United States
Environmental Protection
Agency
Report on the Peer Review
of the U.S. Environmental
Protection Agency's Draft
External Review Document
"Perchlorate Environmental
Contamination:
Toxicological Review and
Risk Characterization"
-------
EPA/63 5/R-02/003
June 2002
Report on the Peer Review of the U.S. Environmental Protection Agency's
Draft External Review Document "Perchlorate Environmental
Contamination: Toxicological Review and Risk Characterization"
U.S. Environmental Protection Agency
Sacramento, California
March 5-6, 2002
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
-------
NOTICE
This document has been reviewed in accordance with U.S. Environmental Protection
Agency (EPA) policy and approved for publication. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
This report was prepared by Eastern Research Group, Inc. (ERG), an EPA contractor
(Contract No. 68-C-99-237, Task Order No. 58), as a general record of discussion for the peer
review meeting. This report captures the main points of scheduled presentations, highlights
discussions among the reviewers, and documents the public comments provided at the peer
review meeting. This report does not contain a verbatim transcript of all issues discussed during
the peer review, and it does not embellish, interpret, or enlarge upon matters that were incomplete
or unclear. Except as specifically noted, no statements in this report represent analyses by or
positions of EPA or ERG.
-------
Table of Contents
List of Abbreviations vi
Executive Summary vii
1.0 Introduction 1-1
1.1 Background 1-1
1.2 Scope of the Peer Review 1-3
1.2.1 Selecting the Reviewers 1-3
1.2.2 Activities Prior to the External Peer Review Meeting 1-5
1.2.3 Activities at the External Peer Review Meeting 1-6
1.2.4 Activities Following the External Peer Review Meeting 1-11
1.3 Report Organization 1-11
2.0 Responses to Questions in Topic Area A: Hazard Characterization and
Mode of Action 2-1
2.1 Charge Question A.I 2-1
2.2 Charge Question A.2 2-4
2.3 Charge Question A.3 2-5
2.4 Charge Question A.4 2-7
3.0 Responses to Questions in Topic Area B: Human Health Effects Data 3-1
3.1 Charge Questions B.I and B.2 3-1
3.2 Charge Question B.3 3-3
3.3 Charge Question B.4 3-7
3.4 Charge Question B.5 3-8
4.0 Responses to Questions in Topic Area C: Laboratory Animal Studies 4-1
4.1 Comments on Developmental Toxicity 4-1
4.2 Comments on Reproductive Toxicity 4-3
4.3 Comments on Endocrine and Neuroendocrine Toxicity 4-5
4.4 Comments on Thyroid Pathology (Including Cancer Effects) 4-7
4.5 Comments on Neurotoxicity 4-10
4.5.1 Comments on Studies of Brain Morphometry 4-10
4.5.2 Comments on Studies of Motor Activity 4-14
4.6 Comments on Immunotoxicity 4-15
in
-------
Table of Contents (Continued)
5.0 Responses to Questions in Topic Area D: Ecological Risk Assessment and
Evidence for Indirect Exposure 5-1
5.1 Charge Questions D.I and D.2 5-1
5.2 Charge Question D.3 5-3
5.3 Charge Question D.4 5-4
5.4 Charge Question D.5 5-4
5.5 Charge Question D.6 5-6
5.6 Charge Question D.7 5-7
6.0 Responses to Questions in Topic Area E: Use of PBPK Modeling 6-1
7.0 Responses to Questions in Topic Area F: Human Health Dose-Response
Assessment 7-1
7.1 Charge Question F.I 7-1
7.1.1 Consistency Between Observed Effects and Mode of Action 7-2
7.1.2 Comments on the Use of Brain Morphometry Effects as the
Bases for the Point of Departure 7-4
7.1.3 Comments on Use of Data Other Than Brain Morphometry for the
Point of Departure 7-7
7.2 Charge Question F.2 7-10
7.3 Charge Question F.3 7-11
7.4 Charge Question F.4 7-15
8.0 Responses to Questions in Topic Area G: Risk Characterization 8-1
8.1 Comments on the Human Health Risk Characterization 8-1
8.2 Comments on the Ecological Risk Characterization 8-2
9.0 Responses to Questions in Topic Area H: General Comments, Conclusions,
and Recommendations 9-1
10.0 References 10-1
IV
-------
Table of Contents (Continued)
Appendices
Appendix A
Appendix B
Appendix C
Appendix D
Appendix E
Appendix F
Appendix G
Appendix H
Appendix I
Appendix J
Appendix K
List of Selected Studies Performed Since the 1999 Peer Review
List of Expert Peer Reviewers
Premeeting Comments, Alphabetized by Author, and Charge to the Reviewers
Index of Public Comments Submitted before the Peer Review Meeting
Index of Public Comments Submitted after the Peer Review Meeting
List of Registered Observers of the Peer Review Meeting
Agenda for the Peer Review Meeting
Public Comments Provided Orally during the Peer Review Meeting
Copies of EPA's Opening Presentation Materials
Post-Meeting Comments Submitted by Peer Reviewers
Index of Written Materials Submitted by Observers at the Peer Review Meeting
-------
List of Abbreviations
AFRL Air Force Research Laboratory
ANOVA analysis of variance
AUCB area-under-the-curve in blood
AUCT area-under-the-curve in thyroid
BCF bioconcentration factor
EPA U.S. Environmental Protection Agency
Revised ERD revised external review draft (i.e., the review document)
ERG Eastern Research Group, Inc.
HPT axis hypothalamic-pituitary-thyroid axis
K,,, Michaelis-Menten affinity constant
IRIS Integrated Risk Information System
LLNA local lymph node assay
LOAEL lowest-observed-adverse-effect level
NCEA National Center for Environmental Assessment
NIEHS National Institute of Environmental Health Sciences
NIOSH National Institute for Occupational Safety and Health
NIS sodium (Na+)-iodide (I") symporter
NOAEL no-observed-adverse-effect level
PBPK physiologically based pharmacokinetic
PWG Pathology Working Group
RAIU radioactive iodide uptake
RIA radioimmunoassay
RED reference dose
STP Society of Toxicologic Pathologists
T3 triiodothyronine
T4 thyroxine or tetraiodothyronine
TBG thyroxine-binding globulin
TPO thyroid peroxidase
TSH thyroid-stimulating hormone
USAF U.S. Air Force
Vmaxc Michaelis-Menten maximum velocity capacity
WPAFB Wright Patterson Air Force Base
VI
-------
Executive Summary
This report summarizes a peer review of the U.S. Environmental Protection Agency's
external review draft document "Perchlorate Environmental Contamination: Toxicological Review
and Risk Characterization" (the Revised External Review Draft, or Revised ERD), and of
associated studies published since 1999 that have not been published in externally-reviewed
scientific literature. During the 2-day peer review meeting, 17 independent experts from a broad
range of relevant scientific backgrounds and affiliations thoroughly discussed and evaluated the
scientific analyses presented in the Revised ERD. The reviewers had favorable feedback on many
issues, such as the proposed harmonized approach for evaluating cancer and noncancer endpoints,
and constructive feedback on others. Reviewers expressed a diversity of opinions on several
critical issues, including the role of human health data in the Revised ERD, the use of reported
changes in rat brain morphometry as a point of departure, and the application of uncertainty
factors.
The peer reviewers' main findings on the Revised ERD are summarized below, organized
into topic areas covered during the peer review meeting. The remainder of this report documents
the extensive discussions that led up to these main findings presented below, as well as
deliberations on additional topics not noted in this Executive Summary.
Topic Area A: Hazard Characterization and Mode of Action (see Section 2). The
peer reviewers generally supported the proposed key event, mode of action, harmonized
approach for characterizing cancer and noncancer toxicity, and approach for low-dose
extrapolations. Some reviewers, however, questioned assumptions EPA made regarding
perchlorate not being metabolized and being actively translocated into thyroid cells.
Topic Area B: Human Health Effects Data (see Section 3). Several reviewers
recommended that EPA consider deriving a reference dose using data from the human
health effects studies, particularly those from a recent clinical study (the "Greer study").
On the other hand, some reviewers cautioned against using these studies, given their lack
of control for confounding factors, limited exposure duration, consideration of only healthy
adults, and focus on a narrow set of toxicologic endpoints.
vn
-------
Topic Area C: Laboratory Animal Studies (see Section 4). The peer reviewers'
comments on laboratory animal studies were made for the individual toxic endpoints:
- The reviewers concluded that the two developmental toxicity studies completed
since the 1999 peer review were scientifically sound and that both studies indicate
that developmental toxicity appears to occur at doses orders of magnitude higher
than those causing effects at other endpoints.
- The reviewer found the two-generation reproductive toxicity study in rats
conducted since the 1999 peer review to be thorough and well conducted and
EPA's interpretations of these studies generally adequate. They recommended that
EPA further investigate apparent dose-dependent decreases in sperm density and
daily sperm production levels.
- The reviewers with expertise in endocrinology noted that most laboratory animal
studies of thyroid hormone levels have detected effects, although not consistently
within and across studies—an outcome attributed primarily to limitations of the
measurement techniques (i.e., decrements in certain hormone levels cannot be
quantified reliably when baseline hormone levels are near the lowest range of the
diagnostic kits' standard curves). The reviewers had different opinions on the
most appropriate statistical approach for analyzing these data: one reviewer
supported EPA's use of analysis of variance, while another advocated another
advocated testing hypotheses with a complete pharmacokinetic and
pharmacodynamic model.
- The reviewer assigned to evaluate thyroid histopathology indicated that
administrating low doses of perchlorate to rats and rabbits produced adaptive
changes in thyroid histopathology: colloid depletion and epithelial hypertrophy.
Higher doses (at least 1.0 mg/kg/day) are needed to produce hyperplasia, which
presumably resulted from upregulation of thyroid-stimulating hormone (TSH).
While dosage at considerably higher levels (30 mg/kg/day) produced follicular cell
neoplasms in rats, several reviewers doubted humans would develop this cancer
from environmental exposure to perchlorate at the currently reported levels.
The reviewers commented on two different types of studies evaluating
neurotoxicity. First, the reviewers indicated that the two studies of motor activity
in rats were conducted using rigorous methodologies. Moreover, they concluded
that EPA's interpretations of these studies were appropriate and defensible,
including EPA's identification of dose-related motor activity effects in the most
recent study (Bekkedal et al. 2000).
Vlll
-------
Second, the reviewers had different opinions on the studies examining brain
morphometry changes in rats. Most reviewers agreed that use of linear
measurements to characterize brain dimensions is subject to artifacts. They had
different perspectives, however, on how EPA should interpret the data given the
limitations. Some reviewers argued that linear measurements of brain dimensions
in rat pups are not reliable indications of brain morphometry changes. Other
reviewers, however, believed that errors introduced by using linear measurements
would be randomly distributed across dosage groups; this would most likely make
it impossible to detect statistically significant effects, not to detect effects that do
not exist. Overall, given the weaknesses in the study methodology and other
concerns described later in this report, some reviewers felt that EPA should
consider the brain morphometry data inconclusive. Other reviewers, on the other
hand, did not support disregarding these data, especially considering that two
studies and several re-analyses of them have all identified brain morphometric
changes in consistent regions of the brain.
- The reviewer who addressed immunotoxicity focused primarily on the two studies
that were performed since the 1999 peer review, including a study of contact
hypersensitivity conducted in response to a recommendation of that peer review.
This reviewer noted that both studies followed standard protocols and used
validated assays to evaluate both the innate and acquired immune responses,
considering most compartments of the immune system. Though this reviewer
generally supported EPA's interpretations of these studies, he questioned the
relevance to humans of the contact hypersensitivity findings, which exhibited no
clear dose-response. This reviewer concluded (and several other reviewers agreed)
that the immunotoxicity studies should not be used as a point of departure for the
reference dose determination and do not provide an adequate basis for applying an
uncertainty factor of 3 to account for database insufficiencies.
- No new genotoxicity studies have been published since the 1999 peer review. The
reviewers supported the findings of the previous peer review panel (that
perchlorate is not genotoxic) and EPA's relevant dose-response interpretations
(that cancer endpoints can be evaluated using a nonlinear dose-response model).
Topic Area D: Ecological Risk Assessment and Evidence for Indirect Exposure (see
Section 5). The peer reviewers generally supported EPA's assimilation and interpretation
of exposure and effects data that were available at the time the Revised ERD was prepared.
The reviewers were concerned, however, by a study published after the Revised ERD was
released that suggests amphibians may be experiencing toxic effects at perchlorate
exposures considerably lower than those EPA previously predicted (Goleman et al. 2002).
Though they identified potential limitations of the recent study, they concluded that its
implications suggest that the current screening-level ecological risk assessment is not
IX
-------
adequate. The reviewers identified several issues that must be evaluated further if
environmental exposure and ecological risk are to be assessed more thoroughly.
Topic Area E: Use of Physiologically Based Pharmacokinetic (PBPK) Modeling (see
Section 6). The peer reviewers found the structure, basic equations, and physiological
parameters in the PBPK models to be generally adequate, though sometimes not
documented in sufficient detail in the Revised ERD. They recommended that the PBPK
models should include more refined descriptions of passive iodide uptake and active
perchlorate uptake and the kinetic representation of these processes. The reviewers had
different opinions on the proposed dose metric. Some concluded that use of area-under-
the-curve perchlorate in serum is the most defensible dose metric and is suitable for
purposes of interspecies extrapolation. The other reviewer, however, advocated the
development of a biologically-based dose response model that would link
pharmacodynamic changes in the thyroid hormones with internal perchlorate dose and
iodide uptake inhibition.
Topic Area F: Human Health Dose-Response Assessment (see Section 7). The
reviewers comments on the human health dose-response assessment primarily addressed
point of departure and the use of uncertainty factors. Consistent with their differing
reviews of the brain morphometry study (summarized above), the peer reviewers had
differing opinions on whether EPA should use the brain morphometry changes as a point of
departure: some reviewers supported using the brain morphometry data in the reference
dose derivation, while roughly an equal number of reviewers did not. Several reviewers,
however, indicated that EPA may be able to justify using the brain morphometry data as a
point of departure if they can be re-scored blindly and the effects still observed.
The reviewers discussed other options for selecting a point of departure. Some suggested
using data from human clinical studies, but others expressed concern about the limitations
of these data sets. Though they acknowledged that EPA could derive a point of departure
based on changes in thyroid hormone levels and iodide uptake inhibition in laboratory
animals, several reviewers questioned whether such effects are adaptive or truly adverse.
One reviewer noted that thyroid histopathology can be defended as a point of departure,
but, regarding neoplastic sequelae, he recommended that EPA only consider hyperplasia as
an adverse effect, with colloid depletion and hypertrophy being adaptive effects.
Regarding uncertainly factors, most reviewers accepted the factors of 10 applied for
intraspecies variability and extrapolating a lowest-observed-adverse-effect level to a no-
observed-adverse-effect level. Nearly every reviewer, however, was against applying an
uncertainty factor of 3 to account for database insufficiencies in immunotoxicity. Many
reviewers supported the use of an uncertainty factor of 3 to account for the limited
exposure duration of the laboratory animal studies, but some found this factor unnecessary.
During this discussion, several reviewers proposed alternate statistical and modeling
approaches to replace EPA's general practice of using discrete uncertainty factors.
-------
Topic Area G: Risk Characterization (see Section 8). The reviewers recommended
that EPA revise the risk characterization to reflect any changes made when addressing the
issues mentioned above. Moreover, they recommended that the risk characterization give
greater context for the proposed reference dose and potential health risks, perhaps by
describing public health consequences of exposure and by acknowledging the uncertainties
associated with the reference dose derivation.
XI
-------
1.0 Introduction
This report summarizes a peer review by 17 experts of documents that the U.S.
Environmental Protection Agency (EPA) prepared or evaluated when assessing human health and
ecological risks associated with exposure to perchlorate. These documents are:
• The January 2002 external review draft of "Perchlorate Environmental Contamination:
Toxicological Review and Risk Characterization" (EPA 2002). (Throughout this report,
this document is referred to as the Revised External Review Draft, or Revised ERD.)
• Relevant studies performed since the 1999 peer review of EPA's perchlorate assessment,
but provided as contractor reports or as preliminary findings (e.g., as an abstract or letter
to the editor) and not yet vetted in the peer-reviewed literature at the time the current peer
review was scheduled.
Eastern Research Group, Inc. (ERG), organized and implemented the peer review under a
contract to EPA. The peer review took place in a meeting open to the public on March 5 and 6,
2002, in Sacramento, California. This introductory section provides background information on
EPA's perchlorate assessment (Section 1.1), the scope of this peer review (Section 1.2), and the
organization of this report (Section 1.3).
1.1 Background
Perchlorate (C1O4") is an anion that contaminates groundwater and surface waters, where it
originates from dissolution of ammonium, potassium, magnesium, or sodium salts. Perchlorate is
exceedingly mobile in aqueous systems and can persist for many decades under typical
groundwater and surface water conditions. A major source of perchlorate contamination is the
manufacture of ammonium perchlorate for use as the oxidizer component and primary ingredient
in solid propellant for rockets, missiles, and fireworks.
EPA issued a provisional toxicity assessment for perchlorate in 1992 and a revised
provisional assessment in 1995, based on the effects of potassium perchlorate in patients with
1-1
-------
Graves' disease (an autoimmune disease that results in hyperthyroidism). In March 1997, an
independent non-EPA external peer review panel determined that the existing toxicologic
database on perchlorate was inadequate for quantitative human health risk assessment. In May
1997, a perchlorate testing strategy was developed. This strategy initiated an accelerated research
program to inform future human health and ecological risk assessment studies.
In December 1998, EPA developed a draft external peer review version (EPA 1998) of a
document that assessed the most current information at the time on perchlorate toxicity. This
document included a human health risk assessment, which incorporated results from health effects
studies available as of November 1998, and a screening-level ecological assessment. The human
health risk assessment presented a model (motivated by the mode of action) that harmonized
noncancer and cancer evaluations to derive a single oral risk benchmark. This benchmark was
based on precursor effects for both altered neurodevelopment and thyroid neoplasia.
In February 1999, an EPA contractor held an external peer review meeting to evaluate
EPA's 1998 draft perchlorate assessment. The review panel endorsed the conceptual approach
presented in the draft assessment, but recommended that new analyses be conducted and that new
studies be planned and performed. After the 1999 external peer review, EPA prepared the
Revised ERD of perchlorate toxicity (EPA 2002), which incorporates data from the studies that
the previous peer review panel recommended. Both the supporting data from these studies and
the Revised ERD are the subject of the current external peer review.
To evaluate whether the assumptions, methods, and conclusions of the Revised ERD are
based on sound scientific principles, EPA decided, as per policy, to obtain an independent, expert
peer review not only of the Revised ERD but also of the relevant studies performed since the
1999 peer review that are not documented in the peer-reviewed literature. Appendix A lists the
studies that the reviewers evaluated during the current peer review. EPA hired ERG to
implement the current peer review.
1-2
-------
1.2 Scope of the Peer Review
ERG managed every aspect of the peer review, including selecting reviewers (see Section
1.2.1); coordinating selected activities prior to, during, and after the peer review meeting (see
Sections 1.2.2, 1.2.3, and 1.2.4, respectively); and preparing this summary report, which describes
the scope and findings of the peer review. The following subsections describe what each of these
tasks entailed.
1.2.1 Selecting the Reviewers
ERG followed its long-standing procedures for conducting expert peer reviews to select 17
highly qualified and independent reviewers. The initial step was to establish reviewer selection
criteria. The specific criteria for this peer review follow:
Reviewers must be senior scientists or researchers with broad experience and expertise (as
demonstrated by peer-reviewed publications, awards, and service to relevant professional
societies) in the following fields: pharmacokinetics, endocrinology, neurotoxicology,
epidemiology, statistics, immunotoxicology, thyroid pathology, developmental toxicology,
reproductive toxicology, genetic toxicology, ecotoxicology, and environmental transport
and biotransformation.
• Some reviewers should have working knowledge of EPA's risk assessment guidelines and
methodologies, as well as familiarity with the content, format, and objectives of recent
health assessments included in the Integrated Risk Information System (IRIS) database.
• Reviewers must be available to critique the review materials and present their comments at
the peer review meeting.
• Reviewers must have no conflicts of interest in performing the peer review.
To implement the fourth selection criterion, ERG distributed a conflict-of-interest
screening form to all candidate reviewers. ERG used the self-reported responses on the form to
eliminate from consideration any candidates who have real or perceived conflicts of interest. For
instance, ERG did not consider any candidates who have a vested interest, financial or otherwise,
1-3
-------
in the outcome of the peer review or those who have conflicts of interest with EPA on pending
scientific issues pertaining to this review. Further, ERG did not consider candidates who
prepared or edited any section of the Revised ERD or other federal documents related to
perchlorate. Finally, ERG did not consider candidates who have worked on Superfund sites at
which perchlorate is a contaminant of concern, who have worked for potentially responsible
parties for such sites, or who have worked for companies that are members of the Perchlorate
Study Group.1
After establishing the selection criteria, ERG began to identify a large pool of highly
experienced candidates. To identify qualified candidates, ERG conducted literature reviews to
identify widely published researchers, contacted reviewers from the 1999 external peer review,
and performed various other searches for experts in relevant disciplines. Overall, ERG contacted
more than 250 candidate peer reviewers. ERG carefully reviewed the expertise and credentials of
these candidates and selected the 17 most qualified individuals.
Appendix B lists the names and affiliations of the 17 peer reviewers, and Appendix C
includes brief biographies that summarize the reviewers' areas of expertise. Recognizing that few
individuals truly specialize in every technical area specified by the first reviewer selection
criterion, ERG ensured that the collective expertise of the selected peer reviewers covers the
required technical areas (i.e., at least one reviewer has expertise in immunotoxicity, at least one
reviewer has expertise in neurotoxicity, at least one reviewer has expertise in reproductive
toxicology, and so on). Moreover, ERG selected peer reviewers with a broad range of affiliations
(e.g., academia, consulting, industry, other federal agencies), in hope that the expert panel would
offer a balanced perspective on the scheduled discussion topics. ERG instructed the reviewers to
remain independent throughout the peer review process, and therefore refrain from discussing the
scientific merit of the Revised ERD with any of the identified stakeholders. ERG had copies of
1 During the opening conflict-of-interest disclosures, one reviewer (Dr. Gary Williams) indicated that he
had worked for Kerr-McGee (a member of the Perchlorate Study Group), but that he had learned that Kerr McGee
has an interest in the Revised ERD only upon seeing that the company submitted public comments on EPA's
document.
1-4
-------
the peer reviewers' resumes and conflict-of-interest disclosure forms on display at the peer review
meeting.
1.2.2 Activities Prior to the External Peer Review Meeting
ERG took several steps to ensure that the peer reviewers had the information necessary to
conduct thorough, informed, and unbiased reviews of the Revised ERD. The specific activities
that ERG conducted prior to the peer review meeting follow:
Prepare a charge to the reviewers. ERG first worked with EPA to prepare written
guidelines (commonly called a "charge") for the technical review. Specifically, EPA
identified technical issues that the charge should address, and ERG incorporated these
issues into 30 charge questions, organized into 8 topic areas. The charge included a
question that asked the peer reviewers to comment on any topics not explicitly addressed
by the other questions. Copies of the charge were available prior to the meeting, upon
request, and at the peer review meeting; a copy is included in this report as part of
Appendix C.
In the charge, ERG assigned different responsibilities to the individual reviewers. Every
reviewer was asked to read the entire Revised ERD, focusing on specific sections relevant
to their areas of expertise. In the charge, ERG required almost every peer reviewer to
evaluate some of the studies that were conducted since the 1999 peer review meeting and
that were not published in the peer-reviewed literature at the time the March 2002 meeting
was planned. ERG ensured that at least one expert peer reviewer critically evaluated every
study listed in Appendix A. Finally, the charge identified the peer reviewers who would
lead discussions on the eight topic areas during the peer review meeting.
Distribute review documents and communicate reviewer assignments. On January 23,
2002, ERG sent every peer reviewer a package of review materials. These packages
included the charge, the Revised ERD, copies of the studies completed since the 1999 peer
review (see Appendix A), and logistical information regarding the peer review. Copies of
these documents were made available to observers at the peer review meeting. ERG held
several conference calls with the reviewers to confirm the shipment of the review materials
and to answer any questions about the peer reviewers' assignments. ERG facilitated a
conference call with the peer reviewers prior to the workshop to ensure that the reviewers
understood their assignments. During this call, ERG informed the reviewers of the
procedures they should follow to ask EPA or the various study authors questions of
clarification prior to the external peer review meeting (see next bulleted item).
1-5
-------
Facilitate questions of clarification. When conducting their reviews, some peer reviewers
informed ERG that they had questions of clarification for the authors of the recent studies
and of the Revised ERD. ERG forwarded these questions to the appropriate individuals
and then forwarded the responses to the peer reviewers. To maintain the independence of
the peer review, ERG asked that the peer reviewers refrain from contacting the study
authors or representatives from EPA directly with any such questions.
Obtain and compile the reviewers' premeeting comments. In the weeks after the peer
reviewers received the charge, ERG asked the reviewers to prepare their initial evaluations
of the Revised ERD and the studies listed in Appendix A. ERG compiled these written
premeeting comments, distributed them to the reviewers, and made copies available to
observers during the peer review meeting. These initial comments are included in this
report, without modification, as Appendix C. It should be noted that the premeeting
comments are preliminary in nature. Some reviewers' technical findings changed after the
premeeting comments were submitted. Therefore, the comments in Appendix C should not
be considered the reviewers' final opinions.
Distribute public comments for the reviewers' consideration. After receiving the
reviewers' premeeting comments, ERG sent the peer reviewers two sets of copies of public
comments. First, before the peer reviewers departed for the meeting, ERG sent them
packages with copies of all comments that were received by February 19, 2002. Second, at
the peer review meeting, ERG handed the peer reviewers copies of all public comments
that were received between February 20, 2002, and March 5, 2002. ERG distributed the
entire set of public comments prior to the meeting, so that the peer reviewers could factor
any issues raised into the meeting deliberations. Appendix D includes an index of the
public comments that ERG sent to the reviewers prior to the peer review meeting. Copies
of these comments were made available to observers at the meeting.
1.2.3 Activities at the External Peer Review Meeting
The 17 peer reviewers2 and approximately 150 observers attended the peer review meeting,
which was held at the Holiday Inn hotel in Sacramento, California, on March 5 and 6, 2002. The
peer review meeting was open to the public, and the meeting dates and times were announced in
the Federal Register. Appendix F lists the observers who confirmed their attendance at the
2 Dr. Michael Kohn could not attend the peer review meeting in person, but participated in the discussions
relevant to his area of expertise via conference call. Dr. Michael Aschner attended the first day of the peer review
meeting, but could not attend the second.
1-6
-------
meeting registration desk. The schedule of the peer review meeting generally followed the
agenda, presented here as Appendix G.
The meeting began with introductory comments by the meeting's facilitator (Jan Connery
of ERG) and three representatives from EPA. All of these comments are summarized below.
Before beginning their deliberations, the peer reviewers were asked to introduce themselves by
stating their names, affiliations, areas of expertise, and any potential conflicts of interest they had.
(Table 1, at the end of this section, summarizes the reviewers' specific remarks regarding conflicts
of interest.) For the remainder of the meeting, the peer reviewers provided many comments,
observations, and recommendations when answering the charge questions. ERG ensured that
peer reviewers presented their own opinions on technical topics; no efforts were made to reach
consensus on any issue. The meeting included three designated observer comment periods, when
observers were allowed to offer verbal comments. Appendix H documents all observer comments
presented at the peer review meeting and includes copies of written handouts that were
distributed by those who spoke.
The remainder of this section summarizes the opening remarks provided by ERG and the
three EPA representatives. Copies of Dr. Herman Gibb's and Ms. Annie Jarabek's presentation
materials are included in Appendix I of this report.
Jan Connery, ERG, meeting facilitator. In her opening remarks, Ms. Connery welcomed
the reviewers and observers to the meeting, stated the purpose of the peer review, and
identified the document under review. Later in the meeting, Ms. Connery explained the
procedure observers should follow to make comments, both orally at the meeting and in
writing to EPA. Ms. Connery noted that EPA had extended the public comment period
through April 5, 2002, and that ERG would mail the peer reviewers copies of all public
comments submitted by that date. She also summarized several key aspects of the peer
review, including some activities that took place prior to the peer review meeting (see
Section 1.2.2) and planned future activities (see Section 1.2.4).
1-7
-------
Jane Diamond, Acting Director ofSuperfund in EPA Region IX. Ms. Diamond's opening
remarks addressed the challenges that EPA faces when examining sites with perchlorate
contamination. Ms. Diamond said roughly 10% of Superfund sites in Region IX have
perchlorate contamination, as do several other sites not on the National Priorities List. For
these sites, she noted that the emerging science on perchlorate toxicity has affected how
the Region determines action levels and makes remedial decisions. Given that the states
within Region IX are developing their own "action levels" for perchlorate, and that none of
the action levels are consistent, Ms. Diamond said she looks forward to EPA establishing
an official reference dose (RfD). Finally, Ms. Diamond acknowledged the efforts of
numerous parties (e.g., EPA's Office of Research and Development, the Department of
Defense, other researchers, consultants, and community activists) for their ongoing work
on evaluating perchlorate toxicity.
Dr. Herman Gibb, Acting Associate Director for Health at EPA 's National Center for
Environmental Assessment (NCEA). Dr. Gibb's presentation provided context on EPA's
ongoing activities addressing levels of perchlorate environmental contamination. Dr. Gibb
acknowledged that recent advances in analytical methods have allowed scientists and
regulators to characterize perchlorate environmental contamination at much lower
concentrations than could be achieved with other methods. These advances helped reveal
that perchlorate contamination is more widespread than previously thought—an
observation that has heightened concern for potential toxicity of perchlorate exposures.
Dr. Gibb noted that EPA has since sought more detailed information on the occurrence of
this contamination, fate and transport of perchlorate in various media, the potential for
both direct and indirect exposures, adverse effects on human health and ecosystems, and
effective treatment technologies. Dr. Gibb indicated that many advances have been made
in these fields, due largely to contributions from multiple parties, including the Department
of Defense, the Perchlorate Study Group (a consortium of defense contractors), multiple
EPA Offices and Regions, private researchers, and numerous local, state, federal, and tribal
agencies.
Focusing on the current peer review, Dr. Gibb explained that the Revised ERD not only
accounts for comments raised and recommendations made during the 1999 peer review,
but also incorporates data from studies conducted since that time. He noted that the goal
of the current peer review is to critique the scientific studies and approaches in the Revised
ERD and the supporting documents, not to address potential regulatory actions. Dr. Gibb
added that the external peer review is one of many steps that EPA takes in reviewing and
finalizing documents for IRIS. Dr. Gibb mentioned other steps in this process (e.g.,
disposition of comments, internal EPA consensus review) and presented a proposed
schedule for these and other future activities in the perchlorate assessment.
Ms. Annie Jarabek, Chemical Manager for the Revised ERD, EPA NCEA. Ms. Jarabek's
presentation reviewed EPA's scientific analyses of perchlorate toxicity and highlighted
notable milestones from the 1999 peer review through release of the Revised ERD. Ms.
1-8
-------
Jarabek first summarized the content of the previous external review draft (EPA 1998).
She presented the proposed mode of action, which linked exposure to the key event of
iodide uptake inhibition with early biological effects (e.g., decrements in thyroid hormones)
that ultimately lead to clinical disease. In the previous draft, EPA's dose-response analyses
were based largely on thyroid histopathology findings observed in rat pups, which EPA
viewed as biomarkers for adverse hormonal changes believed to occur in utero. Ms.
Jarabek also indicated that the previous draft included a screening-level ecological risk
assessment.
Summarizing the findings of the 1999 peer review, Ms. Jarabek noted that the panelists
endorsed EPA's proposed mode of action and conceptual model, shared the Agency's
concern about potential neurodevelopmental effects, and made several recommendations to
conduct new toxicity studies and to perform additional statistical analyses of the existing
data. Ms. Jarabek identified several of the peer reviewers' recommendations. For
instance, the 1999 peer review panel recommended that EPA convene a Pathology
Working Group (PWG) to review thyroid tissue slides and histopathology diagnoses, that
additional laboratory animal studies be performed or repeated, and that EPA and other
researchers conduct additional studies to characterize the environmental fate and transport
of perchlorate. Ms. Jarabek then reviewed the many research projects completed since
1999, many of which were done specifically in response to the peer reviewers'
recommendations.
Focusing on the Revised ERD, Ms. Jarabek briefly summarized EPA's analyses of human
health effects data, laboratory animal studies, ecotoxicological studies, and human health
dose-response. Specifically, when addressing human studies, Ms. Jarabek identified the
new studies conducted since the 1999 peer review, clarified EPA's policy on considering
third-party human dosing data in toxicity assessments, and emphasized that the Revised
ERD considers all third-party human data that were available to the agency as of December
14, 2001. Similarly, Ms. Jarabek identified the various laboratory animal studies and
ecotoxicological studies conducted since the 1999 peer review meeting and presented
selected EPA interpretations of the findings. She emphasized that the Revised ERD
presents a screening-level ecological risk assessment.
Ms. Jarabek then reviewed how EPA derived its proposed RfD for perchlorate from the
available studies. After reviewing the effect levels EPA and study authors calculated for
various endpoints, Ms. Jarabek summarized the Agency's weight-of-evidence approach
used to identify a point of departure. The proposed value was a lowest-observed-adverse-
effect level (LOAEL) of 0.01 mg/kg/day, based on effects (e.g., perturbation in thyroid
and pituitary hormones, thyroid histopathology, and brain morphometry effects) observed
in multiple laboratory animal studies at various life stages. Ms. Jarabek highlighted key
features of the four physiologically based pharmacokinetic (PBPK) models that EPA used
to extrapolate the dose-response data observed in the animal studies to humans. Finally,
she identified the composite uncertainty factor (300) that EPA proposed for developing an
1-9
-------
RfD and described the individual factors that constitute this value. Briefly, she noted that
EPA proposes the following:
- An uncertainty factor of 3 for intrahuman variability.
- No uncertainly factor for interspecies extrapolations.
- An uncertainty factor of 10 to extrapolate the LOAEL to a no-observed-adverse-
effect level (NOAEL).
An uncertainty factor of 3 to account for subchronic data extrapolation and
concern for in utero programming.
- An uncertainty factor of 3 to account for database insufficiencies, particularly those
associated with immunotoxic effects.
Based on these factors and the selected point of departure, the Revised ERD proposed an
RfD of 0.00003 mg/kg/day. Ms. Jarabek presented two comparative risk calculations, one
based on the use of the human clinical data from the Greer study and the other based on
thyroid tumors observed in the two-generation study. Both of these alternative derivations
resulted in RfD estimates within the range of the proposed value. Ms. Jarabek also
presented hypothetical calculations of Drinking Water Equivalent Levels (DWELs) that
might result from the proposed RfD, though she emphasized that the proposed RfD is
based entirely on scientific analyses of the available toxicity studies and is not a regulatory
standard.
Following these opening presentations, Dr. Ron Wyzga, a peer reviewer and the designated
chair of the meeting, opened the technical discussions among the peer reviewers. Dr. Wyzga
explained that he would ask the designated discussion leaders to facilitate the reviewers'
discussions for the individual topic areas. The discussion leaders drew from the reviewers'
premeeting comments to initiate discussions. ERG notes that the discussions at the meeting (and
not the premeeting comments) should be viewed as the reviewers' final opinions on the Revised
ERD. The technical discussions among the peer reviewers focused almost entirely on answering
the 30 charge questions. The only instances in which individuals other than ERG or the peer
reviewers spoke were when the reviewers asked EPA questions of clarification, which were
facilitated by either Ms. Connery or Dr. Wyzga, and when the observers gave comments during
the designated observer comment period, as documented in Appendix H.
1-10
-------
1.2.4 Activities Following the External Peer Review Meeting
Following the peer review meeting, ERG's involvement in the peer review was limited to
two activities. First, ERG distributed written public comments received after the peer review
meeting. Specifically, ERG mailed to the peer reviewers copies of all public comments that EPA
received between March 6, 2002, and the close of the extended public comment period (April 5,
2002). Appendix E includes an index of the comments that ERG distributed during that time.
The peer reviewers were given the option to submit post-meeting comments based on the
information presented in these supplemental public comments. Appendix J presents all the post-
meeting comments that the peer reviewers submitted to ERG.
Second, an ERG technical writer who attended the peer review meeting prepared this
summary report. ERG distributed a draft of this report to the 17 peer reviewers and asked them
to verify that it accurately reflected the tone and content of the discussions at the peer review
meeting. After every peer reviewer submitted suggested revisions to the summary report or
indicated that the report was a faithful account of the peer review meeting, ERG submitted the
final peer review report (i.e., this report) to EPA.
1.3 Report Organization
The structure of this report follows the order of the reviewers' discussions during the peer
review meeting. For instance, Section 2 summarizes the reviewers' responses to the charge
questions in topic area A (hazard characterization and mode of action), Section 3 summarizes the
discussions on topic area B (human health effects data), and so on. Finally, Section 10 provides
references for all documents cited in the text. Throughout the main body of the report, the
reviewers' initials are used to attribute technical comments, suggestions, and recommendations to
the peer reviewers who made them. The following key lists the initials used:
1-11
-------
WA = Dr. William Adams, Ph.D.
MA = Dr. Michael Aschner, Ph.D.
NC = Dr. Nancy Carrasco, M.D.
MC = Dr. Michael Collins, Ph.D.
TC = Dr. Thomas Collins, Ph.D.
AC = Dr. Anthony Cox, Ph.D.
TF = Dr. Teresa Fan, Ph.D.
DH = Dr. David Hoel, Ph.D.
DJ = Dr. David Jacobson-Kram, Ph.D.
MK = Dr. Michael Kohn, Ph.D.
LK = Dr. Loren Roller, Ph.D.
KK = Dr. Kannan Krishnan, Ph.D.
MP = Dr. Merle Paule, Ph.D.
MR = Dr. Mehdi Razzaghi, Ph.D.
GW = Dr. Gary Williams, M.D.
RW = Dr. Ronald Wyzga, Ph.D.
TZ = Dr. Thomas Zoeller, Ph.D.
The appendices to this report include extensive background information on the peer review
meeting. This information includes items made available to all meeting attendees, as well as items
generated since the peer review meeting (e.g., a final list of attendees). Specifically, the following
information is included as appendices:
A list of studies performed since the 1999 peer review, but not documented in the peer-
reviewed literature at the time when the March 2002 peer review was organized (Appendix
A).
A list of the peer reviewers (Appendix B).
• Premeeting comments and the charge to the reviewers (Appendix C).
• An index of written public comments submitted before the peer review meeting (Appendix
D).
• An index of written public comments submitted after the peer review meeting (Appendix
E).
• A list of registered observers of the peer review meeting (Appendix F).
The agenda for the peer review meeting (Appendix G).
• Public comments given during the peer review meeting (Appendix H).
Copies of EPA's opening presentation materials (Appendix I).
Copies of post-meeting comments submitted by the peer reviewers (Appendix J).
1-12
-------
An index of written materials that observers handed to peer reviewers during the observer
comment period (Appendix K).
1-13
-------
Table 1
Conflict-of-interest Disclosures Made at the Peer Review Meeting
Peer Reviewer
Conflict-of-interest Disclosure
Dr. William Adams
Dr. Adams noted that a perchlorate groundwater contamination plume is in the
vicinity of, but not associated with releases from, his employer.
Dr. Michael Aschner
Dr. Aschner indicated that he has recently applied for a research grant with the
Department of Defense (a stakeholder in EPA's perchlorate work), but for a
research topic not relevant to perchlorate.
Dr. David Hoel
Dr. Hoel mentioned that he is a member of the EPA Science Advisory Board's
Environmental Health Committee. He added that he advises the Department of
Defense (a stakeholder in EPA's perchlorate work) on issues pertaining to
depleted uranium and nerve agents.
Dr. David Jacobson-
Kram
Dr. Jacobson-Kram noted that the laboratory where he works (BioReliance
Corporation) conducted one of the mutagenicity studies cited in the Revised
ERD.
Dr. Michael Kohn
Dr. Kohn mentioned that he has conducted past projects in which he modeled
changes in thyroid hormone levels. He noted that this past work was relevant to
perchlorate, but he did not specify whether it explicitly considered perchlorate.
Dr. Merle Paule
Dr. Paule indicated that he has recently been pursuing employment
opportunities with EPA.
Gary Williams, M.D.
Dr. Williams indicated that he currently serves as a medical expert in a lawsuit
in which Kerr McGee (a stakeholder in EPA's perchlorate work) is a defendant.
He noted that this work does not involve perchlorate and that he has not
actually met with Kerr McGee employees. Dr. Williams added that he learned
that Kerr McGee has an interest in the Revised ERD only upon seeing that the
company submitted public comments on EPA's document.
Dr. Thomas Zoeller
Dr. Zoeller noted that he serves on the Screening and Testing Workgroup of
EPA's Endocrine Disrupter Screening and Testing Advisory Committee. He
mentioned that he was a panelist for the 1999 peer review of EPA's perchlorate
risk assessment, after which multiple news organizations, other parties, and
various sectors contacted him for comment.
Notes: The ten peer reviewers not listed in this table all stated that they have no known conflicts of interest in
performing their review.
This table summarizes the peer reviewers' disclosures at the meeting, which may or may not present true
conflicts of interest. However, most of the reviewers listed above explicitly stated at the meeting that they
did not perceive their past or ongoing activities as conflicts of interest.
1-14
-------
2.0 Responses to Questions in Topic Area A: Hazard Characterization and
Mode of Action
This section summarizes the reviewers' comments relevant to hazard characterization and
mode of action for perchlorate toxicity. Dr. Thomas Zoeller, the designated discussion leader for
this topic area, facilitated the reviewers' responses to charge questions A. 1 through A.4. The
peer reviewers answered these questions by summarizing their premeeting comments and
discussing additional topics not raised in those comments. A general record of the peer
reviewers' discussions on these questions follows. Readers interested in the peer reviewers'
major findings on hazard characterization and mode of action should refer to the Executive
Summary of this report.
2.1 Charge Question A.I—Have all the relevant data on toxicokinetics and
toxicodynamics been identified and appropriately utilized? Have the similarities and
differences in the toxicity profile across species been adequately characterized?
The peer reviewers offered many comments when responding to this charge question.
Most discussion focused on the nature of the interaction between perchl orate and the sodium
(Na+)-iodide (I") symporter (NTS) and whether perchlorate is translocated into thyroid cells.
Most of the peer reviewers' comments suggested that the Revised ERD present more detailed
information on toxicokinetics and toxicodynamics. A summary of the specific comments,
organized by topic, follows:
Is perchlorate translocated into thyroid cells? One reviewer (NC) stated that perchlorate
is not translocated into thyroid cells, as the Revised ERD currently states. This reviewer
first critiqued a study cited in the Revised ERD that suggests such translocation occurs
(Anbar et al. 1959). In the study, rats and rabbits were dosed with radioactive perchlorate,
and researchers quantified perchlorate accumulation in the thyroid by measuring the
radioactivity released when incinerating the glands. Using this study design, the reviewer
argued, one cannot discern whether perchlorate translocated into thyroid cells or simply
bound to them.
2-1
-------
This reviewer then noted that the recent cloning of NTS has enabled laboratories to conduct
more rigorous research on iodide uptake inhibition. Specifically, this reviewer indicated
that her research group and another research group in Japan have completed studies
showing that perchlorate interacts with (and thus inhibits iodide uptake at) the NTS by
creating a conformational change, but does not translocate into the thyroid cells. She
recommended that EPA incorporate the findings from these studies into its discussion of
perchlorate toxicokinetics.
Given that perchlorate does not translocate into thyroid cells, the reviewers briefly
discussed what terminology most accurately describes perchlorate interaction with NTS.
One reviewer (KK), for instance, noted that a recent study (Eskandari et al. 1997) and
other studies cited in the premeeting comments (see the comments in Appendix C
submitted by Nancy Carrasco, M.D.) suggest that perchlorate is a "blocker" rather than a
"competitive inhibitor," the term EPA uses throughout the Revised ERD. Another
reviewer (NC) noted that both terms characterize perchlorate interaction with NTS. The
reviewers revisited this topic when discussing assumptions and kinetic constants considered
in the PBPK models (see Section 6). Regardless of the terminology EPA eventually
adopts, both reviewers recommended, the Agency should update the Revised ERD to
reflect current findings on whether perchlorate is translocated into thyroid cells.
Is MS inhibition reversible? Is perchlorate metabolized? When discussing the
implications of iodide inhibition at NTS, one reviewer (GW) asked the others if this
inhibition is reversible. One reviewer (NC) responded, noting that preliminary, unpublished
research in her laboratory has shown that some NTS inhibition may be irreversible. This
reviewer added that the potential irreversibility of NTS inhibition raises questions about the
extent to which perchlorate is metabolized. She indicated that EPA's assertion that
perchlorate is "excreted virtually unchanged after absorption" is based in part on a study
(Anbar et al. 1959) in which four humans were dosed with radioactive double-labeled
potassium perchlorate (K36C118O4). Noting that the subjects' urine contained 36C1O4" and
36C1", not only 36C118O4", this reviewer suggested the study implies that some ingested
perchlorate may, in fact, be metabolized. This reviewer noted that her research group has
hypothesized how perchlorate may be oxidized by molecular residues on the NTS molecule,
but she added that these hypotheses are the subject of ongoing studies.
NIS in other tissues. One reviewer (NC) noted that the Revised ERD provides limited
information on other tissues known to contain NIS (e.g., lactating mammary gland,
placenta, stomach, salivary glands, choroid plexus), and whether perchlorate exposure can
lead to adverse effects through iodide uptake inhibition in these tissues. She was
particularly concerned that maternal exposure to perchlorate may inhibit iodide transport
both across the placenta and into breast milk, which is the primary source of iodide for
fetuses and nursing neonates, respectively. Because iodide deficits in infants may decrease
thyroid hormone production, which, in turn, may affect the development of the nervous
2-2
-------
system, this reviewer thought the Revised ERD should provide much more information on
the potential for perchlorate to inhibit NTS in other tissues.
General comments on NIS inhibition and upregulation. The peer reviewers identified
additional areas where more detailed information on toxicokinetics is warranted. One
reviewer (NC), for example, recommended that the Revised ERD more prominently
acknowledge thatNIS's affinity for perchlorate is roughly an order of magnitude greater
than its affinity for iodide, as she demonstrated by comparing published values for the
Michaelis-Menten affinity constant (KjJ for iodide transport via the NIS to published
values for the inhibition constant for perchlorate. Further, this peer reviewer recommended
that the Revised ERD note that the NIS upregulation mechanism triggered by the
hypothalamic-pituitary-thyroid axis (HPT axis) regulates expression of NIS only in the
thyroid, not in the other tissues mentioned in the previous bulleted item.
Comments on toxicodynamics. Regarding toxicodynamics, the peer reviewers offered
several comments. Most of these identified specific topics in the Revised ERD that should
include more detailed information. Summarizing the peer reviewers' premeeting
comments, one reviewer (TZ) noted that the Revised ERD lacks detail both on exactly how
increases in thyroid-stimulating hormone (TSH) lead to cancer and on the similarity
between the specific sequence of events in rodents and humans. Elaborating on this topic,
another reviewer (AC) suggested that the Revised ERD could improve its discussion of
carcinogenesis by describing more steps in the sequence of events, such as how perchlorate
exposure leads to compensating proliferation or excess mitoses per unit time. This
reviewer acknowledged that the background discussions in the Revised ERD present some
detailed information on toxicodynamics, but he added that those details are not
quantitatively incorporated into the PBPK models.
When discussing carcinogenesis, one peer reviewer (GW) noted that none of the laboratory
studies being critiqued at the current meeting were designed to evaluate cancer as an
endpoint. He reminded the peer reviewers that much of the information in the Revised
ERD on carcinogenesis is based on observations of precursor lesions, which do not
necessarily result in cancer. The peer reviewers discussed this issue further when
addressing charge questions relevant to thyroid pathology (see Section 4.4). On a similar
note, another reviewer (AC) suggested that EPA apply the term "precursor" carefully: it
conventionally refers to an event that lies on the causal pathway of an adverse effect. This
reviewer recommended that EPA list, possibly in a table, all events that are considered
precursors and the specific adverse effects that ensue.
Other comments. When summarizing the reviewers' pre-meeting comments, the discussion
leader (TZ) identified other sections of the Revised ERD that EPA could clarify, but these
comments were not discussed in detail. First, he suggested that EPA provide additional
detail on the re-analysis of the radioimmunoassay (RIA) data. The Revised ERD, he noted,
should document how analyses were conducted, what standard curves were used, and
2-3
-------
whether data points were interpolated from the standard curves or extrapolated to levels
below the lowest available standard. Second, he noted that the Revised ERD lacks detail
on the levels of iodide in maternal serum, how perchlorate may affect these levels, and the
potential consequences to fetuses and neonates. Finally, he briefly mentioned the potential
for re-programming of the HPT axis following perchlorate exposure and suggested that
some current research on the hypothalamic-pituitary-adrenal axis might offer perspective
on the issue. Another reviewer (NC) indicated that iodide uptake is an electrogenic
process, not one that is driven by adenosine triphosphate (ATP), as the Revised ERD
states.
2.2 Charge Question A.2—The EPA has framed a conceptual model based on the key
event for the mode of action of perchlorate as inhibition of iodide uptake at NIS. Are
the roles and relative importance of the key event and subsequent neurodevelopmental
and neoplastic sequelae clearly articulated and consistent with the available data on
anti-thyroid agents or conditions and with the physicochemical and biological
properties of perchlorate?
Summarizing premeeting comments submitted by several peer reviewers, the discussion
leader (TZ) indicated that the Revised ERD, while it clearly articulates the mode of action and
identifies the subsequent neurodevelopmental and neoplastic sequelae, does not characterize the
mechanisms by which inhibition of iodide uptake (and subsequent decrements in thyroid hormones
and increases in TSH) eventually alter neurodevelopmental and neoplastic processes. As a
suggested improvement, this reviewer recommended that the Revised ERD discuss more
thoroughly the role of different thyroid hormones (e.g., T3 and T4) on other biological processes.
More detailed information on the various mechanisms, he noted, might enable EPA to relate
adverse effects observed in laboratory animals to those anticipated to occur in humans more
effectively.
The peer reviewers also discussed differential sensitivity to perchlorate exposure across
species, particularly between rodents and humans. Noting that rodents are much more susceptible
to thyroid peroxidase (TPO) inhibitors than humans are, one reviewer (GW) asked if similar data
2-4
-------
have been collected on differential susceptibility to NTS inhibitors. In response, another reviewer
(NC) indicated that transporters, like MS, are among the most conserved proteins across species.
Specifically, citing data collected in her laboratory, she indicated that rat NTS and human NTS are
extremely similar (>93% homologous). This reviewer also noted that the Iv for iodide
translocation at the NTS and the inhibition constant for perchlorate at the NTS are nearly identical
in rats and in humans. Based on these observations, this reviewer concluded that rat MS is
essentially as susceptible to inhibitors as human MS. She and another reviewer (GW)
recommended that the Revised ERD highlight the similarities in the MS protein between rodents
and humans.
During these comments, reviewers raised additional topics that were not discussed
extensively: one reviewer (KK) reiterated that the discussion of mode of action in the Revised
ERD should clarify that MS does not translocate perchlorate, though inhibition of iodide uptake
clearly occurs; another reviewer (NC) summarized studies of how perchlorate inhibition of iodide
uptake affects iodide discharge from the thyroid; and another reviewer (TZ) noted that a
submission in the premeeting comments questioned the relevance of "cold exposure" (i.e., not
radiolabeled) to the Revised ERD (see page 39 of the premeeting comments in Appendix C).
2.3 Charge Question A.3—The 1999 peer review panel agreed with EPA that perchlorate
was not likely to directly interact with DNA. What inferences can be made, based on
consideration of the mode-of-action data, to inform the choice of dose metric and the
approach for low-dose extrapolation?
When responding, the peer reviewers focused on two general topics—the use of thresholds
in low-dose extrapolations (even for cancer endpoints) and the choice of dose metric—as the
following paragraphs indicate:
Use of nonlinear low-dose extrapolations. Three peer reviewers (KK,LK,GW) said EPA's
use of nonlinear low-dose extrapolations is appropriate and adequately defended. Two of
these reviewers (KK,LK) noted that the Revised ERD provides compelling evidence that
2-5
-------
perchlorate is not genotoxic. They supported the idea that cancer endpoints may result
from non-genotoxic mechanisms having anti-thyroid effects. Another reviewer (GW)
acknowledged that proving a chemical has a toxicity threshold is difficult given that studies
test a limited number of subjects at a limited number of dosage levels. In the case of
perchlorate, however, this reviewer indicated that EPA presents clear and convincing
mechanistic arguments supporting that the toxicity resulted from a nonlinear process.
Given the large number of NTS molecules in the thyroid, he argued, iodide deficiency in the
thyroid would not result from a single, low perchlorate exposure, but would rather likely
require sustained, elevated exposures.
One reviewer (AC) suggested that EPA use toxicodynamic arguments or modeling to
support the approach of nonlinear low-dose extrapolations. As one example of defending
the low-dose extrapolation approach, this reviewer indicated that EPA could use PBPK
modeling to quantify the extent of NTS inhibition needed to observe downstream adverse
effects. He added that such models can elaborate the relative importance of various
sequelae. Ms. Annie Jarabek (EPA) clarified that EPA used the models that were available
from the Air Force Research Laboratory (AFRL). She added that it was originally planned
to develop a biologically-based dose response model, but this modeling effort did not
succeed to that extent.
Selection of area-under-the-curve in blood (A UCB) as a dose metric. Several peer
reviewers commented on the dose metric selection, and most offered differing insights.
One reviewer (MK) indicated that the selected dose metric appeared to be simply a
convenient choice, rather than a selection that EPA justified mechanistically. This reviewer
expressed particular concern about using area-under-the-curve of perchlorate in the thyroid
(AUCT) as a dose metric, given that perchlorate is not translocated into the thyroid cells.
Another reviewer (KK) clarified, however, that EPA's proposed dose metric is not the
AUCT, but rather the AUC of perchlorate in the blood (AUCB). These reviewers
discussed the dose metric selection in greater detail when responding to charge question
F.2 (see Section 7.2).
Another reviewer (AC) questioned whether the selected dose metric was the best predictor
of toxic effects. He suggested that measurements of intermediate responses on the causal
pathway (e.g., compensating hyperplasia, excess mitoses per unit time) may be better
indicators of adverse effects, particularly for neoplasia. As evidence of his concern, this
reviewer noted that the studies presented in the Revised ERD imply that iodide uptake
inhibition (a dose metric that EPA considered) appears to be a poor indicator of adverse
effects. Another reviewer (NC) disagreed, cautioning about what inferences can be drawn
from human dosing studies of short duration (e.g., 2 weeks or less). Because humans have
relatively vast reservoirs of iodinated thyroglobulin in the colloid of thyroid follicles, she
noted, humans can produce thyroid hormones for up to a few weeks, even in the absence
of iodide uptake. As a result, this reviewer emphasized that perchlorate exposure clearly
may cause changes in thyroid hormones and TSH levels in humans, even if these effects are
2-6
-------
not observed within the first 2 weeks of exposure. The peer reviewers revisited this
argument when critiquing the human studies (see Section 3).
2.4 Charge Question A.4—A harmonized approach to characterize the potential risk of
both noncancer and cancer toxicity has been proposed based on the key event of iodide
uptake inhibition. Comment on whether the approach is protective of both.
Before asking for responses, the discussion leader summarized the proposed harmonized
approach: by focusing on a key event (iodide uptake inhibition) that precedes neoplasia,
neurodevelopmental effects, and other noncancer effects, EPA's toxicity model protects against
both cancer and noncancer outcomes. Summarizing the premeeting comments, the discussion
leader noted that the peer reviewers generally supported EPA's proposed harmonized approach.
No peer reviewers offered conflicting opinions.
2-7
-------
3.0 Responses to Questions in Topic Area B: Human Health Effects Data
This section summarizes the peer reviewers' comments on human health effects data, as
summarized primarily in Chapter 4 of the Revised ERD. Dr. David Hoel was the designated
discussion leader for this topic area, and he initiated and moderated the peer reviewers' responses
to charge questions B. 1 through B.5. The peer reviewers offered comments on many of the
human studies presented in the Revised ERD, including both ecological epidemiological studies
(e.g., Crump et al. 2001) and clinical studies (Greer et al. 2000; Greer et al. 2002 - In Press;
Lawrence et al. 2000, 2001). The reviewers focused largely on studies that were completed since
the 1999 peer review, but were not yet published in the peer-reviewed literature. This section
summarizes the peer reviewers' comments on human health effects in detail. Readers interested in
the peer reviewers' major findings on human health effects data should refer to the Executive
Summary of this report.
Note: Before the peer reviewers discussed the human health effects studies in detail, Ms. Annie
Jarabek (EPA) clarified that all human clinical data presented in the Revised ERD were
considered in the Agency's toxicity assessment and none were excluded due to the policy
issued on December 14, 2001, on the use of third-party human data. Ms. Jarabek
specifically noted that EPA evaluated the Greer study, including information published in
an abstract (Greer et al. 2000) and the raw data presented in a quality assurance/quality
control report (Merrill 200la). Further, ERG distributed to the peer reviewers the
manuscript of the Greer study (Greer et al. 2002 - In Press), which has been accepted for
publication in a scientific journal.
3.1 Charge Questions B.I and B.2—Review of the Human Clinical Data Published Since
1999 That Have Not Undergone Peer Review
The charge to the peer reviewers identifies relevant publications (Greer et al. 2000;
Lawrence et al. 2001; Merrill 200la) that were completed since the 1999 peer review but were
not published in the peer-reviewed literature when the current meeting was planned. As a general
comment, one reviewer (DH) recommended that EPA update Table 4-5 in future releases of the
3-1
-------
Revised ERD to include findings from the most recent human health effects studies. Specific
comments on the publications follow:
Comments on the Greer study (Greer et al. 2000; Greer et al. 2002 - In Press; Merrill
2001a). The "Greer study" is a clinical study that involved administering perchlorate in
drinking water to 24 euthyroid adults over 14 days and testing for radioactive iodide
uptake (RAIU) and thyroid hormone levels at selected days during and after exposure.
Three doses were considered, with 8 adults in each dosage group. When ERG originally
distributed the peer review materials, the study was only available as an abstract (Greer et
al. 2000) with much of the raw data from this study (plus data from 7 additional adults in a
lower dosage group, 0.007 mg/kg/day) documented in a separate report (Merrill 200la).
Prior to the meeting, the Greer study was accepted for publication in the peer-reviewed
literature, and ERG distributed copies of that manuscript to the peer reviewers (Greer et al.
2002 - In Press).
Summarizing the peer reviewers' premeeting comments, the discussion leader (DH)
indicated that the Greer study was generally well conducted and informative. It suffers
from several limitations, though: lack of control for potential confounders (most notably
dietary iodine intake), small sample size, consideration of only healthy adults, and use of
2-week exposure duration when human thyroid reservoirs can generate thyroid hormones
for weeks after iodide uptake is inhibited. The other peer reviewers expanded on these
concerns. One reviewer (MR), for example, expressed concern about the lack of
information provided on the subjects, other than their age (18 to 57 years old). He noted
that one must consider more detailed information to evaluate the study's findings—perhaps
information on weight, smoking habits, health status, and other potential confounders.
This reviewer also questioned the study's use of a simplistic three-point loglinear
regression dose-response model; he recommended more sophisticated statistical analyses
for modeling the dose-response behavior.
Another reviewer (TZ) had additional concerns, primarily associated with a key conclusion
in the abstract that reads ". . . water supplies containing less than [250 |ig/L perchlorate]
should not affect human thyroid function" (Greer et al. 2000). Noting that the study
considered dosing for only 2 weeks and only among euthyroid adults, this reviewer did not
think the Greer study's data supported such a general finding. He questioned whether the
study would have reached similar findings had more subjects been considered, had the
subjects included pregnant women, neonates, and other potentially susceptible populations,
and had the dosing period been longer than 2 weeks. Later in the discussion, another peer
reviewer (NC) indicated that she also strongly disagreed with this conclusion presented in
the Greer abstract.
Regarding the issue of susceptible populations, one reviewer (TZ) asked if researchers have
observed any differences in iodide uptake inhibition between neonates and adults.
3-2
-------
Another reviewer (NC) responded that iodide uptake kinetics have not been studied in
neonates. This reviewer (NC) was concerned, however, that a study of eight euthyroid
adults does not reflect the variability in thyroid function in humans, as evidenced, she
argued, by the considerable variability observed in the baseline 8-hour RAIU levels in the
Greer study (i.e., 6% to 25% of administered dose). Given this variability, this reviewer
questioned whether the Greer study truly supports a no-effect level.
Finally, one peer reviewer (NC) noted that some findings of the Greer study are generally
consistent with expectations. Specifically, those individuals who experienced iodide uptake
inhibition had perchlorate serum concentrations greater than the inhibition constant for
iodide uptake at the NTS.
Comments on the Lawrence publications (Lawrence et al. 2000; 2001). The "Lawrence
study" is a clinical study with resulted documented in two separate publications: an article
with data presented for some doses (Lawrence et al. 2000), and a letter to the editor that
describes another dose tested in the study (Lawrence et al. 2001). The Lawrence study is a
clinical study that, like the Greer study, administered perchlorate in drinking water for 14
days to euthyroid adults. The study considered nine subjects and reported RAIU,
circulating thyroid hormone levels, urine and serum perchlorate levels, and other
parameters at selected days during and after exposure. In his opening remarks, the
discussion leader (DH) indicated that the results of this study are generally consistent with
those of the Greer study. Like the Greer study, the Lawrence study suffers from several
limitations (e.g., limited sample size, lack of control for dietary intake and other
confounders, short exposure duration). The peer reviewers did not offer any additional
specific comments on the Lawrence study during their deliberations.
3.2 Charge Question B.3—Have the epidemiological studies been adequately summarized
as a basis for the hazard characterization?
Summarizing the peer reviewers' discussions, the discussion leader (DH) indicated that the
reviewers generally thought that EPA adequately summarized the human health effects studies.
Some reviewers commented that the findings from these studies should receive greater attention
throughout the Revised ERD, but others felt that a more cautious interpretation of the studies is
appropriate. The peer reviewers' specific comments follow:
Suggestions that EPA more prominently acknowledge findings from human health effects
studies. When discussing this charge question, two peer reviewers (AC,GW) suggested
that EPA place greater emphasis on the findings from the human health effects studies—a
3-3
-------
suggestion that other peer reviewers (DH,LK) echoed later in the peer review meeting. As
an example of the concerns expressed, one reviewer (AC) indicated that the Revised ERD
adequately outlines the limitations of the studies but does not prominently acknowledge
notable gaps in the dose-response of perchlorate exposure in the rat compared to the dose-
response observed in humans. Another reviewer (GW) agreed, noting that the Crump
study (Crump et al. 2000, see below) suggests that humans can experience perchl orate
exposures much higher than the proposed point of departure without having impaired
thyroid function.
When discussing the relevance of the human health effects studies, a reviewer (DH)
cautioned about inferring insights on causality from ecological studies (see the next
bulleted item). Agreeing that such studies have potential limitations, another reviewer
(AC) suggested that EPA present a balanced overview of these limitations for all ecological
studies, including those with positive findings. He specifically referred to the Schwartz
study (Schwartz 2001, see below), which EPA refers to as "by far the most convincing of
the neonatal studies" (page 4-13, line 11).
Two reviewers (AC,GW) made two recommendations for revising the Revised ERD to
emphasize the findings of the human health effects studies. First, noting that the human
studies appear to show the absence of excess health risks (particularly for thyroid tumors)
among highly exposed populations, one reviewer (AC) suggested that the Revised ERD
highlight the apparent differences in toxic responses observed in laboratory animals and in
humans, the limitations of the human studies notwithstanding. Second, this reviewer (AC)
recommended that EPA use results from the human studies to conduct a sensitivity analysis
on the point of departure the Agency used in the Revised ERD. This reviewer specifically
encouraged comparing the point of departure EPA derived from the laboratory studies to
that which would be based on human studies, rather than comparing RfDs calculated from
those points of departure using multiple uncertainty factors. The other reviewer (GW)
supported these recommendations, indicating that a greater emphasis on the human health
effects data could provide some perspective on toxicity thresholds derived from animal
studies.
Concerns about relying too heavily on the human health effects data. Three reviewers
supported the way in which the Revised ERD currently presents the human health effects
studies and cautioned EPA against basing key conclusions on these studies, given their
limitations. One reviewer (DH), for instance, cautioned EPA against using the results of
the ecological epidemiological studies to draw firm conclusions about perchlorate toxicity.
To illustrate his concern, this reviewer referred to an extensive ecological epidemiological
study that found a negative association between exposure to low levels of radon gas and
incidence of lung cancer—a result that contradicts the findings of many case-control
epidemiologic studies. EPA should remember the potential limitations of ecological
epidemiological studies, this reviewer noted, when interpreting results of such studies,
regardless of whether they have positive or negative findings.
3-4
-------
Two reviewers (NC,TZ) questioned the utility of the human health effects data in the
Revised ERD, given at least three specific limitations in the available studies. First, one
reviewer (TZ) emphasized that the available studies may not have considered the most
sensitive health endpoints. Referring to an earlier comment about thyroid tumors not being
observed in highly exposed human populations, this reviewer noted that the absence of
these tumors does not imply that other toxic effects are not occurring. He added that the
available studies considered only certain endpoints, perhaps not including the most
sensitive ones (e.g., neurodevelopmental effects). Second, noting that most humans can
continue producing thyroid hormones for weeks following iodide uptake inhibition in the
thyroid (see Section 2.3), another reviewer (NC) explained that the human studies with
exposure durations of 2 weeks or less (e.g., Greer et al. 2000; Greer et al. 2002 - In Press;
Lawrence et al. 2000, 2001) likely do not identify toxic effects that may occur for long
exposure durations. Third, the same reviewer indicated that no single metric of human
thyroid function identifies potentially significant thyroid impairment and toxic effects. She
cited a study in which researchers found associations between intellectual deficits in
children and their mothers' having decreased T4 levels during pregnancy, but not increased
TSH levels. (Another reviewer [TZ] later noted that his laboratory has observed changes
in gene expression in the brains of laboratory animals related to decrements in circulating
thyroid hormone levels—the magnitudes of such decrements not resulting in upregulation
by TSH.) These findings, said the reviewer (NC), show that adverse effects may result
from impaired thyroid function, even if evidence of upregulation is not observed.
Synthesizing these comments and providing his own insights, another reviewer (KK) noted
that EPA carefully and systematically reviewed the human health effects studies and
eventually concluded that the studies' limitations preclude derivation of a LOAEL or
NOAEL that can be used as a point of departure. This reviewer supported presenting
comparative risk analyses in Chapter 7, in which EPA calculates an RfD that it might have
derived from human data, assuming a defensible calculation can be made.
Comments on the Schwartz study (Schwartz 2001). The peer reviewers offered various
insights on the findings in, and EPA's interpretation of, the Schwartz study. First, one
reviewer (RW) recommended that EPA eventually consider any publication that may result
from the Schwartz thesis. Ms. Annie Jarabek (EPA) responded, noting that EPA considers
graduate dissertations and theses to be peer-reviewed publications. Second, reiterating his
concern about the inferences that can be drawn from ecological studies, one reviewer (DH)
cautioned EPA against the conclusions that can be drawn from the Schwartz study.
Finally, a third reviewer (MR) recommended that EPA interpret the significance of the
transient changes in T4 levels observed in the newborns (see lines 18 through 20 on page
4-12 of the Revised ERD).
3-5
-------
Comments on the Crump study (Crump et al. 2000). The peer reviewers had various
insights on EPA's interpretation of the Crump study. First, based on comments submitted
by an author of the study, one reviewer (GW) recommended that EPA rewrite its review of
the Crump study (pages 4-7 to 4-8 of the Revised ERD), but this reviewer did not specify
the nature of the necessary revisions. Second, the same reviewer (GW) questioned why
EPA dismisses the negative findings in the Crump study for exposure to perchlorate, which
were observed in a population of school children in Chile, while results of other
epidemiological studies of cohorts in Chile have been widely used to develop dose-
response data for exposure to nitrates and nitrites in drinking water and its associated
effects. Third, another reviewer (NC) noted that the Crump study has paradoxical findings,
namely lower TSH levels observed among more highly exposed individuals. Finally, yet
another reviewer (TZ) found the incidence of goiter among the study population (>20% for
some subsets of school children) to be surprisingly high. No other reviewers commented
further on these topics.
Comments on balanced presentation of human health effects studies. Citing statements
made in the reviewers' premeeting comments, the discussion leader (DH) indicated that
one reviewer (AC) suggested that the findings reported in two publications (Crump et al.
2000; Soldin 2001) deserve more discussion in the Revised ERD. Citing his own
premeeting comments, he further noted that the Revised ERD gives disproportionately
great attention to the Lawrence study, as compared to the Greer study.
General comments. Two peer reviewers offered general insights on iodide uptake
inhibition and dietary iodine deficiency during this discussion. First, when interpreting data
on circulating TSH levels, one reviewer (NC) noted that iodide uptake does not have to be
completely inhibited for thyroid upregulation to occur. Upregulation, she argued, may
likely begin to occur when thyroid uptake decreases by a factor of 2, with far greater
upregulation resulting from any further inhibition. Second, one reviewer (NC) indicated
that iodide uptake inhibition may have serious consequences, especially considering that
iodine deficiency among pregnant mothers is one of the most preventable causes of mental
retardation in the world. Another reviewer (GW) agreed and suggested that humans with
iodine deficiency may be a susceptible population, but he added that most U.S. residents'
dietary intake of iodine is currently far higher than the recommended levels. Though she
agreed that iodine deficiency is not a widespread problem in the United States, another
reviewer (NC) indicated that some sub-populations in the United States do not meet their
iodine dietary intake requirements, particularly during pregnancy and lactation.3
3 When reviewing the draft of this report, one reviewer (TF) recommended that EPA consult with expert
clinical endocrinologists when characterizing the incidence of iodine deficiency among the population.
3-6
-------
3.3 Charge Question B.4—Are the exposure measures constructed from data in the
epidemiological studies sufficient to permit meaningful bounding of the predicted dose-
response estimates derived from extrapolation of the laboratory animal studies?
The peer reviewers discussed several issues when responding to this charge question.
First, the discussion leader (DH) summarized the reviewers' premeeting comments: despite
limitations of epidemiological studies and clinical trials, several reviewers recommended that EPA,
at a minimum, explicitly compare effect levels observed in laboratory animal studies to those
observed in humans. Expanding on this subject, one reviewer (AC) suggested that the Revised
ERD include much more specific summaries and analyses of the human health effects studies (e.g.,
at what levels have perchlorate-related effects been observed in humans? at what levels would
one predict health effects to occur?). Ms. Jarabek (EPA) asked if the peer reviewers could
provide specific suggestions on how such predictions can be made, given that the available
epidemiological studies examined changes in circulating thyroid hormone levels, but did not
consider neurodevelopmental and neoplastic sequelae. Another peer reviewer (KK) noted that
the Revised ERD uses two approaches to construct exposure measures from the epidemiological
studies: calculating actual exposure doses (e.g., as EPA reported on pages 4-15 and 4-18 for
selected human studies) and using PBPK models to calculate internal doses (e.g., as EPA did to
compare doses between selected human and laboratory animal studies). This reviewer supported
these approaches, and indicated that he saw no better alternative for constructing doses to relate
the human and laboratory animal studies.
Second, two reviewers discussed the differential sensitivity of rats and humans to
perchlorate exposure. On the one hand, one reviewer (GW) indicated that rats and humans have
dramatically different thyroid physiology, noting that perturbations in the thyroid economy lead to
far greater effects (in terms of circulating thyroid hormone levels) in rats than in humans. On the
other hand, another reviewer (TZ) cautioned against inferring that rats are more sensitive than
humans to perchlorate exposure. This reviewer explained that no researchers have established the
exact amount of thyroid hormone decrements that result in adverse neurodevelopmental effects in
3-7
-------
rats and humans, and therefore no conclusions should be drawn on whether rats are more
sensitive to perchlorate exposure than humans, at least in terms of neurodevelopmental and adult
neurological sequelae.
Three peer reviewers raised additional issues during this discussion or later in the peer
review meeting. First, one peer reviewer (DH) indicated that additional comparisons can be made
across the occupational and clinical studies regarding the perchlorate doses needed to elicit
changes in thyroid hormone levels. Second, a peer reviewer (GW) recommended that the Revised
ERD include some text on the prevalence of goiter among populations exposed to perchlorate,
considering that goiter is widely observed among individuals with iodine deficiencies. Third,
another peer reviewer (NC) expressed concern about serum perchlorate levels observed in the
human health effects studies, based largely on analyses presented in a review of these studies
(Soldin 2001). The review, she explained, reported that humans exposed to 10 mg of perchlorate
a day had serum concentrations of 0.6 |ig/ml, while perchlorate was not detected in the serum of
humans exposed to 3 mg of perchlorate a day. This reviewer could not understand the non-
detects in this latter group, because she expected that those receiving the 3 mg/day dose would
have serum concentrations of approximately 0.2 |ig/ml—a level about 50 times higher than
detection limits commonly reported for ion chromatography. The other reviewers could not
explain this apparent discrepancy.
3.4 Charge Question B.5—Are the associations observed in the epidemiological data
consistent with the proposed mode of action? Did the experimental design have
sufficient power to accurately ascertain the association between perchlorate exposure
and the specific outcome(s) ? Were confounding factors appropriately controlled?
Summarizing the reviewers' premeeting comments, the discussion leader (DH) indicated
that the findings of the human health effects studies are generally consistent with the proposed
mode of action. He added that EPA identified limitations in these studies, such as limited power
for detecting certain effects and lack of control for potential confounding factors. The discussion
3-8
-------
leader noted that the effect of potential confounders on the selected outcomes has not been
estimated. The peer reviewers did not comment further on this response.
3-9
-------
4.0 Responses to Questions in Topic Area C: Laboratory Animal Studies
This section summarizes the peer reviewers' comments on laboratory animal studies of
perchlorate toxicity, as documented in Chapter 5 of the Revised ERD. The peer reviewers
initially focused on specific toxicologic endpoints (e.g., developmental, reproductive,
neuroendocrine) and then offered general comments on EPA's synthesis of the available data.
During these discussions, the peer reviewers critiqued the protocols, performance, and results of
those laboratory animal studies completed since the 1999 peer review. They also commented on
EPA's interpretations of the studies. This section includes detailed summaries of the peer
reviewers' comments; readers interested in the main findings regarding laboratory animal studies
should refer to the Executive Summary of this report.
4.1 Comments on Developmental Toxicity
The discussion leader (MC) presented his comments on developmental toxicity and
facilitated subsequent discussions among the reviewers on this topic. A summary of the
reviewers' discussions follows:
Review of developmental studies completed prior to the 1999 peer review. The discussion
leader (MC) briefly summarized findings from four relevant developmental toxicity studies
published prior to the 1999 peer review (Postel 1957; Brown-Grant 1966; Lampe et al.
1967; Brown-Grant and Sherwood 1971). For each study, he indicated the exposure dose,
the gestational days over which doses were administered, the endpoints that were
evaluated, and whether or not effects occurred. He emphasized three key points to
consider when interpreting these studies: all four studies considered relatively large doses,
ranging from 100 mg/kg/day in one study (Lampe et al. 1967) to 2,660 mg/kg/day in
another (Brown-Grant and Sherwood 1971); fetuses in the studies were not examined for a
wide range of developmental effects; and the window of exposures in some studies did not
include the most critical time frames for organogenesis. The peer reviewers did not
comment further on these studies.
Detailed comments on the two most recent developmental studies. The discussion leader
(MC) provided various insights on the two most recent developmental toxicity studies.
This included one study (Argus 1998) that was peer reviewed by the panel in 1999 and has
4-1
-------
since been published as a journal article and another study (Argus 2000) that was
completed since the 1999 peer review, but had not been peer-reviewed when this meeting
was organized. In general, the discussion leader indicated that the two studies were
apparently well conducted and considered an appropriate number of animals to detect
effects. But, he questioned the need to evaluate teratological outcomes so thoroughly,
given that adverse neurotoxic outcomes reportedly occur at considerably lower exposure
doses (see Section 4.5). This specific comments on the individual studies follow:
- Regarding the 1998 study of developmental toxicity in New Zealand White rabbits
(Argus 1998a), the discussion leader (MC) identified the dose ranges (0 to 100
mg/kg/day), the time frame over which doses were administered (gestational days
6 to 28), and some study conclusions (e.g., decreases in thyroid weight in dams
that were not statistically significant, decreases in T4 levels, no significant changes
in T3 or TSH levels). He indicated that the study administered doses at the
appropriate gestational time for organogenesis. The study reported a NOAEL
greater than 100 mg/kg/day for fetal developmental toxicity, other than for
potential thyroid effects.
- Regarding the "Segment II Developmental Study" in rats (Argus 2000), the
discussion leader (MC) indicated the range of doses considered (0.01 to 30
mg/kg/day), when they were administered (starting 15 days prior to cohabitation
and ending at sacrifice), and some study conclusions (increases in localized
alopecia in dams in two dose groups, "questionable changes" in pre-implantation
loss, decreases in ossification sites at sternal centers and forelimb phalanges in the
highest dose group). The discussion leader found two elements of the study design
unusual: the decision to begin dosing animals 15 days prior to cohabitation and the
notable gap between the highest dose (30 mg/kg/day) and the second highest dose
(1 mg/kg/day). He noted that EPA and the study authors interpret the observed
effects at the highest dose group differently. While Argus reports 30 mg/kg/day as
a NOAEL for developmental toxicity based on the endpoints considered, EPA
considers this value a LOAEL. The reviewers briefly discussed the different
interpretations. One reviewer (KK) indicated that 30 mg/kg/day is an appropriate
LOAEL. The discussion leader (MC) acknowledged that the distinction is
debatable, but he eventually agreed that 30 mg/kg/day could be viewed as a
LOAEL.
Additional discussions. Following the review of the two most recent studies, one reviewer
(GW) asked the discussion leader to comment on the teratogenicity of perchlorate. The
discussion leader (MC) indicated that teratologists typically consider frank anatomic
malformations as teratogenic endpoints, and not the various other endpoints examined in
the two studies (e.g., changes in thyroid hormone levels). From this perspective, the
discussion leader (MC) said, perchlorate does not appear to be teratogenic, except perhaps
at the high exposures considered in the historical studies (see the first bulleted item in this
4-2
-------
list). Another reviewer (GW) concurred, adding that the malformations observed in the
various developmental studies appear to result from hypothyroidism, and not directly from
perchlorate exposure.
4.2 Comments on Reproductive Toxicity
The discussion leader (TC) presented the majority of comments on the reproductive
laboratory animal studies, and EPA's interpretations of the studies' results. First, he summarized
the design of the main reproductive toxicity study that was conducted since the 1999 peer review
meeting but had not been peer-reviewed when the current peer review was planned. This study
(Argus 1999) examined reproductive effects in Sprague-Dawley rats across two generations: the
P, Fl, and F2 generations were all exposed to perchl orate, with doses ranging from 0 to 30
mg/kg/day, and a variety of endpoints were considered.4 The discussion leader offered generally
favorable comments on the study design and methods. The only general methodological
weakness he identified was waiting until day 1 of lactation to weigh pups; he did not consider this
to be a critical shortcoming. Specific comments on the reproductive toxicity study follow:
Comments on reproductive endpoints. The discussion leader (TC) first summarized the
study's findings regarding reproductive endpoints: nearly every endpoint revealed little
evidence of reproductive effects, even in the high dose group (30 mg/kg/day). Though
some changes were observed in pregnancy rates and the number of stillborn pups, these
and other findings were not statistically significant. Based on these observations, the
discussion leader and another reviewer (KK) agreed with the study's finding that 30
mg/kg/day is the appropriate NOAEL for most reproductive effects considered, with the
possible exception of selected male reproductive endpoints.
The discussion leader (TC) was not convinced, however, that the two-generation
reproductive toxicity study fully evaluated the potential for male reproductive endpoints.
He listed several reasons for this concern. First, a dose-related decrease (not statistically
significant) in sperm density and spermatid density was apparent in the Fl generation, but
not in the P generation. Further, the discussion leader presented his own calculations of
daily sperm production, which also showed a (not statistically significant) dose-related
decrease in the Fl generation, but not in the P generation. Moreover, the daily sperm
4 Preliminary data on the Fl generation were reviewed during the 1999 peer review meeting, but the
entire set of data for all generations were compiled after that meeting was conducted.
4-3
-------
production levels observed in the two highest dose groups in the Fl generation were
notably lower than the levels he routinely measures in rats in his laboratory, using the same
measurement device. Finally, the discussion leader said the number of animals with low
sperm counts increased with dose in the Fl generation. He expressed concern about the
possibility of sperm effects manifesting in later generations, where accumulation of
perchlorate exposure effects might occur.
Based on these observations, the discussion leader recommended further analysis of the
existing sperm data in the Argus study. This reviewer acknowledged that an earlier
publication (Springborn Laboratories 1998) examined selected parameters as a satellite to
another 90-day study. This publication reported no evidence of perchlorate affecting
sperm counts and motility, but he indicated that more detailed review of the Argus study is
warranted. For example, this reviewer suggested that the study authors re-evaluate the
histological testes slides to evaluate appropriate male reproductive endpoints more
thoroughly; he asked that the authors clarify the unexpected, and considerable, difference
in sperm density between the P and Fl generation control groups; and he recommended
that the authors evaluate the reliability of the computer-assisted semen analysis
measurement device used to generate much of the sperm data. These follow-up activities
are needed to satisfy this reviewer that 30 mg/kg/day truly is an appropriate NOAEL for
reproductive toxicity.
Note: After the peer review meeting, the reviewer (TC) who expressed concern about the
sperm data reported in the Argus study had follow-up questions for the study's
authors. ERG forwarded these questions to the study's authors and returned the
responses to the peer reviewer. The peer reviewer submitted a post-meeting
comment (see Appendix J) with additional insights on the sperm data.
Comments on all other endpoints. When evaluating the two-generation reproductive
study, the discussion leader (TC) summarized findings for other endpoints, primarily the
thyroid histology and thyroid and pituitary hormone levels. This reviewer indicated that
the findings for the decreased thyroid hormone levels, particularly in the adult rats, are
generally consistent with EPA's proposed mode of action. The other peer reviewers
discussed the histology and thyroid hormone endpoints in greater detail later in the meeting
(see Sections 4.3 and 4.4). Ms. Annie Jarabek (EPA) asked the peer reviewers if they had
any comments to make on effects observed in the F2 generation. None of the peer
reviewers offered detailed insights on this matter.
4.3 Comments on Endocrine and Neuroendocrine Toxicity
Dr. Tom Zoeller led the discussions on endocrine and neuroendocrine toxicity, which
focused primarily on how perchlorate exposure affected circulating thyroid hormone levels in
rabbits and rats, as reported for two laboratory animal studies (Argus 1998a; Argus 2001). A
4-4
-------
summary of the comments made during this segment of the peer review meeting follows, but the
peer reviewers revisited issues of endocrine toxicity when discussing thyroid pathology and
neurotoxicity (see Sections 4.4 and 4.5, respectively).
Comments on concordance among thyroid endpoints. Because of the mode of action for
perchlorate, the discussion leader (TZ) said, concordance is expected among multiple
thyroid endpoints, such as decreases in T4, increases in TSH, and changes in thyroid
histopathology. The reviewers discussed various findings regarding thyroid hormone
levels, saving their comments on thyroid histopathology for the following presentation (see
Section 4.4). The discussion leader found a high degree of concordance among the thyroid
hormone endpoints in a recent laboratory animal study (Argus 2001) (e.g., dose-dependent
increases in TSH and decreases in T4 and T3), but acknowledged that earlier studies (e.g.,
Argus 1998a) did not observe similar results across all three hormones. The discussion
leader suspected that the lack of concordance across endpoints in the earlier study may
have resulted from poor measurement techniques (see the next bulleted item).
Comments on sources of inconsistencies across studies. The reviewers addressed certain
inconsistencies between the thyroid hormone findings of two laboratory animal studies
(Argus 1998a; Argus 2001). Two reviewers (MA,MP) found these inconsistencies
somewhat troublesome, but the discussion leader (TZ) suspected that the inconsistencies
likely resulted from how the researchers used RIA kits to measure thyroid hormone levels.
He explained that many measurements documented in the laboratory animal studies (e.g.,
Argus 1998a; Argus 2001), particularly for T4, appear to be at levels near or below the
range of the standard curves. Measurements of such trace amounts, he argued, are known
to be highly variable. Furthermore, because the studies did not document inter-assay and
intra-assay variability, the precision of the RIA measurements is unknown, complicating
efforts to interpret results. Because of these concerns, the discussion leader suspected that
the poor measurement techniques caused the lack of concordance among thyroid endpoints
within studies and lack of consistency in outcomes across studies. The discussion leader
added that he can confidently dismiss certain inconsistent results, given the measurement
techniques used and the extensive mechanistic knowledge of how perchlorate exposure
inhibits iodide uptake at the thyroid.
Comments on the shape of the observed dose-response. Three peer reviewers commented
on the dose-response relationship observed for changes in thyroid hormone levels, namely
that monotonic dose-response behavior was not identified. One reviewer (MP) indicated
that many studies over the years have identified non-monotonic dose-response behavior,
such as U-shaped or inverted dose-response curves. This reviewer himself observed such
dose-response patterns when investigating nicotine-related behavioral effects and when
others at his institution conducted studies on the chronic administration of endocrine
disrupters. He added that the animal studies for perchlorate are all based on relatively
4-5
-------
short dosage periods, and the shape of the observed dose-response curves may reflect the
nature of an acute response. As a result, this reviewer cautioned against disregarding any
study's findings only because the observed dose-response is non-monotonic. Another
reviewer (MA) agreed with these observations, but added that the lack of consistency
across studies is more troubling than the reported shape of the dose-response curve.
Though not disagreeing with these comments, the discussion leader (TZ) offered different
insights on the observed dose-response behavior. Based on the proposed mode of action,
he indicated, a monotonic dose-response relationship for changes in thyroid hormone
levels5 is expected. The absence of monotonic dose-response in certain dosage groups and
generations, he reiterated, may simply result from failed application of the RIA kits and
measurement of thyroid hormones (particularly T4) at levels at or below the range of the
standard curves.
Comments on statistical analyses of thyroid hormone levels. Ms. Annie Jarabek (EPA)
asked the panelists to comment on the statistical methods used to evaluate the thyroid
hormone levels, asking specifically if they support a recommendation from the 1999 peer
review that EPA use analysis of variance (ANOVA) for these evaluations, rather than
t-tests that do not take into account the litter of the individual animals. One reviewer
(RW) indicated that he supported the recommendations made at the previous meeting by
Dr. Joseph Haseman. The discussion leader (TZ) agreed, noting that multiple t-tests are
clearly inadequate. He supported use of ANOVA, since repeated measures of thyroid
hormone levels in many studies are not available. Moreover, due to concerns about
inconsistent uses of the RIA kits, he cautioned EPA against pooling measurements from
multiple studies into a single statistical analysis.
Another reviewer (AC) recommended approaches other than ANOVA, particularly if EPA
is most interested in evaluating the nature of the dynamic response to iodide uptake
inhibition. This reviewer stressed that aggregate statistics (e.g., correlations, regression
models) will not adequately capture such a dynamic response. His recommended approach
is for EPA to first develop pharmacodynamic hypotheses (e.g., increases of TSH should
follow decrements in thyroid hormones) and then use non-parametric statistical methods to
test them. Though not disagreeing with this alternate approach, the discussion leader (TZ)
noted that the available data on thyroid hormone levels probably will not support extensive
dynamic response modeling, particularly for hormones released in a pulsed manner (i.e.,
TSH).
5 This reviewer (TZ) emphasized that the monotonic dose-response behavior is anticipated for changes in
thyroid hormone levels, but the nature of the dose-response for the downstream effects of these changing hormone
levels cannot be predicted given that the mechanisms by which such effects occur have not been fully established.
The reviewers revisited this issue when discussing neurotoxicity (see Section 4.5).
4-6
-------
4.4 Comments on Thyroid Pathology (Including Cancer Effects)
Dr. Gary Williams moderated the discussions on thyroid pathology, which considered the
findings EPA reported for colloid depletion, hypertrophy, hyperplasia, and cancer. Considering
all studies together, this reviewer noted that the thyroid pathology analyses focus strictly on
histopathology. He added that perchlorate does not appear to exhibit cellular toxicity, as would
be characterized by cell death, cell degeneration, and other outcomes. The emphasis of the
Revised ERD, therefore, is appropriately on thyroid histopathological changes resulting from
impaired thyroid function (itself caused by perchlorate inhibition of iodide uptake). A summary of
the reviewers' specific comments on thyroid histopathology follow:
• Comments on the diets used in a study reporting thyroid pathology (Argus 2001). The
discussion leader (GW) expressed concern that rats in a key laboratory animal study (Argus
2001) were fed "certified rodent diet 5002"—a diet he said he had never seen used in
toxicity studies. He noted that this particular feed is soy-based, and therefore likely
contains goitrogens. As a result, he wondered if the soy-based diet might have
exaggerated the observed effects of perchlorate exposure, though he acknowledged that no
adverse thyroid pathologies were observed in the control group. Another reviewer (TC)
had a different opinion: he thought this particular feed is cereal-based and widely used in
laboratory animal studies, in accordance with Good Laboratory Practices. This issue was
not resolved at the peer review meeting, but two reviewers (LK,GW) recommended that
EPA investigate this issue further.
• Are colloid depletion, hypertrophy, and hyperplasia adverse effects? The discussion
leader (GW) questioned whether the observed thyroid histopathologies (primarily colloid
depletion, hyptertrophy, and hyperplasia) should be considered adverse effects, especially
when some of the outcomes are apparently reversible and are not associated with
compromised thyroid function. As an example of his concern, this reviewer indicated that
colloid depletion is basically an adaptive effect. This reviewer acknowledged that onset of
hyperplasia suggests that the thyroid has lost its ability to compensate adequately, but he
emphasized that the observed hyperplasia appears to be reversible, based largely on
findings from a laboratory animal study that considered a 90-day dosing period followed by
a 30-day recovery period (Springborn Laboratories 1998). He quoted EPA's
interpretation of this study: "recovery of the thyroid histopathological changes was
essentially complete by 30 days post-exposure . . ." (Page 5-26, lines 21-22). Noting that
rats in a control group from another study (Argus 2001) were diagnosed with hyperplasia,
he expressed further concerns about the biological significance of these diagnoses.
4-7
-------
Because of these observations, the discussion leader (GW) was not convinced that the
observations of thyroid lesions are truly adverse effects. He noted that sustained
hyperplasia in the thyroid gland would likely proceed to thyroid follicular neoplasms, but he
did not think the studies provided evidence of such advanced effects (see the following
bulleted items).
One other peer reviewer (DJ) addressed these comments, noting that reversible hyperplasia
does not necessarily suggest that adverse effects will not occur. He explained that some
non-genotoxic carcinogens (e.g., phenobarbital) may cause transient hyperplasia, with
tumors occurring later. The discussion leader (GW) agreed, but emphasized that these
"delayed" tumors following transient hyperplasia would only occur in the presence of
continued exposure to the non-genotoxic carcinogen.
Comments on the diagnoses of thyroid adenomas. The discussion leader (GW) expressed
concern about whether the thyroid adenomas identified in the two-generation reproductive
toxicity study (Argus 1999) were truly neoplasms, and he asked EPA to identify the criteria
that the Pathology Working Group used to make these diagnoses.6 This reviewer noted
that the STP criteria, which are based strictly on histopathology, may lead to false positive
diagnoses—a concern he based on a previous experience in which he noted ovarian
histopathology in rats that met the STP criteria for a granulosis cell tumor, but the
"tumors" later vanished after the treatment ceased. Questioning whether the thyroid
adenomas identified by the Pathology Working Group may instead be advanced, but
reversible, stages of hyperplasia, this reviewer indicated that the tumor diagnoses for these
rats are not compelling.
Another reviewer (KK) had a different opinion. He indicated that the Revised ERD
presents strong evidence that the highest perchlorate exposure doses (30 mg/kg/day)
produced cancer in the rat. He added that the presence of the tumors only in the highest
dosage group supports EPA's inference that perchlorate exposure leads to neoplastic
outcomes. The discussion leader (GW) later agreed, adding that the presence of tumors in
only the highest dosage group suggests that rats exposed at this level (30 mg/kg/day) for a
lifetime would get thyroid tumors.
Are rodents a good model for neoplastic outcomes in humans? The discussion leader
(GW) questioned whether the thyroid tumors reported for rats are relevant to perchlorate
toxicity in humans. Rodent neoplasia, he indicated, is generally a much simpler process
than human neoplasia. As evidence of this, he noted that in vitro studies of rodent cells
Dr. Doug Wolf (EPA) clarified during this discussion that the Pathology Working Group used the
standard National Toxicology Program criteria to diagnose tumors. These criteria, he explained, use Society of
Toxicologic Pathology's (STP's) Standardized System of Nomenclature and Diagnostic Criteria (SNNDC). He
added that the Pathology Working Group included both external experts and pathologists from the National
Toxicology Program. Finally, Dr. Wolf noted that the diagnoses of adenomas are based on the morphology of the
lesions on the slides, not on suspected biology, which cannot be determined from the morphology alone.
4-8
-------
have identified neoplastic transformations following as few as two gene mutations
(typically in an oncogene and a tumor suppressor gene), while similar studies of human
cells have required between four and seven gene mutations to achieve similar neoplastic
transformations. Focusing specifically on thyroid neoplasia, he commented that thyroid
tumors are relatively easy to induce in rats, while no evidence of perchlorate-related tumors
has been observed in humans. Based on these and other arguments, he concluded that rats
are not useful models for thyroid neoplasia in humans. The other peer reviewers did not
comment on this issue.
Comments on the use ofBayesian statistics for the cancer effects. Several peer reviewers
commented on EPA's Bayesian analysis of tumor incidence. One reviewer (KK) found the
analyses elegant. Another reviewer (MR) commended EPA on its use of the Bayesian
hierarchical model, but commented that the software (BUGS) used for this analysis is not
based on the most sophisticated approach for drawing numbers for the numerical
simulations (i.e., it uses correlated draws, rather than independent ones).7 Two other
reviewers, however, had concerns about the Bayesian analyses of tumor incidence.
First, though he acknowledged that Bayesian analysis is a powerful statistical tool, one
reviewer (AC) cautioned about using Bayesian analyses to detect certain outcomes when
the expected ones are not initially observed. He noted that the Revised ERD does not
explain exactly why Bayesian analyses were conducted and what possible outcomes were
examined. He thought EPA could instead have used Bayesian analyses to test multiple
hypotheses, which would avoid the perception that the Agency was seeking a particular
effect. This reviewer also questioned the utility of control groups in laboratory animal
studies, since EPA's statistical analyses instead considered outcomes observed in historical
control groups.
Second, another reviewer (DH) expressed concern about assumptions EPA made to
compare the tumor incidence observed in rats after 19 weeks in the laboratory animal study
(Argus 1999) to that observed among rats aged 2 years in historical laboratory
controls—assumptions that were needed to compare the cancer incidences over the same
time frame. Citing his experiences extrapolating cancer incidences in laboratory animals
from one age to another, this reviewer emphasized that the reliability of these
extrapolations decreases with increased time frames. Though he indicated that some
researchers have performed reasonable extrapolations of cancer incidence data over
relatively short time frames (e.g., using data observed at 18 months to predict incidence at
24 months), he noted that some efforts to extrapolate incidence over longer time frames
have generated "bogus" results. Based on these concerns, this reviewer viewed the
Bayesian analysis as a modeling exercise and questioned whether EPA can state the
7 Dr. David Dunson, from the National Institute of Environmental Health Sciences (NIEHS), clarified that
the BUGS software was used to analyze the motor activity data because multi-dimension integration was required.
However, S-Plus with independent draws was used to conduct the Bayesian analyses of the tumor data.
4-9
-------
probability of cancers occurring at 19 weeks with as much confidence (p = 0.005) as
reported in the Revised ERD.8
4.5 Comments on Neurotoxicity
Dr. Michael Aschner facilitated the discussions on neurotoxicity, during which the
reviewers commented on relevant laboratory animal studies and EPA's interpretations of these
studies. Specifically, the discussions focused on the studies of brain morphometry (Section 4.5.1)
and motor activity (Section 4.5.2) in rats.
4.5.1 Comments on Studies of Brain Morphometry
The peer reviewers discussed the studies of brain morphometry at length, considering both
the most recent study of this endpoint (Argus 2001) and an earlier study that followed a similar
methodology (Argus 1998b). The discussion leader (MA) initiated the comments by reviewing
the scope of the most recent study. Then he said the study has many potential flaws, though he
acknowledged that it was very extensive, used an adequate number of animals, and was
conducted using thorough quality control procedures. The peer reviewers had differing opinions
on these flaws and the extent to which they may have affected the study's findings, as the
following summary indicates:
Methodological concerns. The discussion leader (MA) identified several aspects of the
most recent study (Argus 2001) that could have biased the measured dimensions of brain
sections. For instance, linear measurements of brain dimensions are subject to artifacts
from fixation, sectioning, and positioning of the grid for viewing sections. The discussion
Dr. David Dunson (NIEHS) provided several clarifications when the reviewers discussed interpretations
of thyroid tumor incidence. First, he clarified that the motivation for conducting the Bayesian analysis was the
strong weight of evidence from previous studies that thyroid tumors are very rare in young animals. Given that the
thyroid is the hypothesized target for the mode of action for perchlorate, Dr. Dunson noted, data reviewers could
not simply discard the diagnoses, especially because they were unanticipated. Second, regarding the Bayesian
analyses, Dr. Dunson noted that evaluating trends among historical controls is a well-established technique for
enhancing the sensitivity of statistical analyses, and relevant controls (i.e., same strain of rats) were appropriately
considered.
4-10
-------
leader indicated that volumetric measurements of brain dimensions are preferred. (Note:
Ms. Annie Jarabek, from EPA, clarified that Agency guidelines for brain morphometric
studies currently require the use of linear measurements. She also noted that the purpose
of the most recent study was to repeat the same measures used in the earlier study.)
Moreover, the discussion leader questioned both whether inconsistent sectioning of the
brains may have biased results and how EPA decided which samples to include and exclude
from its statistical analyses.9 Given the overall dimensions of brains in rat pups, the
discussion leader was particularly concerned that small deviations in sectioning practices
could lead to substantial errors in fine-scale measurements. Finally, he pointed out that the
analyses of brain sections were not blinded.
Though not disagreeing with these potential methodological weaknesses, another reviewer
(TZ) was not convinced that they invalidate the data. This reviewer explained that the
sectioning practices and other aspects of the study methodology surely introduce variability
into the observations, but he found no evidence that these factors introduce any systematic
bias. This alleviated some of his concerns about the study's methods.
Is hypothyroidism expected to increase the size of selected brain regions? The peer
reviewers debated the biological plausibility of hypothyroidism causing increased
dimensions in specific brain sections. The discussion leader (MA), for example, expected
hypothyroidism to result in decreased sizes of brain sections, based on studies published in
the literature. Another reviewer (TZ) did not share this expectation. He cautioned that the
published studies linking hypothyroidism to decreased sizes of brain sections are based
largely on subjects with severe hypothyroidism or on subjects given thyroid hormone
replacement to treat severe hypothyroidism—not on subjects experiencing the impaired
thyroid function believed to result from perchlorate exposure. This reviewer noted that,
because researchers have yet to quantify the dose-response behavior for how changes in
thyroid hormone levels affect linear measurements of brain sections, he has no clear
expectation for what changes in brain morphometry might result from small decrements in
thyroid hormones. In summary, several reviewers commented that a more complete
mechanistic understanding of how hypothyroidism alters central nervous system
development is desired, but one reviewer (TZ) noted that such an understanding currently
does not exist.
Nonetheless, for greater confidence that the observed brain morphometry changes are
indeed related to impaired thyroid function, the discussion leader (MA) suggested that
EPA examine the existing data, or possibly control data from the literature, to determine if
9 Dr. Andrew Geller (EPA) indicated that the Agency used all brain sections for its primary statistical
analyses. He noted that EPA's profile analysis was run using all of the data from all brain sections and using only
the data from the two brain levels that were not purported to show bias (i.e., omitting data from the posterior
corpus callosum and hippocampus structures). Both of these analyses showed dose-related alterations in the
pattern of brain development when compared to controls.
4-11
-------
the observed brain morphometry changes are truly associated with other thyroid endpoints,
such as changes in thyroid hormone levels or thyroid histopathological effects (i.e., colloid
depletion, hypertrophy, hyperplasia). Based on his review of the data, he noted that the
dosage groups that exhibited brain morphometric changes did not consistently exhibit
significant effects in terms of decreased thyroid hormone levels, which made him question
whether the brain morphometric changes can be mechanistically linked to hypothyroidism,
as the proposed mode of action suggests. To address these and other concerns, the
discussion leader (MA) recommended several future actions, such as making specific
toxicological hypotheses for studying specific brain regions and integrating observations
from brain morphometry, thyroid hormone levels, and neurobehavioral endpoints into a
single evaluation.
Variability in measurements. The discussion leader (MA) expressed concern about the
variability in the brain section measurements, particularly among pups in the same litter.
He cited some observations in which the Argus study reports a linear dimension of the
corpus callosum for a pup in one litter to be more than twice as large as that for another
pup from the same litter, sacrificed on the same day. He suspected that this considerable
variability results from the weaknesses in the study methodology (see the first bullet in this
list). Dr. Kevin Crofton (EPA) noted that the data cited by the discussion leader present
the range of measurements and not the variability; Dr. Crofton explained that the
coefficients of variation for most measurements were typically on the order of 15% for
animals within a given dose group.
Consistencies and inconsistencies in observed results. The reviewers had different
opinions on the implications of consistencies, and lack thereof, between laboratory animal
studies. On the one hand, the discussion leader (MA) was very concerned about
inconsistent findings across the two studies of brain morphometry (Argus 1998b; Argus
2001). He acknowledged the two studies had some consistent results, but was troubled by
many inconsistent findings, including the following: the recent study (Argus 2001) found
no significant effect in the size of the corpus callosum in female rats, but the previous study
(Argus 1998b) found a significant—and much larger—effect in the females; some results
differed between the right and left hemispheres10; and the previous and recent studies had
10 Dr. Andrew Geller (EPA) acknowledged that some differences were observed among measurements
taken in the right and left brain hemispheres. He added, however, that no systematic biases were observed and that
the differences between the two hemispheres were generally smaller than the brain size differences that appeared to
result from perchlorate dosing. He also noted several consistencies across the brain morphometry studies: (1) In
the 1998 study, an increase in the size of the corpus callosum was observed; in the 2001 study, an inverted
U-shaped dose response in corpus callosum was observed (this study considered higher doses). (2) In the 1998
study, there was an inverted U-shaped dose response in cerebellar size (A-P dimensions); in the 2001 study, there
was also an inverted U-shaped dose response in cerebellar measures. (3) In the 1998 study, there was a U-shaped
dose response in caudate putamen; in the 2001 study, there was a U-shaped dose response in striatum (a large
portion of which is caudate putamen). (4) In the 1998 study, there was a U-shaped dose response in hippocampal
gyrus; in the 2001 study, there was a U-shaped dose-response in CAS.
4-12
-------
inconsistent results in the different age groups and treatment groups considered. The
discussion leader said that these inconsistencies, coupled with his concerns about the study
methods, leave him little confidence in the observed effects.
On the other hand, other reviewers had different insights on the inconsistencies and offered
reasons why they may have been observed. For example, one reviewer (MP) indicated that
inconsistencies in rodent studies can result simply from studies being conducted in different
seasons. As evidence of this, he noted that some laboratory animal studies have observed
that lethal doses to half a subject population vary from one season to the next by as much
as a factor of 2. This reviewer suggested that EPA consider the seasons when the two
brain morphometry studies occurred when commenting on inconsistent findings.
Are changes in brain morphometry adverse? One reviewer (LK) asked if the observed
brain morphometry changes have been associated with any functional, cognitive, or other
type of adverse effects. The discussion leader (MA) replied that the Revised ERD does not
correlate the brain morphometry findings with observations from any other endpoint; he
suggested that EPA evaluate whether such correlations exist. A third reviewer (GW) cited
the following quote from the most recent study of brain morphometry (Argus 2001),
indicating that it did consider other endpoints: "Detailed microscopic analysis . . . failed to
indicate any evidence of treatment-related neuropathologic effects." Another reviewer
(KK) indicated that EPA considers any alteration in brain structure as an adverse effect,
regardless of whether its potential impacts, if any, have been identified.
General comments. The peer reviewers made additional comments on the brain
morphometry studies that do not fall under the categories listed above. First, the
discussion leader (MA) was concerned that the Revised ERD relies too heavily on the
more recent brain morphometry study (Argus 2001), without fully integrating the findings
from the previous study (Argus 1998b). Second, another peer reviewer (AC)
recommended that EPA's statistical analyses include some adjustment for multiple
comparisons to determine if a dose-related signal exists across the two brain morphometry
studies. Third, though not disagreeing that additional statistical analyses may be helpful,
the discussion leader (MA) emphasized that no statistical analyses can correct for the
methodological weaknesses he identified in the study. Finally, the discussion leader (MA)
wondered if any mechanistic arguments could explain the U-shaped dose-response curve.11
11 Dr. Andrew Geller (EPA) offered some clarifications on the issues of mechanisms. Specifically, he
noted that profound hypothyroidism results in both reduced programmed cell death and reduced myelination in
corpus callosum. He added that literature examples show that increased numbers of fibers absent of myelin result
in smaller structures. He suspected that the dose responses to hypothyroidism for programmed cell death and
myelination differ. However, Dr. Geller noted that it is not difficult to imagine that when multiple mechanisms
contribute to a gross measurement (e.g., brain structure size), these mechanisms' competing or complementary
effects may result in non-linear results.
4-13
-------
Overall conclusions. Two peer reviewers offered their individual conclusions on the brain
morphometry studies and EPA's interpretation of them. The discussion leader (MA)
acknowledged that both brain morphometry studies (Argus 1998b; Argus 2001) provide
evidence suggesting an association between perchlorate exposure and changes in brain
morphometry in rats. But, given limitations of the study methodology and inconsistencies
in results between the two studies, the discussion leader said one cannot be certain that the
observed effects are not the result of sampling error, selection bias, or some other artifact.
As a result, he found the results of the brain morphometry studies to be inconclusive.
Another peer reviewer (KK) offered a different conclusion. Noting that EPA considers any
changes in brain structure as an adverse effect, he indicated that EPA has appropriately
designated the observed brain morphometry changes as a LOAEL, even though some
concerns remain about the study methodology and the shape of the dose-response curve.
This reviewer, and several others, commented further on EPA's interpretations of the brain
morphometry data when responding to charge question F.I (see Section 7.1).
4.5.2 Comments on Studies of Motor Activity
The peer reviewers had fewer comments on studies of motor activity in rats, primarily as
documented in one study completed since the 1999 peer review (Bekkedal et al. 2000).
Summarizing the peer reviewers' premeeting comments, the discussion leader (MA) noted that
the reviewers found the recent motor activity study to be rigorous and EPA's interpretations
appropriate. One peer reviewer (MP) recommended that EPA consult with the study's authors
about the timing with which behavioral observations were collected, suspecting that the
considerable variability observed in certain parameters might result from observations for the
different dosage groups being collected at different times of the day, rather than being consistently
collected during a specified window of time.
This reviewer (MP) also addressed the consistency of findings across the two studies that
examined motor activity. An earlier study (Argus 1998) discounted potential motor activity
effects. EPA, however, believed effects were evident. Given the agency's concern regarding
potential effects, another study (Bekkedal et al. 2000) was conducted. The authors of this study
again also found no statistically significant differences in any motor activity measure. EPA's
Bayesian analyses of the motor activity data, however, identified behavioral effects in both the
4-14
-------
1998 and 2000 studies. The reviewer (MP) found EPA's statistical analyses compelling and
agreed that they indicate that behavioral effects did occur in both studies, thus making the two
studies consistent in a general sense (i.e., they both demonstrated behavioral effects). This
reviewer added that perfect replication across two such behavioral studies is not expected, given
his previous comments on seasonal differences observed in laboratory animal studies and
variability in motor activity with time of day.
4.6 Comments on Immunotoxicity
Dr. Loren Roller, the designated discussion leader for immunotoxicity, provided the
majority of comments on this topic. He had generally favorable comments both on the two
immunotoxicity studies completed since the 1999 peer review and on EPA's interpretations of
these studies, but he did not support EPA's proposed uncertainty factor to account for database
insufficiencies regarding perchlorate immunotoxicity. A summary of the comments related to
immunotoxicity follows:
Comments on study O/B6C3F1 mice (Keil et al. 1999). The discussion leader (LK)
reviewed the design and key findings from this study, focusing primarily on endpoints
relevant to immunotoxicity. He commented on several findings that EPA summarizes in
Table 5-2 of the Revised ERD. For most endpoints considered, either no effects were
observed or the observed effects were not consistent across dosage groups and exposure
durations (i.e., 14 days and 90 days). Of particular note, the discussion leader indicated
that no effects were observed in one of the most sensitive indicators for whether a chemical
is immunosuppressive (i.e., spleen antibody responses to challenges with sheep red blood
cell antigens). He then commented on the one parameter that exhibited some consistency
in its effect—decreased macrophage phagocytosis was observed in all dosage groups,
albeit not in a dose-dependent fashion, for the 90-day experiment conducted during this
study, but the effect was not detected 30 days after perchlorate dosage ceased. Given that
the decreased macrophage phagocytosis was not observed across all experiments and was
not accompanied by any sign of compromised host resistance (e.g., response to challenges
by Listeria monocytogenes\ and that no consistent effects were observed for the many
other parameters considered, the discussion leader concluded that this laboratory animal
study (Keil et al. 1999) indicates that perchlorate exposure results in minimal immunotoxic
effects.
4-15
-------
Comments on study of CBA/JHsd mice (BRT-Burleson Research Technologies
2000a,b,c). To review this study, the discussion leader (LK) first noted that mice had
either no significant change in, or an enhanced, antibody response to sheep red blood cells
using the plaque-forming cell assay—a finding that he considered consistent with the
evaluation of humoral antibody response in the other study of immunotoxicity. He then
addressed evaluations of dermal contact hypersensitivity to 2,4-dinitrochlorobenzene, as
determined by the local lymph node assay (LLNA). Although exacerbated sensitivity was
identified in some dosage groups after 14 days, he noted, no clear dose-response
relationship emerged and consistent findings were not observed in the dosage groups
following 90 days of exposure. Moreover, the study reported a lack of negative controls.
As a result, the discussion leader questioned the relevance of the contact hypersensitivity
findings, as described further in the next bulleted item.
Comments on EPA 's overall interpretations of immunotoxicity. The discussion leader
(LK) indicated that the two immunotoxicity studies were comprehensive in evaluating both
innate and acquired immune responses, followed standard protocols, and used validated
assays. Perchlorate exposure showed no effects for most endpoints. Although negative
effects were observed for some endpoints, no clear dose-response relationship was
identified. Finally, the remaining detected effects could be viewed as protective or
favorable (i.e., there are signs of an enhanced immune response).
The discussion leader generally supported EPA's interpretations of the immunotoxicity
studies, but he did not support the Agency's proposed uncertainty factor to account for
database deficiencies regarding contact hypersensitivity. He listed several reasons why he
found the uncertainty factor unnecessary: the doses at which perchlorate affects thyroid
hormone levels are much lower than the doses where contact hypersensitivity was
observed; the contact hypersensitivity findings have inconsistencies and were
inappropriately controlled, thus leaving questions as to whether perchlorate causes the
observed effect; the relevance of skin rashes and agranulocytosis in Graves' disease
patients being treated with perchlorate is questionable, given that no such effects have been
observed in rodents or in humans receiving lower doses of perchlorate; the only
immunotoxic effect that exhibited some consistency (i.e., decreased phagocytosis) appears
to be reversible; and the contact hypersensitivity effects that occur in rodents may not be a
good model for such effects occurring in humans. Another reviewer (GW) did not agree
with this final argument due to the widespread use of LLNA for assessing contact
hypersensitivity for various beauty products. He suggested that EPA revise a sentence in
Chapter 5 (lines 1-2 on page 5-109) that implies LLNA responses are not physiologically
relevant.
4-16
-------
During these discussions, Dr. Ralph Smialowicz (EPA) offered several clarifications.
Regarding one study's (Keil et al. 1999) finding of decreased macrophage phagocytosis, he
indicated that the 1999 peer review panelists did not think the in vitro assay was an
appropriate test for phagocytic capacity of the macrophages. Though that panel
recommended an in vivo clearance assay be used in future studies, Dr. Smialowicz noted
that this recommendation was not heeded. Further, in response to a question asked by the
discussion leader, Dr. Smialowicz explained that mice, and not rats, are typically used for
these types of immunotoxicity studies.
4-17
-------
5.0 Responses to Questions in Topic Area D: Ecological Risk Assessment and
Evidence for Indirect Exposure
This section documents the peer reviewers' comments on ecological risk assessment and
evidence for indirect exposure—issues that EPA covers primarily in Chapters 8 and 9 of the
Revised ERD. Drs. William Adams and Teresa Fan, the two reviewers with demonstrated
expertise in ecological risk assessment, provided the majority of comments on this topic area.
These reviewers evaluated both the relevant studies conducted since the 1999 peer review and
EPA's interpretations of those studies, where applicable. The following subsections present
detailed accounts of the peer reviewers' discussions and recommendations; readers interested in
the peer reviewers' major findings on this topic area should refer to the Executive Summary of
this report.
5.1 Charge Questions D.I and D.2—Review of the Relevant Studies Published Since 1999
That Have Not Undergone Peer Review
The discussion leader (WA) indicated that only two studies conducted since 1999 have
been published (Goleman et al. 2002; Smith et al. 2001). The remaining studies (Condike 2001;
EA Engineering 1999; EA Engineering 2000; Parsons Engineering Science 2001), he noted, are
either memos, internal reports, draft reports, or laboratory reports that were not conducted in
accordance with Good Laboratory Practices and should be viewed strictly as screening-level,
informational studies. Further, this reviewer indicated that most of these studies did not include
measured test concentrations, which he considered to be a major limitation.
The reviewers' other responses to this question primarily addressed the disparate findings
between the various studies of perchlorate uptake by plants. Specifically, a reviewer (TF) noted
that field studies suggest that perchlorate bioconcentration factors (BCFs) for terrestrial and
aquatic plants appear to be less than one (see the graphs on page 31 in the premeeting comments
in Appendix C), while laboratory studies suggest that the BCFs are much greater than 1, and as
5-1
-------
high as 75, for several different terrestrial plant species (see data tabulated on pages 29-30 in the
premeeting comments in Appendix C). This reviewer (TF) identified reasons why both the
laboratory studies and the field studies may not be characterizing plant uptake accurately.
Regarding the laboratory studies, the reviewer noted that uptake was quantified only from
measurements of the amount of per chlorate depleted from the water in which the plants grew,
without consideration for perchlorate possibly absorbing to soils—an assumption that she
questioned. A more rigorous study design, this reviewer noted, would consider additional
measurements (e.g., concentrations of perchlorate in the plant at the end of the study) to verify
the assumption that all perchlorate depleted from the water is indeed taken up by the plant.
Regarding the field studies of plant uptake, the reviewer was concerned that the sediment
sampling results from a particular study (Parsons Engineering Science 2001) may not be
representative of actual environmental contamination levels. Specifically, she noted that sediment
samples were collected with excess water, not just interstitial water. Noting that the study
evaluated sediment concentrations on a wet weight basis, this reviewer wondered if the field
sampling methodology might have diluted the measured sediment concentrations.
The discussion leader (WA) offered several explanations for the apparent disparity between
the studies regarding plant uptake. The widely ranging BCFs, he said, might reflect differences
among the species considered, the differing matrices (water, soil, and sand) in which the plants
grew, or an effect of the age of the plants considered. He noted, for instance, that the laboratory
studies considered young plants, which were likely growing rapidly.
5-2
-------
5.2 Charge Question D.3—Comment on whether the assays selected for evaluation in the
ecological screening and site-specific analyses can be reasonably expected to identify
potential ecological effects of concern.
Though the reviewers agreed that the existing data provide useful insights into potential
ecological effects of concern, they indicated where data on additional species and specific life
stages are needed. Regarding the existing studies, the discussion leader (WA) noted that some of
the aquatic species considered (e.g., daphnids and fathead minnows) are known to be very
sensitive to exposures to environmental contamination. Given the mechanisms of perchlorate
toxicity, however, the discussion leader suggested that EPA consider broadening its evaluation of
these species—for example, by conducting a longer-term study that evaluates both the
reproductive success of fathead minnows and the ability of the juvenile fish to survive, grow, and
mature. Furthermore, due to concerns raised by the most recent ecotoxicological study (Goleman
et al. 2002), two reviewers (WA,TF) strongly supported further studies of amphibians.
The reviewers identified additional future research areas. The discussion leader (WA), for
example, indicated that EPA should consider evaluating herbivorous avian species, given the fact
that plant uptake of perchl orate has been reported. He emphasized, however, that this suggestion
is not based on any perceived sensitivity of these species to perchl orate. Another reviewer (TF)
noted that study of herbivorous terrestrial wildlife (e.g., voles, harvest mice) may be warranted,
given the results of the laboratory animal studies and the evidence of plant uptake. Finally, the
discussion leader (WA) recommended further screening with other algae species and with
macrophytes.
5-3
-------
5.3 Charge Question D.4—Comment on whether the goals and objectives of this ecological
screening analysis have been adequately described and to what extent these have been
met.
The two peer reviewers who commented on the ecological risk assessment offered different
responses. The discussion leader (WA) indicated that he had thought the goals of the ecological
screening analysis were met until he reviewed the results of the most recent ecotoxicological
study (Goleman et al. 2002), which was published after EPA released the Revised ERD. Because
that study suggests that effects may be occurring at water concentrations two orders of magnitude
lower than the proposed no-effect level (0.6 ppm), this reviewer was no longer certain that the
screening analysis truly achieves these goals (see lines 1-8 on page 8-2 of the Revised ERD).
On the other hand, the other reviewer (TF) listed two reasons why she was not convinced
that the screening-level analysis had met its goals and objectives, even without considering the
most recent data. First, because the laboratory toxicity tests do not include measured body
burdens, and therefore cannot be directly compared to field studies, this reviewer questioned
whether EPA can fully integrate the findings from these two types of studies. Second, she
expressed concern that no studies have considered organisms whose anion transport mechanisms
might be impacted by exposure to perchlorate—an issue she considered particularly important for
organisms requiring higher intakes of silicate or nitrate than do Selenastrum capricornutum.
5.4 Charge Question D.5—Do the analyses support the summary and conclusions
presented? Are relevant and important aspects of uncertainty addressed sufficiently?
The discussion leader (WA) provided insights on this charge question during his opening
presentation for this topic area. He noted that he had originally had a favorable opinion of EPA's
analyses, but, after reviewing a laboratory toxicity study (Goleman et al. 2002) published since
EPA released the Revised ERD, his opinions changed. He now highly recommends that EPA
5-4
-------
revise its analyses to incorporate this study's findings. His specific comments on EPA's analyses
and the recent laboratory toxicity study follow:
• Comments on EPA 's analyses of the data available prior to the publication ofGoleman 's
study. The discussion leader (WA) noted that the Revised ERD integrates all data that
were available on exposure and effects into an initial, screening-level ecological risk
assessment. He added that EPA successfully analyzed the limited data to derive its
conclusions, based largely on its derivation of "Tier II water quality values" (see pages
8-17 to 8-21 in the Revised ERD). This reviewer noted that EPA's proposed values—5
ppm for a secondary acute value and 0.6 ppm for a secondary chronic value—compare
well with values he derived from the same set of data using species sensitivity distribution
techniques (Aldenberg and Slob 1993). He concluded, therefore, that EPA's analyses of
the existing effects data were sound.
• Comments on the implications of more recent data (Goleman et al. 2002). Two reviewers
(WA,TF) indicated that a recent laboratory toxicity study of developing Xenopus laevis
(African frogs), if valid, completely changes their views of the findings reported in the
Revised ERD. They were specifically concerned about the 70-day exposure experiment, in
which multiple endpoints, including hind-limb length, fore-limb emergence, and tail
resorption, all showed effects at water concentrations far lower than the secondary chronic
value (0.6 ppm) that EPA reported in the Revised ERD. For instance, inhibited fore-limb
emergence was observed at water concentrations as low as 0.005 ppm. The two reviewers
emphasized that the endpoints considered (e.g., inability to produce limbs) have the
potential to affect the growth of individuals in this species, which in turn can impact
population levels. Based on these observations and the dramatically lower effect levels
reported in the Goleman study, the two reviewers strongly recommended that EPA
critically evaluate the study in subsequent releases of the Revised ERD.
Though concerned about the implications of the Goleman study, the discussion leader
(WA) noted that EPA should carefully evaluate three aspects of the study before making
any interpretations. First, he noted that the 70-day exposure duration for the frog embryos
is much longer than that which is conventionally evaluated, but he would not speculate on
how this exposure duration might have influenced the study results. Second, he noted that
effects were quite common in the controls (e.g., approximately 40% of the controls had
inhibited fore-limb emergence), which made him question the significance of the effects
observed at low-dosage levels. Finally, he had concerns about the use of a test solution
composed chiefly of deionized water and perchlorate, with non-detectable levels of
pesticides, metals, and organics. Given that some metals are essential for development, he
wondered if the lack of essential elements in the test solution might have accounted for the
effects observed in the controls.
5-5
-------
5.5 Charge Question D.6—Comment on the strengths and limitations of the available data
to characterize transport and transformation of perchlorate in the environment,
including soil, plants, and animals.
The peer reviewers agreed that the available data characterizing the fate and transport of
perchlorate are limited. The discussion leader (WA) summarized the results from the various
studies by showing graphs comparing perchlorate concentrations in one environmental medium
(e.g., water) to those in another (e.g., aquatic vegetation, sediment, fish). Copies of these graphs
are shown on pages 31 through 33 of Appendix C. The peer reviewers' specific comments on
these data, and assumptions regarding transport and transformation, follow:
• Transport of perchlorate. The discussion leader (WA) noted that no studies have
extensively characterized the factors that affect perchlorate transport in soils and
groundwater (e.g., soil partitioning, sorption of perchlorate to organic carbon or other
surfaces). Another reviewer (TF) agreed, and added that the available studies provide
conflicting information on perchlorate transport. For instance, she noted, some studies
report that perchlorate does not sorb to sand, while others suggest that considerable
sorption occurs. Further, she referred to a recent abstract according to which pH and
organic content largely determine the extent to which soil sorption occurs. Based on the
limited and conflicting findings, this reviewer concluded that no study has definitively
documented the extent to which perchlorate absorbs to soils. She cautioned EPA against
inferring that perchlorate does not sorb to soils, simply because the chemical is anionic and
hydrophilic in nature. She explained that her own research has demonstrated that anionic
chemicals sorb to local positively charged clusters that may be present in the organic or
mineral matrices of soils.
• Transformation of perchlorate. The peer reviewers briefly discussed the extent to which
perchlorate is transformed, both biologically and chemically, in the environment.
Regarding biological transformation, the discussion leader (WA) noted that the available
data demonstrate that plants and micro-organisms in anaerobic environments (e.g.,
sediments) reduce perchlorate. Another reviewer (TF) agreed, and added that a study
suggests that perchlorate is transformed in humans (see the bulleted item "Is MS inhibition
reversible? Is perchlorate metabolized?" in Section 2.1). Regarding chemical
transformations, one reviewer (TF) questioned statements suggesting that chemical
reduction is limited, especially considering that similar chemicals (e.g., sulfates) are
reduced in groundwater under certain conditions.
5-6
-------
5.6 Charge Question D.7—Comment on the strengths and limitations of the available data
to suggest sources of perchlorate exposure other than drinking water.
The reviewers provided few comments on this charge question. First, the discussion leader
(WA) indicated that future research is needed to develop more sensitive analytical methods, not
only for water but also for biotic tissues, soils, and sediment. Another reviewer (TF) agreed,
indicating that ion chromatography analyses potentially suffer from matrix interference and
interfering ions. She noted, however, that analytical methods (e.g., liquid chromatography with
mass spectrometry) are already being developed with improved sensitivity and selectivity.
Second, noting that perchlorate uptake by lettuce has been documented, the discussion leader
(WA) noted that humans may be indirectly exposed to perchlorate in vegetables grown on land
irrigated with perchlorate-contaminated water. Finally, another reviewer (TF) recommended that
future studies characterize the mechanisms by which plants uptake perchlorate. With a
mechanistic understanding of uptake, she noted, EPA may be able to predict uptake behavior in
plant species that have not been sampled.
5-7
-------
6.0 Responses to Questions in Topic Area E: Use of PBPK Modeling
This section summarizes the peer reviewers' comments on PBPK modeling for perchlorate,
as presented primarily in Chapter 6 of the Revised ERD. AFRL developed the four PBPK
models, and EPA applied them in its assessment approach. Dr. Kannan Krishnan moderated the
peer reviewers' responses to charge questions E.I and E.2, during which the reviewers critiqued
the structure, parameterization, validation, and application of the four model structures. Detailed
summaries of the peer reviewers' comments on the PBPK models follow; readers interested in the
major findings for this topic area should refer to the Executive Summary of this report. (Note:
Section 7.2 presents additional comments on the PBPK models, primarily on how EPA used these
models for interspecies extrapolation.)
Summarizing the premeeting comments, the discussion leader (KK) indicated that the
PBPK model structures are technically sound and accounts for the major anatomical
compartments and that they are based on standard approaches and equations. However, he and
other reviewers voiced several concerns about certain model representations, particularly that of
perchlorate uptake into cells, and parameter selections. The reviewers' detailed comments,
suggestions, and recommendations on these issues and many others are summarized below:
Representation of iodide uptake into thyroid cells. Several peer reviewers (NC,KK,MK)
questioned why the PBPK models consider passive uptake (i.e., diffusion) of iodide into
thyroid cells. Suspecting that active iodide uptake into cells is the dominant transport
process, these reviewers recommended that the model developers reconsider why passive
iodide uptake is simulated. More specifically, one reviewer (KK) recommended that the
PBPK models resolve the relative importance of these two uptake processes. Another
reviewer (TF) commented further on passive uptake of chemicals into cells when
discussing disposition of perchlorate (see next bulleted item).
One reviewer (MK) indicated that the Revised ERD does not adequately justify the use of
Michaelis-Menten kinetics to model iodide uptake into the thyroid. Another reviewer
(NC), however, was not concerned with this aspect of the PBPK models, noting that
several researchers have demonstrated that iodide transport via NTS follows Michaelis-
Menten kinetics.
6-1
-------
Representation ofperchlorate uptake into thyroid cells. The peer reviewers raised several
issues when discussing perchlorate disposition: whether NTS translocates perchlorate into
thyroid cells, whether perchlorate is translocated into cells by other mechanisms, and to
what extent passive (i.e., diffusive) transport ofperchlorate into cells needs to be
incorporated into the PBPK models.
Based on comments raised earlier in the peer review meeting (see the first bulleted item in
Section 2.1), several reviewers (NC,KK,MK) recommended that EPA verify whether NTS
actively transports perchlorate into thyroid cells—an assumption made in the four PBPK
model structures. One reviewer (NC) noted that she is unaware of any research that
unequivocally demonstrates that NTS translocates perchlorate into thyroid cells, while she
has reviewed several papers that suggest such translocation does not occur. Regarding
recent publications that report concentrations ofperchlorate in the thyroid (e.g., Yu et al.
2001), this reviewer suspected that the perchlorate detected may be bound to cell
membranes, rather than inside the thyroid cells. This reviewer indicated that researchers
can readily design an experiment to determine the extent to which perchlorate interacts
with NTS (i.e., whether it binds to NTS or is translocated by the protein), though she did
not think such an experiment has already been conducted. The reviewers revisited the
issue of active perchlorate transport when discussing the kinetic parameters used in the
PBPK models (see the next bulleted item).
Although several reviewers agreed that NTS apparently does not actively translocate
perchlorate into cells, one reviewer (MK) cited evidence that perchlorate is likely entering
cells by other mechanisms. Referring to the study that administered radioactive double-
labeled perchlorate to humans (Anbar et al. 1959), this reviewer indicated that the presence
of single-labeled perchlorate in the subjects' urine implies that perchlorate may be entering
cells somewhere in the body. He indicated that various other anion exchange mechanisms
may carry perchlorate into cells, even if NTS does not translocate the chemical. Based on
these concerns, this reviewer noted that the Revised ERD does not provide a complete,
convincing account of all cellular uptake mechanisms.
Two reviewers (NC,TF) commented on whether the PBPK models need to consider
passive transport ofperchlorate into cells. Citing observations made for anionic transport
in plants, one reviewer (TF) indicated that passive transport of anions into cells can be an
important process, especially when concentrations of the anions in the extra-cellular matrix
are extremely high (e.g., following an exposure). Another reviewer (NC) agreed in
principle, but added that passive transport in humans would be relevant only when
perchlorate achieves extremely high serum concentrations—concentrations that may not be
physiologically relevant.
Overall, the peer review panel indicated that the PBPK models should have more refined
representations of active and passive uptake into cells, based on the comments summarized
above. Two reviewers (NC,KK) clarified that they have no question that perchlorate
6-2
-------
interacts with NTS, thus inhibiting iodide uptake: their primary concern is to what degree
does perchlorate actually enter cells. Another reviewer (GW) agreed that this issue is
important to resolve, because the extent of cellular uptake affects how EPA should
approach other issues, such as the mutagenicity of perchlorate.
Mathematical representation of iodide and perchlorate uptake. Given that NTS does not
translocate perchlorate into thyroid cells, the reviewers questioned whether the PBPK
models and the Revised ERD should describe the cellular uptake process via NTS as
"competitive inhibition." The discussion leader (KK) explained that competitive inhibition
generally implies that two (or more) molecules are substrates for a given protein or
enzyme. Because perchlorate is not translocated by MS, he said, it technically does not
have a Michaelis-Menten constant (KjJ. He questioned, therefore, how the model
developers could derive a K,,, for the PBPK models (see pages 6-23 to 6-25 in the Revised
ERD). Another reviewer (NC) agreed, noting that perchlorate does not have a K,,,, but it
does have an inhibition constant that has been widely published. The reviewers
recommended that the PBPK models include a revised kinetic description of iodide uptake
inhibition.
Model parameterization. Several peer reviewers evaluated the kinetic parameters assigned
in the PBPK models. In addition to their concern that perchlorate does not have a K,,, (see
the previous bulleted item), the reviewers offered several comments. First, regarding the
Iv value used for the active transport of iodide (4.0 x 106 ng/L), one reviewer indicated
that decades of research have established that this K,,, should fall roughly between 20 and
30 jiM (micromolar). ERG notes that 4.0 x 106 ng/L is equivalent to 31 jiM of iodine,
though these figures were not mentioned at the meeting.
Second, the peer reviewers discussed the derivation of other relevant parameters, including
maximum velocity capacity (Vmaxc) in various tissues, permeability area, plasma binding
coefficients, and clearance values. One reviewer (MK) said that appropriate parameters
were selected for iodide, but he indicated that the Revised ERD does not adequately
describe how these parameters were selected for perchlorate. Another reviewer (KK)
noted, however, the PBPK models were parameterized largely from experimental data for
both chemicals, often by assimilating and integrating multiple data sets. He concluded that
the approach for parameter selection was defensible, both for iodide and perchlorate, based
on the data sets currently available. These reviewers identified types of additional data that
would help improve the confidence in the parameterization (e.g., time-course data of
perchlorate in multiple tissue types).
Overall conclusions. The two peer reviewers who ERG assigned to critique the PBPK
models summarized their overall comments. The discussion leader (KK) indicated that the
overall value of the models depends largely on their ultimate application. For instance, he
indicated that the models will not be useful for quantifying cellular concentrations of
perchlorate until a greater mechanistic understanding of the relevant uptake processes is
6-3
-------
achieved. Nonetheless, he noted that the PBPK models, largely because they adequately
represent urinary excretion and estimate serum concentrations, are useful tools both for
estimating internal doses from environmental exposures (or external doses) and for
estimating human equivalent doses. Though he recommended that the models include
more refined representation of uptake and inhibition processes in the thyroid, he stressed
that these refinements likely will have minimal impacts on predicted serum
levels—predictions that depend more on urinary clearance and volume of distribution than
on uptake into a relatively small physiological compartment (i.e., the thyroid).
The other reviewer (MK) echoed many of these comments and added others. Regarding
the cellular uptake processes, this reviewer recommended that the PBPK models include
more refined representation of these processes based on comments raised earlier at the
meeting. If this cannot be achieved, the reviewer suggested that EPA prominently
acknowledge in the Revised ERD that the PBPK models are not based on a mechanistic
understanding of the uptake processes.
6-4
-------
7.0 Responses to Questions in Topic Area F: Human Health Dose-Response
Assessment
This section summarizes the peer reviewers' comments on EPA's human health dose-
response assessment for exposure to perchlorate. Dr. Thomas Collins moderated these
discussions, which largely focused on the selected point of departure and uncertainty factors in
EPA's proposed RfD derivation. The meeting chair (RW) indicated that peer reviewers expressed
various, and often conflicting, opinions on many issues in this topic area: some reviewers
recommended that EPA use human data from the Greer study for the point of departure, but other
peer reviewers did not think this study is an adequate basis for the RfD derivation; some peer
reviewers supported the decision to base the point of departure in part on the reported changes in
brain morphometry, yet roughly the same number of reviewers did not think this study was an
adequate basis for the point of departure; and the reviewers had various opinions on the proposed
uncertainty factors, though nearly every reviewer agreed that the uncertainty factor of 3 for
database insufficiency on immunotoxic endpoints was unnecessary.
A detailed summary of the peer reviewers' responses to the four charge questions in this
topic area follow. Reviewers interested in a brief summary of the comments on EPA's human
health dose-response assessment should refer to the Executive Summary of this report.
7.1 Charge Question F.I—Are the conclusions and conditions regarding the key event and
the weight of the evidence for effects after oral exposure to perchlorate appropriate and
consistent with the information on mode of action? Have the diverse data been
integrated appropriately and do they support the proposed point of departure? Should
any other data be considered in arriving at a point of departure?
Dr. Tom Collins facilitated the peer reviewers' discussions on point of departure. These
focused on three general topics: the consistency between the mode of action and the toxic effects
(see Section 7.1.1), the proposed use of changes in brain morphometry (0.01 mg/kg/day LOAEL)
7-1
-------
when deriving the point of departure (see Section 7.1.2), and the proposed use of other data and
endpoints when deriving the point of departure (see Section 7.1.3). Though much of the
discussion focused on brain morphometry, EPA based the point of departure on other endpoints
as well (e.g., changes in thyroid hormone levels and thyroid histopathology).
7.1.1 Consistency Between Observed Effects and Mode of Action
Summarizing the premeeting comments, the designated discussion leader (TC) indicated
that a clear majority of the reviewers who responded to this charge question found the proposed
mode of action consistent with the observed neurodevelopmental and neoplastic effects.
Elaborating on this general response, one reviewer (KK) indicated that the Revised ERD clearly
states how perchlorate exposure initiates the perturbation of the HPT axis, which leads to
neoplastic and neurodevelopmental effects, thus supporting the harmonized approach to
evaluating noncancer and cancer toxicity. Other reviewers, however, indicated that EPA could
more convincingly link the mode of action to the observed effects:
• Lack of pharmacodynamic modeling. Two reviewers (AC,MK) acknowledged that the
Revised ERD links perchlorate exposure to the key event, but they argued that the
document does not provide specific details on any mechanisms linking the key event to the
neurodevelopmental or neoplastic endpoints. One reviewer (AC), for example, indicated
that the Revised ERD does not explicitly describe the full sequence of events between
perchlorate exposure and neoplasia, particularly for how this is expected to occur in
humans. He and another reviewer (MK) noted that the lack of pharmacodynamic modeling
leaves the relevance of the mode of action uncertain. The other reviewer (MK) further
explained that, with a detailed pharmacodynamic model, EPA could link iodide uptake
inhibition to reduction in thyroid hormone production and circulation—parameters, he
argued, that are more relevant indicators of whether toxic effects will occur. He said that
EPA could have drawn from existing PBPK modeling applications (reference not cited) to
characterize the pharmacodynamic mechanisms more effectively. This would have
provided a more convincing link between the proposed mode of action and the observed
toxic effects.
• Comments on doses where iodide uptake inhibition occurs. To link the proposed mode of
action to toxic effects, one reviewer (GW) recommended, the Revised ERD should clearly
indicate the doses at which iodide uptake inhibition have been observed in laboratory
animals and humans. Focusing on the 2000 abstract of the Greer study (Greer et al. 2000),
7-2
-------
three reviewers (NC,DH,RW) noted that the abstract reports measurable iodide uptake
inhibition in humans dosed at 0.02 mg/kg/day; they also noted that Table 7-5 in the Revised
ERD presents EPA's estimates of iodide uptake inhibition as a function of dose for the
various PBPK model structures.12 One reviewer (DH) indicated that EPA's estimates of
iodide uptake inhibition in humans (in Table 7-5) appears to be quite consistent with the
low-dose findings reported in the abstract of the Greer study. Noting that EPA's mode of
action ultimately links toxic effects to iodide uptake inhibition, one reviewer (GW)
recommended that the Revised ERD more prominently acknowledge the exposure doses at
which this inhibition has been observed.
Inconsistencies of findings within studies on thyroid hormone levels. Referring to Table
5-4 in the Revised ERD, one reviewer (GW) indicated that some groups of animals in the
laboratory animal studies experienced decreases in circulating T3 levels, while no
significant changes in circulating T4 levels were observed (e.g., see data for post-natal day
22 females). Given that T3 is formed by the deiodination of T4, this reviewer found such
trends confusing and wondered if they suggest that perchlorate may affect thyroid hormone
levels by some mode of action in addition to inhibiting thyroid iodide uptake. Another
reviewer (TZ) agreed that some studies may have inconsistent results, but he noted that
others (e.g., Argus 2001) have results quite consistent with expectations: T4 levels
decrease, TSH levels increase, and T3 levels exhibit various changes. He noted that
inconsistent findings in T3 levels might result from the fact that thyroxine-binding globulin
(TBG) is found in lactating rats and pups, which could give some buffering capacity. No
other reviewers commented further on this topic.
7.1.2 Comments on the Use of Brain Morphometry Effects as the Basis for the
Point of Departure
The peer reviewers discussed at length whether EPA should use the brain morphometry
data (Argus 2001) in deriving a point of departure for perchl orate. Their general concerns
centered on the quality of the linear measurements of brain regions and the biological significance
of any observed effect. The reviewers' comments spanned a broad range, from one reviewer
(MA) finding the brain morphometry studies inconclusive to another reviewer (TZ) indicating that
12 Though they clearly identified 0.02 mg/kg/day as a dosage where iodide uptake inhibition was observed
in humans, the reviewers did not specify whether this comment considered the lowest dosage group (0.007
mg/kg/day) in the Greer study. This lowest dosage group was not documented in the abstract (Greer et al. 2000),
but was documented in the subsequent quality assurance/quality control report (Merrill 200la) and the final
manuscript (Greer et al. 2002 - in press).
7-3
-------
the multiple studies and multiple re-analyses of the studies all show brain morphometric effects
occurring. The following bulleted items present a detailed review of these and other comments,
culminating with the individual peer reviewers' final statements on the brain morphometry data:
• Concerns about the study methodology. Based largely on comments made earlier in the
meeting (see Section 4.5.1), one reviewer (MA) reiterated that he found the brain
morphometry studies inconclusive, due largely, but not entirely, to the methodologies used
to measure the dimensions of brain regions. As a result, this reviewer indicated that EPA
should not consider the results of the brain morphometry study when deriving a point of
departure. Though not disagreeing that the brain morphometry studies have flawed
designs, another reviewer (TZ) suspected that the methodological issues of specific
concern (e.g., sectioning practices, use of linear dimensions) are expected to introduce
random errors into the study, not systematic ones. He said random errors introduced by
the study methodology would most likely make it impossible to detect statistically
significant effects, not to detect effects that do not exist. Two reviewers (MP,TZ) noted
that they found no evidence of systematic errors introduced by the study methodology
(e.g., use of different section practices for different dosage groups), and therefore
recommended that EPA not discard the data due to the random errors that the study design
may have caused.
• Comments on statistical re-analyses of the brain morphometry data. The Revised ERD
documents the results from two laboratory animal studies that evaluated changes in brain
morphometry, as well as statistical re-analyses of these studies. One reviewer (TZ) was
concerned about disregarding all of this information, which provides evidence—albeit with
some inconsistencies between the studies—of brain morphometry changes in animals
exposed to perchlorate. On the other hand, another reviewer (MA) did not find EPA's
statistical re-analyses compelling, due not to any flaws in the statistical approaches but
rather to his concern about the quality of the linear measurements of brain dimensions (see
the previous bulleted item).
• Mechanistic questions. Two reviewers (MA,LK) noted that the changes in brain
morphometry cannot be linked to perturbations in thyroid hormone levels, and presumably,
therefore, to perchlorate exposure. This leaves questions about exactly what causes effects
in brain structure and whether these effects are truly adverse or perhaps compensatory.
One reviewer (MA) also questioned the relevance of brain morphometric changes in rats to
humans. Though other reviewers (MP,TZ) agreed that the absence of mechanistic links is
unfortunate, they did not think the brain morphometry findings should be criticized for this
reason, especially considering there is no complete mechanistic understanding of how
thyroid hormone levels affect all neurodevelopmental processes. Noting that the brain
morphometry studies may be the first toxicological studies ever linking perturbations in
thyroid hormone levels to changes in the sizes of brain dimensions, one reviewer (TZ) felt
7-4
-------
uncomfortable disregarding the data because no previous studies have elucidated the
mechanisms that may cause these effects.
Concerns about inconsistencies in the brain morphometry results. Two peer reviewers
discussed whether one should expect consistency among certain findings. One reviewer
(MA) first listed various types of inconsistencies he observed, such as different results
between the sexes, post-natal days considered, and the two brain morphometry studies
(Argus 1998b; Argus 2001). Another reviewer (TZ) agreed that the lack of concordance
across the two studies is troublesome, but he was not as concerned about the other issues
raised. For instance, noting that certain neurodevelopmental events are known to take
place over distinct (and sometimes narrow) windows of time, this reviewer indicated that it
is not unreasonable to observe inconsistent brain morphometry effects at two different
postnatal days. He further noted that perturbations in thyroid hormone levels may affect
various brain regions differently, and one should not necessarily require that consistent
effects be observed across multiple regions.
Comments on the shape of the dose-response curve. The reviewers had various comments
on the biological significance on the shape of the observed dose-response curve, which,
Ms. Annie Jarabek (EPA) indicated, was an inverted U-shape for the corpus callosum (i.e.,
the smallest effects were observed at the lowest and highest doses and the largest effects
were observed at the intermediate doses). For instance, one reviewer (GW) indicated that
the dose-response curve implies that high doses of perchlorate may protect rats against
neurodevelopmental effects.
Other reviewers had different opinions. Because the mechanisms of thyroid hormone
action on the reported brain morphometry changes have not been identified, one reviewer
(TZ) indicated that he has no basis for dismissing the data because a linear or monotonic
dose-response curve was not observed. Another reviewer (MP) agreed, saying that
inverted U-shape dose-response curves have been documented, particularly in cases where
increased effects initiate compensatory responses, similar to the upregulation of thyroid
hormone synthesis observed following iodide uptake inhibition. This reviewer suggested
that the Revised ERD include specific hypotheses about mechanisms that may account for
the U-shaped dose response. Another reviewer (MA) also found no inherent problem with
non-linear dose-response curves, but he was troubled by the fact that the dose-response
trends are not consistently observed across both sexes.
Integration of brain morphometry data with other endpoints. Given the proposed mode of
action for perchlorate toxicity, one reviewer (GW) said, he would have expected that
changes in brain dimensions in the brain morphometry studies would be accompanied by
changes in thyroid hormone levels. But in the most recent brain morphometry study
(Argus 2001), the dams dosed at 0.01 mg/kg/day showed no significant changes in TSH or
T4 levels, and only marginal changes in T3 levels on gestational day 21. He suggested that
7-5
-------
EPA consider whether such modest perturbations to thyroid hormone levels would result in
altered brain structures.
Another reviewer (MP) was not convinced that the data currently available are sufficient
for integrating the brain morphometry data with other endpoints, particularly thyroid
hormone levels. Specifically, he indicated that the most recent brain morphometry study
reported only "snapshots" of thyroid hormone levels, which may not be representative of
the circulating hormone levels prior to the days when animals were sacrificed. He noted
that the observed changes in brain morphometry may result from decreased thyroid
hormone levels that occurred when these parameters were not measured.
Recommendations. Summarizing the peer reviewers' discussions, the meeting chair (RW)
noted that the reviewers' primary concerns about the proposed point of departure are
whether the brain morphometry data are of acceptable quality and whether the reported
effects are truly caused by perturbations in thyroid hormone levels. Individual reviewers
then offered several additional insights. Two peer reviewers (AC,MP), for example,
suggested that the raw data from the brain morphometry be re-analyzed by a party that is
blinded to the dosage levels. Another reviewer (MA) supported this suggestion, but noted
that such re-analyses would not address his underlying concern regarding the validity of the
linear measurements of brain dimensions. A third reviewer (KK) indicated that EPA could
defend use of the changes in brain morphometry as the point of departure, provided that it
addresses the reviewers' concerns regarding the study methodology and inconsistencies in
the findings. Finally, the meeting chair (RW) noted that EPA might consider the following
three options when making its final decision on the point of departure:
- Not consider changes in brain morphometry when deriving the point of departure.
Consider the changes in brain morphometry when deriving the point of departure,
but address concerns about the quality of the underlying data.
- Base the point of departure entirely on other endpoints, but perhaps account for
database insufficiencies regarding neurodevelopmental effects using an uncertainty
factor.
Final comments. After discussing the various strengths and weaknesses of the brain
morphometry study, the meeting chair asked the peer reviewers to give their final
individual opinions on whether EPA should consider the reported changes in brain
dimensions when deriving the point of departure. Table 2 (at the end of this section)
summarizes the reviewers' final remarks: two reviewers supported EPA's proposed
approach, three reviewers indicated that the brain morphometry findings were either
inconclusive or not compelling, three reviewers offered conditional remarks on the use of
the brain morphometry data, and the remaining nine reviewers did not comment specifically
7-6
-------
on whether or not the brain morphometry data should be considered when deriving the
point of departure.
7.1.3 Comments on Use of Data Other Than Brain Morphometry for the Point of
Departure
Though their comments primarily addressed the proposed use of brain morphometry data
for the point of departure, the peer reviewers also addressed EPA's use of other toxic endpoints
in deriving an RfD. Examples of the peer reviewers' other comments follow:
• Should human studies be used for the point of departure? Though some peer reviewers
suggested many times during the meeting that the Revised ERD more prominently
acknowledge findings from human health effects studies, two peer reviewers (KK,TZ)
questioned the utility of those studies for deriving a point of departure. One of these
reviewers (TZ) gave an example to explain his feeling: he did not think the Greer and
Lawrence studies, which examined iodide uptake inhibition and circulating thyroid
hormone levels for a very small subset of healthy, euthyroid adults, offer any insights on
potentially important sensitive populations (e.g., pregnant women, children, fetuses). The
other reviewer (KK) agreed, and added that the human clinical studies are based on limited
exposure durations and have not investigated important endpoints, such as reproductive
toxicity, neurotoxicity, and developmental toxicity. Because of these data gaps, this
reviewer supported an approach of evaluating the laboratory animal studies for insights on
the endpoints that have not been examined in humans.
On the other hand, three peer reviewers (AC,DH,LK) indicated that EPA can better
integrate data from human health effects studies into the Revised ERD, without necessarily
using those data for deriving the point of departure. One of the three (LK) wondered if
more insights can be drawn from effects observed among humans with Graves' disease
who have been prescribed high doses of perchlorate (e.g., one patient received 3
mg/kg/day for 22 years), though he acknowledged that this dosing was necessary to treat
hyperthyroidism. He also suggested that EPA consider basing the point of departure on
data from the Greer and Crump studies. Reiterating a comment made earlier in the peer
review, the second reviewer (AC) recommended that EPA use the human health effects
data in a sensitivity analysis of the proposed point of departure. The third reviewer (DH)
estimated an RfD based on human health effects data as 0.0001 mg/kg/day, which he
derived using a point of departure of 0.001 mg/kg/day, combined with an uncertainty
7-7
-------
factor of 10; he did not specify what the point of departure and uncertainty factor
represent.13
Comments on basing the point of departure on changes in thyroid hormone levels. The
peer reviewers commented briefly on whether EPA should consider basing its point of
departure on changes in thyroid hormone levels. One reviewer (GW) emphasized that EPA
could choose this endpoint for the point of departure, but should carefully distinguish
changes that are biologically significant from those that are simply statistically significant.
Specifically, recognizing that thyroid hormone levels exhibit considerable diurnal
variations, this reviewer recommended that EPA only consider measured hormone levels
outside a "normal" range as being potentially adverse. Another reviewer (LK) agreed,
indicating that statistically significant changes that fall within "normal" fluctuations should
not be considered adverse effects.
The reviewers briefly discussed whether decrements in thyroid hormone levels, specifically
T4, can lead to adverse neurodevelopmental effects. In a general sense, one reviewer (TZ)
said, decrements in T4 levels clearly can cause neurological dysfunction. This reviewer
added that extensive dose-response data linking these decrements to adverse effects are not
available, though some clinical thyroidologists have said that humans sustaining 10% to
15% reductions in circulating thyroid hormone levels may show symptoms of
hypothyroidism. Another reviewer (NC) agreed, citing a study that found associations
between pregnant mothers with lower levels of T4 during their first trimester (without an
associated increase in TSH) and impaired intellectual function in their children. Moreover,
she indicated that physicians evaluate babies for hypothyroidism very early in life to avoid
potentially irreversible effects of decreased thyroid hormone levels.
Comments on basing the point of departure on thyroid histopathology. Two peer
reviewers (GW,TZ) addressed whether EPA should base the point of departure on
observed thyroid histopathology, namely colloid depletion, hypertrophy, and hyperplasia.
The first of these reviewers (GW) questioned, by way of an example, whether EPA should
view thyroid colloid depletion as an adverse effect. He indicated that this effect is better
characterized as adaptive. Another reviewer (TZ) made similar comments, noting that
colloid depletion may demonstrate a perturbation of the HPT axis, but the biological
significance of this perturbation is questionable in the absence of the reported changes in
brain morphometry.
The first reviewer (GW) did propose that 1.0 mg/kg/day may be an appropriate point of
departure for thyroid tumorogenesis. He explained that this was the dosage required to
observe signs (i.e., hypertrophy) that the thyroid was being stimulated and stressed, which
13 This reviewer (DH) indicated that the RfD value he stated at the peer review meeting was only an
estimate he quickly made during the discussions. This reviewer's post-meeting comments (see Appendix J)
include a more detailed calculation he offered for an RfD based on the human health effects studies.
7-8
-------
he considered a departure from homeostasis. This reviewer suggested that EPA apply an
interspecies uncertainty factor of 0.1 to this point of departure, noting that perturbations of
the HPT axis apparently have far different consequences in rats than in humans. As
evidence of this, the reviewer noted that rats exposed to certain proton pump inhibitors
readily develop gastric neuroendocrine tumors, whereas no evidence of such effects has
been observed in humans.
Comments on basing the point of departure on iodide uptake inhibition. The peer
reviewers briefly discussed whether EPA should base its point of departure on any
particular level of iodide uptake inhibition. One reviewer (AC) questioned this approach,
wondering what specific adverse effects result at specified levels of iodide uptake inhibition
and how these effects differ between rats and humans. Another reviewer (GW) agreed,
noting that only marginal inhibition occurs among rats and humans dosed at 0.01
mg/kg/day, the inhibition appears to be reversible after short-term dosage periods end, and
the thyroid hormone levels are not considerably altered at this exposure level. Moreover,
he indicated that the existing data on short-term dosages are insufficient for evaluating
iodide uptake inhibition over chronic exposure durations, unless EPA's models account for
upregulating mechanisms.
Another reviewer (NC) agreed with some of these comments, but cautioned about some of
the inferences drawn. She indicated, for instance, that researchers have not yet established
the extent to which NTS inhibition is reversible. Though she acknowledged that the
kinetics of NTS active transport are strikingly similar across species, she was hesitant to
make premature judgments on how iodide uptake inhibition affects humans and rats
differently, especially considering that MS is expressed in fewer thyroid cells in humans
than in rats.
Other comments. When discussing the appropriate derivation for the point of departure,
the peer reviewers raised additional comments: apparent inconsistencies between thyroid
upregulation in rats and humans, statistical analyses of thyroid hormone levels, and
consistency between the proposed point of departure and findings from recent
ecotoxicological studies.
First, one reviewer (KK) indicated that the laboratory animal studies found evidence of
upregulation, while the human studies did not. On the other hand, another reviewer (NC)
noted that these apparent discrepancies are easily explained by differences in the thyroid
hormone reservoirs in the species. Noting that humans have vast reservoirs of thyroid
hormones compared to rats, this reviewer was not surprised that the 2-week dosage studies
in humans (e.g., Greer et al. 2000; Greer et al. 2002 - In Press) found no evidence of
upregulation.
7-9
-------
Second, regarding benchmark dose calculations (see Appendix 7B in the Revised ERD),
one reviewer (MR) questioned whether the Kodell-West algorithm is an adequate statistical
methodology for evaluating the thyroid hormone data. Because this algorithm assumes
that data are normally distributed, models means of distributions using quadratic functions,
and assumes that variances are equal across dosage levels, this reviewer suspected that the
algorithm is too restrictive to detect effects. He recommended that EPA instead use a
more flexible model for evaluating the thyroid hormone data.14
Finally, one reviewer (KK) noted that the point of departure that EPA proposed is similar
to the effects level reported in selected ecotoxicological studies—a factor the agency may
wish to consider in the Revised ERD.
7.2 Charge Question F.2—Comment on the use of the PBPK models for interspecies
extrapolation and the choice of the dose metric.
The peer reviewers generally supported the use of PBPK models for conducting
interspecies extrapolations, though some suggested that development of pharmacodynamic
modeling may help identify dose metrics more closely linked to adverse effects than serum
concentrations of perchlorate. Commenting specifically on the options EPA considered for dose
metric, one reviewer (KK) indicated that AUC perchlorate in blood is the most reasonable
measure of internal dose that allows for defensible extrapolations across species and different life
stages. He added that other dose metrics (e.g., AUC perchlorate in thyroid, circulating thyroid
hormone levels) would not be appropriate until the mechanisms of perchlorate uptake into cells
and the kinetics of upregulation processes have been adequately characterized. Another reviewer
(NC) agreed, and recommended that EPA verify that the Iv selected for translocation of iodide
through the apical cellular channel is consistent with that documented in a recent publication
(Golstein et al. 1995).
Though not disagreeing that AUC in blood allows for defensible interspecies
extrapolations, two reviewers (AC,MK) advocated use of a dose metric more predictive of toxic
14 Dr. Andrew Geller (EPA) explained that the statistical analyses in question were conducted by
Toxicology Excellence for Risk Assessment and submitted to EPA for review. EPA did not use the Kodell-West
algorithm in its statistical analyses.
7-10
-------
effects. More specifically, one reviewer (MK) noted that decreases in thyroid hormone levels or
increases in TSH may be better indicators of adverse effects than circulating perchlorate levels.
Another reviewer (AC) agreed, adding that pharmacodynamic modeling can help differentiate
metrics more related to adverse effects (e.g., excess cell mitoses per unit time) from those with no
risk implications. Both reviewers, therefore, advocated pharmacodynamic modeling to predict the
impact that perchlorate exposure has on thyroid function.
The reviewers' only other relevant comments addressed the use of PBPK modeling to
interpret effects observed in rats on post-natal day 4. Noting that impaired thyroid function at this
life stage would most likely result from decreased transport of iodide across the placenta, one
reviewer (GW) wondered how the interspecies extrapolations with PBPK models accounted for
any potential differences in placental physiology between rats and humans. One reviewer (KK)
noted that human PBPK models were not developed to evaluate pregnancy, fetuses, or neonates
and instead only a rat model was developed to evaluate these life stages. He indicated that EPA's
approach for using outputs from the rat models to extrapolate between different human life stages
is adequate, and he saw no other defensible approach for estimating human equivalent exposures
for various life stages. Another reviewer (NC) noted that NTS is expressed in the placentas of
both rats and humans, despite notable physiological differences across these species. Other
reviewers did not comment on this issue further.
7.3 Charge Question F.3—Are there other data which should be considered in developing
the uncertainty factors? Do you consider that the data support the values proposed or
different values for each? Do the confidence statements accurately reflect the
relevancy of the critical effects to humans and the comprehensiveness of the database?
Do these statements make all the underlying assumptions and limitations of the
assessment apparent? If not, what needs to be added?
The peer reviewers discussed the proposed uncertainty factors at length. They had widely
varying opinions on the matter. The designated discussion leader for this topic area (TC)
7-11
-------
indicated that the Revised ERD proposes a composite uncertainty factor of 300, which is derived
from several individual factors. The peer reviewers' comments on the individual components of
the uncertainty factor follow. Some of these comments addressed the uncertainty factors EPA
used in deriving the proposed RfD, while others addressed appropriate uncertainty factors should
the Agency base its RfD derivation on human health effects data. General comments on
uncertainty factors follow, and the reviewers' final statements on the proposed uncertainty factors
are listed at the end of this section.
• Extrapolation from aLOAEL to aNOAEL (proposed uncertainty factor = 10). The peer
reviewers had few comments on this element of the composite uncertainty factor. Three
reviewers (MC,KK,LK) acknowledged that EPA typically applies this factor when
extrapolating from a LOAEL to a NOAEL. Aside from expressing concerns about the
general practice of assigning exact numerical figures to this type of uncertainty (see the
bulleted item below titled "General comments"), none of the reviewers questioned the
proposed use of this uncertainty factor.
In the event that EPA bases its RfD derivation on human data, one reviewer (DH) said this
uncertainty factor may not be necessary, given that the authors of the Greer study report
identifying a NOAEL.
• Intrahuman variability (proposed uncertainty factor = 3). The peer reviewers generally
supported EPA's proposed uncertainty factor of 3 for intraspecies variability. One
reviewer (KK) noted that the Agency often assigns a factor of 10 for interindividual
uncertainty, but instead proposed a factor of 3 based on ". . . the variability observed in the
data and PBPK modeling for the adult humans . . ." (see lines 5-6 on page 7-20 of the
Revised ERD). This reviewer commended EPA for using the PBPK modeling to derive an
appropriate uncertainty factor, though he noted that the Revised ERD does not describe
exactly how EPA arrived at the factor of 3.15
Regarding EPA's comment that data from human subjects "... do not represent kinetic
data for the potentially susceptible populations of the hypothyroid and hypothyroxinemic
pregnant women and their fetuses," one reviewer (GW) recommended that the Agency
refer to recent publications from the National Academy of Sciences' Committee on
Reference Dietary Intakes for alternate approaches regarding consideration of medically
disadvantaged groups receiving treatment as potentially sensitive populations. The
reviewers discussed susceptibility to perchlorate exposure in greater detail when
responding to charge question F.4 (see Section 7.4).
15 When reviewing a draft of this report, one reviewer (MC) indicated that EPA can rationalize using an
uncertainty factor of 10 for intraspecies variability, but he did not make this comment at the peer review meeting.
7-12
-------
Inter species extrapolation (no uncertainty factor proposed). The reviewers expressed
differing opinions on whether an uncertainty factor for interspecies extrapolation is
necessary. One reviewer (MP), for example, noted that EPA often applies a 10-fold factor
for this element of uncertainly and commended the Agency for using the PBPK modeling
results to justify its decision not to use an interspecies uncertainly factor. Another
reviewer (KK) agreed, and supported EPA's proposed approach for interspecies
extrapolation.
Two reviewers (AC,GW) wondered if rodents are more sensitive to perchlorate exposure
than humans; if so, they said an interspecies uncertainty factor less than one might be
warranted. One reviewer (GW), for instance, said the rat is a poor model for humans in
terms of thyroid physiology. Noting that thyroid hormone function in rats may be at least
10-fold more active than thyroid function in humans, this reviewer suggested that an
interspecies uncertainty factor of 0.1, or even lower, may be defensible. Another reviewer
(NC) acknowledged that rats and humans have notable differences in thyroid physiology,
but she cautioned that researchers have not clearly established differential sensitivity to
perchlorate exposure.
Database insufficiency (proposed uncertainty factor = 3, based on concerns of
immunotoxicity). The peer reviewer with expertise in immunotoxicity (LK) did not
support EPA's proposed uncertainty factor of 3 to account for database insufficiencies
relevant to potential immunotoxic effects, particularly effects of contact hypersensitivity.
Because the LOAEL reported for contact hypersensitivity (0.06 mg/kg/day) is already
more than 3-fold higher than the proposed point of departure (0.01 mg/kg/day), this
reviewer saw no basis for applying the additional uncertainty factor.16 Moreover, he
questioned the relevance of the skin rashes observed in Graves' disease patients being
treated with high doses of perchlorate, noting that the patients received extremely high
doses of perchlorate and that their autoimmune condition may have contributed to the
observed rashes. This reviewer concluded that the current database on the immunotoxicity
of perchlorate is sufficient and application of any uncertainty factor due to database
insufficiencies is unwarranted. As Table 7-2 shows, a majority of the peer reviewers
agreed that this uncertainty factor is unnecessary.
Another reviewer (KK) viewed two of EPA's proposed uncertainty factors—the factor of
3 for database insufficiency on immunotoxicity and the factor of 3 for lack of data on
chronic exposures (see the next bulleted item)—as a single factor addressing overall
database insufficiency. He said that, given the number of laboratory animal experiments
that have now evaluated a variety of toxic endpoints (e.g., reproductive, developmental,
16 Ms. Annie Jarabek (EPA) clarified that EPA proposed the uncertainty factor due to inadequate
characterization on the immunotoxicity endpoints, primarily that for contact hypersensitivity. The factor is not
based on consideration of the relative magnitude of the LOAELs.
7-13
-------
neurotoxic), EPA should instead use an overall database insufficiency uncertainty factor of
1, if the point of departure is based on laboratory animal studies.
Subchronic to chronic exposure duration (proposed uncertainty factor = 3). Some peer
reviewers supported the proposed uncertainty factor for the lack of a chronic exposure
study, while others did not. The discussion leader (TC), for example, indicated that the
longest-duration exposure study (i.e., the 90-day study) has not provided convincing
evidence that exposures over longer durations will not reveal additional effects. Another
reviewer (DJ) agreed and added that the presence of tumors following a 19-week study
was of concern. Though he was not concerned that the tumors observed in rats are
relevant to humans, another reviewer (GW) indicated that no data have convinced him that
in utero programming of the HPT axis does not occur. As a result, he indicated that
EPA's proposed uncertainty factor may be justified.
One reviewer (LK), on the other hand, was not convinced that the presence of tumors in
two laboratory animals was biologically or statistically significant. Given that the tumors
occurred at dosage levels (30 mg/kg/day) several orders of magnitude higher than the
proposed point of departure (0.01 mg/kg/day), this reviewer questioned whether a 3-fold
uncertainty factor for the thyroid tumors is meaningful. Finally, given that exposures
occurred in utero during the study of concern, this reviewer suspected that further effects
of in utero programming would not be identified if the study duration had been longer than
19 weeks. For these reasons, he concluded that the uncertainty factor for subchronic to
chronic exposure duration is not justified.
Three reviewers (TF,DH,KK) commented on whether EPA should consider an uncertainty
factor for exposure duration if it chooses to base its point of departure on human health
effects data. Two reviewers (TF,KK) indicated that a 10-fold uncertainty factor would be
warranted if human data are used, given that the longest exposure duration in a controlled
study was 14 days. Another reviewer (DH) noted that a lower uncertainty factor may be
appropriate, particularly if occupational epidemiological studies provide perspective on the
implications of chronic exposures.
General comments. Several reviewers (AC,MC,MK,MR) indicated that they prefer
approaches other than applying simple, multiplicative factors to address uncertainty in RfD
derivations—a comment, they emphasized, that applies to all chemical risk assessments,
and not only to EPA's perchl orate analyses in the Revised ERD. As an example of this
concern, one reviewer (MK) thought use of generic uncertainly factors implies that risk
assessors lack an understanding of the toxicity mechanisms. Two other reviewers
(AC,MR) agreed, but added that they prefer more sophisticated uncertainty modeling,
rather than application of default factors. For instance, one reviewer (AC) indicated that
EPA could use Bayesian model averaging or Monte Carlo modeling to derive a probability
distribution for the point of departure, rather than applying 10-fold and 3-fold factors that
do not appear related to any physiological process. Another reviewer (MR) identified
7-14
-------
additional approaches to consider, such as establishing confidence intervals for uncertainty
factors so that the composite factor can be expressed as a range, rather than as a single
number.
Final comments. When the discussions on uncertainty factors ended, the meeting chair
(RW) asked the peer reviewers to summarize their individual opinions on the composite
uncertainty factor and its components. Table 3 summarizes the peer reviewers' specific
comments. Every reviewer who specifically addressed the proposed uncertainty factors for
deriving an RfD indicated that EPA should eliminate the uncertainty factor of 3 for
database insufficiency, but some reviewers indicated that this uncertainty factor may be
justified for database insufficiencies other than those relevant to immunotoxicity. No other
clear trends emerged from this discussion, though 3 of the 17 reviewers also suggested that
the uncertainty factor for subchronic to chronic exposure duration may be unnecessary.
7.4 Charge Question F.4—Have all the factors influencing susceptibility been clearly
described and accounted for in the assessment?
The reviewers had multiple responses to how EPA identified susceptible populations and
whether additional ones should be considered. Regarding EPA's approach, two reviewers
(AC,MK) indicated that they would have preferred identifying susceptibilities based on
mechanistic arguments. One of the two (MK), for instance, suggested that EPA should have
identified susceptibilities from insights on the most relevant biochemical events and how these
differ among subpopulations. The other reviewer (AC) added that EPA's account of
susceptibilities would have been more convincing if it were based on a systematic evaluation of
specific factors (e.g., interspecies differences in TBG levels and thyroid tissue growth rates).
Similarly, a third reviewer (MR) noted that EPA could address potential susceptibilities directly in
its benchmark dose calculations by using mixture models that explicitly account for susceptibilities
in their calculations.
Other reviewers (NC,TC,TF,DJ) identified the following potential susceptibilities for EPA
to consider in the human health dose-response assessment: genetic variations in NTS across the
population, the elderly, fetuses and neonates who depend on iodide transport across the placenta
or into breast milk, smokers, and people with dietary iodide insufficiencies (particularly pregnant
women). The reviewers briefly discussed the potential implications of developing health
7-15
-------
guidelines that protect against all susceptible populations—a discussion that focused more on
general risk management issues and is not summarized here.
7-16
-------
Table 2
Peer Reviewers' Final Comments on EPA's Proposed Use
of the Brain Morphometry Data as the Point of Departure
Peer Reviewer
Comment
Comments supporting use of the brain morphometry data as the point of departure
Dr. Merle Paule
Dr. Tom Zoeller
Agreed with EPA's choice of the point of departure.
Agreed with EPA's choice of the point of departure.
Comments not supporting use of the brain morphometry data as the point of departure
Dr. Michael Aschner
Dr. Loren Roller
Dr. Gary Williams
Found the brain morphometry studies inconclusive.
Disregarded the reported changes in brain morphometry.
Not convinced that the brain morphometry study identified adverse
effects.
Conditional remarks on EPA 's use of the brain morphometry data
Dr. Michael Collins
Dr. Kannan
Krishnan
Dr. Ronald Wyzga
Did not suggest that EPA reject the brain morphometry data, but was
not convinced that the selected numeric value (0.01 mg/kg/day) was the
actual point of departure. (Note: After reviewing a draft of this report,
this reviewer indicated that he finds the brain morphometry data to be
inconclusive based on the opinions that were expressed at the meeting.
This reviewer did not make this comment at the peer review meeting.)
Suggested that EPA not base the point of departure on brain
morphometry data, unless the Agency can adequately address the
concerns raised at the peer review meeting.
Would have greater confidence in the study if a blinded re-analysis
found the same effects, but did not specify whether the existing data are
an adequate basis for the point of departure.
Note: This table summarizes the peer reviewers' specific summary statements made at the end of the discussions
on the point of departure. The following peer reviewers either did not comment specifically on whether the
brain morphometry data should serve as the basis of the point of departure or commented instead on
whether human data should be used: Dr. William Adams, Nancy Carrasco, M.D., Dr. Thomas Collins, Dr.
Anthony Cox, Dr. Teresa Fan, Dr. David Hoel, Dr. Michael Kohn, Dr. David Jacobson-Kram, and Dr.
Mehdi Razzaghi. Refer to Sections 7.1.2 and 7.1.3 for a more complete discussion of the peer reviewers'
specific comments on the proposed point of departure.
7-17
-------
Table 3
Peer Reviewers' Specific Recommendations on Uncertainty Factors That
EPA Proposed in the RfD Derivation
Peer Reviewer
Nancy Carrasco, M.D.
Dr. Thomas Collins
Dr. Anthony Cox
Dr. David Jacobson-
Kram
Dr. Loren Roller
Dr. Kannan Krishnan
Dr. Merle Paule
Dr. Gary Williams
Dr. Ronald Wyzga
Dr. Tom Zoeller
Comment
Did not support the uncertainty factor for database insufficiency.
Did not support the uncertainty factor for database insufficiency.
Noted that an interspecies uncertainty factor less than 1 might be
defended. Did not support the uncertainty factor for database
insufficiency.
Did not support the uncertainty factor for database insufficiency.
Did not support the uncertainty factor for database insufficiency or
the uncertainty factor for subchronic to chronic exposure.
Did not support the combined uncertainty factor of 10 for database
insufficiency with regards to immunotoxicity (factor of 3) and
subchronic to chronic exposure duration (factor of 3).
Did not support the uncertainty factor for database insufficiency, if
it is based strictly on lack of information on immunotoxicity.
Did not support the uncertainty factor for database insufficiency.
Did not support the uncertainty factor for database insufficiency.
Did not support the uncertainty factor for subchronic to chronic
exposure.
Note: This table summarizes the peer reviewers' specific summary statements made at the end of the discussions
on uncertainty factors. The table identifies recommended changes in the uncertainty factors that EPA
specifically proposed for deriving an RfD. The following peer reviewers did not comment specifically on
the uncertainty factors EPA proposed in Section 7.1.4 of the Revised ERD: Dr. William Adams, Dr.
Michael Aschner, Dr. Michael Collins, Dr. Teresa Fan, Dr. David Hoel, Dr. Michael Kohn, and Dr. Mehdi
Razzaghi. Refer to Section 7.3 for a more complete discussion of the peer reviewers' specific comments on
the proposed uncertainty factors.
7-18
-------
8.0 Responses to Questions in Topic Area G: Risk Characterization
The peer reviewers briefly discussed the human health and ecological risk characterization
statements that EPA presents in Chapters 10.1 and 10.2 of the Revised ERD, respectively. Their
comments on these chapters appear in Sections 8.1 and 8.2, below.
8.1 Comments on the Human Health Risk Characterization
Summarizing the relevant premeeting comments, the designated discussion leader (RW)
indicated that the peer reviewers generally thought the human health risk characterization
adequately summarizes the information originally presented in the Revised ERD, though he noted
that EPA should eventually revise the risk characterization to reflect the peer reviewers' various
findings listed throughout this report.
The peer reviewers made relatively few specific suggestions for improving this section. To
ensure that the risk characterization reflects the current understanding of perchlorate toxicity, one
reviewer (RW) recommended, Chapter 10.1 should acknowledge the diversity of opinion
regarding how to interpret key toxicity studies, particularly for the studies reporting changes in
brain morphometry in rats. For greater perspective on whether perchlorate-related toxicity is
believed to occur in humans, two reviewers (KK,RW) suggested, Chapter 10.1 should include
more information on current human exposure levels, trends in these exposures (e.g., are levels of
perchlorate in drinking water supplies increasing or decreasing?), and relevant effects observed in
humans at various dosage levels. Another reviewer (GW) suggested that Chapter 10.1 document
data on the prevalence of goiter, noting that an increased prevalence of goiter would likely be one
of the first detectable thyroid effects in populations exposed to perchlorate.
8-1
-------
8.2 Comments on the Ecological Risk Characterization
Two reviewers (WA,TF) addressed EPA's characterization of ecological risks, drawing
mainly from comments they raised earlier in the peer review meeting (see Section 5). Though
both reviewers initially found EPA's screening-level risk assessment adequate, their views
changed upon reviewing an ecotoxicological study published after the release of the Revised ERD
(Goleman et al. 2002). The two reviewers' comments addressed three general issues:
• Comments on exposure data. One reviewer (WA) indicated that, although the Revised
ERD correctly focuses on environmental media where perchlorate is expected to occur, the
data available for evaluating environmental exposures are limited. He suggested that future
research efforts focus on characterizing potential exposures more broadly, particularly
exposures in the range over which ecotoxicological effects are observed. Another reviewer
(TF) recommended that EPA clarify its statements on chemical transformation of
perchlorate, rather than asserting that the contaminant is extremely stable, and that more
detailed information on biological transformation may be necessary, particularly as it
applies to potential phytoremediation strategies. Two reviewers (WA,TF) suggested that
EPA characterize the extent to which humans are exposed to perchlorate by consuming
agricultural produce grown in areas with perchlorate contamination, whether domestically
or abroad.
• Comments on aquatic effects assessment. Two reviewers (WA,TF) questioned the
adequacy of EPA's screening-level ecological risk assessment for perchlorate, given that a
recent study (Goleman et al. 2002) suggests that adverse effects may be occurring at
exposure concentrations considerably lower than the threshold (0.6 ppm) EPA originally
proposed for aquatic toxicity. The reviewers recommended that EPA critically review
potential limitations of this study (e.g., implications of the extended duration of the
experiment, presence of considerable adverse effects in the control groups, and relevance
of de-ionized water as an exposure matrix) to determine if its proposed toxicity threshold is
scientifically sound. Based on concerns raised by the recent study, one of the two
reviewers (WA) recommended that EPA's ecological testing strategy focus on life stages
and organisms that may be affected by changes in iodide uptake inhibition. This reviewer
specifically suggested that EPA shift its focus in future studies from invertebrates to
vertebrates. The other reviewer (TF) agreed, and recommended that future studies
examine rooted macrophytes and detritus, which she indicated may be important for dietary
exposures in the aquatic food chain.
• Comments on terrestrial effects assessment. Two reviewers (WA,TF) indicated that the
Revised ERD lacks extensive detail on ecological exposures and risks associated with soils
contaminated with perchlorate, and with the contamination of plant tissues that may result.
8-2
-------
These reviewers recommended that future studies focus on dietary exposure routes for
terrestrial organisms that feed on aquatic vegetation and that have developmental stages
influenced by thyroid hormone production (e.g., mice, voles, ducks).
8-3
-------
9.0 Responses to Questions in Topic Area H: General Comments, Conclusions,
and Recommendations
In the final discussions, 14 peer reviewers presented their overall impressions of the
Revised ERD. Three peer reviewers (Dr. Michael Aschner, Dr. Anthony Cox, and Dr. Michael
Kohn) were not present during these final discussions. A summary of the peer reviewers' final
comments, organized by topic area, follow:
Representation of perchlorate uptake and metabolism. Nancy Carrasco, M.D., suggested
that EPA further research two specific issues regarding perchlorate toxicokinetics. First,
she questioned the validity of EPA's assumption that NTS translocates perchlorate into
thyroid cells. She noted that none of the references cited in the Revised ERD provide
compelling evidence that active translocation of perchlorate occurs. Dr. Carrasco indicated
that even the most recent studies reporting concentrations of perchlorate in the thyroid
(e.g., Yu et al. 2001) are not convincing, largely because the studies do not distinguish
whether perchlorate detected in the thyroid is inside cells or simply bound to them.
Second, Dr. Carrasco recommended that EPA reconsider its assumption that perchlorate is
not metabolized and is "excreted virtually unchanged," because studies that administered
double-labeled radioactive perchlorate to humans suggest that some of the amount ingested
is metabolized.
Human health effects data. Dr. David Hoel made several recommendations for how EPA
can better integrate findings from human health effects studies into the Revised ERD.
First, he recommended that EPA more thoroughly evaluate data from certain human
clinical studies (Greer et al. 2000; Greer et al. 2002 - In Press; Lawrence et al. 2000, 2001)
and perhaps use these data as a basis for the proposed point of departure, provided that it
adequately addresses concerns regarding confounding factors (e.g., dietary iodide levels,
smoking, body weight) and notes the limitations associated with the 14-day exposure
duration. Second, Dr. Hoel recommended that EPA determine whether the remaining
epidemiological studies (i.e., ecological and occupational studies) offer further insight into
the clinical studies, particularly regarding long-term exposures. Finally, Dr. Hoel
suggested that EPA calculate human equivalent exposures for the effect levels observed in
the laboratory animal studies and assess whether humans experience comparable effects at
these equivalent exposures.
Laboratory animal data: immunotoxicity. Dr. Loren Roller indicated that the two
immunotoxicology studies conducted since the 1999 peer review used standard protocols
and validated assays to evaluate both the innate and acquired immune responses,
considering most compartments of the immune system. Dr. Roller noted that the effects
9-1
-------
observed in these studies were generally immuno-stimulatory, or protective, with regards
to host resistance to infectious disease and neoplasia. Though he acknowledged that two
separate experiments (i.e., 14-day and 90-day evaluations) identified contact
hypersensitivity effects, Dr. Roller noted that the observed effects did not follow a clear
dose-response signal. He questioned whether these observed effects are expected to occur
in humans, especially considering that skin rashes have only been observed in patients being
treated for Graves' disease with high doses of perchlorate. Dr. Roller concluded that the
immunotoxicity studies do not provide an adequate basis for deriving a point of departure
or for applying an uncertainty factor for database insufficiency.
Laboratory animal data: thyroid hormone levels. Dr. Thomas Zoeller indicated that the
data on thyroid hormone levels generally support EPA's proposed mode of action for
perchlorate toxicity, even though consistent effects (i.e., decreases in T4 and increases in
TSH) are not observed in all studies. Dr. Zoeller emphasized that the laboratory animal
study with the most rigorous design (Argus 2001) reported the most consistent, dose-
dependent changes in thyroid hormone and TSH levels. For instance, he said, the study
reported hormone levels within typically reported ranges and the researchers properly
applied the diagnostic kits for measuring the serum concentrations. Dr. Zoeller
acknowledged that decrements in T4 were not always observed. He attributed this
unexpected outcome to the measurement limitations of the diagnostic kits: T4 levels in
control groups were near the lowest testing standard, where measurement variability is
greatest. Dr. Zoeller supported EPA's use of ANOVA in its statistical analyses of the
thyroid hormone levels, but questioned the validity of pooling results from separate assays
in the data analysis.
Laboratory animal data: brain morphometry. Dr. Thomas Zoeller indicated that the
recent study of brain morphometry is an important element in EPA's RfD derivation,
because there is only limited evidence that adverse effects occur at the lowest dosage levels
in the other endpoints (e.g., perturbations in thyroid hormone levels, thyroid
histopathology, inhibition of iodide uptake). Dr. Zoeller indicated that most reviewers
questioned the integrity of the brain morphometry data, with overall impressions generally
falling into two categories. Some reviewers concluded that linear measurements of the
brain sections will never be an adequate basis for the analyses of brain morphometry.
Other reviewers agreed that the linear measurements and other aspects of the study design
introduce error into the raw data, but they believed these errors would be evenly
distributed among treatment groups, without any systematic biases; these reviewers
believed that a blinded re-analysis of the linear measurements of the brain slides may
resolve the issue.
Laboratory animal data: neurotoxicity. Building on the comments summarized in the
previous bulleted item, Dr. Merle Paule acknowledged that one peer reviewer (Dr. Michael
Aschner) had serious concerns regarding EPA's proposed use of the brain morphometry
data for the point of departure. EPA's and other peer reviewers' comments on the study
9-2
-------
methodology, however, convinced Dr. Paule that the brain morphometric results are valid.
Further, he noted that EPA's Bayesian analyses of the data on spontaneous motor activity
behavior identified effects in both studies that evaluated this endpoint (Argus 1998;
Bekkedal 2000).
Laboratory animal data: thyroid pathology. Dr. Gary Williams indicated that
administration of low doses of per chlorate to rats and rabbits produced adaptive changes in
thyroid histopathology, namely colloid depletion and epithelial hypertrophy, while
administration of higher doses (i.e., 1.0 mg/kg/day and higher) produced hyperplastic
responses to TSH stimulation. Dr. Williams added that dosage at much higher levels
(30 mg/kg/day) produced follicular cell neoplasms in rats. He noted that the thyroid
physiology in rats renders the species particularly susceptible to such metabolic
perturbations. Dr. Williams concluded that these various observations indicate that
perchlorate is an anti-thyroid agent that induces thyroid neoplasms in rats—an effect that
he believed has a LOAEL of 1.0 mg/kg/day. Noting that other agents with comparable
anti-thyroid effects have induced thyroid tumors in rats but not in humans, Dr. Williams
found no compelling reason to believe that the currently reported environmental exposure
levels to perchlorate would cause thyroid cancer in humans.
Laboratory animal data: reproductive toxicity. Dr. Thomas Collins indicated that a two-
generation reproduction study (Argus 1999) had been completed since the 1999 peer
review—a study that he found scientifically sound, except for some methodological
concerns (see pages 39-41 of the premeeting comments in Appendix C). Dr. Collins
supported EPA's interpretations of the study, but suggested that the Agency further
evaluate the apparent dose-dependent decreases in sperm density and daily sperm
production levels.
Laboratory animal data: developmental toxicity. Dr. Michael Collins briefly summarized
the two studies completed since the 1999 peer review that examined developmental
toxicity. Dr. Collins indicated that the authors of one study reported a NOAEL of 30
mg/kg/day for developmental toxicity in rats (which EPA interpreted as being a LOAEL)
and the authors of another study reported a NOAEL of at least 100 mg/kg/day for
developmental toxicity in rabbits. Dr. Collins noted that the NOAELs for these studies are
orders of magnitude higher than the doses EPA considered as possible bases for the point
of departure. More generally, he emphasized that the available studies suggest that fetuses
are more susceptible to perchlorate toxicity than are the maternal organisms.
Laboratory animal data: genetic toxicology. Dr. David Jacobson-Kram indicated that no
new genotoxicity studies have been published since the 1999 peer review. He agreed with
the findings of the previous peer review panel, which concluded that the battery of
available genetic toxicology tests suggests that perchlorate is not genotoxic. Because of
this finding, he supported EPA's nonlinear model for evaluating perchlorate
carcinogenicity.
9-3
-------
Ecological risk assessment and evidence for indirect exposure. Dr. Teresa Fan indicated
that the data currently available on perchlorate exposure and effects are not sufficient for
conducting a screening-level ecological risk assessment. She listed several issues that
require further research before EPA can state its conclusions (e.g., toxicity thresholds) with
confidence: EPA must further evaluate dietary exposures, particularly from aquatic
macrophytes to herbivorous aquatic organisms and from terrestrial vegetation to
herbivorous mammals; focus ecotoxicological studies on species expected to develop
adverse effects resulting from thyroid iodide uptake inhibition (e.g., amphibian
metamorphosis studies and studies of egg-bearing female fish); gather data on perchlorate
body burdens to allow for extrapolations between laboratory toxicity tests and field studies
(this would enable EPA to resolve apparent discrepancies in the plant bioaccumulation
data); and revise text implying that biological and chemical transformation of perchlorate
does not occur. Because of these and other data gaps, Dr. Fan did not think EPA could
conclude that perchlorate is not expected to have effects on populations and species
richness. Dr. William Adams agreed with Dr. Fan's final comments.
PBPK modeling, selection of dose metric, and inter species extrapolations. Dr. Kannan
Krishnan indicated that the structure, basic equations, and physiological parameters EPA
used in the PBPK models are generally appropriate. However, the peer reviewers
questioned three of the models' assumptions regarding cellular uptake processes and their
associated kinetics: Is passive diffusion of iodide a relevant process? Does perchlorate
actively translocate into cells, or does it simply bind to them? Are the kinetics of active
transport of iodide best described as competitive inhibition? Dr. Krishnan noted that errors
in the representation of these cellular uptake processes and their kinetics may have only
marginal impacts on the predicted serum concentrations of perchlorate. Nonetheless, he
recommended that EPA refine the PBPK models to address the reviewers' concerns
regarding cellular uptake processes. Dr. Krishnan indicated that the AUC of serum
perchlorate is an acceptable dose metric, though he noted that other reviewers were not
convinced that this dose metric is the best predictor of adverse health effects.
Human health dose-response assessment. Dr. Tom Collins said the majority of reviewers
agreed that EPA's proposed mode of action is consistent with the observed effects in
laboratory animals and humans. He also said the reviewers had differing opinions on the
proposed point of departure of 0.01 mg/kg/day and whether it should be based on human
health effects data. Dr. Collins noted that the peer reviewers also had various opinions on
the appropriate selection of uncertainty factors for deriving an RfD from the brain
morphometry studies, though nearly every reviewer agreed that application of an
uncertainty factor for database insufficiencies in immunotoxicity was not warranted. He
listed many additional specific comments that the reviewers raised on finer points of the
dose-response assessment.
9-4
-------
Recommendations for further action. The reviewers listed several recommendations for
future research on perchlorate toxicity:
Conduct a blinded re-analysis of the brain sections from the brain morphometry
studies (multiple reviewers).
- Develop valid and validated endpoints of thyroid hormone action on brain
development (TZ).
Evaluate the relative impacts of anti-thyroid dietary components, the effects of
exposures to lower perchlorate doses (0.001 mg/kg/day), the potential for
progression of tumors induced by exposures to ammonium perchlorate, and the
potential impacts of in utero exposure (GW).
- Conduct replications of laboratory animal studies during the same time of year to
prevent seasonality in rodent physiology from masking notable results; incorporate
more sophisticated neurobehavioral endpoints into future developmental studies
(MP).
Conduct an additional multi-generational developmental study that evaluates a full
suite of neurobehavioral, neurodevelopmental, and thyroid histopathological
endpoints (LK).
Consider conducting an epidemiological study on the prevalence of goiter among
populations exposed to perchlorate (DH).
- Conduct a more extensive chronic exposure study (i.e., with a "womb-to-tomb"
design) and another study of potential in utero programming (TC).
Characterize the pharmacodynamics by which iodide uptake inhibition leads to
neurodevelopmental and neoplastic sequelae (MC).
- Ascertain unequivocally whether active translocation of perchlorate occurs, and
characterize potential adverse effects resulting from prolonged exposure to
perchlorate (NC).
- Verify whether ecological effects recently reported in a laboratory study (Goleman
et al. 2002) are expected to occur in the field (WA).
9-5
-------
10.0 References
Aldenberg, T.; Slob, W. 1993. Confidence limits for hazardous concentrations based on
logistically distributed NOEC toxicity data. Ecotoxicol. Environ. Saf 25:48-63.
Anbar, M.; Guttman, S.; Lewitus, Z. 1959. The mode of action of perchlorate ions on the iodine
uptake of the thyroid gland. Int. J. Appl. Radiat. Isot. 7:87-96.
Argus Research Laboratories. 1998a. Oral (drinking water) developmental toxicity study of
ammonium perchlorate in rabbits [report amendment: September 10]. Protocol no. 1613-002.
Horsham, PA: Argus Research Laboratories, Inc..
Argus Research Laboratories. 1998b. A neurobehavioral developmental study of ammonium
perchlorate administered orally in drinking water to rats [report amendment: July 27]. Protocol
no. 1613-002. Horsham, PA: Argus Research Laboratories, Inc.
Argus Research Laboratories. 1999. Oral (drinking water) two-generation (one litter per
generation) reproduction study of ammonium perchlorate in rats. Protocol no. 1416-001.
Horsham, PA: Argus Research Laboratories, Inc.
Argus Research Laboratories. 2000. Oral (drinking water) developmental toxicity study of
ammonium perchlorate in rats. Protocol no. 1416-003D. Horsham, PA. Argus Research
Laboratories, Inc.
Argus Research Laboratories. 2001. Hormone, thyroid and neurohistological effects of oral
(drinking water) exposure to ammonium perchlorate in pregnant and lactating rats and in fetuses
and nursing pups exposed to ammonium perchlorate during gestation or via maternal milk.
Protocol no. ARGUS 1416-003. Horsham, PA: Argus Research Laboratories, Inc.
Bekkedal, M.Y.V.; Carpenter, T.; Smith, J.; Ademujohn, C.; Maken, D.; Mattie, D.R. 2000. A
neurodevelopmental study of the effects of oral ammonium perchlorate exposure on the motor
activity of pre-weanling rat pups. Report no. TOXDET-00-03. Wright-Patterson Air Force
Base, OH: Naval Health Research Center Detachment, Neurobehavioral Effects Laboratory.
Brown-Grant, K. 1966. Failure of orally administered perchlorate to affect deciduoma formation
or pregnancy in the rat. J. Reprod. Fertil. 12:353-357.
Brown-Grant, K.; Sherwood, M.R. 1971. Viability of the rat blastocyst following the oral
administration of potassium perchlorate or potassium iodide to the mother. J. Reprod. Fertil.
27:265-267.
10-1
-------
BRT-Burleson Research Technologies, Inc. 2000a. Ammonium perchlorate: effect on immune
function. Quality assurance audit: study no. BRT 19990524—plaque-forming cell (PFC) assay;
study no. BRT 19990525—local lymph node assay (LLNA) in mice. Raleigh, NC.
BRT-Burleson Research Technologies, Inc. 2000b. Addendum to study report: ammonium
perchlorate: effect on immune function [with cover letter dated August 31 from G.R. Burleson].
Raleigh, NC.
BRT-Burleson Research Technologies, Inc. 2000c. Ammonium perchlorate: effect on immune
function. BRT 19990524 study protocol: plaque-forming cell (PFC) assay; BRT 19990525 study
protocol: local lymph node assay (LLNA) in mice. Raleigh, NC.
Condike, BJ. 2001. Perchlorate data in fish and plants [letter with attachments to Annie M.
Jarabek]. Fort Worth, TX: Department of the Army, Fort Worth District, Corps of Engineers;
December 21.
Crump, C.; Michaud, P.; Tellez, R.; Reyes, C.; Gonzalez, G.; Montgomery, E.L.; Crump, K.S.;
Lobo, G.; Becerra, C.; Gibbs, J.P. 2000. Does perchlorate in drinking water affect thyroid
function in newborns or school-age children? J. Occup. Environ. Med. 42:603-612.
EA Engineering, Science, and Technology. 1999. Results of algal toxicity testing with sodium
perchlorate. Sparks, MD: EA Engineering, Science, and Technology, Inc.
EA Engineering, Science, and Technology. 2000. Results of chronic toxicity testing with sodium
perchlorate using Hyalella azteca and Pimephalespromelas. Report number 3505. Sparks, MD:
EA Engineering, Science, and Technology, Inc.
EPA. 1998. Perchlorate environmental contamination: toxicological review and risk
characterization based on emerging information. Review draft. NCEA-1-0503. U.S.
Environmental Protection Agency, Office of Research and Development. December, 1998.
EPA. 2002. Perchlorate environmental contamination: toxicological review and risk
characterization. External review draft. NCEA-1-0503. U.S. Environmental Protection Agency,
Office of Research and Development. January 16, 2002.
Eskandari et al. 1997. Thyroid Na+/I" symporter: mechanism, stoichiometry, and specificity. J.
Biol. Chem. 272:27230-27238.
Goleman, W.L.; Carr, J.A.; Anderson, T.A. 2002. Environmentally relevant concentrations of
ammonium perchlorate inhibit thyroid function and alter sex ratios in developing Xenopus laevis.
Environ. Toxicol. Chem. 31: in press.
Gol stein et al. 1995. The iodide channel of the thyroid. Am. J. Phsyiol. 268:C11-C118.
10-2
-------
Greer, M.A.; Goodman, G.; Pleus, R.C.; Greer, S.E. 2000. Does environmental perchlorate
exposure alter human thyroid function? Determination of the dose-response for inhibition of
radioiodine uptake. In: Abstracts of the 12th International Thyroid Congress; October; Kyoto,
Japan. Endocrine J. 47(suppl.): 146.
Greer, M.A.; Goodman, G.; Pleus, R.C.; Greer, S.E. 2002 - In Press. Health effects assessment
for environmental perchlorate contamination: the dose-response for inhibition of thyroidal
radioiodine uptake in humans. In press, Environmental Health Perspectives. Submitted July 24,
2001; revised February 11, 2002.
Keil, D.; Warren, D.A.; Jenny, M.; EuDaly, J.; Dillard, R. 1999. Effects of ammonium
perchlorate on immunological, hematological, and thyroid parameters. Report no. DSWA01-97-
0008. Charleston, SC: Medical University of South Carolina, Department of Medical Laboratory
Sciences.
Lampe, L.; Modis, L.; Gehl, A. 1967. Effect of potassium perchlorate on the foetal rabbit
thyroid. ActaMed. Acad. Sci. Hung. 23:223-232.
Lawrence, J.E.; Lamm, S.H.; Pino, S.; Richman, K.; Braverman, L.E. 2000. The effect of short-
term low-dose perchlorate on various aspects of thyroid function. Thyroid 10:659-663.
Lawrence, J.E.; Lamm, S.H.; Braverman, L.E. 2001. Low dose perchlorate (3 mg daily) and
thyroid function [letter]. Thyroid 11:295.
Merrill, E. 200la. Consultative letter, AFRL-HE-WP-CL-2001-0004, QA/QC audit report for
the study of perchlorate pharmacokinetics and inhibition of radioactive iodine uptake (RAIU) by
the thyroid in humans (CRC protocol #628) [memorandum with attachments to Annie Jarabek].
Wright-Patterson AFB, OH: Air Force Research Laboratory; May 10.
Parsons Engineering Science. 2001. Scientific and technical report for perchlorate biotransport
investigation: a study of perchlorate occurrence in selected ecosystems. Interim final. Contract
no. F41624-95. Austin, TX.
Postel, S. 1957. Placental transfer of perchlorate and triiodothyronine in the guinea pig.
Endocrinology 60:53-66.
Schwartz, J. 2001. Gestational exposure to perchlorate is associated with measures of decreased
thyroid function in a population of California neonates [thesis]. Berkeley, CA: University of
California.
Sol din, O.P.; Braverman, L.E.; Lamm, S.H. 2001. Perchlorate clinical pharmacology and human
health: a review. Ther. Drug Monit. 23:316-331.
10-3
-------
Springborn Laboratories. 1998. A 90-day drinking water toxicity study in rats with ammonium
perchlorate: amended final report [amended study completion date: June 3]. Study no. 3455.1.
Spencerville, OH: Springborn Laboratories, Inc.
-------
United
DC
------- |