United States Science Advisory EPA-SAB-DWC-01 -001
Environmental Board (1400A) December 2000
Protection Agency Washington DC iviviv.epa.gov/sab
&EPA ARSENIC PROPOSED
DRINKING WATER
REGULATION: A SCIENCE
ADVISORY BOARD
REVIEW OF CERTAIN
ELEMENTS OF THE
PROPOSAL
A REPORT BY THE EPA SCIENCE
ADVISORY BOARD
-------
December 12, 2000
EPA-SAB-DWC-01-001
The Honorable Carol Browner
Administrator
United States Environmental Protection Agency
1200 Pennsylvania Avenue, NW
Washington, DC 20460
Subject: Arsenic Proposed Drinking Water Regulation: A Science Advisory
Board Review of Certain Elements of the Proposal
Dear Ms. Browner:
This review was conducted by a panel established by the Science Advisory Board (SAB)
consisting of the twelve members of SAB's Drinking Water Committee (DWC) and five consultants
who were asked to participate in order to provide expertise not held by the members of the DWC.
One of these consultants also served on the National Research Council Subcommittee on Arsenic in
Drinking Water. This review panel will be referred to hereafter in this report as the Panel. The report
was developed in response to interactions with representatives from the Agency's Office of Water
during the June 2000 and August 2000 DWC meetings.
The principal task before the Panel was to consider certain technical issues raised by EPA
relative to its proposed reduction of the Maximum Contaminant Level (MCL) for arsenic in drinking
water from 50 to 5 [ig/L. The Panel commends the Agency for undertaking this proposal.
It has been clear for some time that reconsideration of the arsenic MCL is necessary. The
Panel recognizes the need for a reduction in the MCL, however, individual participants in the review
hold diverse opinions about the most appropriate level for the MCL and about how that level should be
attained. Attachment A to this report entitled, A Minority Report on Arsenic in Drinking Water: The
Unique Susceptibility of Children to Arsenic, was authored by a consultant to the Panel, to present
his analysis of the differential sensitivity of children to arsenic and the rationale for his preference for
Agency action in this regard. Though the majority of the Panel agrees with his general thesis, which
asserts that children can be at greater risk from exposure to contaminants due to their high ingestion of
drinking water per unit body weight, they do not agree with the conclusion in the analysis that indicates
that this has been demonstrated for arsenic. Attachment B provides a statement by a member of the
DWC, indicating support of the minority report. At the request of one member of the Executive
-------
Committee, we also append comments critical of the Panel's report by another member of the National
Research Council Subcommittee on Arsenic in Drinking Water, and a response to these comments
from the Chairman, of the SAB Arsenic Review Panel (Attachment C).
In the majority report, the Panel commented on two aspects important to determining the
proposed MCL: the scientific basis of the health assessment, and some technical issues associated with
the economic and engineering analyses. Regarding the health assessment, it appears to the Panel that
the Agency accepted, as a risk assessment per se, a National Research Council (NRC) analysis that
was intended to determine whether the available human data (especially that from a study in Taiwan)
were sufficiently detailed to support a formal risk assessment. The NRC indicated that its analysis was
not meant to substitute for further investigation of the most appropriate method for assessing the risk
posed by arsenic in drinking water. The NRC also noted a number of factors that likely differ between
the Taiwanese study population and the U.S. population and which might influence the validity of
arsenic cancer risk estimates in the United States. Even though the Agency did its own risk
characterization (i.e., they combined the NRC risk factors with U.S. exposure information and arsenic
occurrence distributions to obtain a range of risks for use in their benefits analysis), they chose not to
quantitatively take any of these factors into account at this time.
The Panel agrees with conclusions reached by the NRC in its 1999 report on arsenic,
especially their conclusion that "there is sufficient evidence from human epidemiological studies...that
chronic ingestion of inorganic arsenic arsenic causes bladder and lung, as well as skin cancer." The
NRC also stated that currently the Taiwanese data are the best available for quantifying risk; however,
they also cautioned EPA about certain issues associated with directly applying that study to the U.S.,
and the Panel agrees and joins the NRC in emphasizing these cautions. In particular, clear deficiencies
in selenium intake in the Taiwanese population, other nutritional factors, genetic differences,
socioeconomic differences between the study area and the general population of Taiwan, and the need
for well designed epidemiology studies which use good exposure measures for individuals in the study
population were identified. We note, however, that this Panel does not believe that resolution of all
these factors can nor must be accomplished before EPA promulgates a final arsenic rule in response to
the current regulatory deadlines. However, resolution of the critical factors noted by this Panel, and the
NRC, should not be put off indefinitely. Resolution in time for the next evaluation cycle for the arsenic
regulation should be considered as a goal.
The Panel also went beyond the NRC report when new information provided reasons for doing
so. In particular, since the NRC report was issued, further analyses of the Taiwanese data have been
performed that show that conclusions from this data are very sensitive to the model used for their
analysis. The analysis also shows a very significant impact depending upon whether one uses
unexposed populations outside the study area for a comparison (control) group or uses relatively less
exposed populations within Taiwan who are likely still exposed to some arsenic but may be similar in
other ways to the study population. The ultimate risk number derived from the Taiwanese study has
proven very sensitive to the decision about the appropriateness of the comparison population. This of
-------
course, has important implications for the use of the data to estimate arsenic risk in the U.S. Also a
study in Utah suggests that some U.S. populations may be less susceptible to the development of
cancer, than those in Taiwan, although the Panel found that study difficult to use in a quantitative way
because of the manner in which the data were presented. Also, a recently published study suggests that
the incremental increases in lung and bladder cancers observed in the Taiwan study are of roughly the
same magnitude, rather than the NRC's inference of a potentially two- to five-fold greater rate of lung
cancer relative to bladder cancer.
As noted by the NRC, the mechanisms associated with arsenic-induced cancer most likely
have a sublinear character, which implies that linear models, such as those used by the Agency,
overestimate the risk. Similar advice was provided to EPA in an SAB/DWC report as early as 1989
(SAB, 1989) and in a peer review conducted for the Agency in 1997 (ERG, 1997). Nonetheless, the
Panel agrees with the NRC that available data do not yet meet EPA's new criteria for departing from
linear extrapolation of cancer risk
In summary, the Panel recommends that in future considerations of the risks posed by arsenic in
drinking water, (that is, following the finalization of the current proposed rule), the Agency should
generate a formal risk assessment that thoroughly explores, to the extent possible: a) the impact of
probable differences between the Taiwanese study population and the U.S. population; b) the
sensitivity of available data to a wider range of alternative risk extrapolation models; and c) findings
from other epidemiological and lexicological studies that may be completed by that time.
The Panel discussed at some length the Agency's proposal to issue a Health Advisory to alert
mothers who prepare formula using drinking water that such water might contain arsenic. This advisory
is intended as an interim measure that would apply during the time between promulgation of a final rule
and its implementation. The Agency provided no details to the Panel on the form or method of issuing
the proposed Health Advisory. However, from the Agency's general discussion, it appears to the
Panel majority that the envisioned advisory is different from past health advisories that have been issued
by the Office of Ground Water and Drinking Water. Because the audience, in this case, differs from
those for most such Drinking Water Health Advisories, the methods of communication of the advisory
might need to be different. Therefore, the Panel provided advice on certain issues that it believes the
Agency should consider should it decide to issue such a Health Advisory. The minority report
mentioned earlier supports release of a Health Advisory by EPA without reservation.
The Panel has some concerns about the economic and engineering assessment. In part, this is
because of the limited information provided to the Panel on the Agency approach to determining the
benefits of a decreased MCL, and because of differences between the Agency projections and those
of other organizations. In addition, there were several assumptions made in EPA's analysis about the
disposal of arsenic residuals that the Panel thought may not be realistic. First, the Panel felt that
assuming that high-salt residuals can be disposed of through publically owned treatment works
(POTWs) is questionable based on the strict limits on total dissolved solids in wastewater. Second, the
-------
Panel questioned the assumption that the residuals resulting from all treatments can be disposed of in
municipal landfills as a non-hazardous waste.
Another problem is that while many of the treatment options identified as best available
technology (BAT) are fairly standard in drinking water treatment, they have not been applied or
optimized for arsenic removal at a large scale. The behavior of arsenic is fairly unusual, and it is not
clear that these technologies can be simultaneously operated efficiently for arsenic removal and for their
other intended purposes.
The Panel also suggested that there should be some further thought given to the concept of
affordability as applied to this new MCL. They are concerned that costs to households served by the
small systems (the systems predominantly impacted by the arsenic rule) could force tradeoffs that might
not lead to the greatest overall public health improvement. Households with lower incomes will pay a
proportionately larger part of their incomes as a result of system compliance with new arsenic control
regulations than will those with higher income levels. This would be further exacerbated by additional
rules, now under consideration, because each new rule will add its own incremental costs to the overall
cost of drinking water for specific households.
The Panel moved beyond the scientific, economic, and engineering issues in the Charge to
provide their insights on some policy matters, based upon their experience and informed observations.
Specifically, the majority of the Panel members felt that there is adequate basis for the Agency to
consider use of its discretionary authority under the Safe Drinking Water Act of 1996 to consider
MCLs other than the proposed 5 ng/L. In light of the continuing uncertainties in the risk estimates,
technology, and significant implementation costs, the Panel majority felt that the Agency could consider
a "phased rule" that would be applied, first to a subset of potentially affected systems with the highest
exposures. Such an approach would effectively an adaptive management strategy that couples
immediate action with future flexibility to respond to results from both experience and research. The
minority report mentioned earlier does not support this approach.
Thank you for the opportunity to review these elements of the arsenic proposal. We would be
happy to continue to engage with EPA as it pursues this action. We look forward to your response to
this report.
Sincerely,
/s/ /s/
Dr. Morton Lippmann , Interim Chair Dr. Richard Bull, Chair
Science Advisory Board Drinking Water Committee
Science Advisory Board
-------
NOTICE
This report has been written as part of the activities of the Science Advisory Board, a public
advisory group providing extramural scientific information and advice to the Administrator and other
officials of the Environmental Protection Agency under the procedures detailed in the Federal Advisory
Committee Act (FACA). The Board is structured to provide balanced, expert assessment of scientific
matters related to problems facing the Agency. This report has not been reviewed for approval by the
Agency and, hence, the contents of this report do not necessarily represent the views and policies of the
Environmental Protection Agency, nor of other agencies in the Executive Branch of the Federal
government, nor does mention of trade names or commercial products constitute a recommendation for
use.
Distribution and Availability: This Science Advisory Board report is provided to the EPA
Administrator, senior Agency management, appropriate program staff, interested members of the
public, and is posted on the SAB website (www.epa.gov/sab). Additional copies and further
information are available from the SAB Staff.
-------
U.S. Environmental Protection Agency
Science Advisory Board
Drinking Water Committee
Arsenic Review Panel
CHAIR
DR. RICHARD BULL, MoBull Consulting, Inc., Kennewick, WA 99336
MEMBERS
DR. DAVID B. BAKER, Heidelberg College, Water Quality Laboratory, Tiffin, OH
DR. MARY DAVIS, West Virginia University, Morgantown, WV
DR. RICARDO DE LEON, Metropolitan Water District of Southern California, Water Quality
Laboratory, La Verne, CA
DR. YVONNE DRAGAN, Ohio State University, Columbus, OH
DR. JOHN EVANS, Harvard Center for Risk Analysis, Boston, Massachusetts
DR. BARBARA L. HARPER, Yakima Indian Nation, Richland, WA
DR. LEE D. (L.D.) MCMULLEN, Des Moines Water Works, Des Moines, IA
DR. CHARLES O'MELIA, The Johns Hopkins University, Baltimore, MD
DR. CHRISTINE MOE. The University of North Carolina, Chapel Hill, NC.
DR. GARY A. TORANZOS, University of Puerto Rico, San Juan, Puerto Rico
DR. RHODES TRUSSELL, Montgomery-Watson, Inc. Pasadena, CA.
CONSULTANTS
DR. MICHAEL DEBAUN, Washington University of St. Louis, St. Louis, MO.
PROF. JANET BERING, California Institute of Technology, Pasadena, CA.
DR. ISSAM NAJM, Montgomery-Watson, Inc., Pasadena, CA.
-------
DR. JOHN F. ROSEN, Albert Einstein College of Medicine, Montefiore Medical Center, Bronx,
NY
DR. LOUISE RYAN, Harvard School of Public Health, Dana-Farber Cancer Institute, Boston, MA.
SCIENCE ADVISORY BOARD STAFF
MR. TOM MILLER, Designated Federal Official, US EPA Science Advisory Board (1400A),
1200 Pennsylvania Avenue, NW, Washington, DC
MS. DOROTHY M. CLARK, Management Assistant, US EPA Science Advisory Board (1400A),
1200 Pennsylvania Avenue, NW, Washington, DC
111
-------
TABLE OF CONTENTS
1. Executive Summary 1
2. Introduction and Charge 7
2.1 Introduction 7
2.2 Charge 7
2.2.1 Arsenic Health Effects Charge to the SAB 8
2.2.2 Arsenic Treatment Charge to the SAB 8
3. Health Effects Issues 10
3.1 Comments on the Evaluation of Health Effects and Risk Issues 10
3.1.1 Charge Question 1: Inorganic Arsenic as the Principal Form Causing
Health Effects 10
3.1.2 Charge Question 2: Implications of Natural Arsenic Exposure
Through Food 12
3.1.2.1 Arsenic Exposure Through Food 12
3.1.2.2 Arsenic Concentrations in Drinking Water 13
3.1.2.3 Comparison of Arsenic Intake From Food and Drinking Water .... 14
3.1.2.4 Effects of MCL Choice on Drinking Water and Total Intake of
Inorganic Arsenic 14
3.1.2.5 Influence of MCL Choice on Estimated Health Benefits 17
3.1.2.6 Discussion and Conclusions 21
3.1.3 Charge Question 3: Health Advisory on Low Arsenic Water and Infant Formuli2
3.2 Comments on EPA's Interpretation of the NRC Report 25
3.2.1 General Comments 25
3.2.1.1. Short-comings of the Taiwanese Data 29
3.2.1.2 Effects of Nutrition and Preexisting Disease in Populations
That Have Been Studied 30
3.2.1.3 Modes of Action Attributed to Arsenic are Sublinear 32
3.2.1.4 Use of Experimental Data That Were Available and the Need for
Further Research 33
4. Treatment Technology and Cost Issues 35
4.1 Comments on Treatment Technology Issues 35
4.1.1 Charge Question 4: Disposal Options 35
4.1.2 Charge Question 5: Decision Tree for Treatment Technologies 36
4.2 Other Issues Associated with Cost 37
4.2.1 Affordability and Risk Tradeoffs 37
4.2.2 Need for Performance Data on Arsenic Control Technologies and the Possibility
of Adaptive Management 38
iv
-------
Attachment A: A Minority Report on Arsenic in Drinking Water: The Unique Susceptibility of Children
to Arsenic A-1
Attachment B: Endorsement of the Minority Report on Arsenic and Children B-l
Attachment C: Response to Comments Entered into the Record of the DWC's EC-Review Draft of the
"Arsenic Report" at the September 22, 2000 EC Teleconference Meeting C-1
-------
1. EXECUTIVE SUMMARY
The Science Advisory Board (SAB) met from June 5-7, 2000 and again on August 8, 2000
to consider components of the Agency's proposal for a new Maximum Contaminant Level (MCL) for
arsenic in drinking water. The review was conducted by a panel (referred to in this report as the Panel)
composed of the twelve members of the SAB's Drinking Water Committee (DWC) to which was
added five consultants who provided expertise to supplement that possessed by the DWC members.
The current MCL for arsenic is 50 |ig/L, and the proposed rule would lower that to 5 |ig/L.
The proposal also requests comments on alternatives of 3, 10 and 20 jig/L. The lowering of a national
standard by a factor often is a major change having significant cost impacts. In the case of arsenic, the
costs involved are substantial, but somewhat problematic, because it demands a level of treatment not
ordinarily utilized in the small community water systems that are the principal focus of the rule.
This report has two parts. The basic report provides the majority opinion supported by most of
the Panel members and consultants. The second is a minority report prepared after the Panel
attempted but was unable to agree on a single document that would provide a combined message giving
both the majority and minority views. The minority report (Attachment A) entitled A Minority Report
on Arsenic in Drinking Water: The Unique susceptibility of Children to Arsenic, was prepared by
Dr. John Rosen, a pediatrician and consultant to the Panel for the arsenic review. That document
provides his analysis of issues relevant to the differential sensitivity of children and the rationale for his
preference for Agency action in this regard. Dr. Rosen reviews the impact of toxicants on children's
central nervous system, cardiovascular development, reproductive and developmental organs, and
carcinogenesis and concludes that children are at greater risk of harmful effects from arsenic than are
adults and that additional safety factors are needed to protect children. The Panel majority does not
disagree with many of the statements in that minority report, especially the reasonableness of the general
thesis that children are at greater risk from toxicants because of their greater water ingestion per unit of
body weight. However, on some specific issues and in his conclusion that susceptibility has been
specifically demonstrated for arsenic, the majority and minority views departs.
Dr. Barbara Harper, a lexicologist with the Yakima Indian Nation and a member of the
Drinking Water Committee, also provided a statement of support for the minority report's dissenting
view (Attachment B). Dr. Harper concludes that children are a generally vulnerable or sensitive
population and she prefers precaution when data are substantially suggestive of increased effects in
children, as she believes to be the case for arsenic. She further states that Native American Tribes and
migrant workers constitute unique populations who should not be last in line for arsenic reduction
because of economic or technologic reasons.
The major source document on arsenic's health effects used by the Panel was the National
Research Council's report on arsenic in drinking water (NRC, 1999). In recognition of the importance
1
-------
of this report to the Panel's deliberations, Dr. Louise Ryan, a member of the NRC Subcommittee was
asked and did serve as a consultant on the SAB Panel both to ensure an adequate understanding of the
NRC effort as well as to provide expertise on modeling issues that are key aspects in understanding
arsenic risk from drinking water in the United States. In addition, the Panel considered additional
material that it identified in order to answer the charge questions about the forms of arsenic that are
responsible for its adverse effects and the influence of dietary arsenic sources on the risks projected
from the arsenic studies conducted in Taiwan. The Panel also responded to specific EPA questions
about the necessity for issuing a Health Advisory to communicate to mothers who may be using tap
water for the preparation of infant formula.
The Panel agreed with the major conclusions in the 1999 NRC document. These are noted
throughout the SAB Panel's report. The Panel did go beyond the NRC conclusions in a few instances
where new information provided additional insight since the NRC review was completed. These
instances, too, are noted in the Panel document. The major conclusions shared by both the SAB Panel
and the NRC Subcommittee include:
a) The Panel agrees that the existing national arsenic standard for drinking water (50 |ig/L)
is too high and should be decreased;
b) The Panel agrees that setting a specific standard involves factors beyond just science
issues, therefore, it is not appropriate for the science advisors to determine such levels;
c) The Panel agrees that data from the ecological study conducted in Taiwan, though not
ideal for risk assessment, are the best available at this time for determining arsenic's
carcinogenic dose-response;
d) The Panel agrees that the Agency should conduct a formal risk assessment that
considers additional epidemiology studies and population factors to the extent
practicable, in order to improve the validity of the U.S. assessment of arsenic risk from
drinking water;
e) The Panel agrees that there is not now sufficient evidence for the Agency to abandon
the linear-at-low-dose model, although most data suggest that mechanisms that have
been associated with arsenic are indeed sublinear.
It is important to note that as the Panel's Arsenic Report was being discussed by the SAB
Executive Committee (EC), one EC member entered into the record comments made by Dr. Alan
Smith, another member of the NRC Subcommittee on Arsenic, in response to EPA's proposed arsenic
rule-making, in which he commented on the cover letter of the SAB Panel's draft report. These
comments raised objections to three points in the Committee's draft which was being reviewed by the
EC:
-------
a) The DWC misinterpreted the NRC report
b) The DWC incorrectly asserted that if the risk were as high as 1 in 100, the effect should
be more evident in the U.S. than it apparently is
c) The DWC inappropriately accepted the analysis of Morales, Ryan, et al, which
presented results without the use of a comparison population known to be unexposed.
These comments are included in and have been responded to in a separate document which is included
as Attachment C to this report.
In general, the Panel concludes that determining the forms of arsenic responsible for producing
adverse effects has become more complex since the publication of the NRC report. It can no longer
be concluded that inorganic forms are the only active forms responsible for the carcinogenic effects
associated with arsenic. However, because arsenic in drinking water is largely of the inorganic form,
that then is the appropriate form for EPA's regulatory focus. Recent findings have also complicated
comparisons of the relative importance of food and water sources of arsenic. As long as the agency
relies upon linear extrapolations of arsenic's cancer risk, these problems can be minimized by simply
considering drinking water arsenic as an incremental risk superimposed on a more complex and less
understood background of total arsenic in food. However, this approach does not resolve the fact that
arsenic levels in food are several times that in drinking water. In fact, the Panel concluded that,
reducing drinking water arsenic exposure to levels below that found in food may reach a point of greatly
diminished return in terms of substantial reductions in risk from arsenic in the environment in general and
in the impacted communities. Nevertheless, actions to reduce the MCL for arsenic will provide the
largest benefit to communities with unusually high levels of arsenic in their drinking water.
The NRC report noted that mechanisms associated with arsenic-induced cancer likely have a
sublinear character. Similar advice was provided to EPA in an SAB/DWC report as early as 1989
(SAB, 1989) and in a peer review conducted for the agency in 1997 (ERG, 1997). Nonetheless, the
Panel agreed with the NRC conclusion that the available data do not yet meet EPA's new criteria for
departing from linear extrapolation of cancer risk.
In commenting on the Agency's interpretation of the NRC's arsenic report, the Panel noted its
belief that EPA took the modeling activity in the NRC report as being prescriptive despite the clearly
stated NRC intention that their efforts were illustrative, not actual risk assessments (see page 295-296,
NRC 1999). In addition, the Agency has not yet conducted an updated risk assessment for arsenic in
the U.S.
The Panel also considered some issues on the nutritional status of the Taiwanese study
population that were highlighted in the NRC report (page 295, NRC, 1999) and the issue of lung
cancer risk. An analysis available since the NRC report (Morales, Ryan, et al., 2000) led the Panel to
-------
conclude that the contribution of lung cancer to overall risk is about the same as that of arsenic's
bladder cancer risks. The Panel focused on and reemphasized the NRC's cautions about the selenium
status of the study population in Taiwan. Studies in the U.S. and in Holland have documented
significant elevations of bladder and lung cancer in individuals with low selenium intake. The selenium
status of the general U.S. population is much higher than that of the studied Taiwanese population. This
is not to dismiss the possibility of certain populations in the U.S. having a similar deficiency nor the
possibility that those living below the poverty line in the U.S. might have sensitivities linked to that
situation.
The Panel noted that the NRC subcommittee reviewed the Taiwanese studies and its limitations
at length, noting that, "No human studies of sufficient statistical power or scope have examined whether
consumption of arsenic in drinking water at the current MCL results in an increased incidence of cancer
or non-cancer effects."(NRC, 1999, p.7). The NRC also noted that epidemiological studies in Chile
and Argentina have observed arsenic-related risks of lung and bladder cancer of the same magnitude as
those reported in Taiwan, at comparable levels of exposure (several hundred micrograms/liter-p.2, 7
and 292). However, with respect to estimating risk, the NRC stated that "In the absence of a well-
designed and well-conducted epidemiological study that includes individual exposure assessments, the
subcommittee concluded that ecological studies from the arsenic endemic area of Taiwan provide the
best available empirical human data for assessing the risks of arsenic-induced cancer." The Panel
agreed.
The majority of the SAB Panel concluded that an analysis published since the NRC report
(Morales, Ryan, et al, 2000) provides important additional insights on the use of the Taiwanese data
for risk estimation in the U.S. As noted in the NRC report, "...the choice of the model used for
statistical analysis can have a major impact on estimated cancer risks at low-dose exposures,...."
(NRC, 1999, p.8). Morales, Ryan, et al., support this conclusion and also demonstrate by applying
several models to the Taiwanese data, that the conclusion one draws from the data is very sensitive to
the type of comparison population with which the study population is compared. The Morales, Ryan,
et al. paper does not select a particular model as most appropriate; however, the SAB Panel, after
discussions in their meetings which involved Dr. Ryan, believe that the model which does not use an
unexposed comparison population group should be relied upon by EPA for its risk calculations. This
conclusion is controversial, as noted in the comments made by one member of the SAB Executive
Committee, citing the comments of Dr. Alan Smith (referenced above) and that member and Dr. Smith
both disagree with this Panel's conclusions about the Morales, Ryan, et al., study (see Attachment C).
The Panel recommends that in the future (i.e., following the fmalization of the current proposed
rule) the Agency make a stronger effort to assess the risks of arsenic exposure by conducting a formal
risk assessment that, to the extent possible, quantitatively considers well-designed epidemiology studies
that appropriately measure exposure and those additional issues mentioned by the NRC as being
necessary to improve the validity of the assessment of risk in the U.S. (e.g., selenium intake, other
nutritional factors, socioeconomic differences). Such a risk assessment should also consider, to the
-------
extent possible, important characteristics of children that might increase their risk to arsenic, e.g.,
differences in diet, metabolism, body weight, specific age groups, consumption of water, toxic effects of
arsenic in a rapidly growing organism, and exposure estimates per unit of body weight. The Agency
should also address the full suite of both cancer and non-cancer effects associated with arsenic.
The Panel discussed the need for EPA's issuance of a Health Advisory to mothers who might
use arsenic containing drinking water to mix formula for their young infants. This would apply during the
interval between promulgation of the final rule and its full implementation. While most participants
thought an advisory was potentially valuable, the lack of a clear description from EPA of what an
advisory would contain or on how it would be implemented kept them from fully endorsing this
concept. The majority view reflected a concern that an Advisory could be issued without providing
information on appropriate actions, or without advising mothers about how to contact public health
officials for assistance in their decisions on appropriate actions. The Panel noted that the decision about
whether to release a Health Advisory or not is an EPA policy decision. However, research in the area
of risk communication, as practiced in the pediatric and public health communities, might provide
important guidance on how such an Advisory should be framed if the Agency decides to move in that
direction. The goal should be to inform in such a manner as to achieve an appropriate response,
without leading to overreaction. One member and one consultant to the Panel disagreed and endorsed
the need for EPA to issue a health advisory for arsenic in drinking water, and indicated that past
Agency practices would be suitable for an advisory in the case of arsenic contamination as well (see
Attachments A and B).
The Agency also directed questions to the Panel on the cost of compliance with the proposed
rule, with particular attention directed at disposal options for brines and other residuals from treatment
and factors used in the selection of alternative treatment technologies. In addition to these specific
charge questions, the Panel chose to comment on aspects of affordability and its interaction with risk
tradeoffs.
The Panel agrees that EPA addressed the spectrum of residual disposal alternatives; however,
they felt that certain alternatives may not be viable in some cases due to potential constraints placed on
utilities. The Panel questions whether the disposal of high-total dissolved solid (TDS) brines to a
publicly owned treatment work (POTW) is viable due to regulatory limits on TDS and dilution of
organic wastes in many systems, particularly in the western U.S.
Generally, the Panel believes that the costs estimated by the Agency for the rule appear to be
low. However, the panel notes that it had only limited information from the Agency on its complex
approach to identifying the costs and benefits for this regulation. In regard to costs, the Panel questions
whether the technologies identified as best available technologies (BAT) have been implemented or
optimized for arsenic removal at treatment plant scale. If optimization of these technologies for arsenic
removal reduces their effectiveness for other purposes for which they have been designed, the actual
costs of compliance could be underestimated.
-------
Despite the uncertainties attending the arsenic regulatory issue, there seems to be a growing
consensus among those familiar with the issue in support of a meaningful reduction in the current MCL
for arsenic. Certainly, this is a conclusion common to both the NRC report and this SAB Panel report.
Even so, individual participants in this SAB Panel review vary considerably on where they believe the
actual MCL should be set. Because of the technological uncertainties, and uncertainties in the
assessment of arsenic risk in the U.S., the Panel moved beyond its traditional role as technical advisor
and provided its insights on this policy matter based upon their experience and their informed
observations. Specifically, the majority of the Panel members felt that there is reason to suggest that the
Agency could consider using an adaptive management approach (e.g., a phased rule) which would
couple immediate action with future flexibility. However, those endorsing the idea believed that initially
setting the MCL at a level intermediate between the current MCL and the ultimate target MCL would
result in treatment by a smaller, but representative, number of community water systems which also are
the ones with the highest arsenic contaminant levels. Their experience would then provide needed
data to actually plan for the much larger number of systems that would be required to treat if a lower
MCL were later identified as the ultimate target. A minority of the Panel disagreed. Those opposed to
such an approach were concerned that it would not protect children as expeditiously as possible.
The Panel also discussed the issue of affordability for this rule, both alone and in combination
with other drinking water regulations that are being developed. The possibility of the co-occurrence of
factors such as small communities, along with high arsenic levels, poverty, and special populations
concerned the Panel. This was especially of concern to the two Panelists who authored Attachments A
and B. As they noted there are more than 13 million Americans living below the poverty line. Further,
they hold particular concerns for poor people in the Southwestern U.S., such as Native American
tribes, who both suffer from poor nutrition and live in areas with high arsenic concentrations. As such,
they believe that a co-occurrence of these factors might create population groups in the U.S. that are
similar to those in the Taiwanese study population that was the source of the dose-response information
used in the Agency's risk determination. The Panel believes that this situation could have implications
both for the risk assessment in the U.S. (i.e., sensitive subpopulations) and for the risk management
decision as well (i.e., in terms of overall use of resources to maximize public health gains).
-------
2. INTRODUCTION AND CHARGE
2.1 Introduction
EPA's Office of Ground Water and Drinking Water (OGWDW) proposed a new Maximum
Contaminant Level (MCL) for arsenic of 5 |ig/L on June 22, 2000 (EPA, 2000). This is a substantial
change from the current MCL of 50 |ig/L. The existing MCL was based on concerns related to arsenic
carcinogenicity with a primary focus on skin cancer. In considering the revision of the MCL, new data
on several issues were considered. These included:
a) The quality of the available epidemiological data;
b) Consideration of internal cancers and other health effects attributed to arsenic in
continuing analyses of the data from Taiwan and other populations having drinking
water with elevated arsenic levels;
c) The applicability of the data from Taiwan to the U.S. population;
d) Whether the mechanisms involved require linear or non-linear risk extrapolation; and
e) Practical limitations on the measurements of low levels of arsenic in drinking water.
Since the arsenic MCL was last considered, there have been new analyses conducted on the
available epidemiological data (some new studies have at least qualitatively supported the findings in
Taiwan), and the focus of the analyses have turned from skin cancer to internal cancers, particularly
cancers of the bladder and lung (NRC, 1999). There are now data that allow us to begin to consider
whether risk extrapolation for low doses should be linear or non-linear. However, studies now suggest
that the mechanisms involved in arsenic-induced cancer are more complex than previously recognized.
This led the NRC to conclude that although there are data that support non-linear risk extrapolation,
they are not sufficiently clear for identifying a point of departure based on alternative modes of action.
Finally, there are now data that support much lower practical quantitation limits for arsenic in drinking
water.
2.2 Charge
The Agency charge to the SAB Panel concerned both health effects and treatment technology
issues. The specific questions from the Agency follow.
-------
2.2.1 Arsenic Health Effects Charge to the SAB
Charge Question 1: Concentration on inorganic arsenic as principal form
causing health effects. EPA has identified inorganic arsenic as the principal form
causing health effects, and the literature indicates that most arsenic in drinking water is
inorganic. EPA's Maximum Contaminant Level Goal (MCLG) and MCL do not
distinguish between arsenate and arsenite. Does the SAB have perspectives on this
issue that it believes EPA should consider in developing its risk assessment?
Charge Question 2: Implications of natural arsenic exposure through food. The
1999 NRC report estimated the daily inorganic food intake by assuming that 10% of
the arsenic in seafood is inorganic, and all other foods are 100% inorganic arsenic.
NRC noted that these assumptions set an upper bound on the contribution from food,
which is about 10 jig a day for adults. Does SAB agree with the implied NRC
perspective that relative source contribution of food should be taken into consideration
in the setting of the drinking water standard and how might we consider this and
communicate it to the public?
Charge Question 3: Health Advisory on low arsenic water and infant formula.
The NRC report was inconclusive about the health risks to the pregnant woman,
developing fetus, infants, lactating women, and children. Given the potential for
cardiovascular disease (as evidenced by EPA's Utah studies and extensive other data)
and uncertainty about risks to infants, EPA plans to issue a health advisory to
recommend use of low-arsenic water in preparation of infant formula. Is this
precautionary advice appropriate given the available information?
2.2.2 Arsenic Treatment Charge to the SAB
Charge Question 4: Decision tree for waste disposal options for arsenic
treatment brines and spent media. EPA identified waste disposal options that will
likely be used for arsenic treatment residuals. EPA assigned national selection
probabilities to each of option in a decision tree. Some people are concerned that after
the drinking water MCL is lowered, the Toxicity Characteristic for arsenic will be
lowered and that many drinking water treatment residuals will be subject to the costly
hazardous waste management regulations. EPA believes that its analysis shows that
residuals should be nonhazardous, under the current TC of 5 mg/L and even if the TC
were revised to 0.5 mg/L. EPA suggests that important questions relating to waste
disposal do not relate to hazardous waste disposal. Rather, for brines, they relate to
questions such as TDS (total dissolved solids) restrictions in waters receiving brine, and
restrictions on sanitary sewer discharge due to TBLLs (technically based local limits).
-------
For sludge disposal, they relate to restrictions that may be placed on land application,
which may result in more systems using landfills.
Based upon a review of the attached materials, does the SAB believe that the EPA
produced an accurate projection of the likely disposal options for arsenic residuals and
the distribution of these options by treatment type? What are the SAB's views on the
advantages and the limitations of the various waste disposal options? What effect, if
any, would the SAB's analysis of these advantages and limitations have on the
probabilities assigned? What are the SAB's views on which options will be more likely
used by small systems (less than 10,000 people), and which will be more likely used by
larger ones?
Charge Question 5: Decision tree for ground water treatment technologies.
EPA has identified treatment technologies that will likely be used to treat arsenic in
groundwater systems. These include ion exchange, activated alumina, reverse osmosis,
coagulation-assisted microfiltration, greensand filtration, and point-of-use and point-of-
entry devices. The EPA has also identified non-treatment options such as
regionalization and alternate source. EPA consulted with small utilities and AWWA in
order to identify issues which would affect selection of treatment technologies for small
systems, which included cost, complexity of operation, chemical handling issues, and
frequency of maintenance on point-of-use devices. EPA has assigned selection
probabilities to each of these options in a decision tree that form the basis for the
Agency's overall cost projections. The portions of the preamble that explain this
decision tree as well as certain other relevant documents are attached.
Does the SAB agree with the principal "branches" of EPA's decision tree described in
the attached documents and the likelihood that these options will be used for systems of
various sizes with various source water characteristics? What views does the SAB
have on EPA's description of the advantages and limitations of these treatment
technologies? Would the SAB's views on the these advantages and limitations affect
the probabilities assigned?
-------
3. HEALTH EFFECTS ISSUES
3.1 Comments on the Evaluation of Health Effects and Risk Issues
3.1.1 Charge Question 1. Inorganic arsenic as the principal form causing health
effects. EPA has identified inorganic arsenic as the principal form causing health
effects, and the literature indicates that most arsenic in drinking water is inorganic.
EPA's MCLG and MCL do not distinguish between arsenate and arsenite. Does the
SAB have perspectives on this issue that it believes EPA should consider in developing
its risk assessment?
Because of the emergence of new data in the literature, the identity of the form(s) of arsenic
responsible for health effects is not clear. The long-held hypothesis that inorganic forms are solely
responsible for the carcinogenic effects of arsenic has been challenged by new experimental evidence
that is discussed in the following portion of the report. However, because arsenic in drinking water is
largely of the inorganic form, the Panel believes that it is appropriate for the Agency to make this its
regulatory focus.
Studies available since the 19999 NRC report indicate that organic arsenicals are of interest as
carcinogens (Wei et al., 1999; Arnold et al, 1999). In addition, the +3 valence state of monomethyl
arsenic was found to be much more cytotoxic than inorganic forms (Petrick et al. 2000). On the one
hand, methylation aids in the elimination of arsenic from the body, but on the other, it appears that it
may generate chemical species that are responsible for adverse effects in some target organs or cells
(Aposhian et al., 1999).
A carcinogenic response was observed in the bladder of rats administered dimethylarsinic acid
(DMA) for their lifetime (Wei et al., 1999). These studies did not detect an increased incidence of
urinary bladder tumors at 12.5 mg/L administered in the water; however, an increase was observed at
doses of 50 and 200 mg/L DMA. Further, studies by Arnold et al. (1999) indicated a lack of a
carcinogenic response to DMA in mice at concentrations of up to 100 mg/kg in the diet indicating that
mice are resistant to bladder carcinogenesis by arsenic. However, Arnold, et al. (1999) confirmed that
at 40 and 100 mg/kg DMA in the diet proved carcinogenic to the uroepithelium of the rat, while 2 and
10 mg/kg did not. The carcinogenic action is greater in female than male rats and is dose related at 40
and 100 mg/kg in female rats. In female rats exposed to DMA at 40 and 100 mg/kg in the diet,
cytotoxicity was observed in the urinary bladder epithelium. These are the only animal studies
performed in the absence of a co-carcinogen that demonstrate an induction of bladder cancer by
arsenic or one of its metabolites when administered chronically. The doses of arsenic required are very
high levels compared to the amount of DMA expected to be formed following ingestion of inorganic
arsenic in drinking water. As a result these data point to the potential that another form(s) of arsenic is
responsible. However, it is improbable that this carcinogenic result would be explained by conversion
of DMA to inorganic arsenic (Carter et al., 1999).
10
-------
The Panel recommends that future refinements of the risk assessment for arsenic consider the
concentrations of various arsenic metabolites in the urine, the serum, and uroepithelial cells of rats
treated with these same doses of DMA to the extent possible. This exercise would establish the
relationship between bladder carcinogenesis and urine concentration and speciation of arsenic. This
could be used to predict the concentrations of these arsenic metabolites that would be produced
following exposure to inorganic forms of arsenic. In turn, this would inform the Agency on the intake of
inorganic arsenic needed to produce carcinogenic concentrations of specific arsenic metabolites in the
human bladder under a variety of different exposures. These models should consider the bladder as a
reservoir for arsenic with the concentration varying over the course of the day.
This does not suggest that inorganic arsenic cannot also play a role in other target organs. Ng
et al. (1999) found increased incidences of tumors in the lung and gastrointestinal tract when sodium
arsenate was administered at 0.5 mg As/L in drinking water to female C57B1/6J mice (corresponding to
67 ng/kg body weight per day). Lung cancer is also implicated in the human population exposed to
arsenic (NRC, 1999). The varying responses of different test animals can reflect differences in genetic
susceptibility. However, it is also consistent with the possibility that the development of tumors in
different tissues could result from different metabolites or metabolite combinations.
Adding to the complexity are recent findings that a +3 valence state of organic arsenic is much
more toxic to cells in culture than the +5 organic forms that have been previously studied. Petrick et al.
(2000) found that monomethylarsonous acid, a +3 valence form of organic arsenic, is much more
cytotoxic to cells in culture than inorganic forms of arsenic, as well as the +5 forms of methylated
arsenic. Styblo et al. (1999; 2000) have very similar findings in cultures of rat hepatocytes and human
cells derived from the liver, skin, urinary bladder, and cervix with the +3 form of DMA as well as
MMA. These forms of arsenic are likely to be a short-lived intermediates in vivo., and as a
consequence would be found only at low concentrations compared to the +5 forms. At any rate, it is
no longer clear that the inorganic forms are the most toxic either (as opposed to being carcinogenic).
Consequently, in the future, it will be much more important to specify the dose, endpoint and target
organ when speaking of arsenic's toxicity, because the form responsible may well vary. Therefore, it is
probable that human responses are determined by a variety of conditions and may involve interactions
between metabolites.
Although there are exceptions, the principal forms of arsenic in drinking water are inorganic
forms, and the Agency is setting a standard for arsenic as it appears in drinking water. Because the
available data do not meet the Agency's criteria for abandoning the linear default assumption in
estimating risk, it is best to deal with the incremental risk of arsenic in drinking water. For this reason
alone, the Agency needs to focus on the inorganic forms of arsenic rather than attempting to deal with
all potential forms of arsenic. It is not possible to consider contributions of different forms of arsenic to
the overall response based on the data that are available today.
11
-------
3.1.2 Charge Question 2: Implications of natural arsenic exposure through food. The
1999 NRC report estimated the daily inorganic food intake by assuming that 10% of
the arsenic in seafood is inorganic, and all other foods are 100% inorganic arsenic.
NRC noted that these assumptions set an upper bound on the contribution from food,
which is about 10 jig a day for adults. Does SAB agree with the implied NRC
perspective that relative source contribution of food should be taken into consideration
in the setting of the drinking water standard and how might we consider this and
communicate it to the public?
Ideally, consideration of relative source contribution would place drinking water exposures into
a practical context in which specific forms of a chemical would be weighted by the potency of the form
of chemical that is present in producing the effect of interest. The above conclusions on the increasing
uncertainty about which forms of arsenic are toxic, and the specific toxicity attributed to specific forms,
makes it difficult to do such an evaluation of the comparative risks of arsenic in drinking water versus
that in food with any great confidence. Thus, neither the NRC nor EPA were able to consider the
kinetics of the formation of different arsenic species, and the Panel can only note that the lack of such
data increases the uncertainty about the relative contribution of drinking water to cancer induced by
arsenic relative to that in food.
In this section, the Panel provides an analysis to illustrate the relative contributions of arsenic in
drinking water and food and their relative contributions to the health benefits achieved at a variety of
alternative MCLs. This analysis concludes that for the populations consuming drinking water at average
levels (1 liter/day), the assumption that is used as standard practice by EPA's Office of Water in its
benefits assessment, the benefits rapidly reach a point of greatly diminished returns in terms of
predicted reduction in risk. The Panel did note, however, that as long as the agency relies upon linear
extrapolation of arsenic's cancer risk, these problem of food versus drinking water source contribution
are minimized because the focus is upon incremental risk associated with drinking water alone. Even
so, it appears that the intake of arsenic from food is several times that which is ingested in drinking
water. The Panel analysis does reinforce the NRC conclusion about the sensitivity of the cancer risk
assessment to the extrapolation model used to characterize low dose effects. It should be clear that the
analysis that follows is not a risk analysis. Rather, it is an analysis of the benefits from risk reductions
that are likely from arsenic decreases in drinking water in comparison to the levels associated with
arsenic in food. As such the analysis focuses on the average case instead of the high-end case that
would be typical of a risk assessment.
3.1.2.1 Arsenic Exposures Through Food
The NRC report summarized available information on arsenic in food supplies. These estimates
were based on combining information on average diets by sex and age groups with data available on
the total arsenic content of the foods included in the diet. The average diets are based on FDA Total
Diet Study for Market Baskets Collected for various time periods. For the 1991-1997 period, total
arsenic intake ranged from 2.15 ng/day for 6-11 month infants to 99.1 ng/day for 60-65 year males
(NRC, 1999, Table 3-6).
12
-------
Total arsenic consumed in foods is not directly comparable to total arsenic in drinking water in
terms of toxicity. Seafood contributes about 90% of the total arsenic intake from food. Much of the
arsenic in seafood is in two organic forms - arsenobetaine (AsB) and arsenocholine (AsC). These two
forms are considered nontoxic, although, their carcinogenic potential has not been fully evaluated. In
contrast, drinking water primarily contains inorganic arsenate and arsenite, both of which are
considered toxic.
Comparisons between arsenic in food and arsenic in drinking water were made by assuming
that 10% of the total arsenic in seafood is inorganic and that 100% of the total arsenic in food of
terrestrial origin is inorganic. For adults, the average inorganic arsenic intake from foods, based on the
above percentages, is 10 |ig/day (NRC, 1999, p. 47). Average inorganic arsenic intake from food
ranged from 1.34 |ig/day for 6-11 month old infants to 12.54 |ig/day for 60-65 year old males (NRC,
1999, Table 3-6).
EPA cited work Macintosh, et al. (1997) to indicate the individual variability in inorganic
arsenic intake from food. Macintosh studied 785 adults and found a mean inorganic arsenic intake of
10.22 jig/day, with a standard deviation of 6.54 ng/day and a range of 0.36-123.84 ng/day, using semi-
quantitative food surveys. This variability is apparently due to variations in diet rather than variations in
the inorganic arsenic content of individual foods.
The NRC (1999) and the EPA (2000) documents do not contain information on the regional
variability of arsenic content in foods within the United States. Generally speaking, the food supply
within the United States is considered to be rather homogeneous (Schoof et al., 1999). Nevertheless,
some individuals could have substantial differences in their arsenic intake via food.
3.1.2.2 Arsenic Concentrations in Drinking Water
Using compliance monitoring data from 25 states, the EPA estimated the numbers of ground
water and surface water Community Water Systems (CWS) with treated water falling in various ranges
of arsenic concentrations (EPA, 2000, Table V- 3 and V-4). Using this information, the Panel
prepared concentration exceedency curves for CWSs using ground and surface waters (Figure 1). The
EPA tables provided information whereby percentages of CWSs having concentrations exceeding 2.0,
3.0, 5.0, 10.0, 15.0, 20.0, 30.0 and 50.0 |ig/L could
be determined. These concentrations represent the points plotted in Figure 1. EPA states that the
distribution of arsenic concentrations in CWSs is independent of the size of the CWS (EPA, 2000).
Consequently the plots of concentration exceedency curves for CWSs also
represent concentration exceedency curves for the entire population served by ground water and
surface water CWSs. It is evident from the curves that groundwater has much higher arsenic
concentrations than surface water. It is also evident from the curves that, while most CWSs
have concentrations below the proposed MCL, treated water from some CWSs has arsenic
concentrations considerably in excess of the proposed MCL.
13
-------
The above data can be used to estimate the population-weighted average concentrations of
arsenic in drinking waters from ground and surface water CWSs. For this estimate, the percentage of
the CWSs in each interval (e.g. >10.0 to 15.0 |ig/L) was multiplied by the midpoint concentration for
that interval (e.g. 12.5 |ig/L) to obtain a weighted concentration for that interval. Summing the weighted
concentrations for all intervals gives the average concentration. For ground water supplies the average
concentration of arsenic is 2.85 |ig/L, while for surface water supplies the average is 1.46 |ig/L. Since
surface water serves 66% of the total population on CWSs and ground water serves 34%, the overall
weighted average concentration of arsenic in drinking water is 1.93 |ig/L.
3.1.2.3 Comparison of Arsenic Intake from Food and Drinking Water
Figure 1 also represents the relative intake of arsenic from food and water. The Panel's
analysis followed the standard EPA-Office of Water practice for benefits analyses in that it used the
average of drinking water consumption levels for CWSs of 1.0 L/day (EPA, 2000). This contrasts with
the standard practice used in risk assessment in which a drinking water consumption rate of 2 liters/day
for an adult, which actually approximates the 90th percentile intake (EPA, 2000). Because the arsenic
intake in food is the average dietary intake, the Panel decided to calculate drinking water intake based
on average drinking water consumption. As previously stated, use of average consumption values in
these calculations is consistent with approaches used in benefits assessments.
With an average drinking water consumption of 1.0 L/day, the Y-axis in Figure 1 represents
both the concentration in jig/L and the dose in ng/day. The food intake is represented by the gray area
on the graph under the dashed line at 10 |ig/day. Comparisons of the area under the drinking water
exceedency curve with the area under the food "curve" reflect the relative contributions of each
pathway to the total intake of inorganic arsenic in the diet.
The relative contributions of drinking water and food to total arsenic intake at the current MCL
of 50 jig/L are shown in Table 1. Data are included for ground water and surface water supplies, as
well as for the weighted total for the entire population of CWSs (both ground and surface). On
average, drinking water contributes 16.3% of the inorganic arsenic intake and food contributes 83.7 %.
Thus, water treatment to reduce drinking water concentrations has limited potential to reduce total
arsenic intake on average, in the general population. However, for the part of the population consuming
drinking water with high arsenic concentrations, water treatment can result in substantial reductions in
combined food and water intake. For individuals consuming water at 50 |ig/day, arsenic intake in water
is five-fold higher than average food intake and would be substantially reduced.
3.1.2.4 Effects of MCL Choice on Drinking Water and Total Intake of
Inorganic Arsenic
The Panel used the EPA data from which Figure 1 was produced to calculate the reductions in
drinking water concentrations and total arsenic intake that would accompany various choices for the
MCL. Those data, along with ancillary data, are shown in Table 2 and
14
-------
t
.a
.a
Community water sup plies using ground water
—B— Drinking water concentration
0 20 40 60 80 100
Percent of C"WS s with, Goncentrattons exceeding value on Y-axis
t
•8
M
O
60
50 -i
40-
30-1]
20
i
I 10-
Community water sup plies using surface water
—B— Drinking water concentration
ifta^iige wejii; :bapkpqiiaci iii fpofl •
0 20 40 60 80 100
Percent of CTWSs with concentrations exceeding value on Y-axis
Figure 1. Aisenic concentration exceedency curves for community vrater supplies using ground
and surface waters in relation to average aisenic intake in food. Community water supply data
from tables V-3 and V-4 in the EPA proposed arsenic rule. (see text for explanation)
15
-------
Table 1. Average contributions of drinking water and food to total arsenic doses for
ground water CWSs, surface water CWSs, and weighted average for all CWSs for an
MCLof50ug/L.
Water Source
Ground Water (GW)
Surface Water (SW)
Weighted Average
(0.34 GW and 0.66 SW)
Pathway
Water
Food
Total
Water
Food
Total
Water
Food
Total
Average Arsenic dose
(|ig/day)
2.85
10.0
12.85
1.46
10.0
11.46
1.93
10.0
11.93
Percent by
Pathway
22.2
77.8
100
12.7
87.3
100
16.3
83.7
100.0
Figures 2 & 3. The effects of treatment to reduce drinking water concentrations can be viewed as
truncating the portion of the area under the drinking water exceedency curve (Figure 1) at the level of
the MCL. The reduction in the area under the curve reflects the effect of treatment on average drinking
water concentrations. For these calculations, we assumed that the arsenic concentrations for all
supplies having higher concentrations than the proposed MCL would be reduced to 80% of the MCL
value. This is the same assumption as that made by the Agency (EPA, 2000). The effects of MCL
choice on average drinking water concentrations are shown in Figure 2. Imposing an MCL of 5.0 |ig/L
would reduce the average drinking water concentrations from its current value of 1.93 |ig/L to 1.40
The reductions in peak concentrations and the percentage reductions in peak concentrations
associated with each proposed MCL are also shown in Table 2. It should be noted that reductions of
these sizes would only occur for individuals consuming water at or near the current 50 |ig/L MCL value.
The percent reductions in drinking water doses and total doses (DW plus food) for various
MCL choices are shown in Figure 3 A and Table 2. At an MCL of 20 |ig/L, drinking water and total
doses are reduced by 8.3% and 1.3% respectively, while at an MCL of 3 jig/L , they are reduced by
36.5% and 5.9%. Because drinking water currently comprises only 16.3% of the total inorganic
arsenic intake (average consumption levels) and most of the population already consumes drinking
water with arsenic concentrations less than 3.0 |ig/L, the potential for reducing total arsenic intake is
only 5.9% at the lowest MCL we are to consider.
16
-------
The number of treatment plants required to achieve the reductions in arsenic intake associated
with MCL choices is shown in Figure 3B. The slopes of these curves represent the efficiency
(reduction/CWSs requiring treatment) of reducing arsenic exposure associated with various MCLs.
The efficiency in reducing exposure for going from an MCL of 50 |ig/L to 20 |ig/L is 3.5 times greater
than the efficiency in going from 5 |ig/L to 3 |ig/L.
3.1.2.5. Influence of MCL Choice on Estimated Health Benefits
As noted by the NRC, estimates of cancer risk from arsenic in drinking water are sensitive to a
number of factors, including at least the selection of the model used to represent the dose response
curve, the implications of exposure measurement and grouping that exist in the ecological studies of
arsenic's effects in Taiwan. Therefore, the Panel's consideration of the contribution to risk from arsenic
in drinking water relative to arsenic in food reflects some of the same problems. Even though both the
NRC and this Panel consider the most-likely arsenic dose-response curve to be sublinear, there are not
yet sufficient quantitative data available to link key events in arsenic's cancer induction to the dose-
response curve and thus permit a departure from linear cancer risk estimation approaches. Because of
this, the linear extrapolation default was used in EPA's earlier risk assessment (EPA, 1988) to estimate
cancer risks. The linear default was also used by EPA as the basis for estimating bladder cancer risk
reduction benefits in support of the current arsenic proposal. This is supported by conclusions from
the NRC Subcommittee that conducted the arsenic review (NRC, 1999), as well as this SAB Panel.
If the dose response curve is linear, the health benefits are directly proportional to average
doses and changes in health benefits associated with changes in the MCL would be directly
proportional to the changes in drinking water and total doses presented in Table 2 and Figure 3. For
example, at an MCL of 3.0 jig/L, the adverse health effects associated with drinking water arsenic
would be reduced by 36.5% while adverse health effects associated with total arsenic intake (food plus
water) would be reduced by 5.9%. Other MCL choices are accompanied by smaller reductions in
adverse effects.
As discussed earlier, we do not really know what the equivalency of the various forms of
arsenic are in their intrinsic contribution to the development of cancer. Given that the Agency is relying
on linear extrapolation, the Panel recommends that the Agency simply look at the incremental risk
associated with drinking water, as that is the controllable risk. Nevertheless, progressing to ever lower
arsenic levels, below those levels found in the U.S. diet, provides an ever diminishing return in mean
arsenic exposure per dollar invested in water treatment.
The Panel agrees with the NRC Arsenic Subcommittee in concluding that a linear dose
response curve is a practical interim measure even though existing information on arsenic's mode of
action suggests that the dose response would exhibit sublinear characteristics. If the dose-response
curve is sublinear, reductions of arsenic levels in the high ranges of exposure would have larger health
benefits than those estimated using linear risk estimation techniques,
17
-------
Table 2. Effects of arsenic MCL selection on average concentrations in drinking water,
on percent reductions in average arsenic doses in drinking water and in food plus
drinking water, on the reductions and percent reductions in peak concentrations, and on
the number of CWSs requiring treatment to reach the MCL.
Parameter
Average Ground Water
Concentration (ng/L)
Average Surface Water
Concentration (ng/L)
Weighted Average DW
Concentration (ng/L)
Percent Reduction in DW
Concentration/Dose
Percent Reduction in DW Plus
Food Total Dose
Reduction in DW Peak
Concentration (ng/L)
Percent reduction in DW peak
concentration
Number of CWSs Requiring
Treatment
Current MCL
(^g/L)
SO^g/L
2.85
1.46
1.93
0
0
0
0
0
Proposed MCLs
20ng/L
2.52
1.38
1.77
8.3
1.3
30
60
929
lO^g/L
2.15
1.34
1.62
16.3
2.6
40
80
2,455
S^g/L
1.71
1.24
1.40
27.7
4.5
45
90
5,621
S^g/L
1.39
1.14
1.23
36.5
5.9
48
94
9,330
and reductions in exposure in the low range of exposures would have smaller health benefits, than their
corresponding reductions in average concentrations. Under a linear dose-response curve, an MCL of
3 jig/L reduced adverse health effects by a factor of almost 4.5 times that achieved by an MCL of 20
|ig/L (reduction in average concentrations by 5.9% and 1.3% respectively). Under a sublinear dose-
response curve, the ratio of the health benefits at an MCL of 3.0 jig/L to that for an MCL of 20 jig/L
would be less than 4.5 fold, possibly much less. Under a sublinear dose-response curve, the curves in
Figure 3B would shift such that the slope (efficiency) between MCLs of 50 jig/L and 20 jig/L would be
relatively steeper and the slope between 5 jig/L and 3 jig/L would be relatively flatter.
Under a sublinear dose-response curve there will be a region where exposures to arsenic from
food and drinking water begin to interact significantly. The benefits of water treatment depend on
where the reductions in total dose (food plus water) occur along the dose-response curve. This
calculation would require a well-defined sublinear dose-response curve. The generation of such a
curve requires a more quantitative understanding of how the mechanisms by
18
-------
.-4
p-
?
3y
*>-•
oV
-3— Good
-£— 3 'jrSf s uste:
* AU CVni
n n
=n
?HtK J. Ettxrts of MCL l*wL en
K$, smi
. all OWC
19
-------
•a
•B
1
e
"H
a>
B
0>
PH
•S
a
% reduction in drinking water dose
% reduction in total arsenic dose
A
30-
20-
% reduction in diinMng water dose
% reduction in. total arsenic dose
2500 5000 7500
Number of CWSs requiring treatment
10000
Figure 3. Relationship between percent reduction in drinking water and in total
arsenic dose and choice of MCL (A) and number of CWBs requiring treatment (B).
20
-------
which arsenic contribute to its adverse health effects. Further, generation of any specific sublinear
dose-response curve from current epidemiological studies would likely be very difficult to justify due to
the general lack of measurable health effects in the low dose range.
If data were available to demonstrate that the dose-response curve includes a threshold value
then the arsenic doses from food and water would interact in the consideration of risk levels. If the total
dose exceedency curve spans the threshold value, only that portion of the exceedency curve above the
threshold is accompanied by adverse health effects. The procedures for estimating the health benefits
of various MCL choices would be similar to those listed above for a sublinear dose-response curve.
Information on the shape of the dose-response curve in the portion of the curve above the threshold
value would be needed as well as the threshold. However, data do not currently exist to allow such an
evaluation.
3.1.2.6 Discussion and Conclusions
In attempting to reduce the frequency of bladder cancers by reducing drinking water exposure
to arsenic, the Agency is faced with minimal marginal risk reduction opportunities. For example, there
are approximately 53,000 new cases of bladder cancer in the U.S. each year with over 12,000 bladder
cancer fatalities (American Cancer Society, 2000). The number of bladder cancers attributable to
arsenic can be estimated under the default assumption of a linear response of bladder cancers to total
inorganic arsenic dose. At an MCL of 3.0 |ig/L, EPA estimated an annual reduction in bladder cancers
of 22-42 cases and a reduction in fatal bladder cancers of 5.7 to 10.9 per year (EPA, 2000). At an
MCL of 3.0 |ig/L, the total dose of arsenic is reduced by 5.9%. If a 5.9% reduction in arsenic dose
results in a 22-42 case reduction in bladder cancer occurrence, then a 100% reduction would result in
a 373 to 745 case reduction under a linear dose response relationship. Under this reasoning arsenic
would be responsible for 0.7% to 1.4% of all bladder cancers in the United States. At an MCL of 3.0
|ig/L, the accompanying 5.9 % reduction in total arsenic dose would reduce bladder cancers in the
United States by 0.04% to 0.08%. At the proposed MCL of 5.0 |ig/L, EPA states that there would be
16 to 36 fewer occurrences of bladder cancer. These represent reductions 0.03% to 0.07% in the
annual occurrence of bladder cancers in the United States. Thus, it must be recognized that the impact
of reduction of arsenic in drinking water at these levels is likely to have a very small impact on the
overall incidence of bladder cancers in the country. Smoking and occupational exposures are thought
to be the major causes of bladder cancer (American Cancer Society, 2000).
The above analyses indicate that average arsenic ingestion via food is considerably larger than
average arsenic ingestion via drinking water even at the current MCL of 50 |ig/L. For the limited
populations where drinking water concentrations are at or near the current MCL, considerable
reductions in total arsenic exposure can be achieved by reducing the MCL. By assuming a linear dose-
response curve, the EPA was able to calculate the marginal benefits of drinking water treatment, even
though food represents the major pathway of arsenic intake. If the mode of action supported a
nonlinear response to total inorganic arsenic intake, food and water pathways would both have to be
considered in calculating treatment benefits. Such calculations would require a well-defined nonlinear
21
-------
dose-response curve and more information on the distribution of food intakes, neither of which are
currently available. Consequently, the Panel concurs that EPA had no choice other than to proceed
with marginal risk reduction calculations based solely on considerations of drinking water cancer risk
reduction as calculated using a linear response model and ignoring food intake. However, it is also
clear from the above analyses that there is a limit to the benefits that can be realized by reducing arsenic
in drinking water. It should be kept in mind that regardless of the MCL chosen between 5 and 20 |ig/L
only the extreme levels will be reduced in drinking water. Conversely, the extremes in arsenic content
of foods will remain unaltered by this regulatory action.
That said, it should be noted that studies show the potential for arsenic to cause non-cancer
effects. Most of these studies do not yet provide sufficient data for quantitative risk assessment at this
time, therefore, they can in general only be addressed by EPA qualitatively. The implication of the
possibility of such additional risks is that a given level of arsenic exposure reduction that would come
from lowering the MCL would also provide some additional level of benefit to the populations involved.
The Panel and the NRC both agree that additional studies are needed to refine our understanding of not
just arsenic's cancer effect, but its potential to cause other health effects.
3.1.3 Charge Question 3: Health Advisory on Low Arsenic Water and Infant Formula.
The NRC report was inconclusive about the health risks to the pregnant woman,
developing fetus, infants, lactating women, and children. Given the potential for
cardiovascular disease (as evidenced by EPA's Utah studies and extensive other data)
and uncertainty about risks to infants, EPA plans to issue a health advisory to
recommend use of low-arsenic water in preparation of infant formula. Is this
precautionary advice appropriate given the available information?
EPA plans to issue a health advisory to recommend the use of low-arsenic water in the
preparation of infant formula. This advisory would be active during the period covering the interval
between promulgation of the final rule and its full implementation, a period from 3 to 5 years. The Panel
held extensive deliberations on the implications of such a health advisory. During the discussion, the
Agency provided more detail about the type of health advisory envisioned and how it would be
disseminated. The advisory would note that the exposure standard has been lowered but that
implementation will be delayed for a period of years and in the interim parents concerned about arsenic
risk to infants should consider using low arsenic water to prepare infant formula. Although most of the
Panel agreed generally with the assertion that special circumstances pertain to infants that make it
reasonable to consider them unique in regard to their response to contaminants in drinking water, and
that this could require additional attention by the Agency during the implementation of a new arsenic
drinking water regulation, the Panel was not able to reach consensus on an endorsement of EPA's
intent to issue a Health Advisory whose purpose, content, and approach was not clear to the Panel. As
a result, most of the members favored a response to the charge question that provided a series of
cautions to the Agency as it moved forward to decide upon and develop their advisory.
22
-------
The "Minority Report on Arsenic in Drinking Water" discussed earlier in this report, and
contained in Attachment A, disagrees with certain elements of the Panel's reasoning for not giving a full
endorsement to a Health Advisory for this purpose. The minority report written by a consultant to the
Panel for the arsenic review provides his analysis of issues relative to the differential sensitivity of
children to arsenic that departs from the majority opinion contained in this report in strength of its
conclusion if not the general reasonableness of the need for increased concern for children, which is
also held by the Panel. As noted earlier, one member of the Drinking Water Committee supported the
minority opinion in regard to actions thought to be necessary to address concerns about the differential
sensitivity of children to arsenic (this is discussed more fully later in this section).
The action contemplated by the Agency is different from the health advisories issued in the past
and with which many of the DWC members are familiar. First, prior Health Advisories issued by the
Office of Ground Water and Drinking Water were meant to provide advice to states and utilities to
address infrequent occurrences of contamination, usually as the result of a spill. They are not
enforceable standards nor are they used to directly inform the public about particular health hazards to
potentially sensitive groups in the population. Second, as the Panel came to recognize during the
discussion of the issue at the meeting, the Health Advisory would be an interim measure, and not a
lower exposure recommendation for infants. The intent of this advisory is to alert parents that they may
want to take early action to protect their children against this potential risk in the period before the
standard is fully implemented. The Panel's greatest concern was that the Agency did not address how
the target audience was to be reached and how they were to obtain information about alternatives to
community drinking water for infant use that would allow them to effectively address any concerns
parents might have.
EPA's motivation for issuing an advisory is concern about health risks to the developing child
and uncertainty about cardiovascular risks to infants (that could be expressed later in life) as well as the
higher per unit exposure to infants. It recognizes that children differ in many ways from adults.
Differences in size, maturity of biochemical and physiological functions in major body systems, and
variation in body composition (water, fat, protein and mineral content) all can modulate the severity of
toxicity to any toxicant in a rapidly developing fetus-infant-child. Because newborns are the group
most different anatomically and physiologically from adults, they could exhibit the most pronounced
quantitative differences in sensitivity and susceptibility to environmental toxicants. The majority of the
committee did not feel that data available to them on arsenic had demonstrated an increased sensitivity
to arsenic in children. (This is discussed more fully below.) However, the Panel did note that infants
consuming formula made from drinking water could reasonably be expected to receive a higher dose
per unit body weight than adults based on information available on drinking water consumption in the
U.S. (EPA, 1999).
The Panel noted that while the decision to release a Health Advisory or not is an EPA policy
decision, research results in the area of risk communication as practiced in the pediatric and public
health communities, can provide important guidance on how such an advisory should be framed if the
Agency decides to move in that direction. The goal would be to inform in such a manner as to achieve
23
-------
an appropriate response without leading to overreaction. The Panel also discussed whether the
recommendation should solely focus on infants or if pregnant women should also be included in the
warning and whether the advisory should contain specific recommendations for actions by the target
audience (e.g., identify informed sources for obtaining follow up information, provide guidance on how
to obtain sources of low arsenic drinking water, etc.).
As noted earlier, the Panel effort to reach consensus on fully endorsing the health advisory was
impeded by the lack of specific information on the envisioned Health Advisory. As a result, the Panel
chose to identify those concerns that were voiced at the review meeting and to suggest that in deciding
on whether to go forward with an advisory which is a policy decision, that the Agency consider these
concerns and develop and implement the advisory in a way that would not be counterproductive.
a) The Panel was of one mind that it was not the proper entity to design the envisioned
Health Advisory and that its comments should not be viewed as comprehensive.
However, the Panel recognized that the Health Advisory could have unintended
consequences if it is not carefully designed and implemented. The contemplated
advisory is not analogous in intent and likely content to other health advisories issued by
the Office of Ground Water and Drinking Water in the past. The Panel suggested that
it might better be identified as something other than a health advisory to avoid
confusion.
b) Any health advisory of the type contemplated should focus on health professionals
(pediatricians and public health officials) and not only be issued broadly to the public at
large. These are the people in the community that can be depended upon to find
alternatives within that community.
c) Alternatives must be identified that are reasonable for the community that is being
notified. Bottled water will not have to be in compliance with the new arsenic standard
until the regulation is effective, and consequently may contain levels similar to or greater
than those in the public system. Hopefully, the bottled water industry will cooperate
with the intent of this advisory as well as public water systems. However, the Panel
noted that EPA has no direct jurisdiction over bottled water.
d) The advisory should include information on what is known about arsenic levels in baby
foods and prepared formula.
e) The Panel felt that if an advisory is to be developed it should inform without alarming.
Information should be provided that ensures that as a result of an advisory behavioral
changes do not lead to inadequate fluid consumption by children, inadequate nutrition,
or other unanticipated risks.
f) This is largely a policy issue. The available science does not speak clearly on the
question of whether the sensitivity to arsenic is greater at an early age than as an adult.
24
-------
During its discussion of the value of issuing an advisory, the Panel received the recent
Hopenhayn-Rich et al. (2000) study and there was disagreement within the Panel on its interpretation.
The dissenting view of one consultant, supported by one member, is provided in Attachment A to this
report. The Panel's majority view is that this study suggests that the links between arsenic in drinking
water and any observed increases in still birth or neonatal mortality are associated with exposures to a
very high concentration of arsenic (860 |ig/L). While the Panel believes that it is generally reasonable to
consider that children are generally at greater risk for a toxic response to any agent in water because of
their greater drinking water consumption (on a unit body weight basis), they do not believe that this
study demonstrates such a heightened sensitivity or susceptibility to arsenic.
For example, in the study, the concentrations that were possibly associated with adverse
reproductive outcomes are also associated with toxicity in adults. There are significant uncertainties in
the risks for developmental toxicity, cancer, and vascular disease at exposures in the 5 to 50 jig/L
range. (The current MCL alternatives fall within this range and these would be the levels that a Health
Advisory would address.) The Panel noted that in the Hopenhayn-Rich, et al. study the increases in still
birth or neonatal mortality between Antofagasta and Valparaiso disappeared once the arsenic
concentration in the Antofagasta drinking water fell to 110 |ig/L or less (Figure 4). At that point, the
stillbirth or infant mortality experienced in Antofagasta (that previously had high arsenic concentrations
in water) was similar to that of the control town, Valparaiso, with arsenic concentrations less than 5
|ig/L. Sometimes the effects in the exposed town were less than and sometimes greater than the effects
in the control population (these were small differences and not in a consistent pattern and appeared to
be due to randomness in the data, which were averaged in the original study over 4-year periods
because of the considerable variation from year to year). Furthermore, it is not clear how the proposed
putative effects of arsenic can account for the substantially elevated prenatal mortality seen prior to the
increase in drinking water arsenic content that began in 1954 in the study population. Moreover, there
was a downward trend in both the exposed and control populations over the period of observation that
was apparently not related to arsenic in drinking water. In short, the Hopenhayn-Rich study appears to
be an hypothesis generating study that, in light of the limitations just described, merits and requires
further study before drawing final conclusions.
3.2 Comments on EPA's Interpretation of the NRC Report:
3.2.1 General Comments
This section of the report discusses some uncertainties associated with the Taiwanese study
data used to characterize risks to U.S. populations from arsenic in drinking water. The NRC discussed
the challenges that are presented to EPA in preparation of a risk assessment for arsenic in drinking
water stating that, "In the absence of a well-designed and conducted epidemiological study that includes
individual exposure assessments, the subcommittee concluded that ecological studies from the arsenic
endemic area of Taiwan provide the best available empirical human data for assessing the risks of
arsenic-induced cancer." (p. 7, NRC, 1999). In noting, however, that ecological data might be the
only choice in the absence of such data the NRC stated that, "Such analyses must be conducted with
caution, keeping in mind the potential for measurement error and confounding to bias the results. It is
25
-------
important to remember that any risk assessment based on ecological data must be cautiously
interpreted because of the inherent uncertainty in the exposure-assessment methods used for such
studies." (Page 294, NRC 1999). They also noted that model choice has an impact on estimates of
low-dose risks when the analysis is based on epidemiological data. (p. 294). Finally, the NRC noted
that other factors might affect the risk assessment in Taiwan or extrapolations to the U.S. such as poor
nutrition and low selenium concentrations in Taiwan, genetic and cultural characteristics, and arsenic
intake from food." (p. 295)
The Panel notes its belief that the Agency may have taken the modeling activity in the NRC
report as prescriptive despite NRC comments about possible limitations in the existing knowledge base
and their intention that their efforts be seen as illustrative and not as actual risk assessments (see pages
264 and 295-296, NRC, 1999). The Agency did do a risk characterization using factors from the
NRC report and occurrence information from their own efforts as the basis for their assessment of the
benefits associated with risk reduction. They did not conduct the formal risk assessment integrating
additional factors called for by NRC.
The Committee also learned of a broader set of analyses that have been published since the
release of the NRC report (Morales, Ryan, et al., 2000) which may have important implications for risk
levels associated with arsenic exposure. Ordinarily, epidemiology studies compare the disease
incidence in an exposed population with a group that is unexposed (i.e., a comparison population).
However, if there are substantive differences in the characteristics of the comparison population, the
differences noted may not be valid. (This issue was also a concern to the NRC Subcommittee, see
pages 285-288, NRC 1999.) It is possible to model dose-response without a comparison population
by simply looking at the response rates within the exposed population as a function of the gradation of
exposure.
Morales, Ryan et al., derived arsenic risk estimates using the comparison populations or simply
basing the estimate simply on the dependence of cancer incidence within the population on exposure to
arsenic. They concluded that the risk estimates were extremely sensitive to the use of comparison
populations that were outside the study area (i.e., comparison with the whole of Taiwan, or the
remaining portions of southwestern Taiwan that surround the study area). The panel focused on
evidence provided in the NRC (1999) report that indicate that the population in the study area differed
substantially from these comparison populations socioeconomically and in diet. At least one of these
differences has been found to significantly associated with rates of bladder and lung cancer in other
populations (these issues are discussed in detail in sections 3.2.1.1 and 3.2.1.2). Consequently, most
members of the Panel came to the conclusion that the comparison populations were not appropriate
control groups for the study area.
Morales, Ryan, et al. (2000) focused extensively on the impact of including a comparison
population in the analysis of the Taiwanese data. As discussed in the NRC report, the available internal
cancer data are based on 42 villages from the arsenic endemic region, hence all have
26
-------
Year (1950-1996)
Year (1950-1996)
Year (1950-1996)
Figure 4. Arsenic Concentration and Reproductive Outcomes.*
*Note: Effects data are drawn from Table 3 in Hopenhayn-Rich, et al. (2000). The results were reported as the average of 4 year
periods. Arsenic concentration data for the town of Antofagasta are from their Table 2 and were reported as the average of existing
measurements for groupings of specified years (1950-1957, 1958-1970, 1980-1987, and 1988-1996). The year groupings for
concentration data did not coincide with the year groupings for outcomes data. To depict the results, the average value are shown for
each of the years in that group with shading in the figure. The increase in arsenic in 1958 was caused by using a new drinking water
source for Antofagasta. In 1970, an arsenic-removal plant was installed and the arsenic concentration in the drinking water fell. The
control town, Valparaiso, has no historical evidence of high arsenic concentration (noted in the text as below 5 ug/L in recent surveys
and below the detection limit (20 ug/L) in more recent monitoring (1990-1994).
27
-------
non-zero exposures to arsenic. Benchmark doses (BMDs) can be computed from a dose response
model fitted to the village data and extrapolated to zero, or from models that use population-based data
to specify expected cancer rates at the zero level of exposure. In going beyond the analysis in the NRC
report, the expanded analyses of Morales, Ryan, et al. suggests that estimated BMDs are highly
sensitive to inclusion of a comparison population. In addition to data on the whole of Taiwan, as was
done in the NRC report, Morales, Ryan, et al. also considered a smaller comparison population based
only on the southwestern region of Taiwan.
The expanded analyses of Morales, Ryan, et al., which included log and square root
transformations of exposure, suggested that BMDs derived from models that include a comparison
population could be an order of magnitude lower than those based on models that do not include a
comparison population. For example, they found a 1% BMD of 23 |ig/L for male bladder cancer from
the best fitting model which includes the whole of Taiwan as a comparison population. The analogous
result based on using only the southwestern region was 54 |ig/L. In contrast to the variability of BMD
estimates based on models that included a comparison population, Morales, Ryan, et al. (2000) found a
high degree of stability for models fitted without use of a comparison population: 1% BMD estimates
were consistently found to be around 400 |ig/L.
Morales, Ryan, et al. (2000) also extended the NRC analysis to consider additional classes of
dose response models and by including lung in addition to bladder cancer. An important finding was
that arsenic associated risks from lung cancer are of a similar magnitude to those for bladder cancer.
For example, the 1% BMD for lung cancer based on the best fitting model (no comparison population)
was 343 jig/L for males and 256 jig/L for females.
The Panel had extensive discussions about the validity of estimates based on models that do not
include a comparison population. Most members believe that possible reasons for not using models that
include Taiwan-wide data as a comparison population include differences in lifestyles between the poor
and rural population in the Taiwan arsenic endemic region and the general Taiwanese population and
the influence of arsenic in food on risk in the population living in the arsenic endemic region. This could
mean, for example, that someone classified as being exposed to 40 jig/L in water might actually have
received a total exposure of 80 |ig/L. BMDs calculated from models that include a comparison
population will be particularly sensitive to bias in this setting, since the general population will not have
the same background levels of arsenic or the same nutritional status as the study population. In contrast,
analyses that use only data from the arsenic endemic region should provide fairly accurate estimates of
the risk associated with incremental increases in the amount of arsenic in drinking water.
The Panel noted that issues related to choice and inclusion of a comparison population are also
problematic for the Utah study (Lewis, et al., 1999). The study population in Utah was based on
records from the Church of Jesus Christ of Latter-Day Saints. Just as in Taiwan, there are good
reasons to believe that the analysis could be confounded by lifestyle differences between the study
population and the general population in Utah. Indeed, the variability seen in the standardized mortality
ratio (SMRs) reported in the study emphasizes this concern. The Panel recommends that additional
28
-------
analyses be performed using data only from the study population and focusing on dose response within
that population. The Panel is also concerned about the way in which exposure levels were categorized
in the Utah study: by classifying subjects in terms of ppb-years of exposure, the study induces an
association between exposure and age. The Panel recommends that the analysis be done with
exposure represented by concentration in drinking water, not ppb-years of exposure. Then
adjustments for cumulative exposure can be made separately.
Although the data provided in published results of the Lewis, et al. (1999) study imply that there
was no excess bladder or lung cancer in this population, the data are not in a form that allows dose-
response to be assessed dependably. In the public comment period during its June 5-7, 2000 meeting,
the Panel learned that some of these data were being reanalyzed and that some changes in the results
could occur. In terms of the cancer risk, this analysis is important for establishing the range of
uncertainties that come from attempting to adapt data from Taiwan to estimating dose-response
characteristics for the U.S. The completion of these analyses are important to the longer term
consideration of arsenic risk in drinking water. However, the Panel does not think that the Agency
should delay currently planned actions to decrease the MCL while the reanalysis is completed.
In summary, the Committee concludes that the Morales, Ryan, et al. (2000) paper is a useful
expansion of the analysis provided in the NRC report. Risk estimates based on the use of population-
based comparison groups appear to be unstable and lead to risk estimates that are unrealistically high.
There is good reason to rely on the estimates that use only the data from the study area (i.e., no
comparison population). These estimates consistently and stably predict a risk of a magnitude of one in
one-thousand for both bladder and lung cancer at the current MCL of 50 |ig/L using a linear model.
In the comments introduced into the record by one EC member as the Panel's Arsenic Report
was being reviewed by the SAB Executive Committee (comments originating from Dr. Alan Smith, also
of the NRC Arsenic Subcommittee, objected to the Panel's conclusions about the Morales, Ryan et al.
model without a comparison population. He disagrees that EPA should rely upon this model for its
evaluation of arsenic carcinogenicity. These comments are included in Attachment C.
The remainder of this subsection focuses more specifically on some of the factors that might
account for differences in apparent susceptibility of the Taiwanese population to cancer and other
adverse health effects relative to the U.S. population.
3.2.1.1 Shortcomings of the Taiwanese data.
As pointed out by NRC (1999), the Taiwan data has serious limitations for use in a quantitative
assessment of risk in the U.S. (e.g., see comments on pages, 2 regarding improving the validity of risk
assessment, p. 8 on other factors, p. 294 on cautions about exposure measurement and grouping).
Below, the Panel reinforces these cautions from the NRC and briefly describes uncertainties that make
it impossible to determine the extent to which well known risk factors for lung and bladder cancer might
have contributed to the observations in Taiwan and therefore, have implications for arriving at an
MCLG. This is not simply a question of inadequate knowledge of the smoking habits, endemic disease
29
-------
and nutritional factors within the population, but of whether these factors might significantly modulate the
arsenic effect. Considerable evidence has accumulated in recent years that arsenic has more marked
properties as a cofactor (e.g. co-carcinogen, promoter, and perhaps progressor) than as the sole
initiator of cancer. The area in Taiwan in which the arsenic exposed population lives is rural, quite
poor, and has varying degrees of evidence of nutritional deficiencies that may be reasonable
contributors to the observed effect either directly or indirectly enhancing the effects of arsenic. The
sensitivity of the risk estimates to the use of comparison groups of the whole of Taiwan, or even the
Southwestern area of Taiwan that includes the study population, highlights this issue. Comparing these
results to the assessment of risk over gradients of arsenic suggests the possibility that these other factors
also contribute to cancer risk in the area. This is because the dose-response curve can be viewed as
having a non-zero intercept on the Y-axis when only the study population is considered. Despite its
limitations, the results of the Utah study also suggest there are potential differences between the affected
population in Taiwan and the U.S.
Considering the above factors leads to a conclusion that transferring the dose response curves
describing the cancer risk in this section of Taiwan to the U.S. is likely to bias U.S. risk estimates
towards overestimates. The magnitude of this bias could be large, but the Panel does not have the
resources to resolve these issues more definitively.
3.2.1.2 Effects of nutrition and preexisting disease in populations that have
been studied
A number of mitigating circumstances were identified in the NRC (1999) report that suggest
that risk levels calculated from the Taiwanese data should not be rigidly extrapolated to the general
U.S. population. Poor nutritional status is known to be characteristic of this population and others
(Chile, India) that have been studied. A recent cohort study in Utah (Lewis et al., 1999), found no
evidence of either bladder or lung cancer where mean drinking water concentrations of arsenic
approached 200 |ig/L. While these concentrations are up to an order of magnitude lower than found in
sites where positive associations with cancer have been obtained, these results give rise to significant
questions about whether the Taiwan data apply quantitatively to those U.S. populations that have a
more adequate nutritional status.
Experimental work in animals establishes that deficiencies in selenium substantially increase the
toxicity of arsenic (Pan et al., 1996). The NRC (1989) report summarizes the results from a survey of
urinary selenium concentrations in Taiwan and other parts of the world. Essentially, the study
population in Taiwan was estimated to have selenium intakes that were only 25% of the recommended
dietary intake (NRC, 1989). Their intakes are less than 50% of the safe range identified by the World
Health Organization (WHO, 1996). For this reason NRC recommended that the selenium status of the
Taiwan population be taken into account in transferring the data to populations that are selenium
sufficient. Neither NRC or EPA made an attempt to make these adjustments. The Panel identified a
number of studies that have documented substantial effects of smaller selenium decrements on cancer of
the bladder (Helzlsouer et al., 1989) and lung (Salonen et al. 1985; van den Brandt et al., 1993). The
Panel strongly recommends that the Office of Water take this factor into account in its risk assessment
30
-------
supporting an MCL. The Panel expects that additional data exists that could be used for estimating the
extent to which nutritional deficiencies of the magnitude identified in Taiwan have on arsenic toxicity in
general and on carcinogenicity specifically.
Another nutritional issue that has been identified in the Taiwanese population from which the
data were obtained, was the potential for less than optimal intakes of methyl donors in the diet, such as
methionine or choline. There are data to suggest that hypomethylation (as well as hypermethylation) of
DNA does occur with exposure to various forms of arsenic (Zhao et al., 1997 as quoted by NRC,
1999). Choline deficiency has long been used experimentally as a tumor-promoting regime in animals
(Lombard! et al., 1994; Saito et al., 1994), therefore it is probable that substantive deficiencies in the
diet could increase sensitivity to arsenic induced cancer in humans. However, the NRC Panel did not
indicate whether such deficiencies were documented in the Taiwanese population studied. There were
indirect indications of what made up a substantial portion of the diet (rice and sweet potatoes), but the
estimates were not quantitative, nor were other constituents of the diet discussed in quantitative terms.
Other characteristics of the Taiwanese diet may also have contributed to the increased
susceptibility to cancer. Zinc insufficiency was postulated to occur in the blackfoot region of Taiwan,
but subsequent estimates do not substantiate this. Zinc can protect against acute arsenic toxicity,
although its influence on the chronic effects of arsenic are not known (NRC, 1999, p. 241). For the
study population the diet consisted of 9% protein, while fat contributed 5% of the caloric intake. Poor
diets may have also involved limiting levels of folate, methionine, cysteine, and B12.
In some Asian countries endemic infectious hepatitis has been known to be important in
sensitizing populations to the hepatocarcinogenic effects of aflatoxin (Caselman, 1996). It would be
useful to consider the incidence of infectious hepatitis in the study area of Taiwan to determine if that
might contribute to the increased risk for liver cancer found in some studies. The NRC did not consider
this recognized risk factor in its deliberations. The DWC suggests that EPA attempt to find out whether
the area studied in Taiwan also has high rates of hepatitis which is known to act as a co-carcinogenic
factor in liver cancer.
In summary, the characteristics of the Taiwanese population studied for arsenic carcinogenesis
are not typical of the characteristics of the general U.S. populations, but there may be segments of the
U.S. population which have one or more of the same potential co-risk factors as the Taiwanese
population. For example, poor U.S. subpopulations, particularly in the rural Southwest, may have
some of the nutritional deficiencies of concern (except selenium) in the Taiwan arsenic endemic region in
terms of nutrition, etc., and they may be exposed to high levels of arsenic in their drinking water as
well. The prevalence of the risk factors mentioned above need to be evaluated in both the Taiwanese
and U.S. populations. These differences raise significant uncertainties about the accuracy of risk
estimates that are based on the Taiwanese data. Unfortunately, the DWC cannot be more quantitative
in its own assessment for lack of resources and time. For this reason, we join with the NRC (1999)
recommendation that the Agency make a stronger effort to quantify risks in a way that attempts to take
these factors into account. However, this should not significantly delay promulgation of a rule that
-------
makes a significant reduction in the MCL for arsenic as there may be populations with similar nutritional
deficiencies within the U.S.
3.2.1.3 Modes of action attributed to arsenic are sublinear.
Cancer has been produced in experimental animals with arsenic, but a measurable response has
been most readily observed when combined with other treatments, such as the use of a tumor initiator
(see Shirachi et al, 1983; Laib and Moritz, 1989). Yamamoto et al. (1995) using several different
initiators, found DMA to be an effective promoter in the lung, bladder, kidney, liver and thyroid gland
of the rat. The site concurrence for these tumors with the human data should be of interest. Wei et al.
(1999) and Arnold et al. (1999) demonstrated that high doses of dimethylarsinic acid can produce
bladder carcinogenesis in the rat and Ng et al. (1999) found that tumors of the lung, gastrointestinal
tract, and liver were produced with 0.5 mg As/kg water as sodium arsenate.
Studies of arsenic's effects at the cellular and molecular level support a sublinear dose-response
model (NRC, 1999). Its apparently non-linear effects in producing structural and numerical
chromosomal abnormalities through apparently indirect mechanisms are one example. Similar
arguments would be developed for the "comutagenic" activity of arsenic. Other plausible modes of
action include modification of DNA methylation (presumably caused, in part, by arsenic's competition
for methyl donors) are associated with altered gene expression. It would be anticipated that these
effects behave in a sublinear way at low doses (i.e. possess effective thresholds).
There are abundant data that associates various forms of arsenic with a variety of mechanisms
or modes of action. If they could be shown to uniquely or collectively account for human tumors, the
dose-response curve could be viewed as being sublinear at low doses. NRC (1999) pointed out,
however, that none of these alternative modes of action have been clearly demonstrated as essential in
the development of arsenic-induced tumors. In most cases, even dose-response information showing
parallels between those that produce tumors and those that activate these other mechanisms have not
been explored. Therefore, the NRC concluded that the prudent course would be to use linear
extrapolation. However, the data derived from studies attempting to identify mechanisms that are
outlined in much more detail in the NRC report, suggest that applying linear models for low dose
extrapolation may be conservative. In future risk assessments for arsenic in drinking water, the Panel
suggests that the Agency explore additional models, [one such model would be the Moolgavkar,
Venzon, Knudson (MVK) model].
3.2.1.4 Use of experimental data that were available and the need for further
research
Because the Agency is presumed to be acting under a new set of cancer risk assessment
guidelines, the Panel was somewhat surprised that the Agency did not at least provide some summary
of the data that are available and how they inform the current risk assessment decision made by the
Agency. There have been substantial breakthroughs in the development of animal models of arsenic
carcinogenesis in the past several years. In part these data point up the weakness of some of the
32
-------
arguments that have been made to attribute cancer risk to inorganic arsenic alone. The relative ease of
producing bladder cancers in rats with dimethylarsenous acid may require some shift in the paradigm.
The Panel recognizes that in the case of arsenic where the margins between actual exposures
and effects are small compared to most other contaminants, the final decisions will involve many non-
scientific issues. However, these data are essential for identifying susceptible populations and need to
be forcefully pursued before the arsenic MCL undergoes its next review cycle. The fact that they are
not acknowledged in the documentation put forward by EPA provides no further encouragement to
pursue these issues in the future. Consequently, we have taken this opportunity to comment on some of
the research results that the Panel feels provide direction for future research in characterizing the risks of
arsenic in drinking water.
Arsenic is not a classical direct acting carcinogen. It does not cause DNA adducts, nor does it
induce point mutations although it can replace some of the phosphates in the sugar-phosphate
backbone of DNA (Dixon, 1997). This can result in the cleavage of the sugar-phosphate/sugar-
arsenate backbone and potentially in single strand breaks. This mode of action might account for the
occurrence of deletions and translocations in the absence of point mutations in arsenic-induced cancer.
Patrick (1964) demonstrated the incorporation of arsenic into DNA, protein, and lipid at the same rate
as PO4. Furthermore competition between these two ions has been said to uncouple oxidative
phosphorylation (Frost et al, 1968). These data suggest a mode of action not yet fully explored.
Asm can interact with thiols. This type of interaction may be important in the interaction with
lecithin cholesterolacyl transferase (Jauhiainen et al., 1988) of potential relevance to vascular changes
indicative of atherosclerosis.
Interactions with thiols are also significantly involved in the metabolism of arsenic. In serum,
arsenic is transported bound to sulfhydryl groups of proteins, GSH, and cysteine. AsHI can form a
complex with GSH (Delnomdedieu et al., 1994) and is more generally reactive with tissue than AsV.
Moreover, arsenite (not ionized at physiological pH) can be taken up by liver cells and methylated, but
not arsenate (ionized at physiological pH). In the kidney, however, arsenate is taken up, reduced,
methylated, and released into the urine. The reduction of AsV can be accomplished by sulfhydryls.
For example, glutathione can provide a reducing equivalent for AsV and the resulting Asm can then
oxidatively add a methyl through SAM (S adenosylmethionine) to produce the methylarsenic V.
There are only a few animal studies performed in the absence of a co-carcinogen that
demonstrate induction of neoplasms by arsenic or one of its metabolites. The induction of bladder
cancer in rats by DMA as reported by Wei et al. (1999) and Arnold et al. (1999) are of particular
interest, since bladder cancer is one of the principal sites of concern in humans. These studies must be
followed up. Arnold et al. (1999) have indicated that a non-linear mode of action is appropriate for
DMA in the rat and hence for cacodylic acid (DMA) in the human. Genotoxic effects of arsenic
appear to require substantially higher concentrations of DMA than would be observed systemically in
animals provided DMA, since Moore et al. (1997) found concentrations of 5 mg/ml necessary to obtain
a positive result in the MOLY assay. Also of interest is the observation that mice given a relatively low
33
-------
level of arsenate, develop tumors in several organs that have counterparts in human epidemiology
studies (but not in the bladder) (Ng et al, 1999). With these animal models there is now a much more
reasonable path to pursue the mode(s) of action of arsenic and PB-PK analyses that may be
responsible for these tumors.
The level of various arsenic metabolites in the urine, the serum, and in the target tissues and cells
should be determined in these newly developed animal models. The most important questions are the
concentrations and forms of arsenic responsible for each of the tumor types identified in human studies.
When that information is obtained, truly useful pharmacokinetic models can be developed that will be
very important in identifying the concentrations and species of arsenic formed in humans that are
associated with carcinogenesis. This would help to determine the level of inorganic arsenic intake
required to obtain the effective levels of arsenic metabolites under different conditions of human
exposure. This will provide a much more comprehensive basis on which to determine the relative
importance of drinking water and food sources of arsenic.
There are also genetic factors that increase the susceptibility to bladder cancer that might
contribute to the background rate of tumor incidence that is independent of arsenic exposure. The
prevalence of the GST-mu null and certain polymorphic forms of NAT2* (Bell et al., 1993; Eaton and
Bammler, 1999) in the Taiwanese population (Chiou, et al. 1997) were not discussed by the NRC
(1999) and, we presume they were not explored. These variables have been important modifiers of
risks in smoking populations (Wen et al., 1994; Salagovic et al., 1999).
34
-------
4. TREATMENT TECHNOLOGY AND COST ISSUES
4.1 Comments on Treatment Technology Issues
4.1.1 Charge Question 4: Disposal Options. Based upon a review of the attached
materials, does the SAB believe that the EPA produced an accurate projection
of the likely disposal options for arsenic residuals and the distribution of these
options by treatment type? What are the SAB's views on the advantages and
the limitations of the various waste disposal options? What effect, if any, would
the SAB's analysis of these advantages and limitations have on the probabilities
assigned? What are the SAB's views on which options will be more likely used
by small systems (less than 10,000 people), and which will be more likely used
by larger ones?
The Panel believes that, based on the information provided to it by EPA, that the Agency
appears to have considered the spectrum of residual disposal alternatives. However, the Panel
questions whether certain alternatives will be viable due to potential constraints placed on utilities. For
example, the Panel believes that disposal of ion exchange (IX) or activated alumina (AA) treatment
residuals to a publically owned treatment work (POTW) might not be acceptable in the majority of
systems because of the high Total Dissolved Solids (TDS) concentration in those residuals. This is
especially problematic in the southwest where treated wastewater is reused for irrigation and
groundwater recharge and salt concentration is very important. Additionally, POTWs generally are
opposed to receiving dilute organic wastes that can reduce the efficiency of biological treatment. This is
the case in systems such as in Des Moines, Iowa. This would reduce the probability of selection for
those alternatives which rely on these disposal options to near zero.
Additionally, the Panel feels that the assumed non-hazardous classification of the waste brines
and sludges is questionable in the economic analysis. It is clear that in many cases in California the
wastes would be classified as hazardous because of the waste characterization procedures used their
and this could result in a public water supply choosing another alternative. Furthermore, the Panel has
concerns related to the Toxicity Characteristic Leaching Procedure (TCLP) test which is used in other
areas of the U.S. as the standard test for hazardous waste determination. The TCLP is designed to
maintain a pH of 5-6, which represents the best cast scenario for arsenic binding to sludge. Therefore,
while an arsenic-laden sludge may pass the TCLP test it may still leach arsenic into the groundwater
under normal pH conditions found in some landfills. Additionally, characterization of lime softening (LS)
sludge by the TCLP test is suspect because the target pH of the test (pH = 5) is likely to be
overwhelmed by the acid neutralizing capacity of LS sludge.
35
-------
4.1.2 Charge Question 5: Decision Tree for Treatment Technologies. Does the SAB
agree with the principal "branches" of EPA's decision tree described in the attached
documents and the likelihood that these options will be used for systems of various sizes
with various source water characteristics? What views does the SAB have on EPA's
description of the advantages and limitations of these treatment technologies? Would
the SAB's views on the these advantages and limitations affect the probabilities
assigned?
It was very difficult for the Panel to address this charge question without having the detailed
documentation on the decision tree that was used by EPA to predict technology selection and cost. As
a result the Panel was not able to follow, nor comment extensively upon, the "decision tree." Generally,
the Panel feels the cost estimates predicted for the rule, on the basis of the Agency's decision tree
analysis of applicable technology, appear to be low. From the limited information provided and from
presentations to the Panel at the June 5-7, 2000 DWC meeting, the model seems to have certain
deterministic and probabilistic components that make it quite complex.
In spite of the limitations noted above, the Panel does provide the following observations on
some of the assumptions used in the model:
a) The list of best available technologies (BATs) seems to overstate the real situation. It is
the opinion of the Panel that none of the technologies listed as BAT have been
demonstrated in full-scale operation for arsenic removal. While it is true that some of
the technologies are used in full-scale water treatment, they have not been operated
optimally for arsenic removal. This optimization may result in a substantially different
control strategy from the traditional operation.
b) The Panel is concerned that the list of BAT technologies may bias technology selection
by community water systems (CWSs), and particularly to bias selection against some of
the more promising emerging technologies [e.g., granular ferric hydroxide (GFH)].
c) The model does not appear to account for land acquisition cost. For groundwater
systems using multiple entry points, this may be a substantial cost when wells are
located on small lots of land within developed portions of a city.
d) It appears that the cost of replacement chemicals is not included in the cost of removing
arsenic. In particular the cost of fluoride replacement when the resulting concentration is
below optimum should be included in the cost of arsenic removal.
e) It is not clear that the monitoring burden and costs associated with point-of-use (POU)
and/or point-of-entry (POE) systems is adequately represented in the costs for these
technologies.
36
-------
f) It is not clear that EPA has considered the need for increased training and certification
of operators (or even availability of personnel) for a large number of very small systems.
The Panel is concerned that this might lead to closure of some of these systems and
increased reliance on private wells.
g) It is clear to the Panel that there are uncertainties contained in the model that result in
uncertainties in the output. It would be more appropriate to present the output with a
range of results than a discrete number. It is the feeling of the Panel that the range of
uncertainty is larger for an MCL of 5 |ig/L as compared to a value of 10 |ig/L or 20
4.2 Other Issues Associated with Cost
4.2.1 Affordability and Risk Tradeoffs
In the previous section of this report, the Panel discussed its concerns with EPA's designated
best available technologies (BAT) for arsenic removal. A related issue was raised during these
discussions which focuses on whether the identified technologies are affordable to small community
water systems. Even though the Agency notes that the listed BAT pass the affordability criterion and
thus will not result in the use of variance technologies, the Panel still is concerned that if the technologies
are too costly, they might force tradeoffs that do not maximize the gains to public health for persons in
those communities where a co-occurrence of small system size, high arsenic water concentrations, and
poor/susceptible population groups might exist.
The SAB previously commented on issues related to this concern when it commented on the
Agency's efforts to develop a national affordability criterion for the U.S. (SAB, 1998). In that advisory
the SAB called attention to the fact that there was no Agency definition of "national affordability" and it
anticipated difficulties in utilizing a national criterion in a rule that disproportionately impacts small
systems. The DWC comments here relate more to concerns about how affordability might affect
community and individual behavior and potentially force trade-offs that do not seem to have been
considered by EPA.
The Panel is encouraged that under the approach used to determine the need for variance
technologies that inequitable situations will not be created across systems of various sizes, that is, that
some systems will not be required to use technologies that lead to a greater risk than in other systems.
However, the Panel believes that the situation might be more complex than that which is addressed
under the affordability criterion itself. The Panel concern is with how individual households at the lower
end of the income distribution within an area might be limited in their ability to make tradeoffs that
influence their overall health status because of the involuntary allocation of additional income to lower
arsenic levels in their drinking water.
37
-------
A number of scenarios are possible. For example, it is possible that these same households
might, as a consequence of nutritional or other factors resulting from their economic situation or because
of an increased susceptibility, sustain a greater risk from elevated arsenic levels. In this case, it would
be reasonable to expect that the result of any disproportionate cost they would sustain in paying for
decreases in drinking water arsenic would be a greater than average gain in benefits and the two might
offset one another in a cost-benefit analysis. However, there could also be a situation in which some of
the increased arsenic risk, or other risks might be linked to nutritional factors. In these cases, allocation
of income to arsenic might preclude addressing nutritional factors that could possibly result in even
greater gains to the individual's health. It is important to note that this situation would apply to large as
well as small communities, because not all of the 13 million Americans that live below the poverty line
live in communities served by small water system. The Agency should consider the impact of these
multiple tradeoffs as it promulgates a final MCL and decide how they influence the final risk picture in
such communities.
The key to the above concern is who eventually has to pay for arsenic removal and what
actions they will take as a result. In most areas, households will have to pay for the cost of arsenic
removal. Too high treatment costs could shift these populations away from a small CWS to untreated
sources of lesser quality and treatment that could influence risks realized from microbial and other
contaminants, including arsenic. To avoid such outcomes the Agency and States might need to
consider how their loan and grant funds are used to help communities comply with the rule.
It is also not apparent to the Panel how the economic impact of other regulations promulgated
or in place might influence this problem. Arsenic control is but one of many issues that face drinking
water system as a result of new and potential rules (e.g., radon, disinfection byproducts, long-term
enhanced surface water rule, filter backwash, etc.). Compliance with this constellation of rules presents
additional tradeoffs that need to be considered in the evaluation of risks and in the use of the funds
available for drinking water treatment.
The Panel encourages the Agency to consider the issue of affordability in a broader context
than that addressing the limited issue of the affordability criterion and the use of variance technologies.
4.2.2 Need for Performance Data on Arsenic Technologies and the Possibility
of Adaptive Management
The Panel recognizes that selection of the MCL is within the policy domain of EPA and that it is
not just a scientifically derived number. In setting a Maximum Contaminant Level, EPA must consider
costs of implementing the rule and the benefits of decreased health risks. Considerable uncertainty
exists in the calculation of both costs and benefits for this rule. Even so, it was the unanimous opinion of
the DWC that the MCL needs to be reduced from its current level of 50 |ig/L. Each member of the
DWC has his/her own views about why the MCL should be more or less restrictive.
38
-------
Uncertainties in the health data exist and cause some to suggest that an even more cautious
MCL selection is appropriate while others suggest a higher MCL is warranted because the Agency
neglected many cautionary statements provided in the NRC document about the shortcomings of the
Taiwanese data for estimating risks in the U.S. The Panel believes that the uncertainties in applying risk
estimates from the study population in Taiwan to estimation of cancer and non-cancer risks from
arsenic in the U.S. are very large. Some of this concern over the uncertainty could be resolved if the
Agency would extend the analysis of the potential effects of the uncertain risk factors, thus following the
advice of the NRC to perform additional formal risk assessments on these data that consider how these
factors may have modified the responses to arsenic.
The Panel believes that there is uncertainty in both the health effects and the limited technology
and cost analysis information that it was provided by EPA. This could result in substantial uncertainty in
the final national cost and benefit estimates for reducing the arsenic MCL from 50 |ig/L to 5 |ig/L.
Further, this uncertainty is substantially greater at low values of the MCL (see the illustration in Fig 5
which uses an uncertainty factor of 30% as an example). As such, it appears that the outcome of
mandating such technologies across the country before reliable information on their performance is
available will be difficult to predict. The Panel suggests that the Agency consider whether it might be
appropriate to gather performance data for technologies identified by their decision tree as it has done
in some earlier situations. For example, such a rationale was used in the Information Collection Rule
(ICR) which required collection of data on the performance and cost of certain treatment technologies
during the Microbial/Disinfection Byproduct regulatory process. These data, which were not available
on a national basis, were needed before these treatment technologies were to be implemented by
utilities across the country.
When the uncertainties that arise from reliance on the Taiwanese data are combined with the
apparent overestimation of the lung cancer risks in the Agency's benefit analysis and the substantive
costs of implementing the lowest MCLs considered by EPA the Panel believes that EPA could judge
that it has sufficient grounds to consider an alternative MCL for arsenic under the discretionary authority
of the 1996 SDWA Amendments. There are technological uncertainties and they impact on
implementation costs. These considerations could be addressed by implementing the reduction in a
phased manner, that is allow a lesser reduction of the MCL, initially.
The recommendation for a stepwise approach could be supported by the following rationale:
a) Setting the initial MCL at a level intermediate between the current MCL and a long
term target would require the utilities with the highest level of arsenic to implement
arsenic treatment first. The resulting reduction in the number of systems that have to
initially comply will greatly reduce the cost of this rule. More importantly, it would
allow for the gathering of "real life" data on the performance and cost of various
technologies for arsenic removal without establishing a regulation that runs the risk of
39
-------
imposing very substantial costs on the nation prior to determining the full impact of
doing so.
b) It is noted that these arsenic treatment systems will utilize the same technologies set as
BAT by the EPA, and most of them (such as IX and AA) will produce waters that have
arsenic levels at or less than 3 jig/L virtually all the time. Therefore, if the later
rulemaking activity sets a lower MCL, these systems will be able to comply with these
lower MCLs without further incremental costs. Any subsequent lowering of the MCL
will simply increase the numbers of systems that will be required to treat to control
arsenic.
c) In addition, the installation of these treatment systems will also allow the EPA and the
industry to evaluate the validity of the assumption that the solid residuals from these
technologies can be disposed in municipal landfills. This issue has a significant impact on
the national cost of an arsenic MCL.
d) The feasibility of financing, designing, constructing, and commissioning of a large
number of treatment systems is in question. A stepped rule will allow for a more
practical implementation schedule.
e) Under the assumption of linearity, the efficiency of risk reduction is greater (i.e., the
costs per unit risk reduction are lower) at the higher arsenic concentrations, in other
words, the first increments of arsenic reduction below 50 jig/L are more cost-effective
than further reductions, at least for small systems (Figure 3B). If future research
establishes parameters for a sublinear dose-response curve, then the differences in
efficiency at the upper ranges of the exposure distribution relative to the lower ranges of
the exposure distribution would be even greater than the differences under a linear-dose
response curve.
40
-------
$900
^ $SOO
*-^
5 $700
"en $600
D
H $500
C $400
_ $300
O $200
IS
Z $100
5 10 15 20
Arsenic MCL, ug/L
Figure 5: Cost of a 30% Uncertainty (as an example) in EPA's National Cost Estimate
41
-------
REFERENCES
American Cancer Society. 2000. "Bladder Cancer - Overview." American Cancer Society Bladder
Cancer Resource Center on the ACS Website. See URL:
http://www3.cancer.org/cancerinfo/! oad_cont.asp?ct=44&doc=40&Language=English.
Aposhian, H.V., R.A. Zakharyan, E.K. Wildfang, S.M. Realty, J. Gailer, Radabaugh, G.M. Bogdan,
L.A. Powell, and M.M. Aposhian. 1999. How is inorganic arsenic detoxified? In: Arsenic
Exposure and Health Effects. W.R. Chappell, C.O. Abernathy, and R.L. Calderon, Eds.
Elsevier, Amsterdam, pp. 289-297.
Arnold, L.L., M. Cano, M. St. John, M. Eldan, M. van Gemert and S.M. Cohen. 1999. Effects of
dietary dimethylarsenic acid on the urine and uroepithelium of rats. Carcinogenesis 20: 2171-
2191.
AWWARF (American Water Works Association Research Foundation). 2000. Cost Implications
of a Lower Arsenic MCL. May, 2000. Pub. AWWA Research Foundation and AWWA.
Denver, CO and Washington, DC.
Bell, D. 1993. Genetic risk and carcinogen exposure: a common inherited defect of the carcinogen-
metabolism gene GSTM1 increases susceptibility to bladder cancer. JNCI&5: 1159-1164.
Carter, D.E., M. A. Peraza, F. Ayala-Fierro, E. Casarez, D.S. Barber, and S.L. Winski. 1999.
Arsenic metabolism after pulmonary exposure. In: Arsenic Exposure and Health Effects.
W.R. Chappell, C.O. Abernathy, and R.L. Calderon, Eds. Elsevier, Amsterdam, pp.299-309.
Caselmann, W.H. 1996. Trans-activation of cellular genes by hepatitis B virus proteins: A possible
mechanism of hepatocarcinogenesis. Advanced Virus Research 47:253-302.
Chiou, H-Y, Y.M. Hsueh, L.L. Hsieh, L.I. 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 CJ. Chen. 1997. Arsenic methylation
capacity, body retention, and null genotypes of GST Ml and Tl among current arsenic
exposed residents in Taiwan. Mutation Res 386: 197-207.
Clewell, H. J., A.M. Shipp, M.E. Andersen, J. W. Yager, K.S. Crump. 2000. Background
information for a June 6, 2000 presentation to the DWC entitled Application of the risk
assessment approaches in the USEPA proposed cancer guidelines to inorganic arsenic.
Cullen, W., B.C. McBride, and J. Reglinski. 1984. The reaction of methylarsenicals with thiols: some
biological implications. JInorgBiochem21: 179-194.
R-l
-------
Delnomdedieu, M., M. Basti, J. Otvos, and D. Thomas. 1993. Chemical Res. Toxicol. Vol. 6:598-
602.
Dixon, H. 1997. The biochemical action of arsonic acids especially as phosphate analogs. Adv Inorg
Chem44: 191-227.
ERG (Eastern Research Group). 1997. Report on the Expert Panel on Arsenic Carcinogenicity:
Review and Workshop. Prepared for the National Center for Environmental Assessment,
USEPA, Washington, DC by Eastern Research Group, Inc. Lexington, MA. EPA Contract
No. 68-C6-0041. August 1997.
Eaton, D. and T.K. Bammler. 1999. Concise review of the GSTs. Tox Sci 49: 156-164.
EPA. 2000. National Primary Drinking Water Regulations: Arsenic and Clarifications to Compliance
and New Source Contaminants Monitoring. Notice of proposed rule making. Federal
Register. Vol. 65:121. 38888-38983.
D.V.Frost. 1967. Arsenicals in biology-retrospect and prospect. Fed Proc 26: 194.
Helzlsouer, K.J., G.W. Comstock and IS. Morris. 1989. Selenium, lycopene, alpha-tocopherol,
beta-carotene, retinal, and subsequent bladder cancer. Cancer Res. 21:6144-6148.
Hopenhayn-Rich, C., S.R. Browning, I. Hertz-Piccotto, C. Ferriccio, C. Peralta, and H. Gibb. 2000.
Chronic arsenic exposure and risk of infant mortality in two areas of Chile. Envir. Health
Persp. 108:667-673.
Jauhiainen, M, K.J. Stevenson and PJ. Dolphin. 1988. Human plasma lecithin-cholesterol
acetyltransferase. The vicinal nature of cysteine 31 and cysteine 184 in the catalytic site. JBiol
Chem 263: 6525-6523.
Laib, RJ. and H. Moritz. 1989. Investigation of tumor initiating and/or cocarcinogenesis properties of
arsenite and arsenate with the rat liver foci bioassay. Exp. Pathol. 12:231 -233.
Lewis, D.R., J.W. Southwick, R. Ouellet-HellStrom, J. Rench, and R.L. Calderon. 1999. Drinking
water arsenic in Utah: A cohort mortality study. Environ. Health Persp. 107:359-365.
Lombard!, B. and M.L. Smith. 1994. Tumorigenesis, protooncogene activation, and other gene
abnormalities in methyl deficiency. J. Nutr. Biochem. 5:2-9
Macintosh, D.L., J.W. Southwick, D J. Hunter, L.A. Sampson, S.C. Morris, W.C. Willett, and E.B.
Rimm. 1997. Evaluation of a Food Frequency Questionnaire-Food Composition Approach for
Estimating Dietary Intake of Inorganic Arsenic and Methylmercury. Cancer Epidemiology,
Biomarkers, and Prevention. 6:1043-1050.
R-2
-------
Moore, M.M, K. Harrington-Brock, and C.L. Doerr. 1997. Relative genotoxic potency and its
methylated metabolits. Mutation Research. 386:279-290.
Morales, K.H., L. Ryan, K.G. Brown, T-L. Kuo, M-M. Wu, and C-J. Chen. 2000. Risk of internal
cancers from arsenic in drinking water. Environ. Health Persp.
Ng, J.C., A.A. Seawright, L. Qi, C.M. Garnett, B. Chiswell and M.R. Moore. 1999. Tumors in mice
induced by exposure to sodium arsenate in drinking water. In: (Chapell, W.R., Abernathy,
C.O. and Calderon, R.L. Eds.) Arsenic Exposure and Health Effects. 1999 Elsevier
Science B.V., Amsterdam.
NRC (National Research Council). 1989. Recommended Dietary Allowances. 10th Edition,
National Academy Press. Washington, DC.
NRC (National Research Council). 1999. Arsenic in Drinking Water. National Academy Press.
Washington DC:
Pan, T.C., Y.L. Chen, and WJ. Wu. 1996. Serum trace metals in Blackfoot disease patients. Kao
Hsiung I Hsueh Ko Hsueh Tsa Chic. 12(10):555-560.
Patrick, H., R. Voitle, H. Hyre and W.G. Martin. 1964. Incorporation of arsenic into cock sperm.
PSEBM117:365.
Petrick, J.S., F. Ayala-Fierro, W.R. Cullen, D.E. Carter and H.V. Aposhian. 2000.
Monomethylarsonous acid (MMATTT) is more toxic than arsenite in Chang human hepatocytes.
Toxicol. Appl. Pharmacol. 163:203 -207
SAB (Science Advisory Board). 1998. An SAB Advisory on the National-Level Affordability
Criteria and Technologies for Small Systems Under the 1996 Amendments to the Safe Drinking
Water Act/EPA-SAB-DWC-ADV-99-OOl, Dec. 21, 1998
Saito, R., E. Jahnke-Spinnenweber, H. Sinozuka and B. Lombardi. 1994. On the role of
compensatory mitogenesis in the hepatocarcinogenicity of choline and multiple-lipotrope devoid
diets. Carcinogenesis 15:1413-1419.
Salagovic, L, I. Kalina, V. Habalova, M. Hrivnak, L. Valansky, and E. Biros. 1999. The role of
human glutathione S-transferases Ml and Tl in individual susceptibility to bladder cancer.
Salonen, J. T., R. Salonen, R. Lappetelainen, P.H. Maenpaa, G. Alfthan and P. Puska. 1985. Risk
of cancer in relation to serum concentrations of selenium and vitamins A and E: matched case-
control analysis of prospective data. Br. Med. J. (Clin Res Ed) 290:417-420.
R-3
-------
Schoof, R.A., LJ. 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 and Chemical Toxicology.
37:839-846.
Shirachi, D.Y, M.G. Johansen, J.P. McGowen and S.H. Tu. 1983. Tumorigenic effect of sodium
arsenite in rat kidney. Proc. Western. Pharmacol. Soc. 26:413-415.
Styblo, M., L. Vega, D.R. Germolec, M.I. Luster, L.M. Del Razo, C. Wang, W.R. Cullen, and D. J.
Thomas. 1999. Metabolism and toxicity of arsenicals in cultured cells. In: Arsenic Exposure
and Health Effects. W.R. Chappell, C.O. Abernathy, andR.L. Calderon, Eds. Elsevier,
Amsterdam, pp.311-323.
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 toxi city of trivalent and
pentavalent inorganic and methylated arsenicals in rat and human cells. Arch. Toxicol. In
Press.
van den Brendt, P.A., R.A. Goldbohm, P. van't Veer, P. Bode, E. Dorant, RJ. Hermus and F.
Sturmans. 1993. A prospective cohort study on selenium status and the risk of lung cancer.
Cancer Res. 20:4860-4865.
Wei, M., H. Wanibuchi, S. Yamamoto, W. Li, and S. Fukushina. 1999. Urinary bladder
carcinogenicity of dimethylarsinic acid in male Fisher 344 rats. Carcinogenesis. September.
20(9): 1837-44.
Wen, C.P., S. Tsai, and D. Yen. 1994. The health impact of smoking in Taiwan. Asia J. of Public
Health. Vol. 7:206-213.
WHO (World Health Organization). 1996. Trace Elements in Human Nutrition and Health. Food
and Agriculture Organization of the United States - International Atomic Energy Agency -
World Health Organization. Geneva: World Health Organization.
Yamamoto, S.Y., T. Konishi, T. Matsuda, T. Murai, M.A. Shibata, S. Matsui-Yuasa, K. Otani, K.
Kuroda, G. Endo and S. Fukushina. 1995. Cancer induction by an organic arsenic compound,
dimethylarsinic acid (cacodylic acid) in F344/DuCrj rats after pretreatment with five
carcinogens. Cancer Res. 55:1271-1276.
Zhao, C.Q., M.R Young, B. Diwan, T.P. Coogan and M.P. Waalkes. 1997. Association of arsenic-
induced malignant transformation with DNA hypomethylation and aberrant gene expression.
Proc. Natl. Acad. Sci. (USA) 94:10907-10912.
R-4
-------
ATTACHMENT A
A MINORITY REPORT ON ARSENIC IN DRINKING WATER: THE UNIQUE
SUSCEPTIBILITY OF CHILDREN TO ARSENIC
DR. JOHN F. ROSEN
PROFESSOR OF PEDIATRICS
HEAD, DIVISION OF ENVIRONMENTAL SCIENCES
THE CHILDREN'S HOSPITAL AT MONTEFIORE
THE ALBERT EINSTEIN COLLEGE OF MEDICINE
THE BRONX, NY 10467
*****SUBMITTED TO THE US EPA'S SCIENTIFIC ADVISORY BOARD ON ARSENIC
IN DRINKING WATER ON SEPTEMBER 1, 2000*****
I. SOME PRINCIPLES OFPEDIATRIC AND DEVELOPMENTAL TOXICOLOGY.
Some basic principles of pediatric-developmental toxicology are presented below. In the
following section(H), the unique susceptibility of young children to arsenic is outlined, within the context,
phrased simply, that children are not young adults; differences in diet, metabolism, body weight,
variable age groups, consumption of water, toxic effects of metal pollutants in a rapidly growing
organism, and exposure estimates per unit of body weight are essential ingredients of risk assessment in
A-l
-------
children. In brief, determination of safe levels of exposure to arsenic should take into consideration
physiological factors that can place the fetus, infant and young child at greater risk of harmful health
effects than adults. Differences in size, maturity of biochemical and physiological functions in major
body systems, and variation in body composition(water, fat, protein and mineral content) all can
modulate the severity of toxicity to any toxicant in a rapidly developing fetus-infant-child, including
arsenic. Because newborns are the group most different anatomically and physiologically from adults,
they can exhibit the most pronounced quantitative differences in sensitivity and susceptibility to
environmental toxicants, including arsenic(\-6).
Furthermore, uncertainty factors are widely used to establish guidelines for human exposure.
This is often accomplished by dividing the no-observed-effect-level(NOEL) by an uncertainty factor of
100 in animal studies. This factor comprises two separate factors of 10-fold each: one allows for
uncertainty in extrapolating experimental data to humans; and the other accommodates variation within
human populations. To provide added protection during early development, a third uncertainty factor of
10 is applied to the NOEL to develop the RfD. Because there exist uncertainty factors relating to
susceptibility and vulnerability during early fetal, neonatal and childhood developmental toxicity, an
additional 10-fold factor is used by EPA and FDA when testing data relative to children is incomplete.
This is not a new or additional uncertainty factor but an extended application of uncertainty factors
routinely used by agencies of the U.S. Government(1-9). In risk assessment, when there is some
level of uncertainty relating to the overall quality of available data, an additional factor, typically 3-
fold, is included as a "modifying" factor. In summary, there are unique risks and increased
susceptibility of the fetus, young infant and child to damage from environmental chemicals, including
arsenic. These risk assessment paradigms are recognized broadly by the U.S. government. Arsenic fits
directly into the above paradigms.
II. THE UNIQUE SUSCEPTIBILITY OF YOUNG CHILDREN TO EXCESSIVE
TOXICANT EXPOSURES:
The NRC/NAS(10) recognized that a margin of safety may be needed when conducting risk
assessments of arsenic, because of variations in the sensitivity of individual subpopulations.
Some general concepts followed by specific examples in different organ systems are provided below.
Children are a unique population; and their risks can differ qualitatively and quantitatively from
those in adults(10,ll) These differences include organ systems that are
primarily affected by arsenic, such as the central nervous system, cardiovascular development,
reproductive and developmental organs and cancinogenesis(lO).
Physiologically, respiratory and circulatory flow rates, as well as cellular proliferative rates, in
many organs, are greater in children versus adults. From a metabolic standpoint, some enzymatic
pathways are more efficient in the young(the P450s peak in adolescence) and others are far less
effective in young children, such as glucuronidation. Developmental changes in cell permeability,
binding and storage modulate the distribution and excretion of xenobiotics. The amount of water
A-2
-------
intake(see below) and dietary status differ in young children compared with adults. From an
environmental standpoint, living space and habits, specific to neonates and young children, are highly
specific for these young age groups. Clearly, variations in chemical sensitivity and exposure exist in
children in contrast to adults; and developmental changes from fetal to newborn to postneonatal to
adolescence periods are superimposed on genetic and environmental variables in the young child, which
are evidently different from those in adults. Moreover, early exposure in infancy to toxicant metals,
such as lead(12) or arsenic(13) can lead to latent adverse health effects that become manifest later
during adulthood.
Excretory capacity, in relation to the kidney, undergoes a considerable amount of maturation
with aging. Renal clearance is reduced at birth and gradually matures over the first few years of life;
and similar maturation in the liver, in the metabolism of xenobiotics, also occurs with aging.
In terms of water intake, body mass and cellular proliferation, in the brain, for instance, the
differences between young children and adults are marked. 1) The body surface:body mass ratio
declines by about 66% from infancy to the adult years; 2) Brain growth is extremely rapid in the first
two years of life. About 75% of all brain cell types are present by the age of two years; and the brain
represents a considerably larger portion of an infant's body mass compared to that in adults. Cerebral
blood flow is also far more robust: a ten year-old has a flow rate of about SOL/Kg brain weight
compared to about 40L in a 65 year-old adult. Thus, the younger the child, both brain mass and
cerebral blood flow are considerably greater in contrast to adult values; 3) The Tolerance Assessment
System(TAS), used by the US EPA, indicates unequivocally that infants and young children
consume the highest amount of water per unit body weight during their entire lifetimes. An
infant, a 1 to 6 year-old, a 7-12 year-old children consume 28 grams of water/Kg body weight/day, 30
grams of water/Kg body weight/day and 17 grams of water/Kg body weight/day, respectively, in
contrast to an adult who typically consumes about 10 grams of water/Kg body weight/day. Obviously,
the intake of arsenic from drinking water will be greatest within the pediatric age group (11).
In addition to all the above differences between children and adults, it can be concluded
that infants and young children, in contrast to adults, have different exposures to toxic metals
in water, have different life expectancies, absorb and maintain unique internal doses of a toxic
metal from similar external exposures and respond differently and specifically to the same internal dose
of a toxic metal.
IIL THE CENTRAL NERVOUS SYSTEM(CNS):
Development of the human CNS involves the production of 100 billion nerve cells and 1 trillion
glial cells(5). Once produced, these cells undergo migration, synaptogenesis, selective cell loss and
myelination. This progression occurs unidirectionally. Thus, inhibition at one developmental stage can
cause alterations to subsequent processes. Developmental stages occur in temporally distinct time
frames across different brain regions thereby making the brain heterogeneous in response to agents that
interfere with specific processes. Unlike other organ systems, the unidirectional nature of CNS
development limits the capacity of brain tissue to compensate for environmentally induced cell loss.
A-3
-------
Maintenance of this rigid temporal and spatial schedule allows the brain to develop its various functions.
It is this developmental complexity that underlies the sensitivity of the CNS to environmental insults and
emphasizes the unique characteristics of development which place children at special risk from
environmental exposures. In 1984, the US EPA indicated that children may be more susceptible to
arsenic-induced CNS damage(14). For example, severe CNS deficits were observed in children
exposed as babies to arsenic-contaminated powdered milk formulas. Follow up of these babies into
childhood revealed an increased incidence of severe hearing loss, abnormal EEG patterns, and an
increased prevalence of mental deficiency, seizures and other indices of severe brain damage.
IV. THE REPRODUCTIVE SYSTEM:
Toxicant exposure is known to affect critical events in the development of the reproductive
system(15). Once exposure has sufficient influence on essential reproductive events, adult reproductive
competency is reduced or abrogated. Critical windows of development are limited temporally and
characterized by occurrence of sets of organizational events that constitute periods during which
exposure can have effects on later reproductive competency. Environmental exposures can influence
fertility to early embryo loss. Although early embryogenesis is a critical target of toxicants,
preconceptional and even postnatal exposures may also adversely affect the reproductive system and
progeny outcomes.
V. THE CARDIOVASCULAR SYSTEM:
Three to eight weeks following fertilization is the critical time period of organ(heart) formation.
At that time, stem cell populations for organ morphogenesis are established and inductive events of
differentiation occur(16). During this time, structural defects in the heart can ensue. It is this very
narrow time frame when the anlage of the heart is first established. Moreover, later adverse effects can
occur as various cell types begin to differentiate.
VI. CARCINOGENESIS:
The relevance of carcinogenesis to at risk children is discussed on pages 5 and 6 (Section
VII.4)
VII. THE IMPACT OF ARSENIC ON CHILDREN:
1) The NAS/NRC document on arsenic(lO) concluded that the current MCL of 50 ug/L does
not achieve EPA's goal for public health protection and, therefore, requires downward revision as
promptly as possible. Similarly, Morales, Ryan et al, (13) also concluded that 50 ug/L of arsenic is
associated with a substantial increased risk of cancer; and this MCL is not sufficient to protect the
public's health. For adults, EPA found that the safe level of arsenic, to avoid non-cancer diseases, is
0.3 ug/kg/day(U.S EPA. IRIS online: Arsenic, inorganic: 4/10/98, 0278, off the internet 5/15/00). To
extrapolate this value to young children requires dividing the above intake by a factor of at least 6-
10(see above and below). Morever, if the daily intake of arsenic by an adult is about 21 ug/day, and
A-4
-------
the water intake is about two liters, this equals arsenic at a concentration of about 10 ug/L. For
Superfund sites of which I am familiar, (Palmerton and Throop, PA and Kellogg, ID), this places
arsenic in the category of an hazardous toxicant at Superfund sites for adults, without any
consideration for the increased susceptibility of young children. As defined by EPA, any MCL above 5
ug/L is, at the very least, an hazard to human health and even more so for children(see below).
2) In children, the arsenic dose per unit of body weight is about 6-fold times higher than in
adults(17). Calderon et al.(17) concluded that the age-dependent difference in arsenic urinary
concentrations can be attributed to the higher dose per unit body weight in children versus adults.
3) In a broad range of ages, children do not detoxify arsenic as efficiently as in adults (18, 19).
The net result in children is that increased amounts of arsenic are available to produce toxic effects.
4) Toxic exposures to the fetus and in childhood are recognized determinants of cancer in
adulthood(20-23); and such periods of latency have been demonstrated for hepatitis B exposure in
infancy leading to hepatocellular carcinoma in adults(20). Morever, models exist of multi-step
carcinogenesis incorporating initiation and progression to latent expression of disease. Toxicant
exposure during conception or pregnancy can provide the initial mutational event that provides
increased risk of cancer during adulthood. An adult will be at higher risk for cancer, once a germline
alteration occurs; toxicant exposure can lead to somatic alterations postnatally with long latency
periods(20-23).
Children have a general sensitivity to carcinogens that can be demonstrated by early biomarkers
of cancer; and this may foretell an unique sensitivity in childhood even when cancer latency is
long(22,23). It is important to point out that cancer biomarkers in young children vary considerably
with ethnicity(22); and this observation may place specific ethnic groups of children at higher risk for
developing arsenic-induced cancer as adults(22). Although there is presently an absence of
longitudinal studies of excessively exposed young children to arsenic, ultimately leading to cancer in
adulthood, the pathophysiological frame work exists in the fetus, infant and young child for such events
to occur. Thus, the fetus, infant and young children should be considered to be at increased risk for
developing arsenic-induced cancers after long latency periods.
5) To dismiss the Taiwanese data(24) in young and older children in this country is a simplistic
approach to this country's pediatric population. The majority report posits that all American children,
exposed to arsenic, have a nutritional status that is complete compared to the Taiwanese population.
"If individuals in the Taiwan endemic zone were at added risk for arsenic effects by virtue of poor
nutritional status, then individuals anywhere with this risk factor are of concern(25)." There are about
13 million American children who are living below the poverty line today in the United StatesQSjew
York Times. 8/13/00) of diverse ethnicity(African-American, Hispanic and Native American children);
and these subpopulations of American children, except, perhaps, for selenium, are more likely than not
to be in poor nutritional status. Indeed, Smith(26) found the prevalence of skin lesions among men and
children in a relatively small population, who had been drinking water containing excessive quantities of
arsenic for decades within Northern Chile, was similar to the prevalence of these arsenic induced skin
A-5
-------
lesions reported from Taiwan and West Bengal(27). However, the North Chilean population was
nutritionally sound in contrast to malnutrition reported from Taiwan and West Bengal. Although the
sample size was small in the North Chilean population, the findings were robust.
The above reality is especially operative for arsenic and nutritionally at-risk children in America.
Many of the areas of the United States, which contain relatively high-risk fractions of particularly at-risk
children, including Native American children, are also those areas where water arsenic levels are high,
such as in desert areas of the Southwest. Arsenic-laced water consumption among risk groups in the
Western United States would parallel the case for the Taiwanese, even if one were to accept that
nutritional deficiencies were pivotal and determinative for carcinogenic and other non-cancer outcomes
in the Taiwanese people.
Welch et al.(28) recently summarized data from the United States Geological Survey at an
International Conference on arsenic exposure and health effects. Analyses were based upon over 17,
000 analyses of arsenic recorded by the USGS National Water Information System(NWIS). NWIS
data revealed that Western areas of the United States have significantly higher rates of exceedences
using any standard cut-point for arsenic (current or proposed EPA, current WHO) compared to water
supplies in the East. These data are also in agreement with a related survey, namely, the National
Arsenic Occurrence Survey(29).
Native American children living in the Western U.S., particularly reservation populations in
desert areas of the Southwest are subjected to the poorest and most risk-producing factors for adverse
effects of arsenic in America. They have higher rates of poverty and typically have higher rates of
nutritional deficiencies in contrast to other demographic and socioeconomic subpopulations in the
United States. Furthermore, a number of Native American tribes have contemporary diets that are
clinically recognized as predisposing to diverse chronic diseases, for example, cardiovascular and
cancer-based adverse health effects. These factors pose additive and, perhaps, synergistic risks
together with excessive arsenic intake from water. This is especially relevant to areas of the
Southwestern United States.
Ballew(30) recently described data for the Navajo in a "Navajo HANES"(Health and Nutrition
Survey) a demographic and ethnic spin-off of the NHANES surveys, similar to the "Hispanic HANES"
carried out in the 1980s as an adjunct to the 1976-1980 NHANES n. Navajo diets are typically low
in important sources of vitamins and minerals(30), as are the diets of the Hopi and the Pima(31,32).
The U.S. EPA and SAB Committee cannot claim ignorance of the potential consequences of
nationally high-risk Native-American children(and adults), because the 1997 Exposure Factors
Handbook, widely used by risk assessors in various U.S. regulatory scenarios, includes coverage of
this issue of intakes and diets of Native Americans(33). This multi-volume EPA document presents
information for Native American tribes based upon data from four of these.
A-6
-------
Overall, the SAB Committee and U.S. EPA must address the fact that, when one looks at high
water levels of arsenic that simultaneously serve as a potential water source for nutritionally-deprived
and otherwise at risk Native American children and adults(who are also likely to have increased intakes
of water arsenic and consume animal herds who also drink arsenic-contaminated water), these
populations would be similar to the Taiwanese in terms of exposure and nutrition. The article by Harris
and Harper(34) should be consulted to examine the extent to which intakes of toxicant-contaminated
media are remarkably different and much higher than non-Native American populations.
My own direct experience as a clinician working with Navajo includes recently published
findings that arsenic and other chemical toxicants, in tandem with radionuclides, will, in fact, produce
toxic harm in utero and post-natally in Navajo children(35,36). A detailed clinical and risk assessment
evaluation of two Navajo sisters, exposed in utero and in early life to arsenic and other toxicants in pit
waters was carried out. They were excessively exposed to arsenic during their pastoral family activities
of herding the family's sheep on the Navajo reservation in Arizona. The net CNS result was a severe
toxic peripheral neuropathy and CNS cortical disease.
Through direct interviews with Navajo family members, it was ascertained that many of the intake
parameters(water, in particular), described by Harris and Harper(34) as potentially elevated, were in
fact markedly elevated. This Navajo family spent its herding existence within an highly arid environment
coming into contact and consuming higher amounts of arsenic contaminated water than typical children
or even Native American children, who did not engage in pastoral activity.
In view of the above discussion, it is reasonable to conclude that the subpopulation of
American children are at higher risk for arsenic-related disease than others from a nutritional standpoint,
if the postulate(relating to malnutrition) is as strongly supported as it is in the majority report.
6) The recent article by Hopenhayn-Rich et al.(37) reported elevated late fetal, neonatal and
postnatal mortality in a Chilean town(Antofagasta) with high levels of arsenic in drinking water
compared to a control town(Valparaiso), where arsenic levels in drinking water were less than 5 ug/L.
Similar results, reported from Bulgaria, including congenital malformations(38), from Texas(39) and
from other parts of the United States, including congenital cardiovascular malformations, and
spontaneous abortions, collectively support the view of increased susceptibility of the fetus and neonate
to arsenic (40-42). Drinking water levels of arsenic decreased in Antofagasta from 1961 on, so did the
prevalences in fetal mortality rate, neonatal mortality rate and postnatal mortality rates. In contrast,
when arsenic water levels were elevated pre-1961, the combined mortality rate was 68 deaths per
1000 births.
More specifically, these data reflect a dose-response curve that is typically found in the field of
toxicology(43-50). From 1974-1977, although mortality rates were still elevated in Antofagasta, a
gradual decline in these rates was observed as drinking water concentrations of arsenic decreased from
860 to 110 ug/L. In the period of about 1978-1982, the mortality rates in both towns were similar; but
the rate of decline in Antofagasta was far more pronounced over the preceding years than that in
A-7
-------
Valaparaiso. Moreover, the rate of decline into the 1980s was the most pronounced for postneonatal
mortality in Antofagasta, with more gradual reductions in
neonatal and late fetal mortality. Arsenic levels in this town were decreased to 40 ug/L from the
previous value of 70 ug/L. Statistically, Poisson regression analyses(with relatively few data points)
were used to fit the mortality rates as a function of the estimated exposure to arsenic by log-linear
models adjusting for location and calendar time(37). Arsenic values for Valparaiso were measured by
the Chilean government and the authors; levels less than 5 ug/L were
reported. Data collected from water companies during 1990-1994 found that arsenic levels in
Valparaiso were below the analytical detection limit of 20 ug/L.
In the report by Hopenhayn-Rich and co-workers(37), data were analyzed in 4-year blocks of
time, because there was considerable variation in annual mortality rates. Nonetheless, in the majority
report, these originally reported data were analyzed year-by-year; and it was concluded that a dose-
response curve was absent. It is scientifically unsound to re-calculate original data by artificially creating
unreported data. Using the original data, as reported, analyses were carried out by linear regression,
Spearman's rank correlations, and ANOVA. The p values from these three statistical methods ranged
from O.044 to <0.01 thereby indicating a typical toxicological dose-response curve as stated
above(Rosen: Unpublished observations).
Hopenhayn-Rich(37) did acknowledge the possibility of confounders in ecologic studies, such
as the design of their study. However, the distinct temporal pattern of infant mortality rates in
Antofagasta compared to Valparaiso argued strongly against individual-level confounders; and the
changes in the arsenic levels in the water was an "indisputable" event. The authors concluded that "the
results of this study indicate that exposure to inorganic arsenic from public water supplies may be
associated with increased risk of infant mortality. Specifically, these data suggest that arsenic exposure
may represent a greater risk for late fetal mortality with lower, but still elevated, risk for neonatal and
postneonatal mortality."(3 7).
The findings of Hopenhayn-Rich(37) are consistent withe the report of Concha et al.(51),
which showed that ingested arsenic crosses the placenta during pregnancy, producing fetal exposure, as
indexed by levels of arsenic in cord blood. The levels of arsenic in cord blood approached those
measured in maternal samples. While this study appeared to show that arsenic metabolites were
present, at this time, it cannot be ruled out that these metabolites were lexicologically inconsequential.
In fact, these data gain increased support for their toxicological significance from the findings of
Hopenhayn-Rich(37), which are consistent with animal studies
summarized in the NAS/NRC report(lO). As noted(lO), animal species do show reproductive and
developmental effects of arsenic evidenced by birth defects, impaired fetal growth, and reduced survival
rates for fetal and newborn animals.
Collectively, it can be concluded that the above data indicate that young children are an
uniquely susceptible population for adverse health effects of arsenic. Safety information based upon
data from adults, in view of all the aforementioned differences between young children and adults, are
A-8
-------
highly unlikely to effectively protect children, as a subpopulation most at risk. In the interests of public
health, the population of the developing fetus, neonate and young infant
should be rigorously protected by considerable lowering of the MCL for arsenic to the very lowest
level that is analytically reliable. A step-wise "phase-down" of the MCL will not protect this susceptible
population.
VIII. CONCLUSIONS.
From a public health point of view, establishing new guidelines in drinking water for a potent
toxic metal, namely, arsenic, requires protecting the most susceptible population. In this instance, the
population includes the developing fetus, neonate, infant and young child. To protect this susceptible
population now, the MCL for arsenic should be as low as analytically feasible. Any type of "phased-
in" approach, above that which is analytically possible, will fail to protect a large population of
susceptible young children.
IX. THE MEDICAL AND PUBLIC HEALTH NEED FOR AN EPA DRINKING WATER
HEAL THAD VISOR Y.
I strongly endorse the need for the U.S. EPA to issue a health advisory for arsenic in
drinking water. I do so within my knowledge of the current data base supporting the need for such an
advisory. It is my understanding that EPA has issued such advisories on numerous occasions. These
can be documented by anyone on the SAB Committee examining the online IRIS file for the many
contaminants contained therein. An explicit section in each of these files refers to health advisories. In
my informed opinion, the evidence for the need of such an advisory is compelling, as is my
understanding of EPA's requirement to do so.
The evidence that compels such an advisory, particularly for those regions in the United States
where water supplies of arsenic are elevated, is clear from the voluminous evidence for arsenic within its
toxicological and epidemiological context. This evidence indicates that the current MCL is inadequate;
and that currently available science dictates a drastic downward revision. While a substantial revision
must follow along a feasible track for implementation, arsenic does not await imparting toxic effects
while various regulatory frameworks become operative. Children specifically will continue to be
exposed while control measures are put into place by EPA. Therefore, the U. S. EPA must take
cognizance of the public health realities-that between on-going intoxication and practical needs for
implementation time frames-by using
the advisory as a mechanism of public health awareness and education. The mechanisms for how an
advisory is issued are, generally, in place and have been used extensively in the past. No deviance from
this process is necessary.
A-9
-------
BIBLIOGRAPHY:
1. Developmental Toxicology, edited by Hayes, Thomas and Gardner. Raven Press, pp. 15-189,
1994.
2. Neuropsychological Toxicology: Identification and Assessment of Human Neurotoxic
Syndromes, edited by D.E. Hartman. Plenum Press, pp.9-44,1995.
3. Handbook of Developmental Toxicology, edited by R.D. Hood. CRC Press, pp.597-667, 1996.
4. Goldman, LR and Koduru, S.: Chemicals in the environment and developmental toxicity to children:
A public health and policy perspective. Environmental Health Perspect. 108: 443-448, 2000.
5. Faustman et al., Mechanisms underlying children's susceptibility to environmental toxicants.
Environmental Health Perspect. 108: 13-21,2000
6. Environmental Health Perspect. SUPPLEMENT 3, 108,2000
7. U.S Congress. Titles 7 and 21.Food Quality Protection Act, 1996.
8. The "Boxer"Amendment to the 1996 Safe Drinking Water Act 1412(6)(1)(C),(6)(3)(C)
(5),1457(a).
9.Clinton, WJ: Executive Order 13045. Protection of Children from Environmental Health and Safety
Risks. Wasinghington, D.C., 1997).
10. Arsenic in Drinking Water. National Research Council. National Academy of Sciences. 1999.
Washington, D.C.
11. Similarities and Differences Between Children and Adults. P.S. Guzelian, CJ. Carol, S.S.
Olin(eds). 1992. International Life Sciences Institute, Washington, D.C.
12. Rosen, J.F. Adverse health effects of lead at low expossure levels: Trends in the management of
childhood lead poisoning. Arch. Toxicol. 97:11-17, 1995.
13. Morales et al., Risk of internal cancers from arsenic in drinking water. Environmental Health
Perspect. 108: 655-661, 2000.
14. Inorganic Arsenic: Final Report. U.S. EPA, 1984
15. Pryor, J.L. et al., Critical windows of exposure for children's health: The reproductive system in
animals and humans. Environmental Health Perspect.108: 491-503, 2000.
16. Sadler, T.W. : Susceptible periods during embryogenesis of the heart and endocrine glands.
Environmental Health Perspect.108: 555-562, 2000..
17. Calderon et al. Excretion of arsenic in urine as a function of exposure to arsenic in drinking water.
Environmental Health Perspect. 107: 663-667, 1999
18. Concha et al., Metabolism of inorganic arsenic in children with high arsenic exposure in northern
Argentina. Environ. Health Perspect. 106: 355-359, 1998.
19. Kurttio et al., Urinary excretion of arsenic species after exposure to arsenic present in drinking
water. Arch. Environ. Contam. Toxicol. 34: 297-305, 1998
20. Anderson, L.M. et al.: Critical windows of exposure for children's health: Cancer in human
epidemiological studies and neoplasms in experimental animal models. Environmental Health
Perspect. 108: 573-594.
21. Olshan, A.F. et al: Workshop to identify critical windows of exposure for children's health: Cancer
work group summary. Environmental Health Perspect. 108: 595-597, 2000.
A-10
-------
22. Tang, D. et. al: Molecular and genetic damage from environmental tobacco smoke in young
children Cancer Epidemiol. Biomarkers, Prev 8: 427-431, 1999
23. Perera, P.P. Environment and cancer: Who are susceptible? Science 287: 1068-1073, 1997.
24. Tseng, W.P. et al.: Prevalence of skin cancer in an endemic area of chronic arsenicism in Taiwan.
J. Natl. Cancer Instit. 40: 453-463, 1968
25. Mushak, P., and Crocetti, A.M.F. Risk and revisionism in arsenic cancer risk assessment.
Environmental Health Perspect. 103: 684-689, 1995
26. Smith, A.H. et al. Arsenic-induced skin lesions among Atacameno people in Northern Chile
despite good nutrition and centuries of exposure. Environmental Health Perspect. 108: 617-620,
2000.
27. Rahman, M. et al. Diabetes mellitus associated with arsenic exposure in Bangladesh. Am. J.
Epidemiol. 148: 198-203, 1998.
28. Welch, A.H. et al.: Arsenic in ground water supplies of the United States. In: Arsenic Exposure
and Health Effects. W.R. Chappell, C.O. Abernathy, and Calderon, R.L.(eds). pp. 9-17, 1999.
Elsevier Science, New York.
29. Frey, M.M., Edwards, M.A.: Surveying arsenic occurrence. J. Amer. Water Works Assoc.
89:105-117, 1997.
30. Ballew, C. et al.: Intake of nutrients and food sources of nutrients among Navajo:Findings from the
Navajo Health and Nutrition Survey. J. Nutr(Suppl 10):2085S-2093S, 1997.
31. Brown, A.C., Benton, B.: Dietary survey of Hopi Native American elementary students. J.
Amer.Diet. Assoc. 94: 517-522, 1994.
32. Smith, C.J. et al.: Survey of the diet of Pima Indians using quantitative food frequency assessment
and 24-hour recall: Diabetic renal disease study. J. Amer. Diet. Assoc. 96: 778-784,
1996.
33. US EPA Exposure Factors Handbook. Volumes I-HI.(An Up-date to Exposure Factors
Handbook: (EPA/600/8-89/043, 3/89). Report No. EPA/600/R-97/006,12/96: Office of Solid
Waste and Emergency Response.
34. Harris, S.G., Harper, B.L.: A Native American exposure scenario. Risk Analysis 17: 789-795,
1998.
35. Rosen, J.F., Mushak, P.: Metal-and radiation-induced toxic neuropathy(TN) in two Navajo sisters.
Pediat. Res. 45: 346A, 1999.
36. Rosen, J.F. and Mushak, P.: Metal and radiation-induced toxic neuropathy in two Navajo sisters.
lexicologist 54: 80, 2000.
37. Hopenhayn-Rich et al. Chronic exposure and risk of infant mortality in two areas of Chile.
Environmental Health Perspect. 108: 667-673, 2000
38. Zelikoff et al. Health risks associated with prenatal metal exposure. Fundam. Appl. Toxicol. 25:
161-170, 1995.
39. Ihrig et al. A hospital-based controlled study of stillbirths and environmental exposure to arsenic
using an atmospheric dispersion model linked to a geographical information system. Epidemiology 9:
290-294, 1998.
40. Engel, R.E. and Smith, A.H. Arsenic in drinking water and mortality from vascular disease: An
ecological analysis in 30 counties in the United States. Arch. Environ. Health 49: 418-427, 1994.
A-ll
-------
41. Zierler et al. Chemical quality of maternal drinking water and congenital heart disease. Int. J.
Epidemiol. 17: 589-594, 1988.
42. Arsenichengrau et al. Quality od community drinking water and the occurrence of spontaneous
abortion. Arch. Environ. Health 44: 283-290
43. Environmental Toxicology and Pharmacology of Human Development. S. Kacew and G.H.
Lambert(eds). 1984. Taylor and Francis. Washington, D.C.
44. Reproductive Toxicology. A.W.Hayes, J.A.Thomas and D.E. Gardner(eds). 1995. Raven
Press, New York City.
45. Human Toxicology. J. Descortes (ed). 1996. Elsevier, Amsterdam, New York, Oxford,
Shannon, Tokyo.
46. Environmental Epidemiology: Public Health and Hazardous Wastes. NAS/NRC.
1991.Washington, D.C.
47. Environmental Epidemiology: Use of the Gray Literature and other Data in Environmental
Epidemiology NAS/NRC. 1997. Washington, D.C.
48. Clinical Epidemiology: A Basic Science for Clinical Medicine. D.L. Sackett, R.B. Haynes,
GH Guyatt and P Tugwell (eds).1991. Little, Brown and Company. Boston, Toronto and London.
49. Handbook of Statistics: Environmental Statistics GP Patil, C.R. Rao (eds) 1994 North-
Holland. Amsterdam, London, New York, Tokyo.
50 . Statistics and Experimental Design for Toxicologists. C. Cad(ed). 1998. CRC Press, Boca
Raton, London, New York, Washington, D.C.
51. Concha, G. et al.: Exposure to inorganic arsenic metabolites during early human development.
Toxicol. Sci. 44: 185-190, 1998.
A-12
-------
ATTACHMENT B
MEMORANDUM
To: SAB Drinking Water Committee
From: Barbara Harper, PhD, DABT
Yakama Nation Toxicologist
Date: September 9, 2000
Subject: Endorsement of the Minority Report on Arsenic and Children
I would like to lend my personal support for this report. The report describes a large body of evidence
showing that children are a generally vulnerable or sensitive population. This has been recognized in a
regulatory context through the use of an additional safely factor used for determining allowable trace
concentrations of pesticides in food products intended for infants and children. Many children's health
initiatives are being developed as well.
There was disagreement among SAB members about the interpretation of some of the arsenic
epidemiology papers, including some of those which involved children. I personally prefer precaution
when the data are substantialy suggestive of an increased effect in children, as I believe is the case with
arsenic, rather than waiting for statistically conclusive evidence of adverse outcomes in children (i.e., I
have a preference for making alpha errors when children's health is at stake).
In a risk ranking system (without economic considerations), the small systems with elevated arsenic
serving the poorest populations would probably rank highest because their children are most vulnerable
for a number of reasons. These systems may also serve unique populations with different exposure
patterns (e.g., higher water intake) and different nutritional status, compounded by less access to health
care, different underlying disease patterns, and so on. Many Native American Tribes fall into this
category, as would migrant workers. The risk management question is whether they should be first in
line to reduce their arsenic because their risk is greatest, or last in line because they can afford it least
and because the water treatment technology may not be fully developed. The assertion that water
treatment is solely self-funded is not strictly true - there are rural water system assistance grants and
other special programs to improve rural and reservation water quality. Therefore, I would argue that a
phased approach based only on economic reasons (large systems first, small systems later) is
inadequately protective of the most at-risk communities, and additional approaches should also be
considered.
B-l
-------
I also support the issuance of a health advisory for selected situations, written with culturally sensitive
language. The mechanisms of outreach may be different for different situations, especially for tribal,
migrant worker, or other ethnic, linguistic, or disadvantaged communities. This may involve
pediatricians in some situations, and other mechanisms in other situations.
This memo reflects my personal opnion and should not be interpreted as official tribal policy.
B-2
-------
ATTACHMENT C
Response to Comments Entered into the Record on the DWC's EC-Review
Draft of the "Arsenic Report" at the September 22, 2000 EC
Teleconference Meeting
November 27, 2000
1. Introduction
The Science Advisory Board's (SAB) established a Panel to review portions of the EPA
proposed rulemaking on arsenic from June to August, 2000. The Panel was comprised of members of
the SAB Drinking Water Committee and five consultants added to provide expertise on special issues
that were known to be important in the review. The Panel's report on "Certain Elements of the
Proposed Arsenic Drinking Water Regulation" notes that the major background document it used as
the source of information on arsenic's health effects was the NRC's Arsenic in Drinking Water
(NRC, 1999) report that was developed by an NRC Subcommittee. In recognition of the importance
of that report to their own deliberations, Dr. Louise Ryan, a member of the NRC Subcommittee on
Arsenic in Drinking Water, was asked to serve on the SAB Panel reviewing the arsenic proposal. Dr.
Ryan was selected because of her knowledge of the NRC report and because of her expertise in
modeling issues that were considered to be key aspects in responding to the Agency's charge to the
SAB. The Panel's conclusions agree with the major conclusions in the NRC report. In a few
instances, the SAB Panel considered additional evidence that support its conclusions, and those of the
NRC, when this new information provided additional insights into key issues.
As the Panel's arsenic report was being discussed by the SAB's Executive Committee (EC) on
September 22, 2000, one member of the EC entered into the record comments on the SAB report that
were originally made by a member of the public in response to EPA's proposed arsenic rule (Greer,
2000). These comments were suggested as carrying special weight because they were provided to the
EC member by Dr. Alan Smith, an epidemiologist who also served on the NRC Arsenic Subcommittee
(see Attachment 1 to these comments).
These comments raised objections to three points in the Panel's draft report, including
statements about: 1) the direct applicability of the Taiwanese ecological data to U.S. risk assessment,
2) the detectability of a 1 in 100 risk level in epidemiological studies, and 3) the use of comparison
populations in risk assessments based on epidemiology studies. The Chair of the SAB Panel
considered these comments and discussed them with various members of the Panel in formulating the
response provided below.
C-l
-------
Comment Number 1:
Dr. Smith stated:
"The letter states that, 'In the opinion of the DWC, the Agency
misinterpreted some of the conclusions of the NRC report.' This point
was elaborated with the statement that 'The NRC (1999) noted, there
are several reasons why the Taiwanese data should not be accepted as
being directly applicable to the U.S.' This is simply not correct and it is
misleading to imply that such a statement was made in the NRC report.
The following are pertinent quotes from the NRC report:
[a] 'Ecological studies in Chile and Argentina have
observed risk of lung and bladder cancer of the same
magnitude as those reported in the studies in Taiwan at
comparable levels of exposure' (page 7).
[b] 'Human susceptibility to adverse effects resulting
from chronic exposure to inorganic arsenic is likely to
vary based on genetics, nutrition, sex, and other
possible factors. Some factors, such as poor nutrition
and arsenic intake from food might affect assessment of
risk in Taiwan or extrapolation of results in the United
States' (page 8)...
[c] 'A wider margin of safety might be needed when
conducting risk assessments of arsenic because of
variations in metabolism and sensitivity among
individuals or subgroups' (page 244)... .
In short, there may indeed be susceptible sub-populations. These
would be present both in Taiwan and also in the United States. Added
margins of safety may be called for, not reduced ones. The DWC has
grossly distorted information in the NRC report without any good
basis."
Response to Comment No. 1:
The comment misinterprets the Panel's statement regarding "direct" applicability and ignores
strong cautions contained in the NRC report about evaluating U.S. risk from arsenic. The Panel
accepted the NRC's evaluation of epidemiological studies as providing strong and corroborating
C-2
-------
evidence that arsenic is carcinogenic. Consequently, the first two statements are not at issue.
However, the NRC was also quite clear about limitations on using such ecological data to predict risks:
"First, there is no question that the ideal basis for risk assessment is a
well-conducted epidemiological study involving accurate assessment of
individual exposures. In the absence of such data, however, ecological
data might be the only choice." Such analyses must be conducted with
caution keeping in mind the potential for measurement error and
confounding to bias the results. It is important to remember that any
risk assessment based on ecological data must be cautiously interpreted
because of the inherent uncertainty in the exposure-assessment methods
used for such studies. In the case of the Taiwanese data, the fact that it
came from a culturally homogeneous area provides some reassurance
that confounding might not be too serious a concern. Our findings also
suggest that additional caution might be needed when exposure
concentrations are grouped into broad exposure categories. It is
important to keep in mind that the considerable variability in the arsenic
concentrations detected in multiple wells within some of the villages
leads to considerable uncertainty about exposure concentrations in the
Taiwanese data" (page 294 NRC).
The quote at item "b," is precisely what prompted the Panel to examine issues related to human
susceptibility. The Panel discussed this in the report in some detail. In this regard it is important to
point out that the NRC also restated the likely influence of these factors to risk in Taiwan and
extrapolation to the United States later in their document on page 295, with a further note that these
factors "...could not be taken into account quantitatively in this chapter." The Panel believes that one
such important factor is the low selenium concentrations in Taiwan.
In this regard, the Panel concluded that:
a. There were several variables that are important for considering bladder and lung cancer
risk, but the DWC recognized early that there were no data available that would allow us to
compare the U.S. and Taiwan population that was studied. This would be an important area to
follow-up on with targeted research.
b. Other sources of arsenic contributing to the risks of cancer in the Taiwan study population
probably do exist and could be important if the appropriate data could be captured. At the
present time, the uncertainties about how the form of arsenic influences lexicological responses
prevents the Panel from resolving this issue. It was primarily for this reason that we suggested
that Agency consider the incremental risk that is posed by arsenic in drinking water. Most Panel
members support the NRC's indication that mechanisms plausibly associated with arsenic-
induced cancer are sublinear (p. 300). However, the uncertainty about forms of arsenic and
C-3
-------
their toxicity forced the Panel to accept the use of a linear model for estimating risks (as low as
1/1000), even though this was unsettling to many on the Panel.
c. The DWC noted and followed up on the data provided in the report on selenium status of
the Taiwanese study population. Based upon four epidemiology reports identified by the Panel,
it concluded that there may be sufficient information to make appropriate adjustments in the
current estimate of risk. Two of these, discussed more fully below dealt specifically with
selenium status and bladder and lung cancer.
In regard to statement "c," the Panel does not take direct issue with this statement. However,
the Panel report expands on this issue in two respects: 1) we clearly accepted the possibility that there
can be sensitive populations and examined these variables in some detail and 2) based on those
variables having either qualitative information or quantitative data, the Committee suggests that the
Taiwanese population studied may in fact itself be a sensitive population. Therefore, at least some of
the additional margin of exposure appears to be captured in the source of the data that was used to
estimate risks at low doses.
The SAB Panel's report does not take the position that there are no sensitive subgroups in the
U.S. population. The Panel at least indirectly took the position that the Taiwanese study population
appeared to be a "susceptible" subpopulation and explicitly recognized the possibility of children as
such a sensitive population in its discussion of the Agency's proposed Health Advisory for mother's
who might use drinking water with high arsenic levels to prepare infant formula. The Panel did not
reject such an Advisory because it believes that these children could obtain higher doses of arsenic due
to their greater intake of drinking water per unit of body weight. The Panel stopped short of fully
endorsing the advisory because specific detailed information was not available on the proposed
advisory and its implementation. The Panel believes that this Advisory will be different from the current
health advisory program administered by the OGWDW because of the intended audience.
Comments Number 2 and 3:
Dr. Smith's comment number 2 stated that "The most serious error in the DWC report
concerns the statement that: 'Further analyses of the Taiwanese data have been performed since the
NRC report was issued that bring into serious question the use of the comparison populations outside
the study area for estimating cancer risks due to arsenic. A study in Utah suggests that some U.S.
populations may be less susceptible to arsenic.'" Further Dr. Smith states that "In the body of the DWC
report it is stated that 'For one thing, if the lifetime cancer risk at the current standard (50ug/L) was
really 1 case in 100 persons in the population, or greater, then there should be more evidence of effects
in the U.S.'"
He goes on to state that "The above demonstrates a serious basic misunderstanding of
epidemiological studies. To start with, the Utah study involved a highly select population from which no
inference can be made about risk assessment." "There are no studies in the U.S., or anywhere else,
conflicting with a 1 in 100 risk estimate. It needs to be understood that "...it is very hard to
demonstrate if a 1 in 100 risk estimate truly exists." He concludes that "In short, the assumption that
risks cannot possibly be as high as 1 in 100 has no scientific basis, and is in fact, very dangerous."
C-4
-------
In comment number 3 Dr. Smith states that
"It is imperative that any good arsenic risk assessment using epidemiological data
should have a comparison population group that is clearly known not to be exposed to
increased concentrations of arsenic in drinking water. While Morales et al. have
conducted a good risk assessment in many aspects, no weight should be given to
findings in their publication which do not include a comparison population known to be
unexposed. Within the endemic area of Taiwan, only single samples from wells taken at
one point in time were available. People migrate, they move to different villages, they
do not drink from the same well for their total life. This means that within the endemic
region, there is no comparison population known to be unexposed. Therefore,
attention should be confined to the risk assessment results that were reported using
external comparison populations."
Response No. 2 and 3:
Although many on the SAB Panel believe that their statement of concern as to why a 1 in 100
risk is not more noticeable in U.S. studies accurately reflects their feelings, they defer to the
epidemiological expertise of Dr. Smith regarding the lack of sensitivity of such studies. Although
deferring on this point, the Committee does not necessarily cede the point, rather the statement has
been removed from the document because it does not materially add to their conclusions about the
need for caution when applying the Taiwanese data to the U.S. situation.
Regarding the Utah study, the Panel did not attempt to use the Utah data for estimating cancer
risks, partly for the same reasons cited by Dr. Smith. More fundamentally, however, we were not
convinced that the population studied in Utah was any more representative than the Taiwanese in the
study area modeled by NRC. Essentially, the Utah population was unlikely to have many of the co-
carcinogenic exposures (e.g. smoking, dietary habits) that would be found in the broader U.S.
population. The Panel came to the conclusion that the two populations may be at opposite ends of the
spectrum in terms of susceptibility. This was the genesis of the statement that "some U.S. populations
may be less susceptible to arsenic."
Specifically, in regard to the Panel's quote "For one thing, if the lifetime cancer risk at the
current standard (50 ug/L) was really 1 case in 100 persons in the population or greater, then there
should be more evidence of effects in the U.S.", Dr. Smith takes the position that the more than 1/100
risk that is projected from this population, based on the use of a comparison population, to a U.S.
population is likely to be real and could be as high as 1/10. He dismisses the lack of parallel findings in
the U.S. on the basis of the well-recognized insensitivity of epidemiological studies in resolving such
issues. Dr. Smith also identifies the difficulties that are associated with the exposure assessments in this
study. In contrast, the Panel took this analysis to suggest that there could be some substantive
differences in the study population and surrounding areas of Taiwan. As indicated above, several
differences were clearly identified in the NRC report (quote above taken from page 295 of the NRC
report). It is important to recognize that these uncertainties call into question the accuracy of the dose-
response evaluation as it is extrapolated to the U.S. population. Nevertheless, the Panel accepted the
projection of an approximate 1/1000 risk of bladder cancer at the current MCL to exercise caution.
C-5
-------
Moreover, the Panel concluded, based on the Morales et al. (2000) paper, that there was a similar
incremental risk for lung cancer. These findings were the basis for the Panel's concurrence that the
MCL should be decreased.
A key advantage in using data that are internal to the study population, is that it decreases the
possibility of confounding. For example, from experimental data, arsenic is much more effective as a
co-carcinogen than it is as a carcinogen. The study population was comprised of rural poor persons
with relatively poor nutrition. Therefore, the Panel felt that it was important to explore potential
differences in the study population in Taiwan and other parts of the world. Several differences were
identified in the NRC report (pp.241-243 of the NRC report as cited above). One was the issue of
how high the level of non-drinking water exposure to arsenic. There was evidence that relatively high
exposures can exist from such sources, but the Panel concluded that for the meantime, this would be
best ignored because of the uncertainties in the relative carcinogenic activity of varying forms of arsenic.
Since the Agency had settled on linear extrapolation, it was suggested that this issue could be simplified
by examining the incremental risk of drinking water for the purposes of the current rule.
One variable that the SAB Panel felt should be considered more closely in the risk assessment
is the poor selenium status of this study population (this is documented in the NRC report at pp. 241-
243). There are data to suggest that selenium status influences bladder and lung cancer rates in human
populations. The Panel was able to identify four publications that bear directly on the influence of
selenium status on cancer risk without pursuing the issue exhaustively. To save time, only the two most
relevant studies are summarized. One was a case control study in Washington County MD in which
selenium status was measured in a cohort of 25,802 people that were followed over time. A nested
case-control study found odds ratios of 2 in those individuals with lower selenium status (Helzsouer et
al., 1989, Cancer Res. 49:6144-6148). The second study was part of a cohort study of diet and
cancer in Holland. In a 3.3 year follow-up 550 incident cases of lung cancer were detected and
selenium status of cases (370) vs. controls (2459) were compared. Again an odds ratio for individuals
having higher levels of selenium was 0.5 with a significant inverse trend across quartiles of selenium
status. These data suggest that the selenium status of the Taiwanese study population put them at
greater risk from these two cancers. To state this differently, if the Taiwanese population is at all similar
to other populations in the world with altered selenium status, they should have at least twice the
background rate of bladder and lung cancer, irrespective of their exposure to arsenic. In turn, this
suggests that the median susceptibility of this population for these cancers is greater than the median
susceptibility of a U.S. population.
The SAB Panel report does not suggest that there are no subgroups in the U.S. population with
greater sensitivity to arsenic. In terms of the data that were before the Panel, it was difficult to conclude
that additional uncertainty factors are necessary to adjust between this Taiwanese population and the
U.S. population. This does seem to be a different conclusion from that arrived at by Dr. Smith.
However, it does not appear to be contrary to the substance of the NRC committee report.
Consequently, the Panel felt that the use of Taiwanese data to estimate risks, without the comparison
population, would capture the apparently higher susceptibility to arsenic-induced cancers in that
population. This approch would provide an additional margin of safety for the general U.S. population.
C-6
-------
The Panel also recognized that new data that are just becoming available will undoubtedly
provide better insight into the appropriate models for low-dose extrapolation of arsenic risks, both
cancer and non-cancer. Such information can be applied to improving on the shortcomings that the
NRC Subcommittee identified in its report which stated:
Regardless of the data set that is ultimately used for the arsenic risk assessment, the
subcommittee recommends that a range of feasible modeling approaches be explored.
The final calculated risk should be supported by a range of analyses over a fairly broad
feasible range of assumptions. Performing a sensitivity analysis ensures that the
conclusions do not rely heavily on any one particular assumption, (page 296 NRC).
Comment on the Figure provided by Dr. Smith
Dr. Smith includes the Taiwanese data in his figure that plots the nominal concentrations of
arsenic in drinking water and tumor responses across several studies. This is an appropriate way to
display these data, if you believe that there are no differences, other than arsenic exposure, in the
Taiwanese population and the referent population. As discussed above, the NRC report also identified
difficulties in using the comparison population (see page 292) which states:
"... the analyses presented in this chapter used age-specific cancer rates reported
for the whole of Taiwan. Bias could be a potential problem, because the
Taiwanese-wide data might not form an appropriate comparison group for the
arsenic endemic region, which is a poor, rural area. Thus, the choice to use
external information on baseline cancer rates represents a trade-off that to some
extent can be explored using sensitivity analysis ".
The Panel concluded from this statement that the NRC Subcommittee was concerned that there
were substantive differences between the study population and the rest of Taiwan, not to mention
differences between this population and the U.S. In essence the validity of the comparison population
as a "control" group that is postulated to differ from the study population by only the single variable
under study must be questioned. The NRC Subcommittee repeatedly identified the unusual character
of this population relative to the rest of Taiwan.
It is instructive to note the diverse behavior of the dose-response curves that Dr. Smith has
plotted from different studies in the low dose range. The graph very strongly reinforces the extent of the
uncertainties at the low end of the dose-response curve (i.e. in the range of 3-20 ug/L). The correlation
coefficient of this line would not approach R2 = 0.8552 if the concentrations above 200 ug/L were
eliminated. In other words the high-dose data does not inform us much with respect to the nature of the
dose response curve in this region. It is difficult to understand how Dr. Smith can argue that
epidemiological studies in the U.S. do not have the power to detect a 1/100 risk, but that it has the
power to resolve much lower risks from the epidemiology studies depicted in the graph. The latter fact
has not been established. What occurs in this range simply depends upon the glasses that are worn.
The Panel simply calls attention to the fact that effects have not been clearly documented in this range.
C-7
-------
The Panel suggests that some serious attempt needs to be made to explore this question under
controlled conditions. The Panel was impressed with the preliminary findings of Ng et al. (1999, In:
Chapell et al. Arsenic Exposure and Health Effects, Elsevier, pp. 217-223) who were able to induce
tumors in the lung, gastrointestinal tract, and liver of female C57B1/6J mice at concentrations of arsenate
of 500 ug/|L of drinking water. The tumors were produced in the same range of concentrations that
were associated with cancer in various sites in the epidemiological studies. Equally encouraging were
the independent findings of two groups that dimethyl arsenic was capable of inducing bladder tumors in
rats. These findings appear to provide the experimental models necessary to explore the
pharmacokinetic and pharmacodynamic variables that will be important for enabling biologically-based
dose-response models envisioned in the SAB's research recommendations in 1989 and which were
reinforced as being desirable in the NRC report (pages 293 and 296).
General characterization of the DWC Report
In summary, the majority of the SAB Panel viewed the uncertainties in the dose-response
relationships associated with cancer and non-cancer effects of arsenic in a substantially differently than
Dr. Smith. While we accept that the current MCL for arsenic is too high (as the NRC report points
out, there is a margin of exposure that is less than 10 between those concentrations that produce effects
and the current MCL based on the information that is available, as flawed as it might be), the Panel also
recognized that this is an expensive rule. For that reason the Panel took the view that the great deal of
existing uncertainty in the areas of risk assessment and in treatment costs in small systems could form
the basis for the Agency's exercise of its discretionary authority in proposing this rule. This view was
the basis for the Panel suggesting that the Agency consider an adaptive management approach to
arsenic regulation (phased rule). There was one member of the committee and a consultant who
dissented from this view. The Panel was careful to point out that this recommendation arose from
considering the relatively large economic impact of this rule and that it was a risk management concern,
not one that comes solely from scientific considerations of the risk.
-------
ATTACHMENT 1
Comments from Dr. Alan Smith, University of California - Berkeley
NOTE: THE HEADER INFORMATION FOR THIS EMAIL WILL NOT COPY
DIRECTLY TO THE WORD PROCESSING SOFTWARE; HOWEVER, IT INDICATES
THAT IT IS FROM:
LGreer@nrdc.org
09/22/00 02:13PM
Don
Here are the comments from Dr. Alan Smith of Berkeley, which were sent to my
colleague Erik Olson at NRDC, who has been a long-time advocate on drinking
water standards. Please pass them along to the rest of the executive committee
as we discussed today.
Thanks, Linda
Forward Header
Subject: Fwd[2]: Arsenic drinking water standard
Author: Erik Olson
Date: 9/21/00 4:34 PM
linda: here's a short comment on the SAB report (and on EPA's proposed rule)
from Dr. Alan Smith from DC Berkeley, who sat on the MAS arsenic committee and
is probably the leading epidemiologist on arsenic in the world. — erik
Forward Header
Subject: Fwd: Arsenic drinking water standard
Author: "Elena O. Lingas"
Date: 9/21/00 10:24 AM
Dear Mr. Olson,
The following comments were sent to the Water Docket earlier this week. Dr.
Smith thought that you would be interested in seeing them.
Elena O. Lingas
> Date: Tue, 19 Sep 2000 16:20:58 -0700
> To: ow-docket@epamail.epa.gov
> From: "Elena O. Lingas"
> Subject: Arsenic drinking water standard
> Cc: ahsmith@uclink4.berkeley.edu
19 September 2000
C-9
-------
> W-99-16 Arsenic Comments Clerk
> Water Docket (MC-4101)
> U.S. Environmental Protection Agency
> 1200 Pennsylvania Avenue, NW
> Washington, DC 20460
»
> In addition to the comments previously submitted, dated 29 August 2000 and an
> erratum dated 1 September 2000, it has been suggested that we provide
> material for the EPA indicating how cancer risks, including lung cancer
> risks, could be estimated. In addition, I wish to comment on the Drinking
> Water Committee (DWC) report of the SAB.
>
> 1 . Cancer risk estimation for arsenic in drinking water.
>
> There is extensive information from several countries for use in lung cancer
> risk assessment, which is the main site for cancer mortality. We previously
> submitted a graph integrating data from Taiwan, Argentina, Chile and Japan.
> This includes results from a case-control study (Chile), a cohort study
> (Japan), and ecological studies (Argentina, Chile, Taiwan and Japan). We
> attach this graph again (Figure 1).
>
> Table 1 presents the risk assessment calculations which were part of our
> report to the Office of Environmental Health Hazard Assessment, California
> EPA. The lung cancer risk estimates for drinking 1 liter of water per day
> containing 50 ug/L were 7.8 per 1000 for men, and 9.9 per 1000 for women.
> Thus, the risk for lung cancer alone is on the order of 1 per 100.
>
> The methods used are standard risk assessment methods with linear relative
> risk extrapolation. This is the standard default. As can be seen in Figure
> 1, there is no basis for the incorporation of sub-linear or threshold models.
>
> The use in calculations of 2.3 liters as the volume of water consumed per day
> was estimated from extensive interview studies in several countries (Table
> As can be seen in Table 3, lung cancer related to arsenic was a more
> important cause of death in Taiwan, Chile and Argentina. This was
> particularly true for Argentina and Chile, both of which have populations
> which are similar to the U.S. population with regard to various
> characteristics, such as nutrition, ethnicity and lifestyle.
>
> Full details of the risk calculations are available in the report submitted
> to the California EPA.
>
> 2. The Drinking Water Committee Report.
>
> Rather than reviewing the full report, I will comment on three points raised
> in the cover letter to The Honorable Carol Browner.
>
> ? The letter states that "In the opinion of the DWC, the Agency
> misinterpreted some of the conclusions of the NRC report".
>
> This point was elaborated with the statement that "The NRC (1999) noted,
> there are several reasons why the Taiwanese data should not be accepted as
> being directly applicable to the U.S.".
C-10
-------
> This is simply not correct and it is misleading to imply that such a
> statement was made in the NRC report. The following are pertinent quotes
> from the NRC report:
>
> "Ecological studies in Chile and Argentina have observed risk of lung and
> bladder cancer of the same magnitude as those reported in the studies in
> Taiwan at comparable levels of exposure" (page 7).
>
> "Human susceptibility to adverse effects resulting from chronic exposure to
> inorganic arsenic is likely to vary based on genetics, nutrition, sex, and
> other possible factors. Some factors, such as poor nutrition and arsenic
> intake from food might affect assessment of risk in Taiwan or extrapolation
> of results in the United States"(page 8) (italics added).
>
> "A wider margin of safety might be needed when conducting risk assessments of
> arsenic because of variations in metabolism and sensitivity among individuals
> or subgroups" (page 244) (italics added).
>
> In short, there may indeed be susceptible sub-populations. These would be
> present both in Taiwan and also in the United States. Added margins of
> safety may be called for, not reduced ones. The DWC has grossly distorted
> information in the NRC report without any good basis.
>
> ? The most serious error in the DWC report concerns the statement that
>
> "Further analyses of the Taiwanese data have been performed since the NRC
> report was issued that bring into serious question the use of the comparison
> populations outside the study area for estimating cancer risks due to
> arsenic. A study in Utah suggests that some U.S. populations may be less
> susceptible to arsenic .". In the body of the DWC report it is stated that
> "For one thing, if the lifetime cancer risk at the current standard (50 ug/L)
> was really 1 case in 100 persons in the population, or greater, then there
> should be more evidence of effects in the U.S."
>
> The above demonstrates a serious basic misunderstanding of epidemiological
> studies. To start with, the Utah study involved a highly select population
> from which no inference can be made about risk assessment.
>
> There are no studies in the U.S., or anywhere else, conflicting with a 1 in
> 100 risk estimate. It needs to be understood that it is very hard to
> demonstrate if a 1 in 100 risk estimate truly exists. One example of this is
> that it took many studies (about 16) before an NRC report could conclude that
> lung cancer risks from passive smoking by non-smokers married to smokers were
> indeed increased and of the order of 1 in 100. In the case of arsenic in
> drinking water, one would need large populations who over many years (at
> least 30) consumed water containing 50 ug/L every day. The background risk
> of cancer mortality is about 20 per 100 (i.e. about 1 in 5 people die from
> cancer). For lung cancer alone it is about 5 in 100. The relative risk for a
> population having an increment of 1 in 100 would be 1.2. Such a relative
> risk is extremely hard to prove. There are simply not enough people in the
> U.S. with long enough exposures at the 50 ug/L level to demonstrate if the
> risk estimate of 1 in 100 is real or not. Even if there were enough people
> in the U.S. with long enough exposures at this level, you would need many
> studies over several years to demonstrate this risk.
C-ll
-------
> In short, the assumption that risks cannot possibly be as high as 1 in 100
> has no scientific basis, and is in fact, very dangerous.
>
> ? It is imperative that any good arsenic risk assessment using
> epidemiological data should have a comparison population group that is
> clearly known not to be exposed to increased concentrations of arsenic in
> drinking water. While Morales et al. have conducted a good risk assessment
> in many aspects, no weight should be given to findings in their publication
> which do not include a comparison population known to be unexposed. Within
> the endemic area of Taiwan, only single samples from wells taken at one point
> in time were available. People migrate, they move to different villages,
> they do not drink from the same well for their total life. This means that
> within the endemic region, there is no comparison population known to be
> unexposed. Therefore, attention should be confined to the risk assessment
> results that were reported using external comparison populations.
>
> Again, for further information, feel free to contact my office at
> 510-843-1736 or the web page at http://socrates.berkeley.edu/~asrg, which
> contains information on our research.
>
> Sincerely,
> Allan H. Smith, MD, PhD
> Professor of Epidemiology
> School of Public Health
> University of California, Berkeley
Dear Mr. Olson,
The following comments were sent to the Water Docket earlier this
week. Dr. Smith thought that you would be interested in seeing
them.
Elena O. Lingas
Date: Tue, 19 Sep 2000 16:20:58 -0700
To: ow-docket@epamail.epa.gov
From: "Elena O. Lingas"
<lingas@uclink4.berkeley.edu>
Subject: Arsenic drinking water standard
Cc: ahsmith@uclink4.berkeley.edu
19 September 2000
W-99-16 Arsenic Comments Clerk
Water Docket (MC-4101)
U.S. Environmental Protection Agency
C-12
-------
1200 Pennsylvania Avenue, NW
Washington, DC 20460
In addition to the comments previously submitted, dated 29 August 2000
and an erratum dated 1 September 2000, it has been suggested that we
provide material for the EPA indicating how cancer risks, including lung
cancer risks, could be estimated. In addition, I wish to comment on
the Drinking Water Committee (DWC) report of the SAB.
1. Cancer risk estimation for arsenic in drinking water.
There is extensive information from several countries for use in lung
cancer risk assessment, which is the main site for cancer
mortality. We previously submitted a graph integrating data from
Taiwan, Argentina, Chile and Japan. This includes results from a
case-control study (Chile), a cohort study (Japan), and ecological
studies (Argentina, Chile, Taiwan and Japan). We attach this graph
again (Figure 1).
Table 1 presents the risk assessment calculations which were part of our
report to the Office of Environmental Health Hazard Assessment,
California EPA. The lung cancer risk estimates for drinking 1 liter
of water per day containing 50 ug/L were 7.8 per 1000 for men, and 9.9
per 1000 for women. Thus, the risk for lung cancer alone is on the
order of 1 per 100.
The methods used are standard risk assessment methods with linear
relative risk extrapolation. This is the standard default. As
can be seen in Figure 1, there is no basis for the incorporation of
sub-linear or threshold models.
The use in calculations of 2.3 liters as the volume of water consumed per
day was estimated from extensive interview studies in several countries
(Table 2).
As can be seen in Table 3, lung cancer related to arsenic was a more
important cause of death in Taiwan, Chile and Argentina. This was
particularly true for Argentina and Chile, both of which have populations
which are similar to the U.S. population with regard to various
characteristics, such as nutrition, ethnicity and lifestyle.
Full details of the risk calculations are available in the report
submitted to the California EPA.
2. The Drinking Water Committee Report.
Rather than reviewing the full report, I will comment on three points
raised in the cover letter to The Honorable Carol Browner.
?The letter
states that "In the opinion of the DWC, the Agency misinterpreted some of
the conclusions of the NRC report".
This point was elaborated with the statement that "The NRC (1999) noted,
C-13
-------
there are several reasons why the Taiwanese data should not be accepted
as being directly applicable to the U.S.".
This is simply not correct and it is misleading to imply that such a
statement was made in the NRC report. The following are pertinent
quotes from the NRC report:
"Ecological studies in Chile and Argentina have observed risk of lung and
bladder cancer of the same magnitude as those reported in the studies in
Taiwan at comparable levels of exposure" (page 7).
"Human susceptibility to adverse effects resulting from chronic exposure
to inorganic arsenic is likely to vary based on genetics, nutrition, sex,
and other possible factors. Some factors, such as poor nutrition
and arsenic intake from food might affect assessment of risk in Taiwan or
extrapolation of results in the United States"(page 8) (italics
added).
"A wider margin of safety might be needed when conducting risk
assessments of arsenic because of variations in metabolism and
sensitivity among individuals or subgroups" (page 244) (italics
added).
In short, there may indeed be susceptible sub-populations. These
would be present both in Taiwan and also in the United States.
Added margins of safety may be called for, not reduced ones. The
DWC has grossly distorted information in the NRC report without any good
basis.
?The most
serious error in the DWC report concerns the statement that
"Further analyses of the Taiwanese data have been performed since the NRC
report was issued that bring into serious question the use of the
comparison populations outside the study area for estimating cancer risks
due to arsenic. A study in Utah suggests that some U.S.
populations may be less susceptible to arsenic .". In the body of
the DWC report it is stated that "For one thing, if the lifetime cancer
risk at the current standard (50 ug/L) was really 1 case in 100 persons
in the population, or greater, then there should be more evidence of
effects in the U.S."
The above demonstrates a serious basic misunderstanding of
epidemiological studies. To start with, the Utah study involved a
highly select population from which no inference can be made about risk
assessment.
There are no studies in the U.S., or anywhere else, conflicting with a 1
in 100 risk estimate. It needs to be understood that it is very
hard to demonstrate if a 1 in 100 risk estimate truly exists. One
example of this is that it took many studies (about 16) before an NRC
report could conclude that lung cancer risks from passive smoking by
non-smokers married to smokers were indeed increased and of the order of
1 in 100. In the case of arsenic in drinking water, one would need large
populations who over many years (at least 30) consumed water containing
C-14
-------
50 ug/L every day. The background risk of cancer mortality is about
20 per 100 (i.e. about 1 in 5 people die from cancer). For lung
cancer alone it is about 5 in 100. The relative risk for a population
having an increment of 1 in 100 would be 1.2. Such a relative risk
is extremely hard to prove. There are simply not enough people in
the U.S. with long enough exposures at the 50 ug/L level to demonstrate
if the risk estimate of 1 in 100 is real or not. Even if there were
enough people in the U.S. with long enough exposures at this level, you
would need many studies over several years to demonstrate this
risk.
In short, the assumption that risks cannot possibly be as high as 1 in
100 has no scientific basis, and is in fact, very dangerous.
?lt is
imperative that any good arsenic risk assessment using epidemiological
data should have a comparison population group that is clearly known not
to be exposed to increased concentrations of arsenic in drinking
water. While Morales et al. have conducted a good risk assessment
in many aspects, no weight should be given to findings in their
publication which do not include a comparison population known to be
unexposed. Within the endemic area of Taiwan, only single samples
from wells taken at one point in time were available. People
migrate, they move to different villages, they do not drink from the same
well for their total life. This means that within the endemic
region, there is no comparison population known to be unexposed.
Therefore, attention should be confined to the risk assessment results
that were reported using external comparison populations.
Again, for further information, feel free to contact my office at
510-843-1736 or the web page at
http://socrates.berkeley.edu/~asrg,
which contains information on our research.
Sincerely,
Allan H. Smith, MD, PhD
Professor of Epidemiology
School of Public Health
University of California, Berkeley
Received: from uclink4.berkeley.edu ([128.32.25.39]) by mail.nrdc.org with SMTP
(IMA Internet Exchange 3.14) id 0007588A; Thu, 21 Sep 2000 13:28:25 -0400
Received: from ehs-204-7 (ehs-204-7.SPH.Berkeley.EDU [128.32.252.91])
by uclink4.berkeley.edu (8.10.1/8.10.1) with SMTP id e8LHTEJ20679
for ; Thu, 21 Sep 2000 10:29:14 -0700 (PDT)
Message-Id: <4.1.20000921102128.OOa29eeO@uclink4.berkeley.edu>
X-Sender: Iingas@uclink4.berkeley.edu
X-Mailer: QUALCOMM Windows Eudora Pro Version 4.1
Date: Thu, 21 Sep 2000 10:24:11 -0700
To: EOIson@nrdc.org
From: "Elena O. Lingas"
C-15
-------
Subject: Fwd: Arsenic drinking water standard
Mime-Version: 1.0
Content-Type: multipart/mixed;
Attachments to the Email from Dr. Smith:
a. File Name: SMITHTABLS
(Originally in the email this was called )
Table 1: Estimate of Excess Lung and Bladder Cancer Risk Due to Arsenic in the Drinking Water
Slope of excess lung cancer relative risk (RR-1) versus exposure per ug per liter
as obtained from Figures 4 and 5
Estimate of background lifetime lung cancer mortality risk per 1000 persons
based on U.S. rates in 1996 from Table 6
Approximate adjustment of average daily water consumption of 2.3 liters per day
to 1 liter per day (1 liters/2.3 liters) from Table 7
Estimate of lifetime added lung cancer risk per 1000 persons exposed to 50 ug/L
Estimate of lifetime added lung cancer risk per 1000 persons exposed to 10 ug/L
Ratio of excess lung cancer plus bladder cancer deaths divided by excess lung
cancer deaths from Table 9
Males
0.0046
79
0.43
7.8
1.6
1.3
Females
0.0076
52
0.50
9.9
2.0
1.6
Estimate of lifetime added lung and bladder cancer risk per 1000 persons
exposed to 50 ug/L
10.1
15.8
Estimate of lifetime added lung cancer risk per 1000 persons exposed to 10 ug/L
2.1
3.2
Estimate of lifetime added lung and bladder cancer risk per 1000 persons
exposed to 50 ug/L for both sexes combined
13.0
Estimate of lifetime added lung and bladder cancer risk per 1000 persons
exposed to 10 ug/L for both sexes combined
2.7
C-16
-------
Table 2: Drinking water consumption in liters per day from various epidemiological
investigations of arsenic outside the U.S.
Country-Reference
Chile-Ferreccio, personal communicationA
Total
Males
Females
Chile-Biggs et al., 1997
San Pedro-high
Toconao-low
Chile-Moore et al., 1997a
High and low exposure groups
Argentina-case control*
Total
Males
Females
India*
Total
Males
Females
Drinking water consumption
(liters/day)
2.4
2.6
2.2
2.5
2.3
2.6
1.9
2.0
1.7
2.4
2.6
2.1
Afindings from participants in the lung cancer case-control study of Ferreccio et al.,
(submitted)
*preliminary findings from an on-going bladder case-control study; data from 28 females
and 149 males
*preliminary findings from an on-going skin cancer case-control study; data from 73
Females and 143 males
C-17
-------
Table 3: Excess Deaths (observed minus expected) from cancers related to arsenic in drinking water
Country
Cancer Site
Argentina*
Lung cancer
Bladder cancer
Chile (Region TI)+
Lung cancer
Bladder cancer
Taiwan*
Lung cancer
Bladder cancer
TOTALS
Lung cancer
Bladder cancer
Men
Excess
deaths
307
70
401
78
228
152
936
300
Ratio of
excess
lung/bladder
cancers
4.4
5.1
1.5
3.1
(Excess lung
cancer deaths plus
bladder cancer
deaths)/Excess
lung cancer deaths
1.2
1.2
1.7
1.3
Women
Excess
deaths
84
12
105
56
177
157
366
225
Ratio of
excess
lung/bladder
cancers
7
1.9
1.1
1.6
(Excess lung cancer
deaths plus bladder
cancer
deaths)/Excess lung
cancer deaths
1.1
1.5
1.9
1.6
"Hopenhayn-Rich et al., 1996; 1998(high exposure group); +Smith et al., 1998; *Chen et al., 1985
b. File Name: SMITHFIG1
(Originally in the email this was called )
Please provide a fax number to obtain this Figure.
C-18
-------
ATTACHMENT D
ACRONYMS AND ABBREVIATIONS
BAT Best Available Treatment
CWS Community Water System
DWC Drinking Water Committee
GFH Granular Ferric Hydroxide
LS Lime Softening
MCL Maximum Contaminant Level
MHI Median Household Income
POE Point of Entry
POTW Publically Owned Treatment Works
POU Point of Use
PQL Practical Quantitation Limit
SAB U. S. EPA Science Advisory Board
SDWA Safe Drinking Water Act Amendments of 1996
TCLP Toxicity Characteristic Leaching Procedure
TDS Total Dissolved Solids
E-l