this update focuses on risk from exposure to arsenic in drinking water, other factors that affect arsenic exposure from all sources are important to consider when interpreting the dose-response modeling based on use of available epidemiological data. This section discusses those sources of variability in exposure estimates and their potential impacts on an arsenic risk assessment. Many of those sources are further explored and discussed in Chapter 5. Per Capita Water Consumption Water Consumption in the U.S. Population One reason a subpopulation might be at higher risk for the adverse effects of arsenic in drinking water is that they have a higher than average water
136 ARSENIC IN DRINKING WATER: 2001 UPDATE
VARIABILI7YAND UNCERTAINTY 13 7 For many of these permutations, EPA (2000b) provided the raw data that could be used to fit empincal distributions to the data set. Table 4-! provides an example ofthe date provided, end Figure 4-1 shows the plot ofthe dis~ibu- tions. The data clearly show vanability in water consumption with age, with the highest daily total water ingestion rates per unit of body weight occurring in infants and young children. In its risk assessment, EPA took the reported ingestion rates (given in milliliters per day) from an earlier draft of the EPA (2000b) report. EPA developed distributions of body weight from national census data to get the distribution of inking-water rates on the basis of per unit of body weight. Ultimately, the EPA (2000b) report on drinking-water TABLE 4-1 Example of the Distribution of the Water Intakes Represented As Se- lected Percentiles Reported by EPA (2000b) for Bow Sexes, Total Water (from All Sources), and Direct and Indirect Pathways Averaged over 2 Nonconsecutive Days (mL/kg/day) F~ne-age Percentile Category (year) Mean 1% 5% 10% 25% 50% 75% 90% 95% 99% <0.5 92 2 7 14 31 87 139 169 196 239 0.5-0.9 65 2 6a 11 26 58 88 120 164 185 1 -3 31 1 4 7 15 26 40 60 74 118 4-6 27 1a 5 8 14 23 36 51 68 97 7-10 20 1a 4 6 10 17 26 36 44 70 11-14 16 1a 3 4 8 14 21 33 40 60 15-19 15 2 4 7 12 19 29 38 66 20-24 18 2 4 8 14 22 34 44 86a 25-54 20 1 4 6 11 17 26 37 46 69 55-64 20 2 6 8 12 18 26 35 42 59 21 4 7 9 13 19 27 34 39 54 Allages 21 1 4 6 11 17 26 38 50 87 a Sample size does not meet the minimum reporting requirements. Source: EPA (2000b, p. IV-19~.
138 ARSENIC IN DRINKING WATER: 2001 UPDATE rates also presented the distribution in per unit of body weight, and the distri- butions were very similar. In its analysis, EPA computed lifetime average water consumption by randomly selecting from ( 1 ) the water consumption and body-weight distributions for each sex and age category, and (2) randomly selecting a sex for the simulated individual, then multiplying the values drawn by the number of years the individual male or female spent in each age cate- gory, and finally summing over the age categories to simulate an entire life- time. This method did not make any assumptions about consistency (or corre- lation) in individual body weights or water-consumption rates over time. The very short sample time of the water intake survey deserves additional discussion. Although the data on water-consumption rate provide a large amount of information about water ingestion for a very short time (2 days), several important uncertainties arise when using these short-term data to repre- sent the long-term average. To compute the average lifetime ingestion rate from those data, the changes in individual drinking-water behaviors over time must be assumed. It must be also assumed that a random cross-sectional sam- ple for 2 nonconsecutive days accurately estimates water consumption over a lifetime. The later assumption might be valid if the 2 days are representative, because it might be expected that intraindividual variability in water consump- tion over time will be lower than interindividual variability over time (i.e., people who drink a lot of water ~ year will probably drink a lot of water other years). However, the data are lacking to test this assumption, and the extrapo- lation from a 2-day sample to a longer time period might be misleading if some people participate in the survey during a hot summer week and others respond in the winter, neither giving an estimate of typical consumption. In general, some regression to the mean should be expected as a longer period of time is sampled. Biases associated with the survey and participant compliance also add to the uncertainties in the data. Despite those caveats, EPA's data are useful for risk analysis and for characterizing the variability in population drinking-water rates per unit of body weight. Those data are used in the risk calculations done by the subcommittee to determine the sensitivity of the risk estimates to drinking-water consumption values (presented in Chapter 5~. Water Consumption in the Taiwanese Study Population In its earlier risk assessment, EPA (1988) did not consider variability in water consumption per unit of body weight in the study populations when interpreting the epidemiological data on arsenic carcinogenicity in Taiwan or
FARIAB~7YAND UNCERTAINTY 139 (a) (b) So% ~ 70% - 50% 40% 30% 20% 10% 0% 100% / Ad/ Of/ 0.5 - 0.9 4-6 7-10 - - - -11 -14 All ages 0 50 100 150 200 250 Water consumed, mL/kg/day 80% Anon /,7 - 15-19 - - 20 - 24 - 25-54 55 - 64 ~ 65~ All ages 0 50 100 150 200 Water consumed, mL/kg/day 250 FIGURE 4-1 Cumulative distributions of water intake data by age. (a) Cumulative distributions at ages less than 0.5, 0.5-0.9, 1-3, 4-6, 7-10, 11-14, and all ages. (b) Cumulative distributions at ages 15-19, 20-24, 25-54, 55-64, 65 and above, and all ages.
~ 40 ARSENIC IN DRINKING WA TER: 2001 UPDA TE the dose-response relationship; the previous subcommittee also did not quan- tify that variability TIC 1999~. EPA (1988) and NRC (1999) assumed that a typical Taiwanese male and female weigh 55 kg and 50 kg, respectively, and drink 3.5 L/day and 2.0 L/day, respectively. Those values correspond to a daily per kilogram intake of water of approximately 60~0 mL/kg/day, much more than the 21-28 mElkg/day that Americans are estimated to consume. The assumptions suggest considerable differences between Taiwanese and American populations, and the assumptions about Taiwanese water intake have been questioned previously (Mushak and Crocetti ~ 995~. Unfortunately, uncertainty remains about how much water the Taiwanese drink, and no ap- propriate data on the distribution of water consumption in the Taiwanese study populations are available at this time. When such data become available, they could be used to better estimate exposures in the Taiwanese study populations, reducing the uncertainty in the risk assessment. However, it seems likely that the water consumption pattern of the people in southwestern Taiwan, where many of the previous epidemiological studies were carried out, has changed with time as the socioeconomic situation has improved. Therefore, informa- tion on current water intakes might not be useful for reevaluation of the situa- tion 20~0 years ago. In the absence of reliable data on water consumption in the Taiwanese study populations, the sensitivity of the risk estimates to those assumptions should be assessed to quantify some ofthe uncertainty in the risk assessment. Some analyses ofthe sensitivity ofthe risk estimates to different Taiwanese water-intake assumptions are presented in Chapter 5. Given that there is variability in daily drinking-water ingestion rates, which can be partially characterized based on available data, the subcommit- tee believes that the EPA risk assessment should explicitly address the impacts of this variability. In particular, the subcommittee notes that the recent ap- proach of EPA (2001) to account for variability in quantity of water consumed by the United States population without simultaneously accounting for vari- ability in water consumption among the Taiwanese would have the result of underestimating the upper bound of risk attributable to arsenic in drinking water in the quantitative risk assessment. It is reasonable to assume that the Taiwanese had a variable intake of water as well, and biostatistical analysis should consider that the southwestern Taiwanese cohort consisted of a signif~- cant proportion of males and females whose direct water consumption was less than 3.5 and 2.0 L/day, respectively.
VARIABILITY AND UNCERTAINTY 141 Consideration of Infants and Children Within a given population, certain individuals or subpopulations can be more susceptible to the toxic effects of a substance, either because they have a higher exposure to the substance or because they are intrinsically more susceptible to it. Infants and children are often considered more susceptible to the adverse effects oftoxic substances than adults (NRC 1993), but suscep- tibility must be assessed on a case-by-case basis (~ST 1992~. Susceptibility from Exposure Differences EPA (2000b) noted that "when considering water-ingestion rates in units of milliliters per kilogram of body weight per day, this analysis shows that the mean per capita ingestion rates for babies younger than one year are estimated to be three to four times higher than the mean rates for the population as a whole." That increased exposure (on a body-weight basis) indicates that infants and young children might be at increased risk. Therefore, the sub- committee considered whether there is any evidence that infants and children are more susceptible than adults to the toxic effects of arsenic in drinking water. Calderon et al. (1999) investigated the excretion of arsenic in urine. They found increased concentrations of arsenic metabolites (measured as total arsenic) in individuals less than 18 years of age compared with individuals greater than ~ ~ years of age and suggested that the difference might be due to differences in water consumption. Although these data might support the hypothesis that children are at higher risk to the effects of arsenic in drinking water, the impact of the increased exposure in infants and young children on the risk assessment depends on what pattern of exposure correlates with the effect and whether the increased exposure is intrinsically accounted for in the design of the epidemiological studies used in the risk assessment. Those issues are discussed further in the section Dose Metrics and Model Uncer- tainty. lit should be noted that the variation in water intake of infants is dependent on the frequency of breast feeding. t
~ 42 ARSENIC IN DRINKING WA TER: 2001 UPDA TE Intrinsic Susceptibility It is also important to consider whether infants and children are intrinsi- cally more susceptible to arsenic toxicity. Differences in the metabolism of arsenic in infants, children, and adults could make one ofthose subpopulations more susceptible to the toxic effects of arsenic. Previous studies indicate that young individuals might have a lower rate of methylation of inorganic arsenic than adults (NRC 1999~. More recently, indigenous Andean children have been found to have about 45% oftheir urinary arsenic as dimethylarsinic acid (DMA) (Concha et al. 1998~. In a study of people (2-83 years of age) exposed to arsenic in drinking water in Finland, the percentage of arsenic in urine as DMA increased with age (from 59% in people less than 30 years of age to 74% in people more than 50 years of age) (Kurttio et al. ~ 998~. Similar results were seen in a study of arsenic-exposed people in northeastern Taiwan (Hsu et al. 1997~. in contrast, a recent study from Mexico showed that children in Region Lagunera had normal fractions of DMA in urine (an average of about 70°/0; De! Razo et al. 1999~. Therefore, the data on variation in arsenic methylation with age are not consistent, and it is not known how such differ- ences in methylation would affect arsenic toxicity. As discussed in Chapter 3, the methylation of inorganic arsenic can affect its toxicity, but that methyla- tion is no longer thought to be entirely a detoxification process. Variability in Arsenic Metabolism Some subsections or subpopulations of a given population could be more or less susceptible to arsenic toxicity because of differences in arsenic metabo- lism. In particular, there are marked differences in arsenic methylation among mammalian species, population groups, and individuals. As discussed in Chapter 3, the valence state and extent of methylation of arsenic can affect its toxicity. The main products in the methylation of inorganic arsenic, pentava- lent monomethylarsonic acid (MMAV) and dimethylarsinic acid (DMAV), are readily excreted in urine. Therefore, more efficient methylation, especially to DMAV, means faster overall excretion of arsenic (Vahter 1999a). The methylation of arsenic, however, is not entirely a detoxifying process. The initial reduction of pentavalent inorganic arsenic (ASV) to trivalent inorganic arsenic (AsIII), as well as the formation and distribution of methylated trivalent arsenic metabolites in tissues, results in increased toxicity. MMAIII is highly .
VAR]ABILI7Y AND UNCERTAIN7Y 143 reactive with cellular constituents, especially sulflhydryl groups, and, in gen- eral, exhibits a greater toxicity than inorganic Astir (see Chapter 3~. Variations in the metabolism of arsenic, therefore, are likely to be associated with vana- tions in susceptibility to arsenic. The extent to which such metabolic varia- tions might affect an individuaT's susceptibility to cancer or systemic toxicity with arsenic exposure is an important uncertainty. Because methylated forms of arsenic are readily excreted in urine, evalua- tion of arsenic methylation efficiency is generally based on the relative amounts of the different metabolites detected in urine (Vahter ~ 999a). Cur- rently, however, there are no data on the association between urinary concen- trations of MMAi~ or DMA~ and their formation and concentrations in tis- sues. Those metabolites are highly reactive and retained intracellularly, whereas DMAV is the main form of arsenic excreted from the cells (Styblo et al. 2000) (see Chapter 3~. Therefore, the concentrations of MA and DMA~ in the urine are less likely to reflect the formation of those metabolites in tissues. However, MA has been detected in human urine, although the site of reduction of MMAV to MMA~i is not known. Because MMAi~T is pro- duced fromMMAV, it seems probable that the tissue concentrations of EMMA would increase with the total production of MMA and, therefore, with increas- ing total MMA in urine. Genetic Polymorphisms Related to Arsenic Metabolism Glutathione (GSH) S-transferases (GST) constitute a large family of de- toxifying enzymes that catalyze the conjugation of reduced GSH with a wide range of compounds. Because GSH levels, which are essential for the reduc- tion and methylation of arsenic, are influenced by GSTs, the latter have been considered as possible candidates for influencing the metabolism and the toxicity of arsenic. A study from the area in northeastern Taiwan where ar- senic is endemic showed that subjects with the null genotype of GST M! had a slightly increased percentage of inorganic arsenic in urine (Chiou et al. 1997~. In addition, there was a tendency for an increased percentage of DMA in urine of subjects with the null genotype of GST T1. Whether those geno- types might influence susceptibility to arsenic is not known. As discussed in Chapter 3, recent studies indicate that human MMAV reductase is identical to human glutathione-s-transferase omega class 1-1 (hGSTO 1-1~. Therefore, polymorphisms in that GST also might affect both cellular protective mecha-
144 ARSENIC IN DRINKING WA TER: 2001 UPDATE nisms and arsenic metabolism. hGSTO I-l has not been extensively studied (see Chapter 3), and it is not known if polymorphisms in that gene occur in human populations. A polymorphism in the N-acetyltransferase-2 (NAT2*) gene can affect susceptibility to bladder cancer. Previous studies have demonstrated that individuals who are slow acetylators as a result of a polymorphism in that gene are at greater risk for developing bladder cancer following a chemical exposure (to arylamines). Therefore, Su et al. (1998) investigated whether that polymorphism also makes individuals susceptible to arsenic-induced bladder cancer. The NAT2~-related slow N-acetylation polymorphism was not associated with increased risk for bladder cancer among arsenic-exposed people in the area of southwestern Taiwan where blackfoot-disease is en- demic. An association was seen, however, in an area of Taiwan without in- creased arsenic in drinking water (Su et al. 1998~. Population Variations in Arsenic Metabolism The average relative distribution of arsenic metabolites in the urine of various population groups exposed to arsenic seems to be fairly consistent: ~ 0-30% inorganic arsenic, ~ 0-20% MMA (generally lower than the percentage of inorganic arsenic), and 60-70% DMA (for review, see Vahter 1999b). There are, however, exceptions. Indigenous people in the north of Argentina and Chile often excrete very little MMA in urine, and in some cases, only a few percent of the arsenic is MMA (NRC 1999~. In a recent study in Region Lagunera, Mexico, Del Razo et al. (1999) found that in children who con- sumed 100-1,250 ,ug arsenic/L in drinking water, less than 10% of urinary arsenic was MMA in about 40% of the children, and less than 5°/O was MMA in about 20% of the children. In contrast, studies from southwestern Taiwan consistently show a higher percentage of MMA (usually 20-30%) than inor- ganic arsenic in urine. In people living in northeastern Taiwan, where arsenic concentrations in drinking water might reach 3,000 ,ug/L, 27% of urinary arsenic was MMA (Chiou et al. ~ 997~; inorganic arsenic constituted only ~ 2% oftotal urinary arsenic metabolites. Because high arsenic intake might inhibit the second methylation step (i.e., from MMA to DMA), increased concentra- tions of MMA might be due in part to the high exposure concentrations. However, the subset of Taiwanese individuals who consumed water with arsenic at less than 50 ~g/L, a concentration at which the second methylation step is unlikely to be inhibited, also had, on average, 31.5°/O and 12% of their
VARIAB~ITYAND UNCERTAINTY 145 urinary arsenic as MMA and inorganic arsenic, respectively. In another study from southwestern Taiwan, where people no longer drink the artesian water with high arsenic concentrations, 21% of the average urinary arsenic was MMA (Hsueh et al. 1998~. In skin-cancer patients from the same area, 31% of urinary arsenic was MMA, and 1 1% was inorganic arsenic; in controls, 23% was MMA, and 11% was inorganic arsenic. Thus, the average percent- age of inorganic arsenic in total urinary arsenic was less than half of the per- centage of MMA. Total urinary arsenic was less than 100,ug/L, which is not likely to markedly influence the percentage of MMA. Among the populations studied, less variation appears to occur in urinary DMA than in MMA (Figure 4-2~. People exposed to arsenic concentrations that would not inhibit methylation generally have more than 60% of their urinary arsenic as DMA. There are marked differences in the percentage of MMA in urine among mammalian species (see Chapter 3~. The observed variations in the pattern of arsenic metabolites in urine might indicate a genetic polymorphism in the regulation of enzymes responsi- ble for arsenic methylation. in particular, the reduction of MMAV to MA and subsequent methylation of MMA~i to DMA~ might be slow in certain people, resulting in high MMA concentrations in urine. The high urinary MMA concentrations might also be the result of low binding affinity of MMA to a carrier protein or more efficient mechanisms for excreting MMA from the cell. A few studies published to date on MA and DMA~ indicate variations in the excretion of these arsenic metabolites in urine. People exposed to arsenic in drinking water in Romania at 28-161 ,ug/L were found to excrete MA at 5-7 ,ug/L, on average, independent of the exposure concentration; in those individuals with very low exposures (2.8 ~Lg/L), MMAii~ was not detectable in urine because ofthe low concentrations (Aposhian et al.2000b). Four groups in Bangladesh, where average arsenic concentrations in water are between 33 and 248 ,ug/L, showed increasing urinary MMA~i and DMAi concentrations with increasing exposures (Mandal et al. 2001~. However, only 42% of those exposed had urinary MA, and 72% had DMA~. The average concentrations of MA for the groups ranged from 3 to 30 ,ug/L, and those of DMAi~ ranged from ~ to 64 1lg/L. in contrast, people in Inner Mongolia, China, exposed to arsenic at about 500 ~g/L of drinking water (Aposhian et al. 2000a), and people in northern Chile, exposed at about 600 ,ug/L (Aposhian et al. 1997), excreted MMAi~ in urine only after treatment with 2,3-dimercaptopropane sulfonic acid (DMPS). It should be noted that trivalent arsenic metabolites are easily oxidized following the collection of
46 ARSENIC IN DRINKING WA TER: 2001 UPDA TE a b c Talwan C a lifornia S. Ana, Moxico Toconao, Chib S.pedro, Chib S.Antonlo, Arg. (Women) Talwan Califomia S. Ana, Me~dco Toconao, Chile S.Podro, Chib S.Antonio, Arg. (Women) Ta~an Ca.tornia S. Ana, Moxico Toconao, Chib S.podro, Chib S.Antonio, Arg. (Womon) 20 3o . . 0 10 . . . . . 20 30 40 50 60 30 40 ~nn ~ . . . 70 80 90 100 50 60 70 80 90 100 FIGURE 4-2 Population variations in the percentage of urinary as various arsenic metabolites. (a) Percentage as DMA, (b) Percentage as MMA, and (c) Percentage as inorganic arsenic. Source: Modified from Vahter 2000.
VARIABILITYAND UNCERTAINTY 14 7 urine samples (Le et al. 2000), and the oxidation might have resulted in underestimation of the amounts of MA and DMAi~ in urine. That might explain some of the differences seen between studies. Therefore, firm conclusions on population variations must await further investigation of the effect of sampling, storage, and analyses on urinary concentrations of MA and MA. Interindividual Variation in Arsenic Metabolism There is considerable interindividual variation in arsenic methylation in humans, possibly due to variations in the activities of the enzymes involved in arsenic methylation. In vitro studies ofthe methylation of inorganic arsenic in human hepatocytes obtained from four donors showed variation in arsenic methylation between 3 and 6 picomoles (pmol) of arsenite per 106 cells per hour (Styblo et al. 1999~. There is also wide interindividual variation in the relative amounts of MMA and DMA in urine (Vahter 1999b, 2000~. The distribution pattern for the arsenic metabolites in the urine of individual Argentinian women exposed to arsenic in drinking water was remarkably stable over a period of about a week, indicating that an individual's methyla- tion of inorganic arsenic is fairly constant over time (Concha 20014. Therefore, interindividual variation cannot be explained by the fact that in most studies there is only one urine sample per person. The variation among individuals seems to be more marked in the amount of urinary MMA than in the amount of DMA. The percentage of MMA might vary 30-fold among individuals in a particular population, whereas the varia- tion in the percentage of DMA is generally no more than 2-fold (Figure 4-2~. The observed differences in the urinary excretion of arsenic metabolites might indicate a genetic polymorphism in the regulation of enzymes responsible for arsenic methylation. Genetic polymorphism has been demonstrated for other human methyTtransferases (Weinshilboum et al. ~ 999~. Effect of Dose on Arsenic Methylation Experimental animal studies have shown that several factors, including dose, influence the methylation of inorganic arsenic (NRC 1999~. Recent studies are consistent with the effect of dose on arsenic methylation reported in the previous report (NRC 1999~. In Region Lagunera, Mexico, children
~ 48 ARSENIC IN DRINKING WA TER: 2001 UPDA TE with an average of 70°/O of urinary arsenic as DMA showed a significant de- crease in the percentage of DMA and a concurrent increase in the percentage of inorganic arsenic and MMA in urine with increasing exposure to arsenic (total urinary arsenic at 100-1,100 ~g/~(Del Razo et al. 1999~. In a few children with some of the highest exposures, only 40-50% of urinary arsenic was DMA, MMA constituted 20%, and inorganic arsenic 25% of total urinary arsenic. It is probable that the decrease in methylation with increasing dose is due to inhibition of methyltransferase. As discussed in Chapter 3, recent experimental studies in vitro confirm previous findings that methylation of arsenic, especially the second methylation step in which MMA is converted to DMA, is inhibited by excess amounts of arsenic (Styblo et al. 1999, 2000; De Kimpe et al. 1999~. Effect of Other Chemicals on Arsenic Methylation Arsenic methylation might also be influenced by simultaneous exposure to other exposures or chelating agents. Experimental studies indicate that the methylation of inorganic arsenic to MMA is inhibited by vanadium, iron, selenite, and cadmium; specifically, selenite, mercury, lead, and chromium inhibit the second methylation step (De Kimpe et al. 1999~. Administration of chelating agents, such as DMPS, has been found to change the urinary arsenic metabolite pattern dramatically (Aposhian et al. 2000a,b; Le et al. 2000~. In particular, urinary excretion of MMA increased. Of total arsenic in urine, the percentage of MMA increased by about 10-fold and the percent- age of DMA decreased following treatment with DMPS. In vitro studies using liver cytosol from the Flemish rabbit demonstrated that British Anti-Lewisite (BAL), dimercaptosuccinic acid (DMSA), and ethylenediaminetetraacetic acid (EDTA), all at concentrations between 0.15 and 15 mM, inhibit the methylation of both arsenite and MMA (De Kimpe et al. ~ 9991. DMPS mainly inhibited the methylation of MMA to DMA. Citrate had a limited stimulatory effect on both steps in the arsenic methylation. Metabolic Variability, Tissue Retention, and Health Effects If the reactive metabolite MMAii~ is formed and distributed to the tissues following exposure to inorganic arsenic, it seems likely that people who ex- crete a relatively higher proportion of MMA in the urine might have a higher
VARIABILITYAND UNCERTAINTY 149 retention of the absorbed arsenic than those who excrete lesser amounts of MMA. Indeed, most animals for which the main product of inorganic arsenic methylation is DMA show a faster overall excretion of arsenic than do humans (Vahter 1999a). in Andean people, who have only a small percentage of urinary arsenic as MMA, the ratio of arsenic in blood to arsenic in urine (ad- justed for variation in total urinary metabolites of inorganic arsenic) was very low (about 0.03) (Concha et al. 1998~. In contrast, in a study that measured total urinary arsenic in people from California and Nevada, that ratio was twice as high (about 0.06) (Valentine et al. 1979~. Although the urinary me- tabolites of arsenic were not reported in the latter study, the urine can be assumed to contain 10-20% MMA, because recent studies of people exposed to arsenic in drinking water in Nevada showed 20% inorganic arsenic, 22% MMA, and 58% DMA in the urine (Warner et al. 1994~. Because arsenic from seafood is readily excreted, seafood arsenic influences urine concentra- tions more than blood concentrations. Therefore, measuring total urinary arsenic, rather than metabolites of inorganic arsenic in the urine, might under- estimate the true ratio of blood-to-urine arsenic from inorganic arsenic. Thus, the ratio in the study by Valentine et al. (1979) might have been even higher. A lower retention of arsenic in people with low MMA excretion is also suggested by data from a number of experimental studies on human subjects receiving specified doses of inorganic arsenic (Vahter2001~. The percentages of urinary arsenic as MMA and inorganic arsenic showed similar negative associations with the overall arsenic excretion, while the percentage of urinary arsenic as DMA showed a positive association. Thus, urinary excretion of all arsenic metabolites (percentage of dose) decreased with increasing percentage of both inorganic arsenic and MMA in urine but increased with increasing per- centage of DMA. A number of recent studies indicate an association between the prevalence of arsenic-related toxic effects and the pattern of arsenic metabolites in urine. In particular, there are indications that people with arsenic-related toxic ef- fects have a higher percentage of urinary arsenic as MMA compared with people without such effects. Mexican people with signs of dermal toxicity due to exposure to arsenic in drinking water had a higher percentage of uri- nary arsenic MMA and a Tower percentage of DMA than those without visible dermal toxicity (Del Razo et al. 19971. The difference was about 5% MMA. In the area of southwestern Taiwan where biackfoot-disease is endemic, skin cancer patients were found to have a higher percentage of urinary arsenic as MMA, 31% on average compared with 23% among the controls (Hsueh et al. ~ 997~. Skin-cancer patients were also found to have Tower serum p-carotene
~ 50 ARSENIC IN DRINKING WA TER: 2001 UPDA TE concentrations than controls. However, the association between percentage of urinary arsenic as MMA and the incidence of skin cancer remained signifi- cant after adjustment for cumulative arsenic exposure and serum ,8-carotene concentrations. In a later study from the same area of Taiwan, a significantly higher percentage of urinary arsenic as inorganic arsenic (about 2%) and MMA (about 2%) and a Tower percentage as DMA were found in people with arsenic-related skin lesions than in matched controls (Yu et al. 2000~. The number of structural chromosomal aberrations in peripheral lymphocytes of Finnish people exposed to arsenic via drinking water (average concentration of 410 ,ug/~) was also found to be associated with increasing percentage of urinary arsenic as MMA and decreasing percentage as DMA (Maki- Paakkanen et al. 1998~. Further studies are needed to clarify the roles of MMA and DMA in arsenic toxicity. SOURCES OF UNCERTAINTY Dose Metrics and Model Uncertainty The association between arsenic exposure and an adverse health effect can be calculated by comparing the response to a range of different dose metrics, including cumulative arsenic dose, average daily intake over a lifetime, and peak arsenic exposure, and using a wide range of different statistical models. The dose metric and model used can affect the calculation of risk estimates, and depending on how accurately exposure is measured by the dose metric, and how closely the metric is related to the end point of concern, more or less uncertainty is introduced to a risk assessment. The end point of concern for arsenic toxicity is cancer (see Chapter 2~. In its risk impact assessment, EPA (2000a) stated, In certain circumstances, the increased daily dose in children can be effectively considered for non-carcinogenic effects because toxicity is evaluated in terms of exposure that can range from relatively short- term to Tong-term exposure. However, carcinogenic effects (i.e., bladder cancer) are evaluated based on a lifetime of exposure, which takes into consideration the elevated dose that occurs in children. Because the health effects measured in this benefits assessment [EPA's Risk impact Assessment] are bladder and lung cancer, a sensi- tivity analysis to consider higher doses of arsenic during childhood was not necessary.
VARIABILI7YAND UNCERTAINTY 151 The subcommittee agrees with that statement but emphasizes that its validity also depends on the lifetime cancer risk focus that uses the lifetime average daily dose as the dose metric. The subcommittee further emphasizes that cancer represents the most sensitive health end point and that the dose- response data are based on a population with lifetime exposure to arsenic. Thus, the dose-response model reflects lifetime exposure (from childhood to adulthood), as expected in any model for lifetime cancer risk. Clearly, uncer- tainty exists about whether this mode! is appropriate, as discussed more in Chapter 5. As noted in Chapter 3, at a mechanistic level, little is known about the impact of prolonged exposure at relatively Tow doses compared with shorter exposures at high doses, and this lack of knowledge arises as an impor- tant source of uncertainty in assessing the risks of ingesting arsenic. The lack of agreement between health effects observed in the epidemiological studies described in Chapter 2 and those observed in animals described in Chapter 3 raises additional uncertainties about the appropriate dose-response model and dose metric for humans when using animal data. if the peak concentration represents the appropriate dose metric of concern, then use of a lifetime cancer risk model would not be valid. Assuming that environmental exposures are essentially constant, peak concentrations likely occur when people are infants because of the higher water intake per unit of body weight. However, if cu- mulative dose is important, or if arsenic acts as a late-stage carcinogen, infor- mation about the timing and amount of exposure is needed. At this time, the mode of action of arsenic and the dose metrics that are best correlated with the carcinogenic effects of exposure to arsenic in drinking water are not known. The dose metric affects the interpretation of an epidemi- ological study. lit is possible that the lack of an association between arsenic and cancer could result from using an inappropriate dose metric. For example, if the duration and magnitude of exposure are both important for arsenic- induced cancer, then cumulative exposure might be associated with cancer, and peak exposure might not be associated with cancer. The issue of dose metrics, therefore, complicates the epidemiological studies and must be con- sidered when interpreting and comparing different studies. Because the most appropriate dose metric for arsenic-induced cancer is still not known, the choice of metric adds uncertainty to arsenic risk assessments. In addition, a wide range of different models can be used to fit the arsenic carcinogenicity data currently available, and no clear biological basis exists for distinguishing among them. The implications of model uncertainty are explored quantitatively in Chapter 5.
52 ARSENIC IN DRINKING WA TER: 2001 UPDA TE Exposure to Arsenic in Food and Effect of Diet and Nutrition In addition to adjusting for an assumed higher drinking-water rate among Taiwanese people compared with the U.S. population in its most recent risk assessment, EPA (200 ~ ~ also adjusted for differences in dietary arsenic intake. This type of adjustment is appropriate since arsenic in the diet represents an important source of exposure, and one that must be considered in the context of understanding the dose-response data observed in the ecological epidemio- logical studies, because it contributes to the total dose. Since the EPA and the subcommittee have focused on use of the Taiwanese epidemiological data as the basis for developing a quantitative dose-response model, the estimation of the proportion of the total Taiwanese arsenic intake that resulted from food emerges as a source of uncertainty. The previous NRC report on arsenic (NRC ~ 999) had an extensive discus- sion. about arsenic in food. It was concluded that the highest concentrations of arsenic are found in products from the marine environment. In fish and shrimp, the major arsenical is arsenobetaine. In most marine algal products, arsenosugars are the principal arsenic species, although up to 50°/O might be the more toxic arsenate. Arsenosugars are metabolized by humans mainly to DMA. Although the subcommittee was not asked to evaluate the exposure to arsenic, it notes that the recent Food and Drug Administration Total Dietary Study for 1991-1997 (Tao and Bolger 1999) indicates that a major source of arsenic in the U.S. diet is of marine origin. In that study, total inorganic and organic arsenic in food was detected in 63 of the 261-264 (24%) foods and mixed dishes analyzed at a detection limit of 0.03 ppm. The highest concen- tration was found in seafood, followed by rice and rice cereal, mushrooms and poultry. Based on the U.S. Department of Agriculture's 1987-1988 Nationwide Food Consumption Survey, the estimated daily total arsenic average intakes were 2 ~g/day for infants, 23 ,ug/day for toddlers, 20 ~g/day for 6-year-old children, 42 ,ug/day for adults (40-45 years of age), and 32 ~g/day for individ- uals 60-65 years of age. Ofthe estimated total arsenic intakes for infants, 42% originated from seafood and 31% from rice and rice cereals. Ofthe estimated total arsenic intakes, seafood contributed 76% to 90% for children (2-10 years old) and 89°/O to 96% for adults (25-30 years old), whereas rice and rice cere- als contributed 4% to 8% for children and 1% to 4% for adults (25-30 years old). I, In a recent market basket survey, 40 commodities anticipated to provide at least 90% of dietary inorganic arsenic intake were identified (Schoof et al.
VARIABILITYAND UNCERTAINTY 153 1999~. Total arsenic was analyzed using NaOH digestion and inductively couplet/ plasma-mass spectrometry. Inorganic arsenic was analyzed using HC! digestion and hydride atomic absorption spectroscopy. Consistent with earlier studies, total arsenic concentrations (all concentrations reported as elemental arsenic per tissue wet weight) were highest in seafood (ranging from 160 ng/g in freshwater fish to 2,360 ng/g in saltwater fish). In contrast, average inor- ganic arsenic in seafood ranged from less than ~ to 2 ng/g. The highest inor- ganic arsenic values were found in raw rice (74 ng/g), followed by flour ( l ~ ng/g), grape juice (9 ng/g), and cooked spinach (6 ng/g). Thus, grains and produce are expected to be significant contributors to dietary inorganic arsenic intake. Based on these data and previous studies on concentrations of arsenic metabolites in urine ARC 1999), the average dietary intake of inorganic arsenic in the United States is likely to be on the order of 10 Friday, as esti- mated by EPA. Also, the dietary intake of inorganic arsenic in the area of southwest Taiwan where arsenic is endemic is possibly higher than that in the United States, because although the water with increased arsenic concentra- tions usually is no longer ingested, it is still used for agnculture and piscicul- ture, as well as washing dishes, cleaning, watering plants, and, occasionally, drinking in dry seasons (Hsueh et al. 1997; Yu et al. 2000~. The few available data on arsenic in rice and yams from Taiwan were reported in the NRC 1999 report. However, there is little evidence on the levels and species of arsenic con- sumed in foods by different individuals and populations, and the role of ar- senic in food remains somewhat uncertain. Furthermore, while the new data should be used in the context of estimating total arsenic exposure and risk to the U.S. population, they are not particularly helpful for the subcommittee's efforts to characterize the amounts of arsenic in the food of the historically exposed Taiwanese population. Such data are needed to appropriately charac- terize the dose-response function for total ingested arsenic. The bioavailabil- ity to humans of arsenic present in raw foods and the bioavailability of arsenic incorporated into foods, such as rice from cooking water, has not been well characterized. In its risk assessment, EPA assumed that food intake of arsenic was higher in Taiwan than in the United States. However, there is little evidence to sup- port the numbers used, and the subcommittee explores this important source of uncertainty in Chapter 5. Also related to diet is the hypothesis that dietary and nutritional aspects of the Taiwanese population render it more susceptible to the carcinogenic
~ 54 ARSENIC IN DRINKING WA TER: 2001 UPDA TE effects of arsenic than the U.S. population and, therefore, that the results of the Taiwanese studies might not be relevant to the U.S. population. This hypothesis is based in part on the fact that cancer has been consistently associ- ated with aspects of diet; both increased and decreased cancer risks have been associated with different dietary parameters (Doll and Peto ~ 981; N~C ~ 982; Steinmetz and Potter l99la,b; World Cancer Research Fund Panel 1997~. Particularly relevant to the arsenic and cancer data are the epithelial cancers, in which the association between diet and cancer is most obvious (WorId Cancer Research Fund Panel ~ 997~. Before discussing the data specific to arsenic, it is important to consider the potential influence of diet in a general sense. Deciding on the strength of the association between an exposure and an outcome is heavily influenced by the precision with which the exposure can be measured. Diet and eating habits are complex and difficult to measure accurately (Willett ~ 990~. Typi- cally, when errors in exposures occur, the derived estimates of risk are biased toward the null; that is, the association, as estimated, is weaker than it would be with perfect measurement. The degree of bias in the effect of food on cancer incidence remains a matter of considerable debate, but it is possible that diet could affect the epidemiological studies of arsenic. There are three general ways by which diet and nutrition could affect the association between arsenic and diseases (including cancers), particularly in the Taiwanese popula- tion studies: (~) confounding by diet; (2) potentiation of susceptibility as a result of poor-quaTity diet; and (3) the difficulty with generalizing the Taiwan- ese findings to other populations because of differences in diet. Confounding occurs when there is a condition that might be a factor in producing the same response as the agent of interest. Although some uncer- tainties remain, the subcommittee believes that based on the existing weight of the evidence, it is unlikely that confounding by diet is responsible for the association between arsenic exposure and cancer. The association between exposure to arsenic and cancer has been seen in several independent investiga- tions conducted in different populations with varying diets, using a variety of study designs. The magnitudes of the observed relative risks for cancer were in general so high that they could not be accounted for by known or hypothe- sized dietary factors. For example, in the analysis by Smith et al. (1992) of the southwestern Taiwanese data, the estimated mortaTityriskratio for bladder cancer in females relative to the general Taiwanese population increased in a dose dependent manner from 11.9 (water arsenic concentration less than 300 Go/, to 25.1 (300 to 600 ,ug/~) to ultimately 65.4 (greater than 600 ,ug/~. In the study by Chiou et al. (2001 ) in northeastern Taiwan that used a prospec- tive study design with individual exposure data and internal controls, there
VARIABILITY AND UNCERTAINTY 155 was a 15-fold increase in risk for bladder cancer in the subjects (male and female combined) with the highest levels of exposure. A more general consideration is that although diet remains unmeasured in most of the epidemiological studies of arsenic and cancer, when it has been measured in other settings, it is almost never a confounder of relationships between specific and well-measured exposures. For example, diet has not been identified as a significant confounder of the relationship between ben- zene, asbestos, or cigarette smoking and relevant specific cancer outcomes. Nonetheless, in the absence of clear evidence of a biological mechanism for arsenic-induced carcinogenesis (see Chapter 3), the possibility of a relatively minor degree of unmeasured confounding remains, even if diet is an unlikely candidate. In addition to confounding, interaction or modification of effects might also occur when the association between exposure and outcome is much stron- ger among those with, versus without, a different characteristic. For example, smokers might have a 4-fold increase in risk for a particular disease, and those exposed to radiation might have a doubling of risk. If smokers who are also exposed to radiation have a 20-fold increase in disease, there is evidence of interaction, which suggests biological synergism of one exposure with an- other. In the case of arsenic, interaction or modification of effects by poor nutri- tion could potentially increase susceptibility. In the southwestern Taiwanese population, the staple foods used to be sweet potatoes and rice, with a Tow intake of vegetables and fruit and their bioactive constituents that can protect against cancer. The impact of nutrition on arsenic susceptibility in the study populations was discussed in detail in the ~ 999 report, and more research has been conducted since that time. In a recent study from West Bengal, India (see Dermal Effects in Chapter 2), it was demonstrated that among people exposed to high concentrations of arsenic in drinking water, those who were below 80% of standard body weight had a 1.6-fold increase in the prevalence of keratosis (Mazumder et al. ~ 998~. In a case-controT study of 24 ~ blackfoot- disease (BFD) cases in southwestern Taiwan, artesian-well-water consump- tion, arsenic poisoning, familial history of BED, and undernourishment were all significantly associated with the development of BED (Chen et al. ~ 988~. Undernourishment was mainly characterized by higher intake of dried sweet- potato chips (the major food for rural residents in southwestern Taiwan before 1960) and Tower intake of rice and vegetables. According to the study by Hsueh et al. (1995), the prevalence of arsenic-induced skin cancer increased with increasing consumption of dried sweet potatoes as staple food. It also increased with chronic liver disease (chronic carriers of hepatitis B antigen
156 ARSENIC IN DRINKING WATER. 2001 UPDATE with liver dysfunction). Those data support earlier findings that malnutrition might increase susceptibility to arsenic (NRC 1999~. Recent studies argue against the assertion that poor nutrition has a major impact on the toxicity of arsenic. The prevalence of arsenic-induced skin lesions in northern Chile was reported to be similar to that for areas of Taiwan and West Bengal. However, although the areas had similar arsenic concentra- tions in the drinking water, the study population in Chile had good nutrition and those in the other areas did not (Smith et al. 2000~. In Chile, 4 of ~ ~ men (36%) and 2 of 23 children (ADO) had skin effects in the form of pigmentation changes and hyperkeratosis. These data were compared with prevaTences of skin effects of 11% among men (30-59 years) and 5% among children less than 20 years of age in West Bengal (Mazumder et al. ~ 9981. They were also compared with an overall prevalence of 18% hyperpigmentation and 7°/O hyperkeratosis in Taiwan, although age distribution was not presented (Tseng et al. 1968; Tseng 1989~. However, it should be noted that the study in Chile was limited to ~ ~ exposed families and ~ control families. Taken together, those data indicate that nutritional status might influence the risk for health effects following arsenic exposure. Thus, the risk estimates based on the population in southwestern Taiwan might have been influenced by the poor nutrition of that population at the time of the increased arsenic exposures. However, the recent studies from northeastern Taiwan (Chiou et al. 2001) and Chile (Ferreccio et al. 2000) that can also be used for quantita- tive risk characterization (see Chapter 5) involve populations with much better nutrition. Some research has investigated whether the status of specific nutrients is associated with an increased susceptibility to arsenic-induced cancer. In the area of southwestern Taiwan where blackfoot-disease is endemic, skin-cancer patients were found to have significantly Tower serum ,6-carotene concentra- tions compared with controls (Hsueh et al. 1997~. The odds ratio for skin cancer decreased from 1 .0 at a serum ,8-carotene concentration of 0.14 mg/L or less to 0.43 (95°/O confidence interval (CI) = 0.06-2.85) at 0.15-0.18 mg/L and to 0.01 (95% CT = 0.00-0.37) at greater than 0.~S mg/L. A synergistic interaction was seen between duration of consumption of artesian well water with high arsenic concentrations and low serum ,8-carotene concentrations in the development of ischemic heart disease (Hsueh et al. ~ 998~. Selenium status has also been suggested as an influence on the toxicity of arsenic (see NRC 1999), and data in laboratory animals show that relatively high doses of selenium can affect arsenic metabolism. Studies on mice fed diets with various relatively high doses of selenium indicate that the animals on the selenium-deficient diet had a slower elimination of single oral doses of
VARIABILITYAND UNCERTAIN7Y 1 5 7 arsenite, arsenate, or DMA compared with selenium-suff~cient mice (Kenyon et al. 1997~. Mice fed a diet with excess selenium showed a decreased methylation of inorganic arsenic. Similarly, in vitro studies using primary rat hepatocytes showed that concurrent exposure to selenite (0. ~ to 6 EM) inhib- ited the methylation of arsenite (Styblo and Thomas 2001~. The second methylation step, MMA to DMA, seemed to be more sensitive to the inhibi- tory effect of selenium as the ratio of DMA to MMA decreased significantly with selenium treatment. Given that a possible mechanism of action of arsenic involves the influ- ence of arsenic on DNA repair, Tow folate intake could cause increased sus- ceptibility to arsenic carcinogenesis (EPA 1997~. Deficiency in folate and vitamin Bl2 might lead to decreased levels of S-adenosy~methy~transferase (SAM), increased levels of serum homocysteine, and possibly hypomethyla- tion (Newman 1999~. Thus, deficiency in those vitamins might result in de- creased methylation of arsenic. Folic acid was found to protect SWV/Fnn mouse embryo fibroblasts against cytotoxicity of arsenite and DMA (Ruan et al. 2000), but the folio acid had less effect on the arsenic-related growth inhi- bition. However, even if deficiencies of folate and other micronutrients in- crease susceptibility to arsenic, intakes of many micronutrients are often low in developed countries, particularly among the poor; therefore, that increased susceptibility might be relevant to a segment of the U.S. population. The final question is whether the Taiwanese findings can be generalized to the United States even though the populations have different patterns of dietary intake. There are two issues here. First, the pathobiology of arsenic -related deleterious outcomes, including carcinogenesis, appears to be consis- tent across human populations, with little evidence of selection for resistance to some consequences even with millenia of exposure (Smith et al. 2000~. Second, although human diets vary substantially, patterns of cancer risk and protection against cancer are also relatively consistent across populations (WorId Cancer Research Fund Panel 1997~. As suggested above, there might be variation across countries in the proportion of the population exposed to carcinogens or protected against those carcinogens by host biology or protec- tive dietary intake. The pathobiology itself is relatively uniform. Other Uncertainties Although the discussion on latency in Chapter 2 mentioned a number of uncertainties, it is worthwhile to reiterate that the subcommittee's understand- ing of the timing of transitions on the path from exposure to disease for ar-
158 ARSENICINDRINKING WATER: 2001 UPDATE sonic remains highly uncertain, even as more is learned about the underlying mechanisms of disease. In addition, the variability in latency of different types of arsenic-related diseases has not been explored. These might arise from a difference in the exposure level, duration of exposure, age of exposure, susceptibility (that might lead some people to respond more quickly than others), or other unknown risk factors. In assessing the risks from arsenic, it is nearly impossible to entirely rule out the possibility that genetics, lifestyle differences (e.g., smoking, food preferences, cooking habits), and exposure to other environmental factors might play a role in explaining variability in the risks. A few investigations have suggested that there might be an interaction between arsenic exposure and smoking in cancer causation. A case-control study by Bates et al. (1995) examined the relationship between arsenic exposure in drinking water and bladder cancer and found a positive association between arsenic and bladder cancer only among smokers. The findings of two studies (Ferreccio et al. 2000; Tsuda et al. 1995) suggest that there might be a synergistic effect be- tween the arsenic ingestion and smoking on the risk of lung cancer. A syner- gistic interaction for lung cancer between smoking and inhaled arsenic has been noted in occupational cohorts (Hertz-Picciotto et al. 1992~. Further confirmation and characterization of an interaction between arsenic ingestion and smoking in the causation of cancer are necessary before this potential effect can be accounted for in a quantitative risk assessment. Nevertheless, it is important to emphasize that there is little reason to suspect that smoking is a significant confounder of the association between arsenic ingestion and lung or bladder cancer. In the case-control study of lung cancer in Chile by Ferreccio et al. (2000) described in Chapter 2, high odds ratios for lung cancer with increasing arsenic exposure were observed in models that adjusted for smoking history. In the cohort study of incident bladder cancer in northeastern Taiwan reported by Chiou et al (2001), the observed association with arsenic exposure also persisted after adjusting for smoking history. As noted by NRC (1999), an ecological study by Smith et al. (1998) reported standardized mortality ratios (SMRs) for lung and bladder cancer in a region of northern Chile (Region IT) for past exposures to high arsenic concentrations in drinking water. The SMR for chronic obstructive pulmonary disease (COPD) in females in that region was 0.6 (95% CI = 0.4- 0.7~. Since COPC is overwhelmingly due to smoking, it is apparent that the region contained few female smokers relative to the rest of Chile. Yet the female lung cancer SMR in that region was 3.1 (95% CT = 2.7-3.7), and the female bladder cancer SMR was 8.2 (6.3-10.5). As nosed by the authors, there was a large risk of arsenic associated lung and bladder cancer that was not
VARIAB~ITY AND UNCERTAINTY ~ 59 subject to confounding by smoking. A similar finding emerged for males. The high SMRs for Jung cancer and bladder mortality in the region of south- west Taiwan where arsenic is endemic were unlikely to be subject to signifi- cant confounding by smoking or other factors (see description of Tsai et al. (1999) in Chapter 2~. Neither mates nor females in the area where arsenic is endemic had elevated SMRs for emphysema, a cause of death almost entirely attributable to smoking. VALUE-OF-INFORMATION APPROACH Future research might reveal that the manifestations of arsenic toxicity are influenced by gene-environment interactions or that some populations might be at relatively increased risk of developing cancer from exposure to arsenic based on their genetics or their behaviors (e.g., smokers). The implications of this type of variability will raise important policy considerations for risk man- agers, andresearch to support the efforts of decision-makers to deal with these issues should be initiated. In a ~ 996 report, the National Research Council discussed the importance of using an analytic-deliberative process in the context of managing risks (NRC ~ 996~. As noted in the report of the Presidential/Congressional Com- mission on Risk Assessment and Risk Management ~ ~ 997), better risk-man- agement decisions will be made when the process has the capacity for itera- tions "if new information is developed that changes the need for or nature of risk management." A value-of-information approach can be used to decide if it is more appropriate to collect additional information or make a decision based on a risk assessment with specified uncertainties. Given the existing uncertainties that remain for arsenic, EPA should explore whether implement- ing a value-of-information approach to prioritizing research efforts would be helpful in supporting its process for regulating arsenic. As discussed by NRC (1996), Value-of-information methods address whether potential reductions in uncertainty would make a difference in the decision; they suggest priorities among reducible uncertainties on the basis of how much difference the expected reduction might make (p. ~ 10~. Although the decision to implement a value-of-information approach reflects a policy choice, it might help to focus research efforts on those key uncertain- ties that would have the largest impact on our understanding of the magnitude of arsenic risks.
60 ARSENIC IN DRINKING WA TER: 2001 UPDA TE SUMMARY AND CONCLUSIONS The subcommittee commends EPA for considering variability and uncer- tainty in its risk assessment for arsenic and hopes that EPA will continue to refine and update its assessment. New data developed since the 1999 report provide more insight into how sources of variability and uncertainty can be addressed in a risk assessment for arsenic. For example, variability in drin~ng-water rates per unit of body weight can now be quantitatively ana- Tyzed. The subcommittee explores the implications of using that type of infor- mation on variability in risk assessment in Chapter 5. However, there are still a number of key sources of uncertainty that have not been addressed. They include the following: · The mechanisms by which arsenic causes cancer are not well under- stood. It is unclear what the shape of its dose-response curve is at low doses and whether the magnitude of the dose or the duration of exposure is more important in cancer risk. · There is a lack of concordance in the results of animal and human studies. · Geographical and cultural differences in arsenic exposures make it difficult to relate consumption habits in Taiwan with those in the United States. To make better comparisons and account for differences, more infor- mation is needed on arsenic content in foods, water-ingestion rates, food- preparation practices, and nutritional status of the populations. Dietary and nutntional variability between population groups might result in a different proportion of the populations being at risk for arsenic-induced health effects, but the risk will be increased in all those who share the same pattern of expo- sure and susceptibility. · It is unclear whether infants and young children might be more sus- ceptible to arsenic-induced health effects, particularly those for noncancer end points where less-than-lifetime exposures are important and children's greater water consumption per unit of body weight might put them at relatively greater risk. · The metabolism and extent of methylation of arsenic is not fully understood. For example, There is considerable interindividual variation in arsenic metabolism, particularly in the production of MMA and DMA. Some data indicate that methylation is influenced more by genetic factors than by environmental factors. Recent studies, however, confirm the previous findings that methyla- tion of arsenic, especially the second methylation step in which MMA is
VARIABILITYAND UNCERTAINTY 1 61 converted to DMA, is inhibited at higher doses of arsenic. Data also indicate less-efficient methylation of arsenic to DMA in children compared with adults, but those data are not conclusive. . Experimental studies indicate that exposure to other pollutants (e.g., vanadium, selenium, cadmium, mercury, and lead) might inhibit the second arsenic methylation step. Administration of chelating agents, such as DMPS, has been found to decrease the urinary excretion of DMA and increase that of MMA. Arsenic metabolites vary considerably in their toxicity; thus, variation in the metabolism of arsenic is likely to be associated with variations in sus- ceptibility to arsenic. Animal and human data indicate that efficient methyla- tion of inorganic arsenic to DMAV results in faster overall excretion of ar- senic. Also, there is increasing evidence that individuals with increased MMA production and retention (mainly EMMA retain more arsenic and are more prone to at least some toxic effects of arsenic. RECOMMENDATIONS · EPA should explore research opportunities to reduce the key uncer- tainties identified and should support research efforts that might resolve im- portant uncertainties. EPA should also explore using an iterative value-of- information approach to prioritize future research efforts targeted at resolving uncertainties in the arsenic risk assessment. When better data and information are obtained, they should be factored in the risk assessment. · In considering how variability in the amount of drinking water con- sumed would affect quantitative risk assessment for arsenic, EPA should adjust for variability in consumption in the population that was the source of the data (e.g., the southwestern Taiwanese), as well as in the general U.S. population. Key sources of uncertainty that may be subject to substantial reduc- tion include the following: Better understanding of the differences in arsenic exposure between U. S. and Taiwanese populations. This information could be obtained through surveys of the Taiwan drinking-water consumption, water usage for cooking, and body weights. However, it is unclear whether data obtained at the present time could be applied retrospectively to a study population whose key expo- sure occurred decades in the past. Improved characterization of the form and bioavailability of the ar- senic present in the raw foods and of the arsenic incorporated into food from
~ 62 ARSENIC IN DRINKING WA TER: 2001 UPDA TE drinking water during cooking and food preparation. Such characterization will enhance knowledge of valid means to account for this arsenic in quantita- tive risk assessment, which is particularly relevant to EPA's exposure assess- ment efforts. Clarification of the mechanisms of action of arsenic and development of an appropriate animal model for studying the effects of arsenic. Such clarification and development would help to reduce the uncertainty about dose metrics and the relevance of animal data for human risk assessments. · Further characterization ofthe long-term drinking-waterconsumption patterns in the general population of the United States could narrow the cur- rent characterization of variability in drinking-water rate s. However, although it might reduce the overall variability in the estimated risks, it is unlikely to dramatically change the risk estimates. · More information is needed on the variability in the metabolism of arsenic among individuals and the effect ofthat variability on an individuaT's susceptibility to cancer and systemic toxicity. In particular, the impact of variability in MA formation and distribution in human tissues is needed. Factors influencing the variability, genetic as well as environmental, should be studied. · More data are needed to better understand the susceptibility of chil- dren to arsenic-induced toxicity, particularly for noncancer effects, and their arsenic-methylation efficiency. · The influence of nutritional and dietary factors on the risk for arsenic- induced health effects should be subject to further research. REFERENCES Aposhian, H.V., A. Arroyo, M.E. Cebrian, L.M. Del Razo, K.M. Hurlbut, R.C. Dart, D. Gonzalez-Ramirez, H. Kreppel, H. Speisky, A. Smith, M.E. Gonsebatt, P. Ostrosky-Wegman, and M.M. Aposhian. 1997. DMPS-arsenic challenge test. I. Increased urinary excretion of monomethylarsonic acid in humans given dimercaptopropane sulfonate. J. Pharmacol. Exp. Ther. 282~1):192-200. Aposhian, H.V., B. Zheng, M.M. Aposhian, X.C. Le. M.E. Cebrian, W. Cullen, R.A. Zakharyan, M. Ma, R.C. Dart, Z. Cheng, P. Andrewes, L. Yip, G.F. O'Malley, R.M. Maiorino, W. Van Voorhies, S.M. Healy, and A. Titcomb. 2000a. DMPS - Arsenic challenge test. II. Modulation of arsenic species, including monomethylarsonous acid AMMAN, excreted in human urine. Toxicol. Appl. Pharmacol. 165(1):74-83. Aposhian, H.V., E.S. Gurzau, X.C. Le. A. Gurzau, S.M. Healy, X. Lu, M. Ma, L. Yip, R.A. Zakharyan, R.M. Maiorino, R.C. Dart, M.G. Tircus, D. Gonzalez-Ramirez,
VARIABILITYA ND UNCERTAINTY ~ 63 D.L. Morgan, D. Avram, and M.M. Aposhian. 2000b. Occurrence of monomethylarsonous acid in urine of humans exposed to inorganic arsenic. Chem. Res. Toxicol. 13~8~:693-697. Bates, M.N., A.H. Smith, and K.P. Cantor. 1995. Case-control study of bladder cancer and arsenic in drinking water. Am. J. Epidemiol. 141~69:523-530. Calderon, R.L., E. Hudgens, X.C. Le. D. Schreinemachers, and D.J. Thomas. 1999. Excretion of arsenic in urine as a function of exposure to arsenic in Wrong water. Environ. Health Perspect. 107~8~:663-667. Chen, C.J., Y.C. Chuang, T.M. Lin, and H.Y. Wu. 1985. Malignant neoplasms among residents of a blackfoot disease-endemic area in Taiwan: high-arsenic artesian well water and cancers. Cancer Res. 45~11 Pt 2~:5895-5899. Chen, C.J., Y.C. Chuang, S.L. You, T.M. Lin, and H.Y. Wu. 1986. A retrospective study on malignant neoplasms of bladder, lung and liver in blackfoot disease endemic area of Taiwan. Br. J. Cancer 53~3~:399-405. Chen, C.J., M.M. Wu, S.S. Lee, J.D. Wang, S.H. Cheng, and H.Y. Wu. 1988. Atherogenicity and carcinogenicity of high-arsenic artesian well water. Multiple risk factors and related malignant neoplasms of blackfoot disease. Arteriosclero- sis 8~5~:452-460. Chen, C.J., C.W. Chen, M.M. Wu, and T.L. Kuo. 1992. Cancer potential in liver, lung, bladder and kidney due to ingested inorganic arsenic in drinking water. Br. J. Cancer 66~5~:888-892. Chiou, H.Y., Y.M. Hsueh, L.L. Hsieh, L.I. Hsu, Y.H. Hsu, F.I. Hsieh, M.L. Wei, H.C. Chen, H.T. Yang, L.C. Leu, T.H. Chu, C. Chen-Wu, M.H. Yang, and C.J. Chen. 1997. Arsenic methylation capacity, body retention, and null genotypes of glutathione S-transferase M1 and T1 among cu~Tent arsenic-exposed residents in Taiwan. Mutat.Res.386~3~:197-207. Chiou, H.Y., S.T. Chiou, Y.H. Hsu, Y.L. Chou, C.H. Tseng, M.L. Wei, and C.J. Chen. 2001. Incidence of transitional cell carcinoma and arsenic in drinking water: a follow-up study of 8,102 residents in an arseniasis-endemic area in nor~eastern Taiwan. Am.J.Epidemiol.153~5~:411-418. Concha, G., B. Nermell, and M. Vahter. 1998. Metabolism of inorganic arsenic in children with chronic high arsenic exposure in northem Argentina. Environ. Health Perspect. 106~6~:355-359. Concha, G.Q. 2001. Metabolism of Inorganic Arsenic and Biomarkers of Exposure. Doctoral Thesis. Institute of Environmental Medicine, Karolinska Institutet, Stockholm. De Kimpe, J., R. Cornelis, and R. Vanholder. 1999. In vitro methylation of arsenite by rabbit liver cytosol: effect of metal ions, metal chelating agents, methyltransferase inhibitors end uremic toxins. Drug. Chem. Toxicol.22~4~:613- 628. Del Razo, L.M., G.G. Garcia-Vargas, H. Vargas, A. Albores, M.E. Gonsebatt, R. Montero, P. Ostrosky-Wegman, M. Kelsh, end M.E. Cebrian. 1997. Altered profile of urinary arsenic metabolites in adults with chronic arsenicism. A pilot study. Arch. Toxicol. 71 (4~:211 -217.
~64 ARSENICINDRINKING WATER: 2001 UPDATE Del Razo, L.M., G.G. Garcia-Vargas, M.C. Hemandez, Gomez-Munoz, and M.E. Cebrian. 1999. Profile of urinary arsenic metabolites in children chronically exposed to inorganic arsenic in Mexico. Pp. 281-287 in Arsenic Exposure and Health Effects, W.R. Chappell, C.O. Abernathy, and R.L. Calderon, eds. Oxford: Elsevier. Doll, R., and R. Peto. 1981. The causes of cancer: quantitative estimates of avoid- able risk of cancer in the United States today. J. Natl. Cancer Inst. 66~6~: 1191- 1308. EPA (U.S. Environmental Protection Agency). 1988. Special Report on Ingested Inorganic Arsenic: Skin Cancer; Nutritional Essentiality. EPA/625/3-87/013. Risk Assessment Forum, U.S. Environmental Protection Agency, Washington, DC. July 1988. EPA (U.S. Environmental Protection Agency). 1997. Report of the Expert Panel on Arsenic Carcinogenicity: Review and Workshop. Prepared by Eastern Research Group, Lexington, MA, for the U.S. Environmental Protection Agency, National Center for Environmental Assessment, Washington, DC. EPA (U.S. Environmental Protection Agency). 2000a. 40 CFR Parts 141 and 142. National Primary Drinldug Water Regulations. Arsenic and Clarifications to Compliance and New Source Contaminants Monitoring. Notice of proposed rulemaking. Fed. Regist. 65~121~:38887-38983. EPA (U.S. Environmental Protection Agency). 2000b. Estimated Per Capita Water Ingestion in the United States: Based on Data Collected by the United States Department of Agriculture's (USDA) 1994-1996 Continuing Survey of Food Intakes by Individuals. EPA-822-00-008. Office of Water, Office of Standards and Technology, U.S. Environmental Protection Agency. EPA (U.S. Environmental Protection Agency). 2001. 40 CFR Parts 9,141 and 142. National Primary Drinking Water Regulations. Arsenic and Clarifications to Compliance end New Source Contaminants Monitoring. Final Rule. Fed. Regist. 66(14): 6975-7066. Ershow, A.G., L.M. Brown, and K.P. Cantor. 1991. Intake of tapwater and total water by pregnant and lactating women. Am. J. Public Health 81~3~:328-334. Ferreccio, C., C. Gonzalez, V. Milosavljevic, G. Marshall, A.M. Sancha, and A.H. Smith. 2000. Lung cancer and arsenic concentrations in drinking water in Chile. Epidemiology 11~6~:673-679. Hertz-Picciotto, I., A.H. Smith, D. Holtzman, M. Lipsett, and G. Alexeeff. 1992. Synergism between occupational arsenic exposure and smoking in the induction of lung cancer. Epidemiology 3 ~ 1 ~ :23 -31. Hopenhayn-Rich, C., M.L. Biggs, and A.H. Smith. 1998. Lung and kidney cancer mortality associated with arsenic in drinking water in Cordoba, Argentina. Int. J. Epidemiol. 27(4):561-569. Hsu, K.H., J.R. Froines, and C.J. Chen. 1997. Studies of arsenic ingestion from drinking-water in northeastern Taiwan: chemical speciation and urinary metabo- lites. Pp.190-209 in Arsenic Exposure and Health Effects, C.O. Abernathy, R.L. Calderon, and W.R. Chappell, eds. London: Chapman & Hall.
VARIABILITYAND UNCERTAINTY ~ 65 Hsueh, Y.M., H.Y. Chiou, Y.L. Huang, W.L. Wu, C.C. Huang, M.H. Yang, L.C. Lue, G.S. Chen, and C.J. Chen. 1997. Serum beta-carotene level, arsenic methylation capability, and incidence of skin cancer. Cancer Epidemiol. Biornarkers Prev. 6~8~:589-596. Hsueh, Y.M., Y.L. Huang, C.C. Huang, W.L. Wu, H.M. Chen, M.H. Yang, L.C. Lue, and C.J. Chen. 1998. Urinary levels of inorganic and organic arsenic metabolites among residents in an arseniasis-hyperendemic area in Taiwan. J. Toxicol. Envi- ron. Health A 54~6~:431-444. Hsueh, Y.M., G.S. Cheng, M.M. Wu, H.S. Yu, T.L. Kuo, and C.J. Chen. 1995. Multiple risk factors associated with arsenic-induced skin cancer: effects of chronic liver disease and malnutritional status. Br. J. Cancer 71~1~: 109-114. ILSI (International Life Sciences Institute). 1992. Similarities and Differences Be- tween Children and Adults: Implications for Risk Assessment, P.S. Guzelian, C.J. Henry, and S.S. Olin, eds. Washington, DC: ILSI. IOM (Institute of Medicine). 2001. Dietary Reference Intakes for Vitamin A, Vita- min K, Arsenic, Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybde- num, Nickel, Silicon, Vanadium, end Zinc. Washington, DC: National Academy Press. Kenyon, E.M., M.F. Hughes, and O.A. Levander. 1997. Influence of dietary sele- nium on the disposition of arsenate in the female B6C3F1 mouse. J. Toxicol. Environ. Health 51~3~:279-299. Kurttio, P., H. Komulainen, E. Hakala, H. Kahelin, and J. Pekkanen. 1998. Urinary excretion of arsenic species after exposure to arsenic present in ~inking water. Arch. Environ. Contam. Toxicol.34~3~:297-305. Le. X.C., M. Ma, X. Lu, W.R. Cullen, H.V. Aposhian, X. Lu, and B. Zheng. 2000. Determination of monomethylarsonous acid, a key arsenic methylation intermedi- ate, in human urine. Environ. Health Perspect. 108~11~:1015-1018. Maki-Paakkanen, J., P. Kurttio, A. Paldy, and J. Pekkanen. 1998. Association be- tween the clastogenic effect in peripheral lymphocytes and human exposure to arsenic through drinking water. Environ. Mol. Mutagen.32~4~:301-313. Mandal, B.K., Y. Ogra, and K.T. Suzuki. 2001. Identification of dimethylarsinous and monomethylarsonous acids in human urine of the arsenic-affected areas in West Bengal, India. Chem. Res. Toxicol. 14~4~:371-378. Mazumder, D.N.G., R. Haque, N. Ghosh, B.K. De, A. Santra, D. Charkraborty, and A.H. Smith. 1998. Arsenic levels in drinking water and the prevalence of skin lesions in West Bengal, India. Int. J. Epidemiol. 27~5~:871-877. Morales, K.H., L. Ryan, T.L. Kuo, M.M. Wu, and C.J. Chen. 2000. Risk of internal cancers from arsenic in drinking water. Environ. Health Perspect. 108~7~655- 661. Mushak, P., and A.F. Crocetti. 1995. Risk and revisionism in arsenic cancer risk assessment. Environ. Health Perspect. 103~7-8~:684-689. Newman, P.E. 1999. Can reduced folic acid and vitamin B 12 levels cause deficient DNA methylation producing mutations which initiate atherosclerosis? Med. Hypotheses 53~5~:421-424.
~ 66 ARSENIC IN DRINKING WA TER: 2001 UPDA TE NRC (National Research Council). 1982. Diet, Nutrition and Cancer. Washington, DC: NationalAcademy Press. NRC (National Research Council). 1993. Pesticides in the Diets of Infants and Children. Washington,DC: NationalAcademy Press. NRC (National Research Council). 1994. Science and Judgment in Risk Assessment. Washington, DC: NationalAcademy Press. NRC (National Research Council) 1996. Understanding Risk: Informing Decisions in a Democratic Society. Washington,DC: NationalAcademy Press. NRC (National Research Council). 1999. Arsenic in Drinking Water. Washington, DC: National Academy Press. The Presidential/Congressional Commission on Risk Assessment and Risk Manage- ment. 1997. Framework for Environmental Health Risk Management. Final Report. Vol. 1. Washington, DC: The Commission. Ruan, Y., M.H. Peterson, E.M. Wauson, J.G. Waes, R.H. Finnell, and R.L. Vorce. 2000. Folic acid protects SWV/Fnn embryo f~broblasts against arsenic toxicity. Toxicol. Lett. 117~3~: 129-137. Schoof, R.A., L.J. Yost, J. Eickhoff, E.A. Crecelius, D.W. Cragin, D.M. Meacher, and D.B. Menzel. 1999. A market basket survey of inorganic arsenic in food. Food Chem. Toxicol. 37~8~:839-846. Smith, A.H., A.P. Arroyo, D.N. Mazumder, M.J. Kosnett, A.L. Hemandez, M. Beeris, M.M. Smith, and L.E. Moore. 2000. Arsenic-induced skin lesions among Atacameno people in Northern Chile despite good nutrition and centuries of exposure. Environ. Health Perspect. 108~7~: 617-620. Smith, A.H., M. Goycolea, R. Haque, and M.L. Biggs. 1998. Marked increase in bladder and lung cancer mortality in a region of Northern Chile due to arsenic in drinking water. Am. J. Epidemiol. 147~7~:660-669. Smith, A.H., C. Hopenhayn-Rich, M.N. Bates, H.M. Goeden, I. Hertz-Picciotto, H.M. Duggan, R. Wood, M.J. Kosnett, and M.T. Smith. 1992. Cancer risks from arsenic in drinking water. Environ. Health Perspect. 97:259-267. Steinmetz, K.A., and J.D. Potter. 1991 a. Vegetables, fruit, and cancer. I. Epidemiol- ogy. Cancer Causes Control 2~6~:325-357. Steinmetz, K.A., and J.D. Potter. 1991b. Vegetables, fruit, and cancer. II. Mecha- nisms. Cancer Causes Control 2~69:427-442. Styblo, M., and D.J. Thomas. 2001. Selenium modif~es the metabolism and toxicity of arsenic in primary rat hepatocytes. Toxicol. Appl. Pharmacol. 172~1~:52-61. Styblo, M., L.M. Del Razo, E.L. LeCluyse, G.A. Hamilton, C. Wang, W.R. Cullen, andD.J. Thomas. 1999. Metabolismofarsenic in primary cultures of human and rat hepatocytes. Chem. Res. Toxicol. 12~7~:560-565. Styblo, M., L.M. Del Razo, L. Vega, D.R. Germolec, E.L. LeCluyse, G.A. Hamilton, W. Reed, C. Wang, W.R. Cullen, and D.J. Thomas. 2000. Comparative toxicity oftrivalent and pentavalent inorganic and methylated arsenicals in rat and human cells. Arch. Toxicol. 74~69:289-299.
VARIABILI7YAND UNCERTAINTY ~ 67 Su, J.H., Y.L. Guo, M.D. Lai, J.D. Huang, Y. Cheng, and D.C. Christiani. 1998. The NAT2* slow acetylator genotype is associated with bladder cancer in Taiwanese, but not in the Black Foot Disease endemic area population. Pharmacogenetics 8~2~: 187-190. Thompson, K.M., and J.D. Graham. 1996. Going beyond the single number: using probabilistic risk assessment to improve risk management. Hum. Ecol. Risk Asses. 2~4~: 1008-1034. Tao, S.S., and P.M. Bolger. 1999. Dietary arsenic intakes in the United States: FDA Total Diet Study, September 1991-December 1996. Food Addit. Contain. 16(11):465-472. Tsai, S.M., T.N. Wang, and Y.C. Ko. 1999. Mortality for certain diseases in areas with high levels of arsenic in drinking water. Arch. Environ. Health 54~3~: 186- 193. Tseng, W.P. 1989. Blackfoot disease in Taiwan: a 30-year follow-up study. Angiol- ogy 40(6):547-558. Tseng, W.P., H.M. Chu, S.W. How, J.M. Fong, C.S. Lin, and S. Yeh. 1968. Preva- lence of skin cancer in an endemic area of chronic arsenicism in Taiwan. J. Natl. Cancer Inst. 40~34:453-463. Tsuda, T., A. Babazono, E. Yamamoto, N. Kurunutani, Y. Mino, T. Ogawa, Y. Kishi, and H. Aoyama. 1995. Ingested arsenic and internal cancer: a historical cohort study followed for 33 years. Am. J. Epidemiol. 141~3~:198-209. Vahter, M. 1999a. Methylation of inorganic arsenic in different ma~runalian species and population groups. Sci. Prog. 82(Pt 1~:69-88. Vahter, M. l999b. Variation in human metabolism of arsenic. Pp. 267-279 in Ar- senic Exposure and Health Effects, W.R. Chappell, C.O. Abernathy, and R.L. Calderon, eds. Oxford: Elsevier. Vahter, M. 2000. Genetic polymorphism in the biotransformation of inorganic ar- senic and its role in toxicity. Toxicol. Lett. 112-113:209-217. Vahter, M. 2001. Mechanisms of arsenic biotransformation. Toxicology. In Press. Valentine, J.L., H.K. Kang, and G. Spivey. 1979. Arsenic levels in human blood, urine, and hair in response to exposure via drinking water. Environ. Res. 20~1~:24-32. Warner, M.L., L.E. Moore, M.T. Smith, D.A. Kalman, E. Fanning, and A.H. Smith. 1994. Increased micronuclei in exfoliated bladder cells of individuals who chron- ically ingest arsenic-contaminated water in Nevada. Cancer Epidemiol. Biomark- ers Prev. 3~7~:583-590. Weinshilboum, R.M., D.M. Otterness, and C.L. Szumlanski. 1999. Methylation pharmacogenetics: catecol O-methyltransferase, thiopurine methyltransferase, and histamine N-methyltransferase. Annul Rev. Pharmacol. Toxicol.39: 19-52. Willett, W. 1990. Nutritional Epidemiology. New York: Oxford University Press. World Cancer Research Fund Panel. 1997. Food, Nutrition and the Prevention of Cancer: A Global Perspective. Washington, DC: World Cancer Research Fund Panel, American Institute for Cancer Research. Wu, M.M., T.L. Kuo, Y.H. Hwang, and C.J. Chen. 1989. Dose-response relation
68 ARSENIC IN DRINKING WA TER: 2001 UPDA TE between arsenic concentration in well water and mortality from cancers and vascular diseases. Am. J. Epidemiol. 130~61:1123-1132. Yu, R.C., K.H. Hsu, C.J. Chen, and J.R. Froines. 2000. Arsenic methylation capacity and skin cancer. Cancer Epidemiol. Biomarkers Prev. 9~11~:1259-1262.