Click for next page ( 2


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 1
Summary The U.S. Environmental Protection Agency (EPA) is required under the Safe Drinking Water Act (SDWA) to establish the concentrations of contaminants permitted in public drinking-water supplies. The SDWA requires EPA to set two specific concentrations for each designated contaminant in drinking waterthe maximum contaminant level goal (MCkG) and the maximum contaminant level (MCL). The MUG is a health goal to be based on the best available, peer-reviewed scientific data. It is to be set at a concentration at which no known or anticipated adverse health effects occur, allowing for adequate margins of safety. The MCL`G is not a regulatory requirement and might not be attainable with current technology or analytical methods. In contrast, the MCL is an enforceable standard that is required to be set as close to the MING as is technologically feasible, taking cost into consideration. Following the 1976 enactment of the SDWA, EPA proposed, as part of the National Interim Primary Drinking Water Standards, an interim MCL of 50 micrograms per liter (high) for arsenic. The U.S. Public Health Service originally set the 50-,ug/L standard in 1942. In 198S, EPA conducted a risk assessment for arsenic in drinking water and, in 1996, requested that the National Research Council (NRC), the operating arm of the National Acad- emy of Sciences and the National Academy of Engineering, independently review the scientific database and evaluate the scientific validity of that risk assessment. In response to that request, the NRC published Arsenic in Drirzk- ing Water in 1999. Following that report, EPA proposed an arsenic standard 1

OCR for page 1
ARSENICIN DRINKING WATER: 2001 UPDATE of 5 ~g/L in the Federal Register. After review by EPA's Science Advisory Board (SAB) and a period of public comment, EPA issued a pending standard of 10 ~g/L on January 22, 2001. That pending standard was primarily based on dose-response models and extrapolation from a cancer study of a Taiwan- ese population exposed to high concentrations of arsenic in its drinking water. On March 23, 2001, EPA published a notice that delayed the effective date of the arsenic rule pending further study of options for revising the MCL for arsenic. To incorporate the most recent scientific research into its decision, EPA's Office of Water subsequently requested that the NRC independently review studies on the health effects of arsenic published since the ~ 999 NRC report. CHARGE TO THE SUBCOMMITTEE In response to EPA's request, the NRC assigned the project to the Com- mittee on Toxicology (COT) and convened the Subcommittee to Update the ~ 999 Arsenic in Drinking Water Report. The members selected by the NRC to serve on this subcommittee have expertise in epidemiology, cellular and molecular toxicology, biostatistics and modeling, risk assessment, uncertainty analyses, and public health. Five of the nine members of the subcommittee also served on the earlier NRC Subcommittee on Arsenic in Drinking Water. The 2001 subcommittee was charged with the task of preparing a report up- dating the scientific analyses, uncertainties, and findings of the 1999 report on the basis of relevant toxicological and health-effects studies published and relevant data developed since the ~ 999 NRC report and to evaluate the analy- ses subsequently conducted by EPA in support of its regulatory decision- making for arsenic in drinking water. The subcommittee was charged and constituted to address only scientific topics relevant to toxicological risk and health effects of arsenic. It was not asked to address questions of economics, cost-benefit assessment, control technology, exposure assessment in U.S. populations, or regulatory decision-making. The subcommittee performed the following tasks in response to its charge: Determine whether data from the 1988, 1989, and 1992 Taiwanese studies remain the best data for dose-response assessment and risk estimation. Assess whether the EPA analysis appropriately incorporates popula-

OCR for page 1
SUMMARY 3 tion differences, including diet, when extrapolating from the Taiwanese study population to the U.S. population. Evaluate whether the dose-response analysis conducted by EPA and any other available analyses of more recent data are adequate for estimating an effective dose for a ID response (EDo~. ~ Determine whether EPA's analysis appropriately considers and char- acterizes the available data on the mode of action of arsenic and the informa- tion on dose-response relationships and uncertainties when assessing the public-health impacts. Determine whether EPA's risk estimates at 3, 5, ~ 0, and 20 ,ug/L are consistent with available scientific information, including information from new studies. TO SUBCOMMITTEE'S APPROACH TO ITS CEIARGE The subcommittee considered several hundred new scientific articles on arsenic published since the ~ 999 NRC report. It also heard presentations from the EPA administrator; other EPA representatives; the EPA Science Advisory Board; other scientists with expertise in arsenic toxicity; federal, state, and local government agencies; trade organizations; public-interest groups; and concerned individuals. The subcommittee evaluated the arsenic hazard assessment conducted by EPA for the pending arsenic standard published in the January 22, 2001, Federal Register and considered the comments made in the EPA Science Advisory Board's December 2000 report on the previously proposed rule. The subcommittee was not asked to assess U.S. population exposures. It addressed scientific issues concerning the hazards from consumption of drink- ing water contaminated with arsenic. It did not comment or make recommen- dations on risk management or policy decisions. By definition, determining an MCL requires policy considerations, including risk-management options and cost-benefit analyses, which are beyond the scope of the charge to this subcommittee. It should also be noted that the NRC was charged with updating the ~ 999 report Arsenic in Drinking Water, not with reviewing its own report. There- fore, the subcommittee has taken that report as a starting point in its evalua- tion of more recent information.

OCR for page 1
4 ARSENIC IN DRINKING WA TER: 2001 UPDA TE TlU: SUBCOMAIITTEE'S EVALUATION Epidem~ological (Human) Studies The 1999 NRC report concluded that arsenic is associated with both cancer and noncancer effects. At that time, there was sufficient evidence to conclude that ingestion of arsenic in drinking water causes skin, bladder, and Jung cancer. The internal cancers (bladder and lung) were considered to be the main cancers of concern, and there was sufficient evidence from large epidemiology studies in southwestern Taiwan of a dose-response relationship between those cancers and exposure to arsenic in drinking water. Since the publication of the 1999 report, evidence has increased that chronic exposure to arsenic in drinking water might also be associated with an increased risk of high blood pressure and diabetes. Pending fisher re- search that characterizes the dose-response relationship for high blood pres- sure and diabetes, the magnitude of possible risk that exists at low levels is not quantifiable. Nevertheless, even small increases in relative risk for these conditions at low dose could be of considerable public-health importance. This potential impact should be qualitatively considered in the risk-assess- ment process. Some evidence also published since the 1999 NRC report shows an association between arsenic ingestion end potentially adverse repro- ductive outcomes and noncancer respiratory effects. However, those data require confirmation. Four major epidemiological studies have been published since the 1999 NRC report in which the association between intemal cancers and arsenic ingestion in drinking water has been investigated. The data from three of those studies (one in Chile, one in northeastern Taiwan, and one in southwest- em Taiwan) confirm the association between intemal cancers and arsenic exposure through drinking water. Another study (in Utah) did not demon- strate such an association. The strengths of the recent studies from Chile and northeastern Taiwan include the evaluation of some potential confounding factors affecting the observed association between arsenic ingestion and cancer in newly diag- nosed cases. Although the recent study in southwestern Taiwan is limited in its exposure assessment, it addresses the issue of lifestyle differences (e.g., diet, smoking) that might have influenced mortality rates in the area where arsenic is endemic. In that study, cancer rates in the area of southwestern Taiwan where arsenic is endemic were compared with cancer rates in counties

OCR for page 1
SUMMARY 5 neighboring the area (where the lifestyle is similar) and with rates for all of Taiwan. The arsenic-related risk estimates based on the two different com- parison populations did not differ substantially, indicating that lifestyle differ- ences between the region of southwestern Taiwan where arsenic is endemic and the rest of Taiwan do not substantially affect estimates of the risk of cancers from ingesting arsenic in drinking water. The study in Utah was the first large-scare study to attempt to consider the association between internal cancers (bladder and lung) and arsenic exposure through drinking water in a U.S. population. However, the subcommittee concluded that the limitations of the Utah study currently preclude its use in a quantitative risk assessment. One limitation was the unconventional method used in that study to characterize exposure. Furthermore, in contrast to the southwestern Taiwan study where lifestyle differences do not appear to influ- ence relative risk of cancer from arsenic in drinking water, the Utah study used a comparison group with differences in lifestyle characteristics from the study population. The study population was composed of individuals with religious prohibitions against smoking, and the unexposed comparison group was the overall population of Utah, where such religious prohibitions are not practiced by all residents. The other recent studies of arsenic in humans, taken together with the many studies discussed in the 1999 NRC report, provide a sound and suffi- cient database showing an association between bladder and lung cancers and chronic arsenic exposure in drinking water, and they provide a basis for quan- titative risk assessment. The subcommittee concludes that the early data from southwestern Taiwan remain appropriate for use in dose-response assessment of arsenic in Winking water. In addition, recent studies increase the weight of evidence for an association between internal cancers and arsenic exposure through drinking water. In particular, data from northern Chile on risk of lung cancer incidence are also appropriate for use in a quantitative risk assessment. Metabolism and Mode-of-Action Studies When evaluating the hazards from arsenic in drinking water, it is impor- tant to evaluate data on the fate of arsenic in the body (i.e., its metabolism) and how it causes its adverse effects (i.e., its mode of action). Arsenic is metabolized in the body by reduction and methylation reactions. The main product of those reactions, dimethylarsinic acid, is readily excreted from the

OCR for page 1
6 ARSENIC IN DRINKING WA TER: 2001 UPDA TE body in the urine, but recent data indicate that reactive and toxic intermediate metabolites may be distributed to tissues and excreted in urine. The mecha- nisms responsible for the adverse effects associated with arsenic, including some types of cancer, cardiovascular disease, and diabetes, probably occur through multiple independent and interdependent mechanisms. The shape of the dose-response curve for one type of adverse effect might have little rele- vance to the shape for a different effect. Likewise, the shape of the dose- response curve for disruption of a specific biochemical pathway by arsenic is not necessarily relevant to the overall shape of the dose-response curve for a complex disease process, such as tumor development following chronic expo- sure. Biostatistical approaches are requ*ed in a dose-response assessment to extrapolate from the lowest concentrations of arsenic at which increases in cancer are observed in a study population to lower concentrations to which the study population of interest is exposed. The mode of action by which a chemical causes cancer can sometimes determine how human or animal data should be extrapolated and used to evaluate allowable drinking-water contam- inant concentrations. If an agent acts directly to cause DNA damage, it is standard practice for the estimated risk ofcancerto tee extrapolated in a linear fashion from the lowest measured exposure to zero (i.e., below the range of observations, risk is assumed to be directly proportional to the exposure.) If an agent acts indirectly, the possibility of sublinear extrapolation is consid- ered (i.e., such extrapolation has sometimes been interpreted to indicate a "threshold" for effects.) In the absence of definitive mode-of-action data, EPA's general policy is to use a linear extrapolation from the observed data range for its carcinogenic risk assessments. After concluding that the mode- of-action data were inadequate to define the shape of the curve, EPA made a policy-based decision to use a default assumption of linearity. Although a large amount of research is available on arsenic's mode of action, the exact nature of the carcinogenic action of arsenic is not yet clear. Therefore, the subcommittee concludes that the available mode-of-action data on arsenic do not provide a biological basis for using either a linear or nonlin- ear extrapolation. Furthermore, in laboratory studies, cellular effects of ar- senic occur at concentrations below those found in the urine of people who had ingested Winking water with arsenic at concentrations as low as 10 Egg. Therefore, even if the curve is sublinear at some point (e.g., if a threshold exists), the available data showing cellular effects at arsenic concentrations in the range ofthose measured in U.S. populations suggest that anyhypotheti-

OCR for page 1
SUMMARY 7 cal threshold would likely occur below concentrations that are relevant to U.S. populations. Variability and Uncertainty in an Arsenic Risk Assessment Variability (differences in outcomes due to factors contributing to risk) and uncertainty (resulting from lack of knowledge in the underlying science) should be considered in an arsenic risk assessment. Differences in the expo- sures of individuals and populations and differences in responses to a given exposure result in variability in a response. Often, that variability can be measured and quantified, but in many cases, assumptions must be made about many of the variables when information is lacking. Sources of variability in an arsenic risk assessment include exposure differences in subpopulations (e.g., infants and children), and variability in arsenic metabolism. Individual exposures to arsenic can be affected by a number of factors, particularly the variability in the amount of arsenic in drinking water, water-ingestion rates, arsenic content in different foods, food- consumption rates, and other characteristics ofthe exposed population, such as sex, age, and body weight. EPA made assumptions with regard to intake of drinking water (including that for cooking) and arsenic through food to account for difference between southwestern Taiwan and the United States when estimating its risks. The basis for those assumptions, however, is not clear and adds to the uncertainty in the risk estimates. It has been argued that poor nutrition might make the Taiwanese popula- tion more susceptible to the effects of arsenic than the U.S. population and that generalizing from the Taiwan population to other populations with differ- ent diets and, possibly, nutritional status is inappropriate. However, the sub- committee concludes that there is no evidence of nutntional factors that could account for the high rate of cancer seen in the arsenic-exposed Taiwanese population. Furthermore, similar increases in risk have been associated with chronic arsenic exposure in many other countnes, including Chile and Argen- tina, where poor nutrition and low-protein diets are not issues. Therefore, the subcommittee concludes that the risk estimates based on the southwestern Taiwanese data are not substantially affected by differences in nutritional status or diet. The subcommittee evaluated data to determine whether there is evidence that infants and children are more susceptible than adults to the effects of

OCR for page 1
~ ARSENICINDRINKING WATER: 2001 UPDATE arsenic. There are no reliable data that indicate heightened susceptibility of children to arsenic. The subcommittee agrees that infants and children might be at greater risk for cancer and noncancer effects because of greater water consumption on a body-weight basis. However, cancer remains the health end point of concern, and the lifetime cancer risk estimates account for the greater childhood exposures by deriving risk estimates from epidemiology studies of cancer among populations exposed to arsenic since birth, as was the case for most of the populations in which the association between arsenic and cancer was studied. Considerable variability in metabolism of arsenic in humans is reflected, in part, by differences in the pattern of excreted arsenic metabolites in the urine. Because arsenic metabolites differ in their toxicity, variation in the metabolism of arsenic is likely to be associated with variations in susceptibil- ity to arsenic. Genetic factors, age, the dose of arsenic received, and simulta- neous exposure to other compounds, such as m~cronutrients, appear to be important considerations in arsenic metabolism. The fact that the metabolism of arsenic varies markedly between individuals should be considered in an arsenic risk assessment; however, at the present time it is uncertain how to account for that variability in a quantitative dose-response analysis. The method used to characterize arsenic dose in a study is a source of uncertainty in arsenic dose-response assessment. The measurement of dose (e.g., cumulative exposure, lifetime average exposure, or peak exposure) that is most closely correlated with cancer outcomes is not well established. If an incorrect measurement of dose is used, then the relationship between dose and effect might be obscured. The choice of the dose measurement affects the interpretation of an epidem~ological study end the choice ofthe dose-response model. Smoking is a well-recognized risk factor for lung and bladder cancer, the two internal cancers mostly strongly associated with arsenic ingestion. There are no data available to indicate that smoking is a significant confounder of the observed association between exposure to arsenic in drinking water and an increase in lung or bladder cancer. However, several of the epidemiologi- cal studies reviewed by the subcommittee suggest the possibility of an interac- tion between smoking and arsenic on the risk of lung cancer or bladder can- cer, but this potential effect requires further confirmation and characteriza- tion. If an interaction between smoking and arsenic were to exist, then differ- ences in smoking prevalence between populations might influence the impact of using relative risks from one population to derive risk estimates in another population. The direction of this impact could be in either direction, that is,

OCR for page 1
SUMMARY 9 it could theoretically either increase or decrease the risk estimates, depending on the relative smoking prevaTences. Quantitative Evaluation of Arsenic Toxicity For the southwestern Taiwanese study, risks can be estimated either by comparing cancer mortality in the human study population exposed to arsenic with cancer mortality in the general Taiwanese or the regional population (i.e., a mostly unexposed external comparison group) or by making compari- sons within the study group between high- and low-exposed individuals (i.e., internal comparison group). The approach of using an unexposed external comparison population is classically used in the analysis of data similar to those available from Taiwan and has the advantage of minimizing exposure misclassification (e.g., classifying low-exposed individuals in the study popu- lation as unexposed). A potential disadvantage of using an external compari- son group is that the analysis can be biased if the study population differs from the comparison population in important ways. Because of concerns about differences between the unexposed external comparison population and the study population in southwestern Taiwan, EPA used an internal compari- son population in its dose-response assessment. As discussed above, how- ever, results of a recent study in southwestern Taiwan indicate that differ- ences in lifestyle factors between the region of southwestern Taiwan where arsenic is endemic and the rest of Taiwan do not appear to affect the risk of cancer from arsenic in drinking water. Therefore, the subcommittee derived its estimates of cancer risk by comparing the arsenic-exposed southwestern Taiwanese population with an external population, and it recommends that approach for arsenic risk assessments. The subcommittee estimated EDIT values (i.e., the exposure dose at which there is a ID response in the study population) for various studies using sev- eral different types of statistical models. The estimated EDIT values from the Chilean study on lung cancer ranged from 5 to 27 vigil, depending on the exposure data used. The EDIT values estimated for the southwestern Taiwan- ese study ranged from 33 to 94 ,ug/L for lung cancer, and from 102 to 443 ,ug/L for bladder cancer, depending on the choice of statistical model. The previous NRC Subcommittee on Arsenic in Drinking Water estimated EDIT values for male bladder cancer mortality of 404 to 450 ~g/L, depending on the model used. Those values are approximately within the range of EDIT values estimated by this subcommittee. However, because the EDIT values reported

OCR for page 1
~ O ARSENIC IN DRINKING WA TER: 2001 UPDA TE by the previous and current NBC subcommittees were derived through differ- ent biostatistical approaches, they are not directly comparable. The EDGE values in the 1999 NRC report reflect a IN increase relative to background cancer mortality in Taiwan, whereas the current subcommittee's approach reports EDGE values based on a IN increase relative to the background cancer mortality in the United States. The differences between these two approaches are discussed in a later section. The subcommittee investigated the extent ofthe variability among differ- ent types of statistical models using a model-weighting approach and also assessed the impact of differences in background incidence rates between different populations when using relative risks in a risk assessment. In addi- tion, statistical analyses were conducted to investigate the sensitivity of the resulting risk estimates to differences in water intakes and measurement error. Research Needs More research is needed on the possible association between arsenic exposure and cancers other than skin, bladder, and lung, as well as noncancer effects, particularly impacts on the circulatory system (high blood pressure, heart disease, and stroke), diabetes, end reproductive outcomes. Future stud- ies of the relationships between arsenic ingestion and both noncancer and cancer outcomes should be designed to have sufficient power to determine risks in potentially susceptible subpopulations, including children; they should consider factors (e.g., smoking, diet, genetics) that could influence susceptibility to arsenic; and they should collect detailed exposure informa- tion, all in an effort to reduce uncertainty in the risk assessment. In addition, more information is needed on the variability in metabolism of arsenic among individuals and the effect of that variability on an arsenic risk assessment. Laboratory and clinical research is also needed to define the mechanisms by which arsenic induces cancer to cIanfy the risks at lower doses. OVERALL CONCLUSIONS There is a sound database on the carcinogenic effects of arsenic in hu- mans that is adequate for the purposes of a risk assessment. The subcommit- tee concludes that arsenic-induced internal (lung and bladder) cancers should continue to be the principal focus of arsenic risk assessment for regulatory

OCR for page 1
SUMMARY I I decision making, as discussed and as recommended in the 1999 NRC report. The human data from southwestern Taiwan used by EPA in its risk assess- ment remain the most appropriate for determining quantitative lifetime cancer risk estimates. Human data from more recent studies cited in this report, especially those from Chile, provide additional support for the risk assess- ment. In view of new data from southwestern Taiwan, the subcommittee recommends using an external comparison population, rather than high- and low-exposure groups within the exposed population, when analyzing the earlier studies from southwestern Taiwan. The observed data should be ana- {yzed, using a model that is biologically plausible and provides a reasonable statistical fit to the data. For the southwestern Taiwanese cancer data, this mode! is the additive Poisson mode] with a linear term used for dose. The available data on the mode of action of arsenic do not indicate what form of extrapolation (linear or nonlinear3 should be used below arsenic concentra- tions at which cancers have been observed in human studies. As discussed previously, there are no experimental data to indicate the concentration at which any theoretical threshold might exist. Therefore, the curve should be extrapolated linearly from the EDGE to determine risk estimates for the poten- tial concentrations of concern (3, 5, 10, and 20 ~g/~3. The choice for the shape of the dose-response curve below the EDDY is, in part, a policy decision. It should be noted, however, that the Taiwanese and other human studies include data on exposures at arsenic concentrations relatively close to some U.S. exposures. Consequently, the extrapolation is over only a relatively small range of arsenic concentrations. The uncertainty associated with the assumptions in the analyses was discussed earlier. The subcommittee's estimates of theoretical lifetime excess risk of lung cancer and bladder cancer for U.S. populations at different concentrations of arsenic in drinking water are presented in Table ESPY. These are maximum- likelihood (central-point) risk estimates, not upper-bound (worst-case) esti- mates. Because a relative risk approach using data from Taiwan and Chile was used to projectrisks in the U.S. population, differences in the background rate ofthe disease can have an important impact on the overall risk estimate. The background incidence of Jung or bladder cancer in Taiwan is lower than that in the United States; therefore, the projected risk estimates for those cancers will also be lower in Taiwan than in the United States. The corresponding risks estimated using Taiwanese background cancer rates would be approxi- mately 2-fold lower for female bladder cancer, 3-fold lower for male bladder cancer, 3-fold Tower for female lung cancer, and 2-fold lower for mate lung

OCR for page 1
12 ARSENIC IN DR1rNK[NG WA TER: 2001 UPDA TE TABLE S-1 Theoretical Maxi Likelihood Estimates of Excess Lifetime Risk (Incidence per 10,000 People) of Lung Cancer and Bladder Cancer for U.S. Popula- tions Exposed at Various Concentrations of Arsenic In Drinking Waterb C Arsenic Bladder Cancer Concentration Lung Cancer Females Males Females Males ~- 3 4 7 5 4 5 6 11 9 7 10 12 23 18 14 20 24 45 36 27 a The maximum-likelihood estimate is the central point estimate from the distribution of risk calculated using a particular statistical model and data set (see note b). b Estimates were calculated using data from individuals in the region of southwestern Taiwan where arsenic is endemic, data from an external comparison group from the overall southwest- ern Taiwan area, and U.S. age-adjusted cancer incidence data. The risks are estimated using what the subcommittee considered reasonable assumptions: a U.S. resident weighs 70 kg, compared with 50 kg for the typical Taiwanese, and the typical Taiwanese drinks just over 2 liters of water per day, compared with 1 liter per day in the United States; therefore, it assumes that the Taiwanese exposure per kilogram of body weight is approximately 3 times that of the United States. It is possible to get higher and lower estimates using other assumptions. Risk estimates are rounded to the nearest integer. All 95% confidence limits are less than +12% of the maximum-likelihood estimate and are not presented. Those confidence limits reflect statisti- cal variability in the population incidence estimates only, a narrow range that primarily reflects the relatively large sample size ofthe date modeled. As such, they are not indicative ofthe true uncertainty associated with the estimates. cIf Taiwanese baseline cancer rates are used instead of U.S. data to estimate the risk, the corre- sponding risk estimates (incidence per 10,000) for arsenic at concentrations of 3, 5, 10, and 20 ~g/L of drinking water are as follows: female bladder cancer, 2, 4, 8, and 15; male bladder cancer, 2, 3, 7, and 13; female lung cancer, 2, 3, 6, and 12; and male lung cancer, 2, 3, 6, arid 11. cancer (see Table ES-I, footnote c). It should tee nosed that standard epidemi- ological practices support the use of the background incidence rate in the country of interest when comparing relative risks across different populations. However, the subcommittee members are divided in opinion on whether using the U.S. background cancer incidence rate was preferable to using the Tai- wanese background rate; some members of the subcommittee felt strongly that using the U.S. background rate was the preferred approach, while others felt that there was not sufficient justification to select one background rate over the other.

OCR for page 1
SUMMARY ]3

OCR for page 1
14 ARSENICIN DRINKING WATER: 2001 UPDATE study of a large population of individuals who consumed drinking water containing arsenic at a concentration of 20 ~g/L over an extended period of time. Detection would be further complicated by variability in the concentra- tions of arsenic in drinking water, the Unmown distribution of other risk factors (including smoking), end the mobility ofthe U.S. population. Because background lung cancer mortality in the United States is almost lO-fold greater than bladder cancer mortality, it would be even more difficult to dem- onstrate an association of arsenic in drinking water with lung cancer risk. Therefore, although the subcommittee's risk estimates are of public-health concern, they are not high enough to be detected easily in U.S. populations by comparing geographical differences in the rates of specific cancers with geo- graphical differences in the levels of arsenic in drinking water. In accordance with its charge, the subcommittee has not conducted an exposure assessment, subsequent risk characterization, or risk assessment. The theoretical lifetime excess cancer risks estimated by the subcommittee and the uncertainties surrounding those estimates as presented in this report should be interpreted in a public-health context that uses an appropriate risk- management framework. In summary, the subcommittee concludes that recent studies and analyses enhance the confidence in risk estimates that suggest chronic arsenic exposure is associated with an increased incidence of bladder and lung cancer at arsenic concentrations in drinking water that are below the current MCL of 50 Egg. The results ofthis subcommittee's assessment are consistent with the results presented in the NEtC's 1999 Arsenic in Drinking Water report and suggest that the risks for bladder and lung cancer incidence are greater than the risk estimates on which EPA based its January 2001 pending rule.