Review Comments on EPA’s Responsiveness to SAB 2007 Recommendations for the Revision of Cancer Assessment of Inorganic Arsenic

ssIt2:
i
-------
cancer risk from exposure to inorganic arsenic. EPA's 2010 draft assessment presents a
comprehensive overview of the epidemiological literature on arsenic carcinogenicity up to 2007;
however, it needs to state more clearly the set of criteria that EPA used in evaluating and
presenting the studies. Where possible, the summaries of the epidemiology studies should
include a quantitative or qualitative presentation of the relative risk point estimates and the
associated confidence intervals. Additionally, EPA should consider including an addendum or
appendix describing major epidemiology studies that were published since 2007 and that could
substantially impact the calculated cancer unit risk estimate.
In 2007, the SAB noted that there was a possibility of a nonlinear dose-response at low
exposures to arsenic, but due to the lack of a complete understanding of the mode-of-action by
which inorganic arsenic causes cancer in humans, the choice of a specific nonlinear model could
not be justified. The SAB supports the agency's choice of using a default linear approach given
the complexity of the mode-of-action of arsenic. Although extensive new research has been done
in this area, there is not enough information in the literature to fully define the multiple modes-
of-action for arsenic carcinogenicity.
The SAB, in 2007, also recommended that EPA consider using alternative dose-response
models and perform a sensitivity analysis of the Taiwanese data with different exposure metrics.
EPA's 2010 draft assessment uses a default linear low-dose extrapolation and evaluates the
differences between a linear model and three non-linear models: quadratic, quadratic exponential
and linear exponential. The SAB finds that, while the sensitivity analysis did respond to the 2007
recommendation, a more detailed description of the data sets used in the risk model is needed.
Providing the distribution of variability of arsenic concentrations in well water and the data and
parameters used in the modeling would help to make EPA's document more transparent. The
SAB notes that, while EPA's choice of a linear approach is consistent with EPA's risk
assessment default procedures, it has produced a calculated upper-bound cancer risk estimate for
arsenic that is of significant public health concern. The SAB suggests that EPA discuss, possibly
in other EPA complete risk assessment documents, how the estimated risks for arsenic should be
interpreted in light of current estimated bladder and lung cancer incidence for the U.S.
population.
In 2007, the SAB recommended that the agency conduct sensitivity analyses to determine
the potential impact of different choices of exposure assumptions (both water and non-water
consumption) for estimating arsenic cancer potency. The SAB finds that the agency was partially
responsive to the previous recommendations. The SAB recommends that the agency revise its
assessment to provide a more detailed and transparent explanation of the scientific rationale for
its choice and use of alternative exposure assumptions. The SAB has also recommended ways to
enhance the rigor and transparency of the sensitivity analysis for the exposure assessment
through further documentation, explanation and analyses.
The SAB appreciates the opportunity to provide advice on EPA's inorganic arsenic cancer
assessment. We look forward to the upcoming review of the IRIS assessment of non-cancer effects
from arsenic exposure. The SAB underscores the importance of developing integrated,
-------
interdisciplinary IRIS assessments and is amenable to conducting a future review of a synthesized
assessment of arsenic health effects, as needed.
Sincerely,
/signed/ /signed/
Dr. Deborah L. Swackhamer, Chair
EPA Science Advisory Board
Dr. Elaine Faustman, Chair
SAB Arsenic Cancer Workgroup
-------
NOTICE
This report has been written as part of the activities of the EPA Science Advisory Board, a public
advisory committee providing extramural scientific information and advice to the Administrator and
other officials of the Environmental Protection Agency. 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. Reports of the EPA Science Advisory Board are posted
on the EPA Web site at: http://www.epa. gov/sab.
1
-------
U.S. Environmental Protection Agency
Science Advisory Board
Arsenic Cancer Review Work Group
CHAIR
Dr. Elaine Faustman, Professor, Department of Environmental and Occupational Health
Sciences, School of Public Health and Community Medicine, University of Washington, Seattle,
WA
MEMBERS
Dr. Timothy Buckley, Associate Professor and Chair, Division of Environmental Health
Sciences, College of Public Health, The Ohio State University, Columbus, OH
Dr. Thomas Burke, Professor, Department of Health Policy and Management, Johns Hopkins
Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD
Dr. Deborah Cory-Slechta*, Professor, Department of Environmental Medicine, School of
Medicine and Dentistry, University of Rochester, Rochester, NY
Dr. George Daston, Victor Mills Society Research Fellow, Product Safety and Regulatory
Affairs, Procter & Gamble, Cincinnati, OH
Dr. Agnes Kane, Professor and Chair, Department of Pathology and Laboratory Medicine,
Brown University, Providence, RI
Dr. Nancy K. Kim, Health Research Inc., Troy, NY
Dr. Jana Milford, Professor, Department of Mechanical Engineering, University of Colorado,
Boulder, CO
Dr. Eileen Murphy, Ernest Mario School of Pharmacy, Rutgers University, Piscataway, NJ
Dr. Stephen M. Roberts, Professor, Department of Physiological Sciences, Director, Center for
Environmental and Human Toxicology, University of Florida, Gainesville, FL
SCIENCE ADVISORY BOARD STAFF
Dr. Sue Shallal, Designated Federal Officer, U.S. Environmental Protection Agency, Science
Advisory Board Staff Office, Washington, DC
*SAB Board member from 2003-2010
11
-------
U.S. Environmental Protection Agency
Science Advisory Board
CHAIR
Dr. Deborah L. Swackhamer, Professor and Charles M. Denny, Jr., Chair in Science,
Technology and Public Policy and Co-Director of the Water Resources Center, Hubert H.
Humphrey School of Public Affairs, University of Minnesota, St. Paul, MN
SAB MEMBERS
Dr. David T. Allen, Professor, Department of Chemical Engineering, University of Texas,
Austin, TX
Dr. Claudia Benitez-Nelson, Associate Professor, Department of Earth and Ocean Sciences and
Marine Science Program, University of South Carolina, Columbia, SC
Dr. Timothy Buckley, Associate Professor and Chair, Division of Environmental Health
Sciences, College of Public Health, The Ohio State University, Columbus, OH
Dr. Patricia Buffler, Professor of Epidemiology and Dean Emerita, Department of
Epidemiology, School of Public Health, University of California, Berkeley, CA
Dr. Ingrid Burke, Director, Haub School and Ruckelshaus Institute of Environment and Natural
Resources, University of Wyoming, Laramie, WY
Dr. Thomas Burke, Professor, Department of Health Policy and Management, Johns Hopkins
Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD
Dr. Terry Daniel, Professor of Psychology and Natural Resources, Department of Psychology,
School of Natural Resources, University of Arizona, Tucson, AZ
Dr. George Daston, Victor Mills Society Research Fellow, Product Safety and Regulatory
Affairs, Procter & Gamble, Cincinnati, OH
Dr. Costel Denson, Managing Member, Costech Technologies, LLC, Newark, DE
Dr. Otto C. Doering III, Professor, Department of Agricultural Economics, Purdue University,
W. Lafayette, IN
Dr. David A. Dzombak, Walter J. Blenko Sr. Professor, Department of Civil and Environmental
Engineering, College of Engineering, Carnegie Mellon University, Pittsburgh, PA
Dr. T. Taylor Eighmy, Vice President for Research, Office of the Vice President for Research,
Texas Tech University, Lubbock, TX
111
-------
Dr. Elaine Faustman, Professor, Department of Environmental and Occupational Health
Sciences, School of Public Health and Community Medicine, University of Washington, Seattle,
WA
Dr. John P. Giesy, Professor and Canada Research Chair, Veterinary Biomedical Sciences and
Toxicology Centre, University of Saskatchewan, Saskatoon, Saskatchewan, Canada
Dr. Jeffrey Griffiths, Associate Professor, Department of Public Health and Community
Medicine, School of Medicine, Tufts University, Boston, MA
Dr. James K. Hammitt, Professor, Center for Risk Analysis, Harvard University, Boston, MA
Dr. Bernd Kahn, Professor Emeritus and Associate Director, Environmental Radiation Center,
School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA
Dr. Agnes Kane, Professor and Chair, Department of Pathology and Laboratory Medicine,
Brown University, Providence, RI
Dr. Madu Khanna, Professor, Department of Agricultural and Consumer Economics, University
of Illinois at Urbana-Champaign, Urbana, IL
Dr. Nancy K. Kim, Senior Executive, New York State Department of Health, Troy, NY
Dr. Catherine Kling, Professor, Department of Economics, Iowa State University, Ames, IA
Dr. Kai Lee, Program Officer, Conservation and Science Program, David & Lucile Packard
Foundation, Los Altos, CA
Dr. Cecil Lue-Hing, President, Cecil Lue-Hing & Assoc. Inc., Burr Ridge, IL
Dr. Floyd Malveaux, Executive Director, Merck Childhood Asthma Network, Inc., Washington,
DC
Dr. Lee D. McMullen, Water Resources Practice Leader, Snyder & Associates, Inc., Ankeny, IA
Dr. Judith L. Meyer, Distinguished Research Professor Emeritus, Odum School of Ecology,
University of Georgia, Lopez Island, WA
Dr. James Mihelcic, Professor, Department of Civil and Environmental Engineering, University
of South Florida, Tampa, FL
Dr. Jana Milford, Professor, Department of Mechanical Engineering, University of Colorado,
Boulder, CO
Dr. Christine Moe, Eugene J. Gangarosa Professor, Hubert Department of Global Health, Rollins
School of Public Health, Emory University, Atlanta, GA
iv
-------
Dr. Horace Moo-Young, Dean and Professor, College of Engineering, Computer Science, and
Technology, California State University, Los Angeles, CA
Dr. Eileen Murphy, Grants Facilitator, Ernest Mario School of Pharmacy, Rutgers University,
Piscataway, NJ
Dr. Duncan Patten, Research Professor, Department of Land Resources and Environmental
Sciences, Montana State University, Bozeman, MT
Dr. Stephen Polasky, Fesler-Lampert Professor of Ecological/Environmental Economics,
Department of Applied Economics, University of Minnesota, St. Paul, MN
Dr. Arden Pope, Professor, Department of Economics, Brigham Young University, Provo, UT
Dr. Stephen M. Roberts, Professor, Department of Physiological Sciences, Director, Center for
Environmental and Human Toxicology, University of Florida, Gainesville, FL
Dr. Amanda Rodewald, Professor, School of Environment and Natural Resources, The Ohio
State University, Columbus, OH
Dr. Jonathan M. Samet, Professor and Flora L. Thornton Chair, Department of Preventive
Medicine, University of Southern California, Los Angeles, CA
Dr. James Sanders, Director and Professor, Skidaway Institute of Oceanography, Savannah, GA
Dr. Jerald Schnoor, Allen S. Henry Chair Professor, Department of Civil and Environmental
Engineering, Co-Director, Center for Global and Regional Environmental Research, University of
Iowa, Iowa City, IA
Dr. Kathleen Segerson, Professor, Department of Economics, University of Connecticut, Storrs,
CT
Dr. Herman Taylor, Professor, School of Medicine, University of Mississippi Medical Center,
Jackson, MS
Dr. Barton H. (Buzz) Thompson, Jr., Robert E. Paradise Professor of Natural Resources Law at
the Stanford Law School and Perry L. McCarty Director, Woods Institute for the Environment,
Stanford University, Stanford, CA
Dr. Paige Tolbert, Professor and Chair, Department of Environmental Health, Rollins School of
Public Health, Emory University, Atlanta, GA
Dr. John Vena, Professor and Department Head, Department of Epidemiology and Biostatistics,
College of Public Health, University of Georgia, Athens, GA
v
-------
Dr. Thomas S. Wallsten, Professor and Chair, Department of Psychology, University of
Maryland, College Park, MD
Dr. Robert Watts, Professor of Mechanical Engineering Emeritus, Tulane University, Annapolis,
MD
Dr. R. Thomas Zoeller, Professor, Department of Biology, University of Massachusetts,
Amherst, MA
SCIENCE ADVISORY BOARD STAFF
Dr. Angela Nugent, Designated Federal Officer, U.S. Environmental Protection Agency,
Washington, DC
vi
-------
Table of Contents
EXECUTIVE SUMMARY 2
Evaluation of epidemiological data 2
Mode-of-action and sensitivity analysis of dose-response modeling 2
Exposure assessment and sensitivity analysis 3
BACKGROUND 5
RESPONSES TO EPA'S CHARGE QUESTIONS 6
Charge Question 1: 6
Response: 6
Charge Question 2: 8
Response: 9
Mode-of-action and dose-response modeling 9
Sensitivity Analysis 10
Charge Question 3: 11
Response: 12
Other Comments 17
REFERENCES 18
APPENDIX A - Minor edits 20
1
-------
EXECUTIVE SUMMARY
Various committees have evaluated the assessment of cancer risk associated with
exposure to inorganic arsenic. They include two National Research Council (NRC) committees
(1999, 2001) and the EPA Science Advisory Board (SAB) in 2007. In 2010, EPA's National
Center of Environment Assessment (NCEA) within the Office of Research and Development
(ORD) requested the SAB evaluate and comment on the agency's implementation of the SAB
recommendations in 2007 regarding EPA's revision of the cancer assessment of inorganic
arsenic. In response, a workgroup of the chartered Board reviewed EPA's draft document
entitled, "Toxicological Review of Inorganic Arsenic: In Support of the Summary Information on
the Integrated Risk Information System (IRIS)" (EPA/63 5/R-10/001) and was asked to comment
on three areas: evaluation of epidemiological literature; dose-response modeling approaches; and
the sensitivity analysis of the exposure assumptions used in the risk assessment. The SAB was
not asked to conduct a full peer review of the assessment, including EPA's calculation of the
cancer risk estimate. A summary of the SAB responses to EPA's charge questions follows with
further details included in the body of the report.
Evaluation of epidemiological data
The NRC in 1999 and 2001 concluded that ecological studies from the arsenic endemic
area of Taiwan provide the best available empirical human data and are appropriate for use in
dose-response assessment of arsenic in drinking water. In 2007, the SAB also supported the use
of the epidemiologic data on the Taiwanese population for estimating human cancer risk for
inorganic arsenic, especially to identify the potential range of responses of human populations.
The SAB agrees with these previous findings and the draft 2010 assessment that, based
on the current data, the Taiwanese data set remains the most appropriate data set for determining
the population risk of cancer from exposure to inorganic arsenic. The SAB also notes that the
EPA's draft 2010 assessment includes a comprehensive listing of the epidemiological literature
on arsenic and cancer up to 2007; however, the set of criteria that were used in evaluating the
studies needs to be better presented. The SAB recommends that, where possible, the summaries
of the epidemiology studies should include a quantitative or qualitative presentation of the
relative risk point estimates for a specific exposure comparison and the associated confidence
intervals. Furthermore, the SAB suggests that EPA consider including an addendum or appendix
describing major epidemiology studies published since 2007 that could substantially impact the
calculated cancer risk estimate.
Mode-of-action and sensitivity analysis of dose-response modeling
The NRC in 1999 and 2001 concluded that the available mode-of-action data on arsenic
did not provide a biological basis for using either a linear or nonlinear extrapolation. In 2007, the
SAB concluded that inorganic arsenic has the potential for a highly complex mode-of-action and
until more was learned about the complex pharmacokinetic and pharmacodynamic properties of
2
-------
inorganic arsenic and its metabolites, there was insufficient justification for the choice of a
specific nonlinear form of the dose-response relationship.
The SAB concludes that there are multiple potential mechanisms for arsenic
carcinogenicity and potential target tissues. The SAB notes that although a large amount of
research is available on arsenic's mode-of-action, the exact nature of the carcinogenic action of
arsenic is not yet clear. Therefore, there is not enough information in the literature to define a
mode-of-action for all of the relevant cancer endpoints for this assessment. The SAB
recommends that this complexity and limited understanding of the mode-of-action of arsenic
should be openly acknowledged in the 2010 draft assessment.
In 2007, the SAB recommended that EPA consider using alternative dose-response
models and perform a sensitivity analysis of the Taiwanese data with different exposure metrics,
with the subgroup of villages with more than one well measurement and using a multiplicative
model that includes a quadratic term for dose. The SAB finds that the sensitivity analysis of
dose-response modeling presented in the 2010 draft assessment was responsive to the SAB
previous recommendations. The SAB agrees with the conclusion that none of the alternative
models (i.e., quadratic, quadratic exponential and linear exponential) evaluated by EPA
materially changed the estimated risk levels versus use of a linear model. EPA also evaluated
whether the models were inordinately affected by the high end of the dose-response curve and
found that they were not. However, the SAB believes that more transparency and a better
scientific rationale for the agency's selection process are needed. To improve the clarity and
transparency of the draft assessment, a number of aspects of the sensitivity analysis should be
described in greater detail. They include the need for a more detailed description of the
Taiwanese data sets used in developing the risk model; a better description of the distribution of
well water arsenic concentrations across and within the 42 exposed villages; and a further
explanation of the sensitivity displayed for female bladder cancer risks. The SAB notes that,
while EPA's choice of a linear approach is consistent with EPA's risk assessment procedures, it
has produced a calculated upper-bound cancer risk estimate for arsenic that is of significant
public health concern. The SAB suggests that EPA discuss, possibly in other EPA complete risk
assessment documents, how the estimated risks for arsenic should be interpreted in light of
current estimated bladder and lung cancer incidence for the U.S. population.
Exposure assessment and sensitivity analysis
The 1999 NRC report noted that the assessment of arsenic exposure via drinking water is
often based on the measurements of arsenic concentrations in drinking water and assumptions
regarding the amount of water consumed. The 2001 NRC report added that the method used to
characterize arsenic dose in a study is a source of uncertainty in arsenic dose-response
assessment. Furthermore, the NRC report noted that the choice of the dose measurement affects
the interpretation of an epidemiological study and the choice of the dose-response model. The
2007 SAB agreed that water consumption (via drinking water, in beverages, or in cooking water)
assumptions could have an impact on the assessment of arsenic's risk. However, the 2007 SAB
did not recommend specific values for EPA to use in evaluating dose-response in the Taiwanese
study nor for levels of exposure in the U.S. population risk estimates. It instead recommended
3
-------
that uncertainty in exposure parameters be evaluated for both the Taiwanese study population
and the U.S. populations through sensitivity analyses. The 2007 SAB recommended that EPA
evaluate the drinking water consumption rate assumptions used with regard to highly exposed
and sensitive subpopulations. Additionally, the NRC (2001) recommended that EPA consider the
background dietary intake of inorganic arsenic and incorporate the adjustment values. The 2007
SAB also concluded that arsenic levels in food are important considerations for EPA's
assessment of lung and bladder cancer risk associated with exposures to arsenic in drinking
water. The 2007 SAB stated that a range of total arsenic food intake values should be included in
the sensitivity analyses.
The SAB finds EPA's revisions to the IRIS assessment to be partially responsive to SAB's
2007 recommendations regarding the exposure assumptions. The SAB provides two primary
general suggestions for improving the responsiveness of the assessment; they include, making
more transparent the scientific basis of the exposure assumptions used; and enhancing the rigor
and transparency of the sensitivity analysis. The basic approach to the sensitivity analysis is
adequate for meeting the minimum requirements for the intended purpose, and is responsive to
the SAB recommendation in that the impact of choice of assumptions is shown in terms of
specific cancer risks (lung and bladder, males and females). There are sufficient data to support
development of variability and/or uncertainty distributions for some inputs, such as drinking
water consumption rates in the United States, but the data are not available to assign
corresponding distributions for the Taiwanese populations.
The SAB notes that much of the documentation addressing the scientific basis of the
exposure assumptions is available through separate documents (e.g. EPA Issues Paper, 2005d)
that, if incorporated within the agency's draft IRIS assessment, will help address the SAB's
concerns. The SAB recommends that relevant information from these documents be integrated
within the current document as appropriate with the goal of enhancing transparency and
scientific credibility. The SAB has provided specific suggestions, within the body of the report,
for making the scientific basis of the exposure assumptions used more transparent and for
enhancing the rigor, and transparency, of the sensitivity analyses.
4
-------
BACKGROUND
Arsenic is a naturally occurring element that is found throughout the environment.
Exposure to inorganic arsenic can result in different health outcomes depending upon the route
of exposure. Arsenic compounds are used as a mordant in the textile industry, for preserving
hides, as medicinals, pesticides, pigments, and wood preservatives. EPA's health effects
assessment for inorganic arsenic was first made available on the Integrated Risk Information
System (IRIS) database in 1988. Various committees have reviewed aspects of the EPA's revised
assessment of cancer risk associated with exposure to inorganic arsenic. They include two
National Research Council (NRC) committees (1999, 2001) that concluded that the cancer risk
for inorganic arsenic should be based on internal cancers (lung and bladder) instead of skin
cancers. In 2005, the SAB was asked to review several EPA documents including:
• Office of Pesticide Programs' (OPP) Draft Science Issue Paper: Mode-of-action for
Cacodylic Acid (Dime thy larsinic Acid) and Recommendations for Dose Response
Extrapolation (U.S. EPA OPP, 2005a)
• Office of Research and Development (ORD) Issue Paper Cancer Risk Assessment for
Organic Arsenical Herbicides: Comments on Mode of Action, Human Relevance and
Implications for Quantitative Dose-Response Assessment (Appendix E of U.S. EPA OPP,
2005, USEPA ORD, 2005b).
• Office of Water's (OW) Draft Toxicologic Review of Inorganic Arsenic (U.S. EPA OW,
2005c).
At that time, the SAB convened a panel of experts to provide advice on the metabolism,
mode of action, dose-response, and approaches to low-dose extrapolation of cancer risk for
V
Dimethylarsinic Acid (DMA ) and inorganic arsenic (iAs). The SAB review report (EPA-SAB-
07-008) was issued in June 2007 and is available at
http://vosemite.epa.uov/sab/sabproduct.nsf/02ad90b 136fc21 ef85256eba00436459/EADABBF40
DED2A0885257308006741 EF/$File/sab-07-008.pdf.
In 2010, ORD's National Center for Environmental Assessment (NCEA) requested the
SAB evaluate and comment on EPA's implementation of SAB (2007) key recommendations in
the 2010 draft assessment entitled, "Toxicological Review of Inorganic Arsenic: In Support of the
Summary Information on the Integrated Risk Information System (IRIS)" (EPA/63 5/R-10/001).
In response to this request, the SAB convened a workgroup of the chartered Board to comment
on the agency's charge questions that focused on three areas: evaluation of epidemiological
literature, dose-response modeling approaches, and the sensitivity analysis of the exposure
assumptions used in the risk assessment. The SAB was not asked to conduct a full peer review of
the assessment, including EPA's calculation of the cancer risk estimate. The SAB workgroup
held a public face-to-face meeting, on April 6-7, 2010, to discuss and deliberate on its responses
to EPA's charge questions. The chartered Board conducted a quality review of the work group's
draft report (May 13, 2010) at a public teleconference on June 16, 2010 and a revised draft was
again reviewed on November 22, 2010. The final SAB report incorporates the SAB quality
review comments and public comments, both written and oral, which were received throughout
the advisory process.
5
-------
RESPONSES TO EPA'S CHARGE QUESTIONS
Charge Question 1:
The SAB concluded that the Taiwanese data set (Wu 1989; Chen et al., 1988, 1992)
remains the most appropriate data set to determine carcinogenic risk due to exposure to iAs.
They recommended that EPA should evaluate other published epidemiology studies using a
uniform set of criteria and document these findings in the assessment. They also stated that if one
or more studies provide potential utility, comparisons should be provided in the assessment.
EPA agreed that the Taiwanese data were the best available for determining the
carcinogenic risk due to exposure to iAs. In response to SAB's recommendation, an extensive
review and evaluation of all available human studies for iAs using the criteria suggested by the
SAB was performed by EPA and is summarized in Section 4.1 of the draft IRIS assessment and
included in tabular format in Appendix B. EPA concluded in the 2010 draft IRIS assessment that
there were no additional epidemiological studies that had comparable utility to the Taiwanese
data set (Wu 1989; Chen et al., 1988, 1992).
Please comment on EPA's response to the recommendations and the conclusions of the
SAB (2007) Arsenic panel regarding the evaluation of the epidemiological literature.
Response:
In 1999, the NRC 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 2001 NRC report also concluded that the data from southwestern Taiwan
remain appropriate for use in dose-response assessment of arsenic in drinking water. In 2007, the
SAB supported the use of the epidemiologic data on the Taiwanese population for estimating
human cancer risk for inorganic arsenic especially to identify the potential range of responses of
human populations. The 2007 SAB urged the agency to consider other epidemiologic studies
from the United States and other countries, utilizing a uniform set of evaluative criteria, as they
develop their risk assessment. The 2007 SAB recommended consideration of the following
additional factors when reviewing the studies:
1. Estimates of the level of exposure misclassification;
2. Temporal variability in assigning past arsenic levels from recent measurements;
3. The extent of reliance on imputed exposure levels;
4. The number of persons exposed at various estimated levels of waterborne arsenic;
5. Study response/participation rates;
6. Estimates of exposure variability;
7. Control selection methods in case-control studies; and
8. The resulting influence of these factors on the magnitude and statistical stability of risk
estimates.
6
-------
The SAB concludes that EPA has been responsive to the 2007 SAB recommendations in
evaluating the epidemiology studies published through 2007. The 2010 draft IRIS assessment
presents a well organized and very comprehensive overview of the epidemiological literature on
arsenic and cancer through 2007. The SAB recognizes that there are limitations that are inherent
in the design of environmental epidemiological investigations, particularly regarding
reconstruction of past exposure levels. EPA has described the limitations of each study in
Section 4.1 of the draft IRIS assessment, and presented a summary of study strengths and
limitations in the tables of Appendix B. The systematic review of the literature, however, needs
to more clearly state the set of criteria that were used in evaluating the studies. Additional
clarification and documentation on how various study design factors were considered and
weighted in the evaluation are needed. In addition, there are aspects of studies that are discussed
in Section 4.1 narrative that are not included in the summary table of Appendix B. The SAB
recommends that the tables in Appendix B be reformatted to present the study summaries more
clearly and in a more consistent format including adding any essential information from
references into the text for clarity.
The SAB supports the 2007 SAB conclusion that the Taiwanese data set (Wu 1989; Chen
et al., 1988, 1992) remains the most appropriate data set for determining the population risk of
cancer to exposure to inorganic arsenic. The limitations of the Taiwanese studies are well
presented, particularly regarding the ecologic study design, use of death certificates, and
assumptions regarding lifetime individual arsenic exposure. The strengths of these studies
include availability of community drinking water exposure levels, large populations and person-
years of follow-up, and consideration of important potential confounders including
socioeconomic status, lifestyle, dietary patterns, and medical care. The SAB acknowledges the
concerns expressed regarding the limitations of the Taiwanese data set; however given the
fundamental mission of EPA to protect public health, these well conducted and extensively
reviewed studies remain the most appropriate critical studies.
The SAB received public comments that suggested when comparing the large number of
epidemiological studies that demonstrate varying results the power calculations for the studies
can provide important insights and should be taken into consideration. The power of an
epidemiological study is the probability of detecting an association of a specified strength
between exposure and disease if one exists. For example, in studies where statistical significance
is not achieved, failure to identify an association may be a reflection of a limitation of the power
of the study. The SAB, however, notes that while the relative power of various studies is important
to convey, this should not be done by presenting only power calculations. Power calculations are
useful in planning a study, but after the study is completed, the most informative presentation of
epidemiologic findings that combines both the observed results and reflects the power of the study is
the relative risk (RR) point estimates for a specified exposure comparison and the associated
confidence intervals. Furthermore, systematic presentation of numbers of individuals in each
exposure stratum provides the reader with a sense of relevant sample size within strata and the
robustness of the exposure contrast. While a restricted range of exposure within a study population
will limit estimation, it is also likely that studies carried out at a lower-level of exposure will be
estimating an effect smaller than that at higher levels for a categorical comparison of higher-exposed
to lower-exposed. For instance, the required sample size should be larger when a smaller range of
exposures is observed (e.g., the U.S. studies), since the expected magnitude of the RR for low-level
exposure is lower. The SAB also recognizes that many published arsenic studies may not present
7
-------
specific power calculations or RR and that a detailed quantitative comparison is difficult. Where
possible, the summaries of the epidemiology studies should include a quantitative or qualitative
presentation of the relative risk point estimates for a specific exposure comparison and the
associated confidence intervals. This should be included both in the study descriptions in Section
4.1 and in the table of studies in Appendix B.
As noted by public comments, the SAB agrees that failure to control potential
confounders or misclassification of study population exposure levels may bias study results. In
the presentation of one of the critical epidemiology studies (Chen et al. 1992), the IRIS
assessment (p.38) states, "a weakness of the study is the assumption that an individual's arsenic
intake remained constant from birth to the end of the follow-up period; this flaw possibly led to
the underestimation of risk." Other epidemiological studies also had similar issues. Indirect
measures of individual exposure were used to estimate population exposure levels for all of the
epidemiology studies. In Section 4.1, the narrative presenting the epidemiology studies should
include a more detailed discussion of bias including literature citations addressing the potential
for bias, both underestimating and overestimating of risk, due to confounders or limitations in
exposure estimation. While the existence of bias can usually be proposed with some certainty, the
key issue is whether the quantitative consequences of bias are of sufficient magnitude to be of
concern. Methods are available for this purpose (see, for example: Lash, Fox, and Fink: Applying
Quantitative Bias Analysis to Epidemiological Data, Springer, 2009). The SAB suggests that the
IRIS assessment include a simple table that identifies potential biases (misclassification of
exposure, misclassification of disease, omitting potential confounders, etc.) and the potential
magnitude and direction of bias in inferences that are drawn from the study data. A simple
summary could then relate these sources of bias to their impact in the data and methods used in the
IRIS assessment.
The IRIS 2010 assessment includes an extensive review of published epidemiology
studies up to and including the year 2007. The SAB recognizes that the assessment cannot be
continually updated with every newly published paper and it is not the purpose of IRIS to
provide real time summaries of advancing science. However, given the large amount of ongoing
research on the health effects of arsenic, the SAB has concerns about the 2007 cutoff. In order to
ascertain if new studies will impact the 2010 assessment, EPA should consider including an
addendum or appendix describing major epidemiology studies published since 2007 (i.e., those
studies that can influence the dose-response assessment due to large sample size or effect estimate
that is substantially different from that estimated by Chen et al. (1988, 1992).
Charge Question 2:
The SAB noted the possibility of a nonlinear dose-response at low exposures, but due to
uncertainty in the mode-of-action (including pharmacokinetics and dynamics) the use of a linear
low dose extrapolation approach to determine the cancer risk for iAs was recommended using
cancer incidence from the Taiwanese data set. In addition, the SAB stated that EPA should
perform a sensitivity analysis for the variables in the cancer modeling with respect to the
Taiwanese data set (i.e., exposure metrics, subgroup of villages with more than one well
measurement, and a multiplicative model that includes a quadratic term for dose). The SAB
8
-------
concluded that overall, EPA had implemented the recommended modeling by NRC (2001). Also,
the SAB made recommendations to perform a sensitivity analysis regarding the robustness of the
model and alternative formulations.
Consistent with the SAB recommendations, EPA used a linear low-dose extrapolation
approach and conducted a sensitivity analysis of nonlinear forms of the dose-response in the
2010 draft IRIS assessment. EPA also explored nonlinear forms of the dose-response from the
Taiwanese data set (Wu 1989; Chen et al., 1988, 1992). Sensitivity analyses using alternative
dose-response models produced potency estimates similar to the linear approach.
Please comment on EPA's response to the SAB's recommendations and conclusions
regarding the approach to modeling inorganic arsenic cancer risks and the corresponding
sensitivity analyses.
Response:
Mode-of-action and dose-response modeling
The 1999 NRC Committee concluded that the mechanism or mode-of-action by which
inorganic arsenic causes toxicity, including cancer, is not well established. This conclusion was
again supported by the NRC in 2001 which noted that although a large amount of research is
available on arsenic's mode-of-action, the exact nature of the carcinogenic action of arsenic is
not yet clear. Therefore, the 2001 NRC report concluded that the available mode-of-action data
on arsenic did not provide a biological basis for using either a linear or nonlinear extrapolation.
In 2007, the SAB concluded that inorganic arsenic has the potential for a highly complex
mode-of-action and until more is learned about the complex pharmacokinetic and
pharmacodynamic properties of inorganic arsenic and its metabolites, there is not sufficient
justification for the choice of a specific nonlinear form of the dose-response relationship. The
NRC (2001) concluded that the most appropriate approach was to base risk assessments on a
linear dose response model that includes the Southwestern Taiwan population as a comparison
group.
The SAB agrees that there are multiple potential mechanisms for arsenic carcinogenicity
and potential target tissues which make it very difficult to do a single risk assessment model.
This complexity and limited understanding of the mode-of-action of arsenic should be openly
acknowledged in the 2010 draft IRIS assessment. While there is an ever increasing literature on
arsenic, there is not enough information in the literature to define a mode-of-action for all of the
relevant cancer endpoints for this assessment. The SAB notes that it is a reasonable hypothesis
that bladder cancer is the result of repeated cell injury, cell death and compensatory proliferation;
but there is not enough specific data at this point to confirm the hypothesis, nor are there
hypotheses to explain the role of arsenic in lung cancer. For these reasons, the SAB concurs with
EPA's rationale for choosing a linear default approach for risk assessment.
9
-------
Sensitivity Analysis
The 2007 SAB recommended that EPA perform a sensitivity analysis of the Taiwanese
data with different exposure metrics, with the subgroup of villages with more than one well
measurement and using a multiplicative model that includes a quadratic term for dose. The SAB
finds that the sensitivity analysis of dose-response modeling presented in the 2010 IRIS
assessment was responsive to the previous 2007 SAB recommendations. Specifically, EPA was
asked to evaluate a model using a quadratic term for dose. EPA evaluated the differences
between a linear model and three non-linear models: quadratic, quadratic exponential and linear
exponential. Results are described on p. 143, which concludes that "within the range of
exposures covered by the epidemiological data, the alternative forms predict very similar risks."
It would be very helpful if the results could be shown graphically, e.g., by showing the dose-
response data and model dose-response curves for selected endpoints and age and gender classes.
The SAB agrees with the conclusion that none of the alternative models materially changed the
estimated risk levels versus use of a linear model. EPA also evaluated whether the models were
inordinately affected by the high end of the dose-response curve. They were not. This was
evaluated by running the models without the highest exposure group. EPA evaluated whether
exclusion of a reference population influenced the dose-response curve. Results of this analysis
(see Fig. 5-2) suggest that exclusion of the reference population did have an effect on risk
estimates. EPA evaluated the pros and cons of including a comparison population in a 2005
issues paper (Issue Paper: Inorganic Arsenic Cancer Slope Factor, Final Draft, July 23, 2005).
The SAB recommends that the rationale from the issue paper be included in the draft IRIS
assessment, and the reference population described in greater detail. This will provide more
transparency and strengthen the scientific rationale for the agency's selection process.
To improve the clarity and transparency of the draft IRIS assessment, there are a number
of aspects of the sensitivity analysis that should be described in greater detail. They include:
• More detailed description of underlying data. The assessment would benefit from a
more detailed description of the Taiwanese data sets used in developing the risk model.
The data sets are briefly described in section 4.1.1 as part of the review of the Chen et al.
1988a, 1992 and Wu et al., 1989 studies, and key features are summarized in Table B-l.
However, readers are required to piece together this information on their own in order to
understand the basis for the risk modeling presented in section 5.3.
• Variability of well water arsenic concentrations. The distribution of well water arsenic
concentrations across and within the 42 exposed villages is not adequately described.
Only medians and ranges across the whole set of villages are presented in Table B-l.
While the assessment mentions that the number ranged from 2-47 measurements, the
variability of measurements both within and across wells within a given village is not
provided. This information needs to be presented to assist in understanding the results of
the sensitivity analysis the 2007 SAB requested. It would also be helpful to see a more
quantitative characterization of how the 1974-1976 well water re-testing results differed
from the results of tests conducted in 1962-1964, on which the risk modeling relied.
Table B-l indicates the results were "similar"; however, it is not clear how to interpret
this.
10
-------
• Upper and lower limits in water concentration. EPA responded to SAB's request for
sensitivity analysis or Monte Carlo analysis with respect to well water concentrations in
the villages with more than a single measurement by re-estimating the model using
minimum and maximum values of the concentrations for each village. Table 5-10
indicates the effect (in terms of estimated cancer incidence) is up to about a ± 30%
change. Although EPA used upper and lower limits, rather than low and high percentile
values or Monte Carlo analysis as SAB had suggested, the sensitivity analysis responds
adequately to the recommendation. As noted above, however, more information on the
variability in the underlying water concentration data is needed to substantiate the
reported models and results.
• Modeling data and parameters. The SAB suggests that EPA publish the data and
parameter tables used in its modeling analysis. As requested by the 2007 SAB report, this
would strengthen the scientific credibility and transparency in the assessment.
• Selection of a reference population. EPA has tested the sensitivity of the risk model
with respect to the choice of reference population (southwest Taiwan, all Taiwan, or no
reference population) and to the value of non-water arsenic intake (i.e., in accordance
with EPA's document, this refers to food intake) for both reference and exposed
populations. Results indicate that the cancer incidence risks are fairly robust, with the
exception of female bladder cancer risks. The sensitivity displayed for female bladder
cancer risks seems to warrant further explanation - the result is described, but not
explained, in the accompanying text (pp. 141-2). Additionally, EPA should examine
whether any combinations of these parameter variations will affect the assessment- e.g.,
using different non-water intake values in combination with a different reference
population.
The SAB notes that there is tremendous interest in the risk associated with consumption
of water that is contaminated with inorganic arsenic and suggests that EPA discuss how their
results should be interpreted in light of existing population-level data on bladder and lung cancer
risk for exposure levels that are relevant for U.S. populations. The idea of providing a "reality
check" on the estimated risk levels was discussed. The SAB recognizes that IRIS toxicological
reviews are not intended to provide a complete risk assessment but rather a summary and
synthesis of the toxicological evidence that supports risk assessment. Hence, an estimation of
risk attributable to arsenic in drinking water in U.S. populations versus the observed incidence of
cancer is not appropriate within the purview of this document. The SAB considers this as a
difficult but important exercise and recognizes that this is probably better suited for inclusion in
other risk assessment and characterization documents developed by the agency.
Charge Question 3:
The SAB did not recommend specific values for the exposure assumptions or parameters
used in the cancer model. They did, however, recommend evaluating the impact on the cancer
11
-------
risk of using a range of values, assessing the variability, and conducting a sensitivity analysis for
exposure parameters (e.g., water intake, background dietary exposure).
EPA evaluated the impact on the estimated cancer risk of using a range of exposure
parameter values (e.g., water intake, background dietary exposure), assessed variability, and
conducted a sensitivity analysis. After the completion of these analyses, values were chosen for
exposure assumptions based upon the best available science taking into account the NRC (2001)
recommendations.
Please comment on EPA's sensitivity analyses and choice of the exposure assumptions used
in modeling cancer risk as recommended by the SAB (2007) Arsenic panel.
Response:
The 1999 NRC report noted that assessment of arsenic exposure via drinking water is
often based on the measurements of arsenic concentrations in drinking water and assumptions
regarding the amount of water consumed. Such data are estimates, the uncertainty of which will
depend on the method used. The 2001 NRC report added that the method used to characterize
arsenic dose in a study is a source of uncertainty in arsenic dose-response assessment.
Furthermore, the NRC report noted that the choice of the dose measurement affects the
interpretation of an epidemiological study and the choice of the dose-response model.
The 2007 SAB agreed that water consumption (via drinking as water, in beverages, or in
cooking water) assumptions have a substantial impact on the assessment of arsenic's risk.
However, the 2007 SAB did not recommend specific values for EPA to use in evaluating dose-
response in the Taiwanese study nor for levels of exposure in the U.S. population risk estimates.
It did recommend that uncertainty in this parameter be evaluated for both the Taiwanese study
population and the U.S. populations at risk. The 2007 SAB recommended that EPA should:
1) Evaluate the impact of drinking water consumption rates associated with more highly
exposed population groups with differing exposures and susceptibilities (e.g., children,
pregnant women);
2) Incorporate variability parameters for individual water consumption into their analysis for
dose-response in the Taiwanese population as they have done for the U.S. population;
3) Conduct sensitivity analyses of the impact of using a range of consumption values for the
Taiwanese population;
4) Provide a better justification for assuming different consumption levels by gender or in the
absence of such a justification, conduct additional sensitivity analyses to examine the impact
of equalizing the gender-specific consumption level;
5) More fully articulate and document how different sources of water intake, as well as
variability, are incorporated into the risk model (e.g. data for intake from beverages and
cooking water).
The NRC (2001) recommended that EPA consider the background dietary intake of
inorganic arsenic and incorporate the adjustment values of 0, 10, 30, and 50 jag per day into the
12
-------
cancer risk calculations. The 2007 SAB also agreed that arsenic levels in food are important
considerations for EPA's assessment of lung and bladder cancer risk associated with exposures
to arsenic in drinking water. However, the 2007 SAB once again did not recommend a specific
value for EPA to use in its base risk assessment. It did recommend a range of values for
consideration by EPA in its sensitivity analysis and the 2007 SAB offered suggestions to EPA
for additional analytical steps to clarify the impact of food levels of arsenic on dose-response and
exposure as it revises its risk estimates. The 2007 SAB recommended that EPA should:
1) Conduct sensitivity analyses using a range of total arsenic food intake values from at least
50 to 100 jag per day to perhaps as high as 200 jag per day to assess the impact of this range
of dietary intakes on risk of lung and bladder cancer from exposure via drinking water in the
Taiwan cohort;
2) Not assume that the control population has an intake value of zero arsenic from food;
3) Apply greater rigor in their discussions of data used in these assessments (e.g., sources,
methodological and analytical issues, bioavailability); and
4) Give immediate research attention to the issue of arsenic bioavailability.
The SAB finds EPA's revisions to the IRIS assessment to be partially responsive to the
2007 SAB recommendations regarding the sensitivity analyses and choice of the exposure
assumptions used in modeling cancer risk. The SAB provides two primary general suggestions
for improving the responsiveness of the assessment. They include, making more transparent the
scientific basis of the exposure assumptions used and enhancing the rigor and transparency of the
sensitivity analysis.
The basic approach to the sensitivity analysis is adequate for meeting the minimum
requirements for the intended purpose, and is responsive to the 2007 SAB recommendation in
that the impact of choice of assumptions is shown in terms of specific cancer risks (lung and
bladder, males and females). In evaluating the consequences of choices regarding modeling
assumptions and intake values, the IRIS assessment states, "The agency felt that the currently
available data were insufficient to support detailed probabilistic uncertainty and variability
estimation." The SAB agrees with this conclusion. There are sufficient data to support
development of variability and/or uncertainty distributions for some inputs, such as drinking
water consumption rates in the United States, but the data are not available to assign
corresponding distributions for the Taiwanese populations.
The SAB notes that much of the documentation addressing the scientific basis of the
exposure assumptions was available through separate documents (e.g. EPA Issues Paper, 2005d)
that if incorporated within the current assessment, will help address the SAB's concerns. The
SAB recommends that relevant information from these documents be integrated within the
current document as appropriate with the goal of enhancing transparency and scientific
credibility.
The SAB is providing the following specific suggestions for making the scientific basis
of the exposure assumptions used more transparent and enhancing the rigor and transparency of
the sensitivity analysis.
13
-------
• Better explanation of what the sensitivity analysis shows. The sensitivity analyses
presented offer insight as to how the cancer potency estimates change as drinking water
consumption and non-water arsenic intake assumptions change. The various non-water
arsenic intake rate assumptions produced modest changes in risk, with the exception of
bladder cancer risk in females. This calculated risk was very sensitive to the non-water
intake rate assumption. The assessment and this analysis will be strengthened by
providing a short explanation for why this is the case.
• Need for better justified default assumptions. Despite some effort to discuss drinking
water consumption rates and sources of information for non-water arsenic intake rates,
the reasons for some of the specific values chosen to be included in the sensitivity
analyses are not clearly justified. For example, the "default" drinking water consumption
rate for Taiwanese males is 3.5 L/day, citing precedent from U.S. EPA (1988), Chen et
al. (1992), and NRC (1999 and 2001). For the sensitivity analysis, alternative values of
2.75, 3.0, and 5.1 L/day were evaluated [along with alternative values for Taiwanese
females]. No rationale is provided for these specific numbers, other than they are thought
by the agency to span a "reasonable range of values" (see page A-6). To enhance
transparency in this example, it would be helpful to know the scientific basis for selecting
the lowest and highest numbers (defining the range). Also, if the intent was to illustrate
effects at the boundaries of the range of drinking water consumption rates, it is unclear
why the lowest estimate for males (2.75 L/day) was not consistent with the lowest
estimate for females (2.0 L/day) (see Table 5-10), especially given the SAB's request to
justify different consumption values for men and women. Furthermore, no values for
drinking water consumption rates for Taiwanese women were evaluated below the
"default" rate of 2.0 L/day, suggesting that the value selected by the agency is at the limit
of the range of reasonable values for this parameter. The effects on risk were determined
based on assumptions that both the reference and exposed populations had non-water
intake rates of 0, 30, and 50 |ig/day arsenic. Although compliant with SAB's 2007
recommendations, better discussion of dietary intake of inorganic arsenic would help the
reader understand whether the various values included in the analysis represent different
interpretations of the existing data, bounding estimates, or something else.
• Consider additional permutations of gender-specific water consumption. The 2007
SAB recommended: "Because data on gender differences in consumption in Taiwan are
limited, a better justification for assuming different consumption levels by gender is
needed, particularly given the lack of sex difference in consumption in United States and
observed in studies from other countries (Watanabe et al., 2004). In the absence of such a
justification, the SAB recommends an additional sensitivity analysis to examine the
impact of equalizing the gender-specific consumption level." The agency complied with
this recommendation to some extent, evaluating the effect on risk of setting the drinking
water consumption rate for both Taiwanese males and females at 2.75 L/day in the
sensitivity analysis. However, the basis for the choice of this particular drinking water
consumption rate is not explained. Also, by examining a single drinking water
consumption rate for both sexes, the influence of selection of different rates on resulting
risk is not illustrated. In order to be responsive to the 2007 SAB recommendation,
14
-------
discussion of the impact of using a single drinking water consumption rate for males and
females for the Taiwanese populations needs to be justified and expanded.
Need to clearly delineate the basis for water concentration assumptions. Based on the
data in tables 5-10 and 5-11, it isn't clear if EPA has completed the calculations that the
SAB requested. Those tables noted that the sensitivity analyses used minimum and
maximum village water arsenic concentration values. It isn't clear if only the villages
with more than one well measurement were used or if all the villages were used. EPA
needs to clarify the water concentration assumptions. This recommendation is also
consistent with recommendations under charge question #2.
Need to address water consumption rates of susceptible groups. The 2007 SAB
recommended that the "EPA should evaluate the impact of drinking water consumption
rates associated with more highly exposed population groups with potentially different
exposures and susceptibilities (e.g. children, pregnant women) in its arsenic exposure
estimates as the agency determines the overall effects of drinking water consumption
rates on arsenic risk." In the current 2010 draft IRIS assessment, the impact of drinking
water consumption rates associated with more highly exposed population groups with
potentially different exposures and susceptibilities (e.g. children, pregnant women) in its
arsenic exposure estimates has not been evaluated. During the April meeting, the agency
indicated that including these populations in the sensitivity analysis would be difficult
and of limited value. So that the response to this 2007 SAB comment is clear, an
explanation of why this aspect of the sensitivity analysis was not conducted should be
included in Appendix A.
More complete and graphical analysis. EPA has responded to the 2007 SAB's
suggested sensitivity analysis with the development of Tables 5-10 and 5-11 along with
Figure 5-2 showing the influence of various exposure assumptions including water
arsenic concentration, non-water arsenic intake, and water consumption on various cancer
endpoint risks. The tables and figure are efficient in providing a "snapshot" of their
influence for various assumed point estimates; however, a more complete description of
their influence can be shown by graphing across the range of plausible values.
Admittedly, the graphical representation will be less efficient (i.e., require more space)
but will provide a more complete depiction. To the extent possible, it would be useful to
illustrate on these graphs the various historically and currently "assumed" values.
Testing the effects of layered assumptions. To further respond to the 2007 SAB's
recommendation, EPA tested the effects of changing assumptions one at a time. This
approach is necessary to clearly show how individual values potentially affect cancer
potency and risk. This approach does not, however, indicate how changes in assumptions
might interact to produce overall changes in potency and risk. Testing all of the various
permutations of changes in assumptions in a sensitivity analysis would be arduous and of
dubious value. Nevertheless, it may be instructive to examine selected sets of exposure
assumptions and their effect on cancer potency. This would provide an indication of the
extent to which a reasonable range of exposure assumptions in the aggregate has the
potential to affect cancer potency estimation.
15
-------
• Clarification of what the exposure assumptions are intended to represent. It is often
unclear in the assessment whether the exposure assumptions (e.g., drinking water
consumption rate) selected are intended to represent best estimates of the mean for the
exposed population, upper confidence estimates of the mean, upper percentile values,
upper confidence limit estimates of an upper percentile value, or something else. This
should be specified in the IRIS assessment. During the April meeting, the agency
indicated that different types of assumptions may be appropriate for different values. The
rationale for why a particular value is used should be provided in the IRIS assessment.
For example, why an upper percentile drinking water ingestion rate is appropriate for the
U.S. population, while an average (or upper bound average) assumption is used for the
Taiwanese population.
• The bases for the exposure assumptions selected are not adequately described. The
SAB in 2007 stated, "Much greater rigor needs to be applied in discussing and presenting
documented data sources and making clear the basis on which assumptions are being
made and the relative strength of those assumptions." That criticism applies to the 2010
version of the IRIS assessment as well. Some examples include:
o For non-water arsenic intake, EPA has selected an assumed intake value of 10 (j,g/day.
Discussion in support of this selection occurs on pgs. 123-124 of the revised
assessment and is based on six references including US EPA (1989), Schoof et al.
(1998), Yost et al. (1998), NRC (1999), NRC (2001), and EPA (2005c). Of these,
there are only two references that relate to the peer-reviewed primary literature,
reflecting the scarcity of data from which to base this estimate. Although EPA does a
reasonable job of discussing these reports, the current assessment lacks a specific
rationale or justification for the selected value. It appears that the US EPA 1989
reference supporting an intake range of 2 (J,g/day to 16 (J,g/day may provide the
rationale for this selection. Since this reference is not easily available, the SAB
recommends that within the IRIS assessment a more complete discussion of data and
evidence supporting this intake range be provided in a manner similar to what has
been provided for Schoof et al. 1998 and Yost et al. 1998. In the current assessment,
it is unclear what the 2 to 16 (j,g/day estimate is based on. Moreover, the current
assessment does not provide a specific justification or rationale for this selection, but
rather makes a broad statement "Based on available information, EPA selected 10 jag
/day as the best estimate for non-water arsenic intake (food sources) in baseline
calculations." The selection of this value can be strengthened by: 1) elaborating on
the lack of data or evidence upon which to base this estimate; 2) distinguishing
between evidence that is primary (i.e., peer-reviewed with data collection) and reports
that provide expert assessment, and 3) providing specific and scientific justification
for the selected value that can be traced to the primary literature. Again, because of
the effect this parameter has on the risk estimates, providing support for the values
chosen for this parameter is important.
o The current dose-response assessment is based on an assumed water intake value of
3.5 and 2.0 L/day for men and women, respectively. As with the assumed values for
16
-------
non-water intake above, justification for these values can be strengthened by
establishing a clear link to data within the primary literature where possible. The
specific relevant findings from Chen et al. 1992 and Chowdhury et al. 2001 should be
provided in relation to the selected values. In the current assessment, it appears that
EPA justified the selected values largely based on precedent (e.g., EPA and NRC
reports) rather than on the data reported in the primary literature. It is unclear why
EPA did not base their estimate on the data of Chowdhury et al. 2001 since it is
unique and relevant. No discussion is provided of the data available from Chen et al.
1992. To the extent that EPA relies on previous EPA and NRC assessments, the link
to the primary data (if available) should be maintained. The problem illustrated by the
2010 assessment is that these assumed values take on a life of their own and the
evidence upon which they are based is lost, i.e., the scientific basis for the
assumptions is no longer discernible.
• The reason for limiting non-water intake to dietary sources is not explained. Non-
water exposure is currently assumed to consist entirely of arsenic in the diet. For
completeness and transparency, EPA should provide a short description of alternate
routes of exposure (e.g. inhalation, non-dietary ingestion, dermal absorption) from other
media such as soil and include arsenic intake estimates using EPA's routine exposure
assumptions for both the Taiwan and the U.S. populations; EPA should provide
justification for why these other exposures were not considered in the current dose-
response assessment. If the reason is that other pathways are assumed to be minor relative
to arsenic intake from diet, some illustration of this should be provided as justification.
Other Comments
• More clear delineation of organic vs. inorganic exposure assumptions. It would be
helpful to provide a paragraph for IRIS users explaining why the organic arsenic
compounds do not affect the risk estimates for inorganic arsenic. The explanation will
probably be fairly straight forward for the seafood organic arsenic compounds. This may
not be as straight forward for any organic arsenic compound in produce (e.g. rice, etc.).
As a related comment, when discussing non-water arsenic intake care should be taken to
distinguish between inorganic and organic or total arsenic in food. The current draft
assessment is in some places ambiguous, referring simply to "arsenic." (see pages 123-
124).
• Value in identifying research gaps. Given the importance and scarcity of data for
purposes of estimating exposure, the SAB suggests that EPA provide a short paragraph
describing the research needs along with suggested designs to produce credible estimates
for water and non-water intake rates. The research needs are not only to provide point
estimates, but data for distribution analysis to support the more credible stochastic
approaches to risk estimation. Maybe 10 years from now, we will not find ourselves in
the position that we are in now of relying on largely the same sparse/inadequate data for
risk estimation that we were in 10 years ago.
17
-------
REFERENCES
Chen, C-J; Chen, C-W; Wu, M-M; Kuo, T-L. (1992) Cancer potential in liver, lung, bladder, and
kidney due to ingested inorganic arsenic in drinking water. Br J Cancer 66(5):888-892.
Chen, C-J; Kuo, T-L; Wu, M-M. (1988a) Arsenic and cancers. Lancet: letter to the editor.
February 20, 1988.
Chowdhury, UK; Rahman, MM; Mondal, BK; et al. (2001) Groundwater arsenic contamination
and human suffering in West Bengal - India and Bangladesh. Environ Sci 8(5):393—415.
Lash, Fox, and Fink (2009) Applying Quantitative Bias Analysis to Epidemiological Data.
Springer.
NRC (National Research Council). (1999) Arsenic in drinking water. National Academy Press,
Washington, DC. Available online at http://vvvvvv.nap.edu/openbook/0309063337/html/Rl .html.
NRC (National Research Council). (2001) Arsenic in drinking water (2001 update). National
Academy Press,Washington, DC. Available online at
http://vvvvvv.nap.edu/openbook/0309076293/html/R 1 .html.
Schoof, RA; Yost, LJ; Crecelius, C; et al. (1998) Dietary arsenic intake in Taiwanese districts
with elevated arsenic in drinking water. Hum Ecol Risk Assess 4:117-135.
U.S. EPA SAB (2007) Advisory on EPA's Assessments of Carcinogenic Effects of Organic and
Inorganic Arsenic: A Report of the US EPA Science Advisory Board. (EPA-SAB-07-008)
Available online at
http://vosemite.epa.gov/sab/sabproduct.nsf/EADABBF40DED2A08852573080Q6741 EF/$File/sa
b-07-008.pdf
U.S. EPA (2005a) Draft Science Issue Paper: Mode-of-action for Cacodylic Acid
(Dimethylarsinic Acid) and Recommendations for Dose Response Extrapolation Office of
Pesticide Programs. Available online at:
http://vvvvvv.epa.gov/pesticides/reregistration/cacodvlic acid/dm a moa.pdf
U.S. EPA (2005b) Issue Paper Cancer Risk Assessment for Organic Arsenical Herbicides:
Comments on Mode of Action, Human Relevance and Implications for Quantitative Dose-
Response Assessment (Appendix E) Office of Research and Development. Available online as
Appendix E at: http://vvvvvv.epa.gov/pesticides/reregistration/cacodvlic acid/dm a moa.pdf
U.S. EPA (2005c) Draft Toxicological Review of Ingested Inorganic Arsenic. Office of Water.
Available online at:
http://vvater.epa.gov/scitech/svvguidance/vvaterqualitv/standards/criteria/aqlife/pollutants/arsenic/
upload/2007 07 12 criteria arsenic sab AsDraft SAB.pdf
18
-------
U.S. EPA (2005d). Issue paper: inorganic arsenic cancer slope factor. Arsenic Cancer Slope
Factor Workshop. Available online at
http://water.epa.gov/scitech/swguidance/waterqualitv/standards/upload/2007 07 12 criteria ars
enic sab ASlssues SAB.pdf
U.S. EPA (1989) Report on Arsenic (As) Work Group meetings. Memorandum. From:
Abernathy, CO; Marcus, W; Office of Drinking Water; and Chen, C; Gibb, H; White, P; Office
of Research and Development; to Cook, P; Office of Drinking Water; and Preuss, P; Office of
Regulatory Support and Scientific Management. February 23.
Wu, MM; Kuo, TL; Hwang, YH; Chen, CJ. (1989) Dose-response relation between arsenic
concentration in well water and mortality from cancers and vascular diseases. Am J Epidemiol
130:1123-1132.
Watanabe, C; Kawata, A; Sudo, N; Sekiyama, M; Inaoka, T; Bae, M; Ohtsuka, R. (2004) Water
intake in an Asian population living in arsenic-contaminated area. Toxicol Appl Pharmacol.
198(3):272-82. Review.
Yost, LJ; Schoof, RA; Aucoin, R. (1998) Intake of inorganic arsenic in the North American diet.
Hum Ecol Risk Assess 4:137-152.
19
-------
APPENDIX A - Minor edits
• Pages 139 and 140: Providing some information in these tables about the range in village
water arsenic concentrations would be useful.
• Type in footnote for Table 5-11. Table 5-8 should probably be Table 5-10.
• Page 141, line 27. Tables 5-6 and 5-9 should be Tables 5-10 and 5-11
• Page 142 - line 3. Should both increased and decreased be there?
20
-------