Science Review Office of Inspector General Scientific Analysis of Perchlorate April 19, 2010


jriCEi S?j



U.S. ENVIRONMENTAL PROTECTION AGENCY

OFFICE OF INSPECTOR GENERAL

Catalyst for Improving the Environment

Science Review

Office of Inspector General
Scientific Analysis of Perchlorate

Report No. 10-P-0101
April 19, 2010

-------
Author and Reviewers

Author

Michael Wilson
Toxicol ogist

U.S. Environmental Protection Agency
Office of Inspector General (OIG)
Office of Program Evaluation (2460T)
1200 Pennsylvania Ave, NW
Washington, DC 20460

Review

Through solicitation number RFQ261223 (dated February 8, 2008), the OIG contracted with ICF
Incorporated, LLC (ICF) under contract number GS-10F-0124J, to conduct a 6-week technical
review of a working draft of the OIG Scientific Analysis of Per chlorate. The purpose of the
technical review was to provide a scientific critique of the essential features of the OIG's
application of a cumulative risk assessment to this public health issue.

-------
U.S. Environmental Protection Agency
Office of Inspector General

At a Glance

10-P-0101
April 19, 2010

Catalyst for Improving the Environment

Why We Did This Review

Stakeholders often dispute
the scientific support for a
particular U.S. Environmental
Protection Agency (EPA)
decision or program action.
We believe there is a particular
need for impartial reviews of
EPA's regulatory toxicology.
Therefore, as a prototype for
this work, we conducted an
independent, scientific review
of the risk assessment process
and procedures used by EPA to
develop and derive the
perchlorate reference dose
(RfD).

Background

On February 18, 2005, EPA
issued a perchlorate RfD that
corresponds to a drinking water
equivalent level of 24.5 parts
per billion. A regulatory
determination is pending on
whether to issue a National
Primary Drinking Water
Regulation.

For further information, contact
our Office of Congressional,
Public Affairs and Management
at (202) 566-2391.

To view the full report,
click on the following links:

www.epa.qov/oiq/reports/2010/
20100419-10-P-0101.pdf

www.epa.qov/oiq/reports/2010/
20100419-10-P-0101 appD.pdf

www.epa.qov/oiq/reports/2010/
20100419-10-P-0101 appE.pdf

Office of Inspector General Scientific Analysis
of Perchlorate

What We Found

EPA should conduct a cumulative risk assessment to reduce the uncertainty in
characterizing the public health risk posed by perchlorate. A cumulative risk
assessment is the current state-of-the-art technique for evaluating the public
health risk from multiple stressors. Over the last two decades, EPA has received
numerous recommendations to improve environmental risk assessments. In
1997, EPA Administrator Carol Browner issued guidance directing EPA to
embrace the cumulative risk assessment approach on all future major risk
assessments. Although directed to improve the environmental risk assessment
process, EPA continues to rely on the outdated single chemical risk assessment
approach, originally developed in 1954, to characterize the risk posed by
perchlorate. The single chemical risk assessment approach fails to address
known sources of scientific uncertainty, which lowers the confidence in the
perchlorate RfD.

Against established EPA risk assessment procedures, EPA derived the
perchlorate RfD from a nonadverse biological effect instead of an adverse effect.
The perchlorate RfD protects against all human biological effects from exposure,
which is a stricter public health criterion than limiting environmental exposure to
protect against adverse effects in humans. This shift in risk management
constitutes a significant change in environmental policy.

Based on our scientific analysis, perchlorate is only one of several chemicals that
stress the thyroid's ability to uptake iodide. The other sodium iodide symporter
(NIS) stressors include thiocyanate, nitrate, and the lack of iodide. All four of
these NIS stressors meet EPA's risk assessment guidance for conducting a
cumulative risk assessment using the dose-addition method. Our analysis
implemented a cumulative risk assessment that found the following: 1) the risk
from each of the four NIS stressors is not equal; 2) EPA's perchlorate RfD is
conservative and protective of human health, and further reducing the perchlorate
exposure below the RfD does not effectively lower risk; 3) increasing maternal
total iodide intake to healthy levels will reduce the frequency and severity of
permanent mental deficits in children; and 4) correcting moderate and mild
iodide deficiency occurring in about 29 percent of the U.S. pregnant and nursing
population is the most effective approach for reducing risk.

-------
^tDsx

* O \
*

UNITED STATES ENVIRONMENTAL PROTECTION AGENCY

%	»	WASHINGTON, D.C. 20460

% £

PRO"**'

OFFICE OF
INSPECTOR GENERAL

April 19, 2010

MEMORANDUM

SUBJECT:	Office of Inspector General Scientific Analysis of Perchlorate

Report No. 10-P-0101

FROM:	Bill A. Roderick 		

Acting Inspector General

TO:	Lisa P. Jackson

Administrator

The Office of the Inspector General (OIG) is providing this science report on our cumulative risk
assessment of perchlorate and the other sodium (Na+)/iodide (I") symporter stressors. This report
represents the opinion of the OIG and does not necessarily represent the final position of the U.S.
Environmental Protection Agency (EPA). EPA managers will make final determinations on
matters in this report. This report contains no recommendations and no further action is required.

The estimated cost of this report is $263,907. This amount incorporates both the OIG staff cost
and the contracting cost for the technical review. The OIG staff cost is $214,322 and was
calculated by multiplying the project's staff days by the applicable daily full cost billing rates in
effect at the time. The contracting cost for the technical review was $49,585.

If you or your staff have any questions regarding this report, please contact me at 202-566-0847 or
roderick.bill@epa.gov, or Eric Lewis at 202-566-2664 or lewis.eric@epa.gov.

-------
Office of Inspector General
Scientific Analysis of Perchlorate

10-P-0101

Table of Contents

Section	Page

Executive Summary		1

1	Defining the Public Health Issue		10

2	OIG's Approach to the Scientific Analysis of Perchlorate		14

2.1	Introduction		14

2.2	Purpose		15

2.3	Calls to Use Cumulative Risk Assessment		16

2.4	Use of Mechanistic Toxicology		19

2.5	EPA Risk Assessment Procedures		20

2.5.1	EPA Risk Assessment Guidance		20

2.5.1.1 Applying the Dose-Addition Method		21

2.5.2	EPA Development of Cumulative Risk Assessments		22

2.6	Implementing a Cumulative Risk Assessment		23

2.6.1 Aspects in Implementing a Cumulative Risk Assessment		24

2.7	Summary of the OIG Approach		28

3	Sodium (Na+)/lodide Symporter (NIS) Inhibitors		30

3.1	Background		30

3.2	Relative Potencies of Other NIS Inhibitors		30

3.3	Human Thiocyanate Exposure		31

3.4	Human Nitrate Exposure		32

3.5	Human Perchlorate Exposure		33

4	Modeling the NIS Stressors' Influence on the Total Iodide Uptake (Till)		35

4.1	Determining the Total NIS Inhibition Load Acting on the Human Body		35

4.2	Serum Perchlorate Equivalant Concentration (SPEC)		35

4.3	Oral Perchlorate Equivalent Concentration (OPEC) 		37

4.4	Tonacchera NIS Model of Competitive Inhibition		38

5	Deriving a Reference Dose (RfD)		40

5.1 Issues with the Single Chemical Risk Assessment of Perchlorate		42

5.1.1	Defining an RfD		42

5.1.2	Evaluation of the Application of Uncertainty Factors (UF)

for Deriving the Perchlorate RfD		45

5.1.3	First Adverse Effect: Hypothyroidism or Hypothyroxinemia		48

5.1.4	Comparison of the NAS Unconventional RfD with a

Conventionally Derived Excess NIS Inhibition RfD		49

5.1.4.1 Calculation of an Excess NIS Inhibition RfD Using
the Adverse Thyroid Health Effects Observed in

Electroplating Workers Exposed to Cyanide in Adult Men		51

i

-------
Office of Inspector General	10-P-0101

Scientific Analysis of Perchlorate

Table of Contents (continued)

Section	Page

5.1.4.2	Calculation of an Excess NIS Inhibition RfD Using

the Adverse Thyroid Health Effects from the Ingestion of

Thiocyanate-Preserved Cow's Milk by Adult Women		55

5.1.4.3	Calculation of an Excess NIS Inhibition RfD Using

the Adverse Effect Hypertrophy Observed in School-aged

Children Drinking Nitrate-Contaminated Drinking Water		61

6	Application of the Tonacchera Model to the Chilean Epidemiological Studies ...	66

6.1	Tellez Study		66

6.2	Crump Study		75

6.3	Conclusions		77

7	Assessment of Total NIS Inhibitor Exposure		78

7.1	Perchlorate Exposure		78

7.1.1	Adults		78

7.1.2	Fetuses		80

7.1.3	Nursing Neonates		81

7.2	Total NIS Inhibitor Exposure		83

7.2.1	Total NIS Inhibition Load Acting on Adults		83

7.2.2	Total NIS Inhibition Load Acting on Fetuses		87

7.2.3	Total NIS Inhibition Load Acting on Nursing Infants		90

7.2.4	Total NIS Inhibition Load Acting on Non-nursing Infants		92

8	Risk Characterization of NIS Stress		94

8.1	Hazard Characterization		96

8.2	Dose-Response Assessment		110

8.3	Exposure Assessment		116

9	Till Cumulative Risk Assessment		129

9.1	Corroboration of the Tonacchera Model with Effects Observed in Humans		130

9.1.1	NAS Hypothyroidism Statement		132

9.1.2	Hypothyroxinemia in Men Exposed to Excess Thiocyanate		133

9.1.3	Hypothyroxinemia in Women Exposed to Excess Thiocyanate		136

9.1.4	Perchlorate PBPK Model and the Greer Perchlorate Exposure Study..	138

9.1.5	Braverman Perchlorate Occupational Exposure Study		145

9.1.6	Qualitative Risk Assessment of NIS Stressors		148

9.2	Assessing the Contribution of Each NIS Stressor on the TIU		152

9.2.1	Evaluation of the Amount of NIS Inhibition Contributed by

Each NIS Inhibitor		152

9.2.2	Evaluation of %TIU as a Function of Total NIS Inhibition Load		153

ii

-------
Office of Inspector General
Scientific Analysis of Perchlorate

10-P-0101

Table of Contents (continued)

Section	Page

9.2.3	Evaluation of the Role of Iodide Nutrition

on the Ability to Tolerate NIS Inhibition		158

9.2.4	Evaluation of the Relative Influence of Each NIS Stressor		161

9.2.5	Assessing the Effectiveness of Lowering the Total

NIS Inhibition Load to Compensate for Poor Iodide Nutrition		163

9.3	Application of UFs in a Multifactorial Cumulative Risk Assessment		164

9.4	Determining %TIU Levels of Concern 		167

9.4.1	Identifying a %TIU(Loael) in Pregnant Women

to Prevent Subtle Mental Deficits in Children		167

9.4.2	Identifying a %TIU(noael) in Pregnant Women

to Prevent Subtle Mental Deficits in Children		168

9.4.3	Identifying a %TIU(RfD) in Pregnant Women

to Prevent Subtle Mental Deficits in Children		170

9.5	Calculating the %TIU in Pregnant Women

at Various Perchlorate Exposures		172

10	Approaches to Address this Public Health Issue		175

10.1	%TIU Exposure Levels of Concern		175

10.2	Remedy Approach		175

10.2.1 Exposure Management of Each NIS Stressor		176

10.2.1.1	Lack of Iodide Stressor		176

10.2.1.1.1	Iodide Supplementation During Pregnancy

and Lactation		177

10.2.1.1.2	Participation of the Medical Community		178

10.2.1.2	Thiocyanate Stressor		180

10.2.1.3	Nitrate Stressor		181

10.2.1.4	Perchlorate Stressor		183

11	Conclusion		185

Appendices

A "Whole Mixture" Approach to the Cumulative Risk Assessment of Perchlorate
B References

C List of Acronyms and Abbreviations

D Scientific Comments Received on the OIG Scientific Analysis of Perchlorate
(External Review Draft)

E OIG Response to Comments on OIG Scientific Analysis of Perchlorate
(External Review Draft)

F Distribution

-------
10-P-0101

Executive Summary
Background

Our science review evaluates the public health risk from the sodium (Na+)/iodide (I")
symporter (NIS) stressors (e.g., perchorate) that can disrupt the uptake of iodide by the thyroid.
The OIG Scientific Analysis of Perchlorate (the OIG Analysis) documents our review. The
following is background information on perchlorate and other NIS stressors:

• Perchlorate induces human toxicity by blocking the uptake of iodide into the thyroid. A
protein called NIS mediates iodide transport from the blood into the thyroid. Since
perchlorate acts to block the uptake of iodide, perchlorate is an NIS inhibitor. The
perchlorate mechanism of toxicity is that excess perchlorate exposure lowers the total
iodide uptake (TIU) of the thyroid. A prolonged low TIU decreases the supply of iodide
in the thyroid and eventually results in the decreased production of thyroid hormones
thyroxine (T4) and triiodothyronine (T3). Maintaining an adequate supply of maternal T4
during pregnancy is critical for proper fetal brain development. A low TIU during
pregnancy is associated with permanent mental deficits in the children of those mothers.
The more severe the maternal TIU deficiency, the greater will be the frequency and
severity of the permanent mental deficits in their children.

• In addition to perchlorate, diet exposes humans to other NIS inhibitors that also lower the
amount of iodide uptake by the thyroid. The other common dietary NIS inhibitors are
thiocyanate and nitrate. All NIS inhibitors share the same mechanism of toxicity as
perchlorate - blocking the uptake of iodide by the thyroid.

• The risk to public health is from the combined amount of NIS inhibition acting on the
body from all three NIS inhibitors (thiocyanate, nitrate, and perchlorate). This combined
total NIS inhibition load acting on the thyroid determines the TIU level and the
subsequent potential for an adverse outcome. The combined NIS inhibition load is
known as the total goitrogen load.

• In addition to NIS inhibitors, the amount of the essential dietary nutrient iodide directly
affects the amount of iodide uptake by the thyroid in humans. For an environmental
pollutant, increased exposure is associated with increased frequency and severity of
adverse effect(s) on the body. Iodide, however, becomes an NIS stressor when there is an
insufficient amount of iodide in the diet to allow for an adequate uptake of iodide by the
thyroid. Therefore, the lack of iodide in the diet (i.e., iodide deficiency) is an NIS
stressor and shares the same mechanism of toxicity as the NIS inhibitors - preventing the
thyroid from taking up an adequate amount of iodide. Therefore, both the combined total
NIS inhibition load acting on the thyroid and the iodide nutritional level determines the
TIU level and the subsequent potential for an adverse outcome.

• Diet continuously exposes all humans to all four NIS stressors (thiocyanate, nitrate,
perchlorate, and iodide). Common vegetables in the diet are a significant and natural
source of thiocyanate. Foods particularly rich in thiocyanate include cabbage, broccoli,

-------
10-P-0101

Brussels sprouts, corn (maize), turnips, rapeseed, mustard seed, cauliflower, cabbage,
radishes, spinach, tomatoes, and milk. Although diet is an important source of
thiocyanate, cyanide produced and absorbed through cigarette smoking is also a
significant source of thiocyanate. Nitrate is also common in food. Nitrate occurs in
green leafy vegetables, is added to processed meats as a preservative, and is found in
milk. In addition to dietary nitrate, both surface and ground sources of drinking water
commonly contain nitrate due to the agricultural use of nitrate fertilizers. According to
the Food and Drug Administration (FDA) Total Dietary Study, 46% to 59% of the total
estimated intake of perchlorate comes from vegetables and dairy foods. Finally, the FDA
Total Dietary Study also identifies dairy and grain as the most significant sources of
dietary iodine for all groups of adults. Therefore, the degree of dietary exposure to each
of the four NIS stressors (not just perchlorate) is needed to determine an individual's TIU
and the likelihood for an adverse outcome.

• Other potential sources of perchlorate exposure are the Department of Defense/aerospace
sector (e.g., rocket propellant), commercial products (i.e., safety flares and some
fireworks (up to 70% content)), agricultural use (i.e., perchlorate is a contaminant in
Chilean nitrate fertilizer (CNF)), and natural atmospheric production. The Department of
Defense/aerospace sector accounts for about 90% of the annual U.S. consumption of
perchlorate. The burning (i.e., use) of rocket propellants, safety flares, and perchlorate-
containing fireworks does not result in a significant perchlorate exposure; 0.05% or less
of the original perchlorate is left after use because the fire destroys the perchlorate.
However, the improper disposal of unused rocket propellants, safety flares, and
perchlorate containing fireworks allows the unused perchlorate to get into the
groundwater and surface water, resulting in human perchlorate exposure upon ingestion
of contaminated drinking water. Although this type of improper disposal into water
sources would generate "hot spots" of perchlorate exposure within the population, "hot
spots" would not generate the relatively uniform background of perchlorate exposure that
is observed across the entire U.S. population. Natural production of perchlorate by the
oxidation of chloride by lightning or ozone is not a significant source of drinking water
contamination because rainfall contains a mean concentration of perchlorate of 0.015
|ig/L (max. 0.2 |ig/L).

In 2002, the U.S. Environmental Protection Agency (EPA) issued a draft perchlorate risk
assessment titled Perchlorate Environmental Contamination: Toxicological Review and Risk
Characterization (NCEA-1-1053) that proposed a perchlorate reference dose (RfD) of 0.00003
milligram/kilogram-day (mg/kg-day). EPA's proposed RfD corresponds to a drinking water
equivalent level (DWEL) of 1 part per billion (ppb). EPA's draft risk assessment relied
primarily on a rat study conducted by Argus Research Laboratories in 2001. Due to the
scientific controversy surrounding the concentration at which perchlorate should be regulated,
EPA and others1 charged the National Academy of Sciences (NAS) Committee to Assess the
Health Implications of Perchlorate Ingestion (the NAS Committee) with assessing the current
state of the science regarding potential adverse effects of disruption of thyroid function by
perchlorate in humans and laboratory animals at various stages of life. Part of the NAS

1 The U.S. Department of Defense (DOD), U.S. Department of Energy (DOE), and the National Aeronautics and
Space Administration (NASA).

-------
10-P-0101

Committee's charge was to determine whether EPA's 2002 draft single chemical risk assessment
of perchlorate was consistent with the current scientific evidence. In 2005, the NAS Committee
rejected EPA's approach of using the results from the Argus rat studies to derive the perchlorate
RfD.

In January 2005, the NAS Committee recommended an RfD of 0.0007 mg/kg-day based
on the results observed in the Greer human perchlorate dose response study.2 The NAS
Committee included a 10-fold uncertainty factor (UF) in the RfD to protect the most sensitive
population, the fetuses of pregnant women who have hypothyroidism or iodide deficiency. On
February 18, 2005, EPA established an RfD of 0.0007 mg/kg-day, which corresponds to a
DWEL of 24.5 ppb. EPA is currently in the process of making a regulatory determination for
perchlorate, which is a formal decision on whether to issue a National Primary Drinking Water
Regulation.

In the May 1, 2007, Federal Register (72 FR 24016), EPA indicated that it needed
additional information on perchlorate exposure to determine a value for the relative source
contribution of perchlorate. EPA must set a perchlorate relative source contribution value before
it can make a regulatory determination on whether regulating perchlorate in drinking water
presents a meaningful opportunity for health risk reduction. The relative source contribution will
determine how much perchlorate exposure comes from food and water.

Purpose of Science Review

The OIG is frequently requested to examine the scientific support for a particular EPA
decision or program action. Stakeholders often want an entity without a political or scientific
bias to independently evaluate the scientific process and procedures used by the Agency. We
believe there is a particular need for this type of independent review of regulatory toxicology.
Generally, we have been unable to provide this service because we have a limited capacity to do
this kind of work. Additionally, cutting edge science is many times a difficult subject to address.
For this project, we had on staff a toxicologist with the requisite experience and training to
deliver this service.

In conducting this scientific analysis, we sought to use the state of the art techniques and
to comply with EPA's risk assessment requirements so that the validity of our evaluation could
be defended. We think we have done that. We expected to be challenged and we were. We
believe we have responded fairly to all scientific comments and reached a reasonable conclusion.
However, we know that other scientific entities will want to contest our findings. In our
response to those organizations and people we explain why we think the data we used was valid
and we identified the short comings of other data for use in an environmental risk assessment.
Stakeholders should carefully consider all points before making a value judgment on the merits
of this report.

2 The Greer study is a human perchlorate dose response study in which 37 male and female volunteers were given
perchlorate in drinking water at 0.007, 0.02, 0.1, or 0.5 mg/kg-day for 14 days. The study results identified a no-
observed-effect-level (NOEL) of 0.007 mg/kg-day. The NAS Committee used this NOEL value to derive its
recommend perchlorate RfD.

-------
10-P-0101

The potential regulation of perchlorate in drinking water by EPA has been a contentious
and divisive environmental issue. The congressional bills H.R. 1747 and S. 150, which would
amend the Safe Drinking Water Act to require EPA to regulate perchlorate in drinking water,
reflect the interest in addressing this public health issue. We analzyed the science supporting
EPA's perchlorate RfD. The completion of a human health risk assessment that derives a
perchlorate RfD is a key element in the process of making a regulatory determination. Of
particular concern is whether EPA's perchlorate RfD is protective of human health at all life
stages. Since EPA's final perchlorate RfD is taken directly from the NAS Committee's
recommended perchlorate RfD of 0.0007mg/kg-day, we analyzed whether the NAS Committee's
recommended RfD is protective of human health at all life stages. Further, we evaluated the
following scientific issues to assess the effectiveness of EPA's risk assessment procedures:

1.	The NAS Committee used an unconventional approach to derive its recommended
perchlorate RfD by using a no-effect-level, and asserted that this was a conservative,
health-protective approach to the perchlorate risk assessment.

2.	The NAS Committee used a single UF of 10 for intraspecies variation to derive its
recommended perchlorate RfD. The NAS Committee stated that its recommended
perchlorate RfD needed no additional UFs.

3.	The NAS Committee stated that the first adverse effect of perchlorate is hypothyroidism.
The Committee stated, "Defining the adverse effect is important because it influences
how the RfD is derived and ultimately the value of the drinking-water standard."

4.	EPA's charge directed the NAS Committee to assess the potential adverse effects from
perchlorate at various stages of life. The NAS Committee derived its recommended
perchlorate RfD based on the TIU in healthy adults observed in the Greer study.

However, since the scientific literature reports that the fetal thyroid is more sensitive to a
deficiency in TIU. Therefore, an uncertainty to consider in addressing this public health
issue is how well the adult thyroid response to perchlorate measured in the Greer study
represents the fetal thyroid response to lack of an adequate TIU.

5.	The NAS Committee indicated that its recommended RfD is protective of the most
sensitive population - "the fetuses of pregnant women who might have . . . iodide
deficiency."

6.	EPA's charge to NAS is written in a single chemical risk assessment approach, which
prevents the evaluation of the impact that the other dietary NIS inhibitors (i.e.,
thiocyanate and nitrate) have on this public health issue.

7.	EPA's charge to NAS is written in a single chemical risk assessment approach, which
prevents the evaluation of the impact that other NIS stressors, such as the lack of iodide,
have on this public health issue.

4

-------
10-P-0101

Results of Review

The OIG Analysis indicates that although EPA used a single chemical risk assessment for
perchlorate, a cumulative risk assessment that assesses and characterizes the combined human
health risk from all NIS stressors would better describe the nature and sources of risk affecting
this public health issue. The major findings, which directly address each of seven scientific
issues identified in the "Purpose" section of this report, are as follows.

1.	The NAS Committee's Unconventional RID Approach Is Conservative and
Protective

By applying the Tonacchera in vitro NIS Model of Competitive Inhibition in the OIG
Analysis, we analyzed the NAS Committee's statement that its unconventional approach of using
a no-effect-level to derive its recommended RfD is both conservative and protective of human
health. As with the NAS Committee, we could not utilize the conventional approach of deriving
a perchlorate RfD because no perchlorate study has observed adverse effects in any human
population. Therefore, we used the principles of cumulative risk and identified in the OIG
Analysis that the other common dietary NIS inhibitors, thiocyanate and nitrate, act through the
same mechanism of toxicity. The toxic levels observed in humans from excessive thiocyanate
exposure allowed us to estimate a perchlorate RfD by converting the observed toxic NIS
inhibition level for thiocyanate into a perchlorate equivalent concentration through the use of a
relative potency factor. This allowed us to use the conventional risk assessment approach of
deriving an RfD from a lowest-observed-adverse-effect-level by utilizing the adverse effect of
hypothyroxinemia observed in women exposed to an excess NIS inhibition load caused by
excessive thiocyanate exposure. When this decreased production of T4 observed in
hypothyroxinemia occurs during pregnancy, subtle mental deficits occur in the offspring.
Therefore, the decreased production of T4 during pregnancy is the same mechanism by which
perchlorate's NIS inhibition potentially causes mental damage during gestation and lactation.

The OIG Analysis confirmed that the NAS-recommended perchlorate RfD is
conservative by a factor of 6.6 times. In other words, a conventionally derived perchlorate RfD
is 6.6 times higher than the NAS Committee's recommended perchlorate RfD of 0.0007 mg/kg-
day. This finding supports the NAS Committee's statement that its unconventional approach of
deriving the perchlorate RfD represents a conservative value. However, the perchlorate RfD
derived from a single chemical risk assessment approach is accurate only for those individuals
with a typical background exposure to the other three NIS stressors. In other words, the
perchlorate RfD is not a static value, but changes depending on the exposure level to the other
three NIS stressors (thiocyanate, nitrate, and lack of iodide).

2.	Increasing Uncertainty Factors Do Not Effectively Protect Public Health

The NAS Committee used a single UF of 10 for intraspecies variation to derive its
recommended perchlorate RfD. EPA defines an RfD as a daily oral exposure likely to be
without an appreciable risk of adverse health effects over a lifetime with uncertainty/variability
factors applied to reflect limitations of the data used. In other words, the purpose of an RfD is to
allow the setting of a regulatory exposure level that avoids the occurrence of adverse effect in

5

-------
10-P-0101

humans. The OIG Analysis identified that the use of UFs in a single chemical risk assessment is
effective only if the chemical that the risk assessors are evaluating (i.e., perchlorate) is one of the
principal agents inducing the adverse effect. We specifically evaluated the effectiveness of
increasing the application of UFs in deriving the perchlorate RfD and measured the effect on
preventing the TIU from becoming too low in pregnant women, fetuses, and nursing infants. We
found that the scientific data indicate that perchlorate has the weakest impact of the four NIS
stressors on the TIU. Lowering the perchlorate RfD has only a minimal effect on maintaining a
TIU within a healthy range. The OIG Analysis identified that:

• No increase in UF value derives a perchlorate RfD that is low enough to prevent mental
damage in children, by itself.

• In contrast, the only way to maintain a maternal TIU within a healthy range is to prevent
the excessive exposure to each one of the four NIS stressors.

• Limiting the exposure to perchlorate alone at or below the RfD cannot compensate for an
excessive exposure to one or more of the other NIS stressors.

3. Hypothyroxinemia Occurs Before Hypothyroidism

As part of the OIG Analysis, we evaluated the NAS Committee statement that the first
adverse effect of perchlorate is hypothyroidism. Hypothyroidism is a stressed thyroid condition
characterized by a low free thyroxine (fT4) level with an elevated thyroid-stimulating hormone
(TSH) level. The OIG Analysis of the scientific data found that the maternal thyroid's initial
response to a diminished uptake of iodide is not hypothyroidism but a less stressful thyroid
condition called hypothyroxinemia (a low fT4 level with normal TSH level). Hypothyroxinemia
is a common condition in pregnant women. Hypothyroxinemia reflects a thyroid condition in
which the pregnant mother has difficulty meeting her own T4 needs and is unable to meet the
fetal demand for T4, needed for proper brain development. We determined that
hypothyroxinemia occurs in pregnant women when the TIU becomes less than or equal to 25%
of normal. The OIG Analysis of the scientific data indicates that a maternal TIU equal to or less
than 25% is associated with these effects in their children:

• Attention-deficit/hyperactivity disorder (ADHD)

• Lower verbal intelligence quotient (IQ)

• Lower overall IQ

• Lower motor performance

The OIG Analysis of the scientific data estimated that up to 6.9% of infants (i.e., 276,000 infants
per year) are born to U.S. mothers with maternal TIU of less than 25%. These children have a
higher risk of suffering from these permanent subtle neurological, psychological, and intellectual
deficits. This frequency of fetal exposure to maternal TIU of less than 25% is consistent with the
estimated occurrence of attention-deficit/hyperactivity disorder (ADHD) at 19 years of age of
7.5%) (6.5%) to 8.4%) at a 95% confidence interval).

-------
10-P-0101

4. Low TIU Adversely Affects the Fetus First

The EPA and others charged the NAS Committee to assess the potential adverse effects
from perchlorate at various stages of life. The NAS Committee derived its recommended
perchlorate RfD based on the TIU of healthy adults observed in Greer study. However, since the
scientific literature indicates that the fetal thyroid is more sensitive to a deficiency in TIU, a
scientific uncertainty to consider in addressing this public health issue is how well the adult
thyroid response to perchl orate in the Greer study represents the fetal thyroid's response to lack
of an adequate TIU. The OIG Analysis of the scientific literature identified that the fetal thyroid
is less able to adapt to a low TIU than the mother's thyroid. Therefore, the first adverse effect
occurs to the fetus before the mother's thyroid performance shows any signs of stress (i.e.,
changes in thyroid hormone levels). Thus, risk assessors should use neither maternal
hypothyroidism nor hypothyroxinemia as the key biological event in a risk assessment. By
contrast, the first adverse effects in the fetus occur when the maternal TIU is between 25% and
50% of normal. The OIG Analysis of the scientific data indicates that a maternal TIU between
25% and 50% of normal is associated with these effects in their children:

• Permanent delayed reaction time to stimuli

• A higher frequency of elevated TSH in newborns (i.e., the cognitive impact of elevated
TSH in newborns has not been studied)

• An increased frequency of mild thyroid dysfunction (subclinical hypothyroidism) later in
childhood (if born with elevated TSH levels)

The OIG Analysis estimated that up to an additional 22% of infants (i.e., 880,000 infants per
year) are being born to U.S. mothers with maternal TIU between 25% and 50%.

5. Perchlorate RfD Alone Does Not Protect Most Sensitive Populations

As part of the OIG Analysis, we evaluated the NAS Committee's statement that its
recommended perchlorate RfD is protective of the most sensitive population - "the fetuses of
pregnant women who might have . . . iodide deficiency." The OIG Analysis found that mild
iodide deficiency decreases maternal TIU by 50% to 75%, while moderate iodide deficiency
decreases maternal TIU by more than 75%. By contrast, the OIG Analysis calculated that
decreasing the perchlorate exposure in drinking water from 24.5 ppb to 6.1 ppb would result in
only a 1.0% increase in the maternal TIU. Clearly, a 1.0% increase in maternal TIU is
insufficient to compensate for a 50% or more decrease in maternal TIU caused by mild to
moderate iodide deficiency. Further, the OIG Analysis found that no amount of reduction in the
body's total NIS inhibition load (i.e., reducing the exposure to all three NTS inhibitors, not just
perchlorate) is capable of compensating for mild or moderate iodide deficiency in humans.

Iodide deficiency has a much stronger effect on lowering the maternal TIU than can be
compensated for by reducing the environmental exposure to perchlorate and/or the other NIS
inhibitors. Thus, to reduce the risk from this public health issue, mild and moderate iodide
deficiency during pregnancy and nursing must be addressed.

-------
10-P-0101

6. Thiocyanate and Nitrate Must Be Included in the Risk Assessment

The NAS charge prevented the NAS Committee from considering the impact of the other
NIS inhibitors (i.e., thiocyanate and nitrate) on this public health issue. The OIG Analysis of the
scientific literature identified that thiocyanate and nitrate have the same mechanism of toxicity as
perchlorate. Therefore, thiocyanate and nitrate must be considered together in the risk
assessment because their exclusion seriously distorts the portion of the total risk attributed to the
principal stressor of concern, perchlorate. Although perchlorate is a strong NIS inhibitor, human
exposure to perchlorate is relatively low so the contribution to the body's total NIS inhibition
load is relatively small. By contrast, although thiocyanate and nitrate are weak NIS inhibitors,
their human exposure levels are much greater than perchlorate. Therefore, thiocyanate and
nitrate account for the vast majority of the body's typical total NIS inhibition load in adults,
fetuses, and nursing infants.

The OIG Analysis of the scientific data identified that the consumption of 2 liters of
drinking water contaminated with nitrate at EPA's maximum contaminant level induces 12 times
the amount of NIS inhibition that perchlorate exposure at the RfD in an adult would cause. This
indicates that lowering the nitrate exposure in drinking water is a more effective approach to
increasing maternal TIU than lowering the perchlorate DWEL below 24.5 ppb.

7. Lack of iodide Is the Dominant NIS Stressor Affecting this Public Health Issue

The NAS charge also prevented the NAS Committee from fully considering the role that
iodide has on this public health issue. The OIG Analysis of the scientific literature identified that
the lack of iodide is a nonchemical stressor that acts through the same mechanism of toxicity as
perchlorate - limiting the TIU. The lack of iodide (a NIS stressor) has the same mechanism of
toxicity as perchlorate and, therefore, must be included in the risk assessment because its
exclusion seriously distorts the portion of the total risk attributed to the principal stressor of
concern, perchlorate. The OIG Analysis found that the lack of iodide is the dominant and
principal NIS stressor that determines a TIU level.

The OIG Analysis of the scientific literature indicates that to reduce the frequency and
severity of permanent mental deficits in children, the excessive exposure to each of the four NIS
stressors is required to prevent low maternal TIU during pregnancy and nursing. Since
thiocyanate, nitrate, and perchlorate are all chemicals, managing their exposure is within EPA's
legislative mandate. However, what would be the most effective action, overseeing the iodide
nutrition level during pregnancy and lactation, is not in EPA's legislative mandate. EPA's
Framework for Cumulative Risk Assessment directs EPA to seek expertise outside of the Agency
to address aspects outside EPA authority. Therefore, EPA would have to work cooperatively
with other federal agencies, such as the FDA and the National Institutes of Health, to safely
correct low maternal TIU. To reduce the frequency and severity of permanent mental deficits in
children from low maternal TIU, our cumulative risk assessment has identified adding iodide to
prenatal vitamins and implementing their use before and during pregnancy and nursing as an
approach to addressing this public health issue.

-------
10-P-0101

Conclusion

The OIG Analysis of the scientific literature identified that the risk from perchlorate
exposure is only part of a larger public health issue that is defined by the subtle mental deficits
occurring in children born to mothers with low maternal TIU during pregnancy and nursing. The
TIU results from the combined biological effect of four NIS stressors acting on the thyroid:
thiocyanate, nitrate, perchlorate, and lack of iodide. Diet constantly exposes everyone to each of
the four NIS stressors, and an individual's TIU level is the result of the combined effect of all
four NIS stressors, not just perchlorate exposure. The OIG Analysis concludes that a single
chemical risk assessment of perchlorate is not sufficient to assess and characterize the combined
human health risk from all four NIS stressors. However, both EPA's draft perchlorate RfD from
the Argus rat study and the NAS Committee's recommended perchlorate RfD from the Greer
study used a single chemical risk assessment approach. Only a cumulative risk assessment can
fully characterize the nature and sources of risk affecting this public health issue. Furthermore, a
cumulative risk assessment allows an informed environmental decision to be made on how to
mitigate the risk effectively

All four NIS stressors meet EPA's risk assessment guidance requirements for conducting
a cumulative risk assessment using the dose-addition method. In the OIG Analysis, we
conducted a cumulative risk assessment and determined that the risk from each of the four NIS
stressors is not equal. The OIG Analysis also confirmed that EPA's perchlorate RfD is
conservative and protective of human health, but limiting perchlorate exposure does not
effectively address this public health issue. Potentially lowering the perchlorate drinking water
limit from 24.5 ppb to 6 ppb does not provide a meaningful opportunity to lower the public's
risk. By contrast, addressing moderate and mild iodide deficiency occurring in about 29% of the
U.S. pregnant and nursing population appears to be the most effective approach to increase TIU
to healthy levels during pregnancy and nursing, thereby reducing the frequency and severity of
permanent mental deficits in children.

9

-------
10-P-0101

1. Defining the Public Health Issue
Background

Perchlorate induces human toxicity by blocking the uptake of iodide into the thyroid. A
protein called the sodium (Na+)/iodide (I") symporter (NIS) mediates iodide transport from the
blood into the thyroid. Since perchlorate acts to block the uptake of iodide, perchlorate is an NIS
inhibitor. The perchlorate mechanism of toxicity is that excess perchlorate exposure lowers the
total iodide uptake (TIU) of the thyroid. A prolonged low TIU decreases the supply of iodide in
the thyroid and eventually results in decreased production of thyroid hormones thyroxine (T4)
and triiodothyronine (T3). Maintaining an adequate supply of maternal T4 during pregnancy is
critical for proper fetal brain development. A low TIU during pregnancy is associated with
permanent mental deficits in the children of those mothers. The more severe the maternal TIU
deficiency, the greater will be the frequency and severity of the permanent mental deficits in
their children.

In addition to perchlorate, diet exposes humans to other NIS inhibitors that also lower the
amount of iodide uptake by the thyroid in humans. The other common dietary NIS inhibitors are
thiocyanate and nitrate. All NIS inhibitors share the same mechanism of toxicity as perchlorate -
blocking the uptake of iodide by the thyroid. The risk to public health is from the combined
amount of NIS inhibition acting on the body from all three NIS inhibitors (thiocyanate, nitrate,
and perchlorate). This combined total NIS inhibition load acting on the thyroid determines the
TIU level and the subsequent potential for an adverse outcome. The combined NIS inhibition
load is known as the total goitrogen load.

In addition to NIS inhibitors, the amount of the essential nutrient iodide in the diet
directly affects the amount of iodide uptake by the thyroid in humans. Unlike an environmental
pollutant where increased exposure is associated with increased frequency and severity of
adverse effect(s) on the body, iodide becomes an NIS stressor when there is an insufficient
amount of iodide in the diet to allow for an adequate uptake of iodide by the thyroid. Therefore,
the lack of iodide in the diet (i.e., an iodide deficiency) is an NIS stressor and shares the same
mechanism of toxicity as the NIS inhibitors - preventing the NIS from taking up an adequate
amount of iodide. Therefore, both the combined total NIS inhibition load acting on the thyroid
and the iodide nutritional level determines the TIU level and the subsequent potential for an
adverse outcome.

Diet continually exposes all humans to all four NIS stressors (thiocyanate, nitrate,
perchlorate, and iodide). Vegetables are a significant and natural source of thiocyanate. Foods
particularly rich in thiocyanate include cabbage, broccoli, Brussels sprouts, corn (maize), turnips,
rapeseed, mustard seed, cauliflower, cabbage, radishes, spinach, tomatoes, and milk. Although
diet is an important source of thiocyanate, cyanide produced and absorbed through cigarette
smoking is also a significant source of thiocyanate. Nitrate is also common in food. Nitrate
occurs in green leafy vegetables, is added to processed meats as a preservative, and is found in
milk. In addition to dietary nitrate, both surface and ground sources of drinking water commonly
contain nitrate due to the agricultural use of nitrate fertilizers. According to the Food and Drug
Administration (FDA) Total Dietary Study (TDS), 46% to 59% of the total estimated intake of

-------
10-P-0101

perchlorate comes from vegetables and dairy foods. Finally, the FDA TDS also identifies dairy
and grain as the most significant sources of dietary iodine for all groups of adults. Therefore,
the degree of dietary exposure to each of the four NIS stressors (not just perchlorate) is needed to
determine an individual's TIU and the likelihood for an adverse outcome.

In 2002, the U.S. Environmental Protection Agency (EPA) issued a draft perchlorate risk
assessment titled Perchlorate Environmental Contamination: Toxicological Review and Risk
Characterization (NCEA-1-1053) that proposed a perchlorate reference dose (RfD) of 0.00003
milligram/kilogram-day (mg/kg-day). EPA's proposed RfD corresponds to a drinking water
equivalent level (DWEL) of 1 ppb. EPA's draft risk assessment relied primarily on a rat study
conducted by Argus Research Laboratories in 2001. Due to the scientific controversy
surrounding the concentration at which perchlorate should be regulated, the National Academy
of Science (NAS) Committee to Assess the Health Implications of Perchlorate Ingestion (NAS
Committee) was charged to assess the current state of the science regarding potential adverse
effects of thyroid function disruption by perchlorate in humans and laboratory animals at various
stages of life. Part of the NAS Committee's charge was to determine whether EPA's 2002 draft
single chemical risk assessment of perchlorate was consistent with the current scientific
evidence. In 2005, the NAS Committee rejected EPA's approach of using the Argus rat studies
to derive the perchlorate RfD.

In January 2005, the NAS Committee recommended an RfD of 0.0007 mg/kg-day based
on the results observed in the Greer human perchlorate exposure study. The NAS Committee
included a 10-fold uncertainty (UF) in the RfD to protect the most sensitive population, the
fetuses of pregnant women who have hypothyroidism or iodide deficiency. On February 18,
2005, EPA established an RfD of 0.0007 mg/kg-day, which corresponds to a DWEL of 24.5
ppb. EPA is currently in the process of making a regulatory determination for perchlorate, which
is a formal decision on whether to issue a National Primary Drinking Water Regulation.

In the May 1, 2007, Federal Register (72 FR 24016), EPA indicated that it needed
additional information on perchlorate exposure to determine a value for the Relative Source
Contribution of perchlorate. EPA must set a perchlorate Relative Source Contribution value
before it can make a regulatory determination on whether regulating perchlorate in drinking
water presents a meaningful opportunity for health risk reduction. The Relative Source
Contribution will determine how much perchlorate exposure comes from food and water.

Utilizing Cumulative Risk Assessment. In 2005, the NAS Committee reviewed the
current state of the science regarding the potential adverse health effects of perchlorate exposure
(NAS 2005). The charge to the NAS Committee limited the context of its assessment to the
potential adverse effects from perchlorate only. This approach to assess the toxicity of a single
chemical is an outdated, conventional approach to characterizing the risk to the public. By
contrast, the state-of-the-art technique to assess risk is to perform a cumulative risk assessment
when multiple chemicals or stressors act through the same mechanism of toxicity. As indicated
by former EPA Administrator Browner in 1997, cumulative risk assessment "provides a platform
for significant advances in our scientific approach to assessing environmental risks" (EPA
1997b).

-------
10-P-0101

Recent Perchlorate Studies Have Different Conclusions. Various published
perchlorate studies draw substantially different conclusions on the toxicity of perchlorate.

• The Argus 2001 rat study and the Blount 2006 urinary perchlorate study suggest the
need for a lower RfD. Based principally on the observations made in the Argus 2001
rat study (Argus 2001), in 2002 the EPA proposed a perchlorate RfD of 0.00003
mg/kg-day, which corresponds to a DWEL of 1 ppb (EPA 2002a).

• The Blount 2006 urinary perchlorate study observed that increased urinary
perchlorate was associated with increased thyroid-stimulating hormone (TSH) and
decreased T4 levels in women with iodide deficiency (i.e., urinary iodide levels less
than 100 |ig/L) (Blount 2006b).

• By contrast, based on the perchlorate dosing of humans in the Greer study (Greer
2002), the NAS Committee proposed a perchlorate RfD of 0.007 mg/kg-day, which
corresponds to a DWEL of 24.5 ppb (NAS 2005).

• The Tellez 2005 study of pregnant women and neonates in Taltal, Chile, which are
exposed to drinking water at 114 ppb, are observed to have no statistically significant
changes in neonatal thyroid function or birth weight (Tellez 2005).

The lack of agreement in the perceived toxicity of perchlorate among these studies
indicates that there is still an unacceptable level of uncertainty in the science of perchlorate to
instill confidence that the current EPA perchlorate RfD is protective of human health.

Cumulative Risk Assessment is Possible. Perchlorate is not the only anion acting on the
thyroid to inhibit the uptake of iodide by the NIS. Humans are exposed nutritionally and
environmentally to NIS inhibitors other than perchlorate that also inhibit the uptake of iodide by
the NIS.

• The other principal NIS inhibitors of concern to humans are thiocyanate (SCN") and
nitrate (NO3").

• Furthermore, since iodine is the signature element in the molecular structure of the
two thyroid hormones, T4 and T3, a sufficient supply of iodide in the diet is essential
for the thyroid to be able to synthesize an adequate supply of thyroid hormones to the
body.

These three additional factors involved in the uptake of iodide are known and often cited
in the scientific literature, but their relative contribution to the uptake of iodide has not been
quantified or used in either EPA's or the NAS Committee's risk characterization. However, to
reduce the uncertainties in the science, the relative contributions of each of the four factors (i.e.,
thiocyanate, nitrate, perchlorate, and the lack of iodide in the diet) to the uptake of iodide must
be quantified and included in the risk assessment before making any effective environmental risk
management decisions. The Tonacchera in vitro NIS Model of Competitive Inhibition
(Tonacchera Model) quantified the relationship to the uptake of iodide from each of these four
factors (Tonacchera 2004). The Tonacchera Model provides the critical information upon which
to conduct a state-of-the-art cumulative risk assessment for this public health issue.

-------
10-P-0101

Summary. The EPA and NAS Committee approach to characterizing the risk from
perchlorate suggests that the public health issue is driven by perchlorate only. This approach
implies that the proposed remedy of lowering the exposure to perchlorate also directly lowers the
risk to the public. In other words, if the perchlorate level is lowered to a "safe" level, the public
health issue is eliminated. However, this single chemical approach and remedy underestimates
the complexity of the public health issue. The actual occurrence of an adverse outcome is
determined by the combined interaction of all four factors on the uptake of iodide by the NIS.
For EPA and the NAS Committee to determine the risk from just one factor without knowing or
managing the risk from the other three factors inadequately captures the amount of risk to the
public. Further, managing only the risk from perchlorate will not effectively address the public
health issue because the OIG Analysis shows that perchlorate contributes the least amount of risk
of the four factors. Therefore, the public health issue must be defined beyond just perchlorate.
The public health issue must consider and address all factors that prevent an adequate supply of
thyroid hormones during gestation and lactation to ensure proper brain development in both
fetuses and nursing infants. All the factors that impact the adequate supply of thyroid hormones
must be quantitatively evaluated in a cumulative risk assessment. The use of cumulative risk
assessment allows the relative contribution of risk from each factor to be identified and an
effective and cost-efficient solution to be determined.

13

-------
10-P-0101

2. OIG's Approach to the Scientific Analysis of Perchlorate
2.1 Introduction

Our science review identified and focused on scientific issues associated with the setting
of the perchlorate RfD, which is a key element in the process of making a regulatory
determination for perchlorate. We focused on the scientific process from EPA's issuance of the
2002 draft perchlorate risk assessment, Perchlorate Environmental Contamination:

Toxicological Review and Risk Characterization, to July 2008. We conducted field work from
April 2007 to July 2008. The OIG Scientific Analysis of Perchlorate (External Review Draft)
(the OIG Analysis) documents our science review. The OIG Analysis cites approximately 145
scientific journals as well as technical references and EPA risk assessment guidance.

The OIG Analysis identified that a single chemical risk assessment determines the risk to
a defined population to the exposure of a single chemical from all routes of exposure (e.g., air,
water, soil, food, etc.). By contrast, a cumulative risk assessment is an analysis, characterization,
and quantification of the combined risks to health from exposure to multiple chemicals or
stressors. Since 1992, experts in the scientific community, the National Academy of Sciences,
the President, Congress, and former EPA Administrator Browner have either identified the need
or directed EPA to use a cumulative risk assessment approach to assess the risk to public health
from exposure to environmental contaminations. Therefore, our OIG Analysis applied a
cumulative risk assessment approach to evaluate the public health risk from exposure to
perchlorate and the other NIS stressors.

We implemented a cumulative risk assessment by following the concepts in EPA's
Framework for Cumulative Risk Assessment (EPA/600/P-02/001F) and the principle of dose
addition described in EPA's risk assessment guidance for chemical mixtures. Further, we used
the NAS Committee on Toxicity Testing and Assessment of Environmental Agents (NAS
Toxicity Testing Committee) recommendation for the development of a quantitative,
mechanistic, dose-response model of the cellular pathway that is perturbed by the environmental
agent. We implemented the NAS Toxicity Testing Committee's vision and strategy by applying
the Tonacchera Model to evaluate the cumulative public health risk from perchlorate and the
other NIS stressors. Therefore, we conducted a cumulative risk assessment on this public health
issue defined as low TIU during pregnancy and lactation.

The Agency's Peer Review policy encourages and expects all scientific and technical
information intended to inform or support Agency decision making (i.e., rulemaking) to be peer
reviewed. Specifically, "influential scientific information" and "highly influential scientific
assessments" should be peer reviewed in accordance with the Agency's Peer Review Handbook
(EPA 2006). Since the OIG does not conduct rulemaking, the OIG Analysis did not undergo a
formal peer review. However, the OIG Analysis did undergo an independent, third-party
technical review. Four senior scientists from ICF International with direct knowledge and
experience in environmental risk assessment reviewed and provided comments on a working
draft of the OIG Scientific Analysis of Perchlorate. The purpose of this technical review was for
the OIG to get an independent scientific critique of our application of a cumulative risk
assessment to this public health issue induced by a low TIU during pregnancy and lactation. In

14

-------
10-P-0101

short, ICF International supported the implementation of a cumulative risk assessment, but
recommended using the whole mixture cumulative risk assessment approach as opposed to the
dose-addition method cumulative risk assessment approach. The OIG Analysis addresses the
whole mixture approach to the cumulative risk assessment of perchlorate in Appendix A.
However, in discussions with ICF International, the lead ICF scientist concurred that the
biochemical properties and mechanistic interaction of the four NIS stressors meet all the EPA
risk assessment requirements for the use of the dose-addition method (see Section 2.5.1.1).

2.2 Purpose

EPA's potential regulation of perchlorate in drinking water has been a contentious and
divisive environmental issue. The congressional bills H.R. 1747 and S. 150, which would amend
the Safe Drinking Water Act to require EPA to regulate perchlorate in drinking water, reflects
the interest in addressing this public health issue. We reviewed EPA's procedures over its
regulatory determination process for perchlorate. The completion of a human health risk
assessment that derives a perchlorate RfD is a key element in the process of making a regulatory
determination. Of particular concern is whether EPA's perchlorate RfD is protective of human
health at all life stages. Since EPA's final perchlorate RfD is taken directly from the NAS
Committee's recommended perchlorate RfD of 0.0007mg/kg-day, we analzyed whether the NAS
Committee's recommended RfD is protective of human health at all life stages. Further, we
analyzed the following scientific issues to assess the effectiveness of EPA's risk assessment
procedures:

• The NAS Committee used an unconventional approach to derive its recommended
perchlorate RfD by using a no-effect-level and asserted that this was a conservative,
health-protective approach to the perchlorate risk assessment.

-------
10-P-0101

•	The NAS Committee used a single UF of 10 for intraspecies variation to derive its
recommended perchlorate RfD. The NAS Committee stated that its recommended
perchlorate RfD needed no additional UFs.

•	The NAS Committee stated that the first adverse effect of perchlorate is hypothyroidism.
The Committee stated, "Defining the adverse effect is important because it influences
how the RfD is derived and ultimately the value of the drinking-water standard."

•	EPA's charge directed the NAS Committee to assess the potential adverse effects from
perchlorate at various stages of life. The NAS Committee derived its recommended
perchlorate RfD based on the TIU in healthy adults observed in the Greer study.

However, since the scientific literature reports that the fetal thyroid is more sensitive to a
deficiency in TIU, an uncertainty to consider in addressing this public health issue is how
well the adult thyroid response to perchlorate measured in the Greer study represents the
fetal thyroid response to lack of an adequate TIU.

•	The NAS Committee indicates that its recommended RfD is protective of the most
sensitive population, "the fetuses of pregnant women who might have . . . iodide
deficiency."

•	EPA's charge to NAS is written in a single chemical risk assessment approach, which
prevents the evaluation of the impact that the other dietary NIS inhibitors (i.e.,
thiocyanate and nitrate) have on this public health issue.

•	EPA's charge to NAS is written in a single chemical risk assessment approach, which
prevents the evaluation of the impact that other NIS stressors, such as the lack of iodide,
have on this public health issue.

To address these scientific issues, the OIG Analysis applies both mechanistic toxicology and a
cumulative risk assessment to this public health issue.

2.3 Calls to Use Cumulative Risk Assessment

We identified that the EPA draft 2002 perchlorate risk assessment, Perchlorate
Environmental Contamination: Toxicological Review and Risk Characterization (NCEA-1-
1053), used a single chemical risk assessment approach to a proposed perchlorate RfD of
0.00003 mg/kg-day (corresponding to a DWEL of 1 ppb). Furthermore, we found that the NAS
Committee also used a single chemical risk assessment approach to recommend a perchlorate
RfD of 0.0007 mg/kg-day (corresponding to a DWEL of 24.5 ppb). Therefore, both EPA and the
NAS Committee used a single chemical risk assessment approach to derive their respective
perchlorate RfDs. However, since as early as 1992, EPA has known that a single chemical risk
assessment is an outdated approach to assessing risk. A single chemical risk assessment
characterizes the potential adverse effect and quantifies the risk from only a single chemical
pollutant; it does not evaluate the combined effects from multiple chemicals acting through the
same mechanism of toxicity.

16

-------
10-P-0101

By contrast, a cumulative risk assessment is defined as "an analysis, characterization, and
possible quantification of the combined risks to health or the environment from multiple agents
or stressors" (EPA 2003, p 6). The application of a cumulative risk assessment has the potential
to expand our understanding of the public health impacts of environmental exposures and
provides a clearer, more complete picture of a public health issue for making risk management
decisions. In 1995, former EPA Administrator Browner stated, "EPA must be responsive and
resolve to more openly and fully communicate to the public the complexities and challenges of
environmental decision making in the face of scientific uncertainty" (Callahan 2007). In 1997,
then-Administrator Browner wrote that a cumulative risk assessment "provides a platform for
significant advances in our scientific approach to assessing environmental risks." Furthermore,
cumulative risk assessments can reduce risks to the extent that EPA can integrate it into
prevention strategies to protect public health.

Experts in the scientific community, various NAS committees, the Science Advisory
Board (SAB), the President, Congress, and former EPA Administrator Browner either have
identified the need or directed EPA to use a cumulative risk assessment approach to assess the
risk to public health from the exposure to environmental contamination:

•	The 1992 Expert Panel on the Role of Science at EPA issued the report titled
Safeguarding the Future: Credible Science, Credible Decisions (EPA/600/9-91/050).
The reported pointed out, "A key role of science at EPA is to reduce uncertainties in
environmental decision-making." One of the uncertainties identified by the 1992
Expert Panel that needed to be addressed was the "... impact of chemical mixtures
and other general stressors [on people and the ecosystem] ..." (EPA 1992, p 14).
However, the 1992 Expert Panel noted that EPA has not developed methods to assess
the health or environmental effects from multiple stressors (EPA 1992, p 14).

•	In 1994, the NAS Committee on Risk Assessment of Hazardous Air Pollutants issued
Science and Judgment in Risk Assessment, which recommended that EPA move away
from single-chemical assessments and to aggregate cancer risks from exposure to
multiple compounds (NAS 1994, executive summary 13; EPA 2000, p x (first
bullet)).

•	Under Section 3-301(b), Presidential Executive Order 12898 (1994) directs federal
agencies to identify "multiple and cumulative exposures" in environmental human
health analyses, whenever practicable and appropriate.

•	In 1996, Congress mandated, under Section 408(b)(2)(C)(i)(III) of the Food Quality
Protection Act of 1996, that EPA assess the cumulative risk to children and infants
from pesticide chemical residues and other substances having a "common mechanism
of toxicity".

•	In 1997, former EPA Administrator Browner directed the Agency "to take into
account cumulative risk issues in scoping and planning major risk assessments and to
consider a broader scope that integrates multiple sources, effects, pathways, stressors,
and populations for cumulative risk analysis in all cases for which relevant data are

17

-------
10-P-0101

available" and to "embrace this cumulative approach" in all major risks assessments
(EPA 1997b).

• In 1997, the EPA Science Policy Council stated EPA risk assessments should evolve
away from a focus on the potential adverse effect of a single pollutant in one
environmental medium.

• In 2007, EPA's Office of the Science Advisor requested SAB to provide
recommendations on ways to advance EPA's human health risk assessment practices.
On February 28, 2007, SAB recommended to advance cumulative risk assessment
methodologies in order to reflect real-world human exposure that includes multiple
stressors (SAB 2007).

• In 2008, the NAS Committee on Improving Risk Analysis Approaches Used by the
U.S. EPA issued Science and Decisions: Advancing Risk Assessment, which
recommended that EPA "... incorporate interactions between chemical and
nonchemical stressors in assessments" (NAS 2008; executive summary, p 9). This
NAS Committee stated, "EPA is increasingly asked to address broader public-health
and environmental-health questions involving multiple exposures, complex mixtures,
and vulnerability of exposed populations issues that stakeholder groups . . . often
consider to be inadequately captured by current risk assessments. There is a need for
cumulative risk assessments as defined by EPA (EPA 2003)—assessments that
include combined risks posed by aggregate exposure to multiple agents or
stressors ..." ((NAS 2008; executive summary, p 9).

• In 2008, the SAB Drinking Water Committee in its draft report on the Third
Drinking Water Contaminant Candidate List (CCL3) (December 3, 2008) proposes
recommending that "EPA should consider addressing the cumulative effects of
chemicals with similar sources and mechanisms (or modes) of action ..."

(SAB 2008).

In contrast to a single chemical risk assessment approach previously used by both EPA
and the NAS Committee to characterize the risk from perchlorate, a cumulative risk assessment
approach better evaluates the public health risk from exposure to perchlorate and the other NIS
stressors. The toxicity pathway identifies that an inadequate supply of iodide (i.e., low TIU)
during gestation and possibly during lactation results in increased risk for subtle mental deficits
in children. Perchlorate is only one of four NIS stressors (thiocyanate, nitrate, perchlorate, and
lack of iodide) that act on the NIS to lower the TIU by the thyroid. The human diet contains all
four NIS stressors; diet universally exposes every human to each of the four NIS stressors.
Therefore, an individual's TIU level is the combined effect of all four NIS stressors, not just
perchlorate exposure. Since a low maternal TIU results in an increased risk for an adverse
outcome, EPA should evaluate, through a cumulative risk assessment, the relative contribution
that each NIS stressor has on inducing a low TIU. The findings from the cumulative risk
assessment produce a broader, more comprehensive understanding and assessment of the factors
involved in this public health issue.

-------
10-P-0101

2.4 Use of Mechanistic Toxicology

On June 12, 2007, the NAS Toxicity Testing Committee released a report titled Toxicity
Testing in the Twenty-First Century: A Vision and A Strategy (NAS 2007). The NAS Toxicity
Testing Committee recommended a major shift in how the toxicity of chemicals is assessed
because of the revolutionary advances in the science (e.g., proteomics, genomics, and
biochemistry) in which the increased understanding of the biological mechanisms allows in vitro
toxicity testing to be applied to environmental risk assessments. The NAS Toxicity Testing
Committee's vision is to move away from the traditional descriptive animal toxicity testing to the
identification of perturbed biological pathways as the basis for dose-response modeling. Classic
descriptive animal toxicity testing determines a dose in which no adverse effects are identified.
Descriptive toxicity testing does not develop an understanding or a characterization of the
cellular mechanism triggering the toxicity. The NAS Toxicity Testing Committee's vision of
toxicity testing is to develop a quantitative, mechanistic, dose-response model of the cellular
pathway that is perturbed by the environmental agent. Subsequent pharmacokinetic modeling
would identify a safe human exposure level that prevents the environmental chemical from
reaching a toxic tissue concentration. The NAS Toxicity Testing Committee states, "A stronger
scientific foundation offers the prospect of improved risk-based regulatory decisions and
possibly greater public confidence in and acceptance of the decisions."

The NAS Toxicity Testing Committee's vision can be observed in the evolving
perchlorate science. In 2002, EPA released the draft risk assessment Perchlorate Environmental
Contamination: Toxicological Review and Risk Characterization, which proposed a perchlorate
RfD of 0.00003 mg/kg-day. EPA's risk assessment relied on observations made in two rat
studies conducted by Argus Research Laboratories in 1998 and 2001. These studies are
examples of classic descriptive toxicity testing in which various perchlorate doses are evaluated
for their potential to produce adverse effects (e.g., brain structural changes) in the most sensitive
population - pregnant females, fetuses, and neonates. Although the lack of thyroid hormones
during early development can lead to brain damage, these rat studies were not designed to
identify or study a specific cellular pathway that is impacted by perchlorate exposure. These
studies were also not designed to develop a quantitative, mechanistic, dose-response model of
the perturbed cellular pathway that could have led to the observed changes in the linear
measurements of several brain regions of the male and female pups (e.g., corpus callosum).

In 2005, the NAS Committee to Assess the Health Implications of Perchlorate Ingestion
was charged to review EPA's 2002 draft perchlorate risk assessment. The NAS Committee
found that it was impossible to conclude whether perchlorate exposure caused these brain
changes because of problems with the method used to make the measurements, and because the
observed changes did not follow a dose-effect relationship. The NAS Committee proposed a
mechanistic step (i.e., a perturbation of a cellular pathway) that initiates a sequence of
subsequent biological events that eventually results in adverse effects in humans. The key
cellular event identified was perchlorate inhibiting the NIS, resulting in the decreased uptake of
iodide by the thyroid's epithelial cells. The NAS Committee used only the inhibitory effect of
perchlorate on this cellular step as the basis to derive a recommended perchlorate RfD. The
evaluation of only perchlorate's inhibition does not represent a quantitative, dose-response model
of the prominent factors (i.e., iodide, thiocyanate, nitrate, and perchlorate) that impact the uptake

19

-------
10-P-0101

of iodide by the NIS. The NAS Committee relied primarily on the Greer study to develop its
recommended perchlorate RfD of 0.0007 mg/kg-day. The Greer study applied descriptive
toxicity testing techniques to develop a dose-response curve in order to identify a perchlorate
dose in human test subjects that did not inhibit the uptake of iodide by the NIS. The NAS
Committee's recommended perchlorate RfD was not based on a quantitative, dose-response
model of the prominent factors that impact the uptake of iodide by the NIS. The NAS
Committee used a single chemical risk assessment approach despite the fact that the uptake of
iodide by the NIS was known to be perturbed by other NIS inhibitors and to be affected by the
amount of available iodide in the blood.

OIG's approach to reviewing the perchlorate science is consistent with the NAS Toxicity
Testing Committee's vision and strategy for toxicity testing and its implications on risk
assessment. In 2004, Tonacchera developed and evaluated an in vitro NIS Model of Competitive
Inhibition (Tonacchera 2004). This model quantitatively determines the dose-response of the
NIS to the presence of all of the following four anions simultaneously: thiocyanate, nitrate,
perchlorate, and iodide. The model predicts the thyroidal iodide uptake by correlating the
amount of iodide (i.e., iodide nutritional status) for any given total NIS inhibitor load (i.e., total
goitrogen load). The Tonacchera Model quantifies the total goitrogen load into a single value by
mathematically combining the amount of inhibition contributed from thiocyanate, nitrate, and
perchlorate by using potency factors. Tonacchera developed the model from in vitro
experiments that exposed Chinese hamster ovary (CHO) cells expressing the human NIS protein
to various combinations of the four anions and measured the amount of uptake of radioactive
iodide by the cells. The use of the Tonacchera Model for human health risk assessment
implements the NAS Toxicity Testing Committee's vision of toxicity testing by using a
quantitative, mechanistic, dose-response model of the cellular pathway that is perturbed by
perchlorate, and then the subsequent use of pharmacokinetic modeling to identify a human
perchlorate exposure level that prevents the perchlorate from reaching a toxic concentration at
the NIS. Therefore, the OIG used the Tonacchera Model to conduct a cumulative risk
assessment of this public health issue.

2.5 EPA Risk Assessment Procedures

2.5.1 EPA Risk Assessment Guidance

EPA risk assessment guidance provides procedures for combining the risk from multiple
chemicals (i.e., a chemical mixture) sharing the same "mode of action." These risk assessment
procedures are identified in the following EPA risk assessment guidance documents: the 1986
Guidelines for the Health Risk Assessment of Chemical Mixtures (EPA/630/R-98/002) and the
2000 Supplementary Guidance for Conducting Health Risk Assessment of Chemical Mixtures
(EPA/630/R-00/002). EPA assumes the toxicity of individual chemicals in a chemical mixture
add together (i.e., dose additivity) when the chemicals have the same mode of action and elicit
the same effects. EPA's risk assessment guidance identifies the use of a dose-addition method
for combining risk from components of a chemical mixture (e.g., multiple agents or stressors).
EPA defines a chemical mixture as the concurrent or sequential exposure of two or more
chemicals regardless of source, timing, or location of exposure (i.e., the chemicals do not have to

20

-------
10-P-0101

physically exist as a mixture outside the body or be consumed as a mixture to be a chemical
mixture for environmental risk assessment purposes).

EPA's own risk assessment guidance instructs the agency to aggregate the risk from
multiple chemicals or stressors sharing the same mode of action. The following sections of
EPA's risk assessment guidance provide details on the dose-addition model for combining risk
from multiple chemicals having the same mode of action using relative potency factors:

•	The 1986 EPA Guidelines for the Health Risk Assessment of Chemical Mixtures
recommends three approaches for the quantitative health risk assessment of a
chemical mixture (e.g., multiple agents or stressors) (EPA 1986; EPA 2000, section
2.2.1). This guidance defines a chemical mixture as the concurrent or sequential
exposure of two or more chemicals regardless of source, timing, or location of
exposure (i.e., the chemicals do not have to physically exist as a mixture outside the
body or be consumed as a mixture to be a chemical mixture for environmental risk
assessment purposes) (EPA 1986, p 1). In section 4.1 of this 1986 EPA guidance
document, the third recommended approach allows for a risk assessment of a
chemical mixture by using a dose-addition model with potency adjustment if the
chemicals act as dilutions or concentrations of each other (EPA 1986, p 16).

•	In 2000, EPA supplemented this chemical mixture guidance by publishing the
Supplementary Guidance for Conducting Heath Risk Assessment of Chemical
Mixtures (EPA 2000). Section 2.6.1 of this supplementary guidance provides details
on using the dose-addition method for combining the risk from multiple of chemicals
having the same mode of action (EPA 2000).

2.5.1.1 Applying the Dose Addition Method

Since 1986, the EPA risk assessment procedures have allowed the amount of NIS
inhibition from thiocyanate, nitrate, perchlorate, and the lack of iodide to be combined through
dose addition by the use of relative potency factors for the purposes of conducting a human risk
assessment. The biochemical properties and mechanistic interaction of the four NIS stressors
meet the following EPA risk assessment requirements for the use of the dose-addition method:

•	Same mechanism of toxicity (all four NIS stressors act on the NIS to limit the amount
of iodide uptake by the thyroid)

•	Similar dose-response3 curves (e.g., same slopes, different x-axis intercepts)

•	Act independently of one other; act as dilutions of one another

•	Known relative potency factors for each NIS stressor

•	Behave similarly in terms of the primary physiologic processes (i.e., uptake,
metabolism, distribution, and elimination)

•	Human exposures of each chemical are known (including at various life stages)

3 A dose-response is the relationship between a quantified exposure (dose) and the proportion of subjects
demonstrating specific biologically significant changes in incidence and/or in degree of change (response).

21

-------
10-P-0101

Therefore, the OIG Analysis conducted a cumulative risk assessment of this public health issue
using the dose-addition method.

By contrast, an alternative approach to conducting a risk assessment is the whole mixture
approach, which risk assessors can accomplish by developing a statistical model on human
epidemiological data. An example of this approach that has received a lot of attention is the
Blount analysis and the subsequent Steinmaus analysis of the National Health and Nutrition
Examination Survey (NHANES) 2001-2002 epidemiological study. We address the whole
mixture approach for conducting a cumulative risk assessment in Appendix A.

2.5.2 EPA Development of Cumulative Risk Assessments

In 1997, the EPA Administrator directed the Agency "to take into account cumulative
risk issues in scoping and planning major risk assessments and to consider a broader scope that
integrates multiple sources, effects, pathways, stressors, and populations for cumulative risk
analysis in all cases for which relevant data are available" and to "embrace this cumulative
approach" in all major risks assessments (EPA 1997b). In May 2003, EPA published
Framework for Cumulative Risk Assessment (EPA/600/P-02/001F), which establishes a
framework to develop guidance on conducting and evaluating cumulative risk assessments and to
provide basic concepts and principles to be addressed in a cumulative risk assessment. The
document's primary purpose is to offer a simple, flexible structure for conducting and evaluating
cumulative risk assessments within EPA. This framework is neither a procedural guide nor a
regulatory requirement within EPA. EPA states it expects cumulative risk assessment to evolve
with experience. The following section of the Framework for Cumulative Risk Assessment
identifies that the risk from stressors with a common mode of action are assessed together and
that the dose-addition method can be applied:

• In section 3.2.2.1, the EPA Framework for Cumulative Risk Assessment identifies that
for situations of toxicological similarity, "the stressors are grouped according to the
common mode of action . . . during the planning and scoping phase [of the risk
assessment]" "For all effects caused by that mode of action, dose addition can be
applied to the stressor group."

In May 2007, EPA issued four white papers on how cumulative assessment differs from
traditional assessment and how EPA could assess cumulative risk (Callahan 2007; deFur 2007;
Sexton 2007; Ryan 2007). The application of cumulative risk assessment has the potential to
expand our understanding of the public health impacts of environmental exposures (Fox 2004)
and provides a clearer, more complete picture of a public health issue for making risk
management decisions (EPA 2003, appendix A). Cumulative risk assessment will reduce risks
to the extent that it can be integrated into prevention strategies to track and protect public health
(Fox 2004).

On June 27, 2008, EPA issued Concepts, Methods, and Data Sources for Cumulative
Health Risk Assessment of Multiple Chemicals, Exposures, and Effects: A Resource Document
(EPA/600/R-06/013F), which identifies specific elements of and approaches for implementing
cumulative risk assessments. This document demonstrates the feasibility of including

22

-------
10-P-0101

combinations of chemicals, exposures, effects, and their interactions into a cumulative risk
assessment, and is meant to assist with the conduct of multichemical, population-focused
assessments. However, this resource document specifically identifies in its preface that it is
neither a regulatory document nor guidance (EPA 2008, p xvii). Although EPA's Office of
Research and Development (ORD) has made progress in developing resource documents to
conduct cumulative risk assessments, ORD has not developed Agency-wide guidance or
regulations for conducting a cumulative risk assessment.

By contrast, as required by Section 405 of the Food Quality Protection Act of 1996 [P.L.
104-170], EPA's Office of Pesticide Programs (OPP) issued on January 14, 2002, Guidance on
Cumulative Risk Assessment of Pesticide Chemicals that Have a Common Mechanism of
Toxicity. This cumulative risk assessment guidance applies to pesticides and begins by
identifying a group of pesticides that induce a common toxic effect by a common mechanism of
toxicity. OPP had conducted four cumulative risk assessments on the following classes of
pesticides sharing the same mechanism of toxicity: n-methyl carbamate, organophosphate,
triazine, and chloroacetanilide.

2.6 Implementing a Cumulative Risk Assessment

Our approach to conducting a cumulative risk assessment was to apply the NAS Toxicity
Testing Committee's strategy and vision to this public health issue. The in vitro modeling is
required because as the complexity of the risk assessment increases (i.e., more factors evaluated),
animal testing becomes neither practical nor sensitive enough to observe and define the
relationship between an increasing number of stressors. For example, four stressors each
administered at three different doses (i.e., high, medium, and low) would require 81 separate
animal studies to evaluate the 34 possible combinations. To conduct 81 perchlorate animal
studies is impractical. Furthermore, no scientific techniques are available to measure subtle
cognitive deficits in rat offspring.

We used TIU as the basis for the cumulative risk assessment because the NAS
Committee identified the TIU as the key biochemical event and recommended the use of TIU as
the basis for a perchlorate risk assessment. The TIU is the combined effect of all four NIS
stressors acting on the thyroid. We identified the Tonacchera Model as a mathematical model
that defines the relationship between the four NIS stressors (thiocyanate, nitrate, perchlorate, and
lack of iodide) and the TIU by the NIS. Therefore, the Tonacchera Model combines the risk
from multiple chemicals into a single variable, the TIU, which measures the cumulative effect on
the thyroid to the simultaneous exposure to all four NIS stressors. Thus, the Tonacchera Model
utilizes both the NAS Toxicity Testing Committee's strategy and vision and EPA's dose-addition
method to conduct a cumulative risk assessment of this public health issue.

The application of an in vitro model to assess environmental risk is unprecedented, but is
required to address the complexity of multiple stressors affecting this public health issue.

Instead of the conventional single chemical risk assessment approach of identifying an external
dose that results in an adverse effect, our approach identified the maternal TIU that is associated
with subtle mental deficits in a developing child. TIU is not an external dose of a single
chemical, but the combined biological effect of all four NIS stressors acting on the thyroid. So

-------
10-P-0101

addressing this public health issue requires preventing the excessive exposure to any single NIS
stressor to assure the maternal TIU does not reach a level that induces toxicity.

Section 3.1.3 of EPA's Draft Guidance on the Development, Evaluation, and Application
of Regulatory Environmental Models (November 2003) establishes that all environmental models
should be corroborated before they are used as the basis of rule-making or regulation. We have
corroborated the Tonacchera Model against several available human exposure studies. The OIG
Analysis found the calculated TIU to be both a good predictor of an adverse outcome and a good
explanation for the observed toxicity of the NIS stressors.

2.6.1 Aspects in Implementing a Cumulative Risk Assessment

Cumulative risk is defined as "the combined risks from aggregate exposure to multiple
agents or stressors" (EPA 2003, p 6). The 1992 Expert Panel on the Role of Science at EPA
points out that EPA has historically focused on chemical-specific impacts (EPA 1992, p 14). In
1994, NAS issued a recommendation to move away from single-chemical assessments (NAS
1994, executive summary p 13; EPA 2000, p x (first bullet)). In 1997, former EPA
Administrator Browner wrote, "For most of our history, EPA has assessed risks and made
environmental protection decisions based on individual contaminants .... with risk assessments
for these chemicals often focused on one source, pathway, or adverse affect" (EPA 1997b).
Likewise, in 1997, the U.S. EPA Science Policy Council stated that EPA's risk assessment
should evolve away from a focus on the potential adverse effect of a single pollutant in one
environmental medium (Callahan 2007). Unfortunately, both EPA's 2002 draft perchlorate risk
characterization (EPA 2002a) and the NAS Committee on perchlorate (NAS 2005) focused
solely on perchlorate and did not include in their assessments the quantitative impact of the other
stressors that are known to affect the uptake of iodide by the thyroid. For this public health
issue, perchlorate is not the only stressor that inhibits the uptake of iodide by the NIS.
Thiocyanate and nitrate are also NIS inhibitors and human exposure to them is common, from
both natural and manmade sources, and is unavoidable in the diet. A cumulative risk assessment
of this public health issue must incorporate the inhibition from all three NIS inhibitors to produce
a broader, more comprehensive understanding and assessment of the factors involved that lead to
the unacceptable adverse effect of permanent, subtle mental deficits in children.

Another critical aspect of a cumulative risk assessment is that a stressor does not have to
be the exposure to a chemical, but the "absence of a necessity" (EPA 2003, p 2). This concept is
critical to this public health issue; the NIS uptakes iodide from the blood in order to make the
iodide-containing thyroid hormones T3 and T4. An adequate supply of iodide in the human diet
is essential for the production of iodide-containing thyroid hormones and their subsequent supply
to the body. If the human diet lacks a sufficient supply of iodide, at specific level of iodide
deficiency for a given NIS inhibitor load, the thyroid becomes unable to meet the body's
minimum requirement for thyroid hormones. Therefore, the lack of iodide in the diet (i.e., the
absence of a necessity) is also a stressor for this public health issue. Subsequent analysis will
show the lack of iodide to be the dominant stressor in this public health issue. The iodide
nutritional level must be quantitatively factored into the human health risk assessment.
Unfortunately, both EPA's 2002 draft perchlorate risk characterization (EPA 2002a) and the
NAS Committee on perchlorate (NAS 2005) did not quantitatively factor in iodide nutrition into

24

-------
10-P-0101

their reviews. By contrast, cumulative risk assessment allows for the lack of this necessity to be
quantitatively factored into the human health risk assessment in order to fully characterize this
public health issue. The relative contribution of the lack of iodide stressor on the uptake of
iodide by the NIS is quantitatively defined in the Tonacchera Model (Tonacchera 2004).
Furthermore, since the lack of iodide can be shown to be the dominate stressor effecting this
public health issue, it is essential that the lack of iodide be quantitatively included in the human
risk assessment.

The fundamental task in a cumulative risk assessment is to combine the toxicity of
individual chemicals that share the same toxicity pathway. The toxicity of the individual
chemicals sharing the same toxicity pathway can be combined additively by using relative
potency factors (Callahan 2007; Sexton 2007). Since 1986, EPA guidance has allowed the
amount of NIS inhibition from thiocyanate, nitrate, and perchlorate to be combined through dose
addition by the use of relative potency factors for the purposes of conducting a human risk
assessment. Under section 2.6.1 of the EPA Supplementary Guidance for Conducting Heath
Risk Assessment of Chemical Mixtures, for chemicals having the same mode of action, similar
dose-response curves, if the chemicals act as dilutions of each other and behave similarly in
terms of the primary physiologic processes (i.e., uptake, metabolism, distribution, and
elimination), and the exposures of each chemical are known, then the procedure suggested for
estimating risk uses relative potency factors (EPA 2000, p 28-29). The relative potency factors
approach scales the relative toxicity of each chemical (i.e., stressor) to the potency of an index
chemical (i.e., typically the best-studied component of the mixture) (EPA 2000, p 29).

Both EPA's 2002 draft perchl orate risk characterization (EPA 2002a) and the NAS
Committee on perchlorate (NAS 2005) did not utilize the dose-addition approach to human risk
assessment, and did not quantitatively incorporate the other NIS inhibitors into their reviews. In
EPA's perchlorate briefing to the NAS Committee on October 27, 2003, EPA stated that the
Agency recognizes the "potential cumulative burden of toxicity" of the other NIS inhibitors, but
describes comparisons of perchlorate toxicity to the other NIS inhibitors as "nonproductive"
(Farland 2003). EPA is referring to the scientific community's comparison of perchlorate
toxicity to the other NIS inhibitors: thiocyanate and nitrate. For example, an article in the
European Journal of Endocrinology indicates that the daily perchlorate equivalent concentration
(PEC) intake from the ingestion of thiocyanate and nitrate in the U.S. Department of Agriculture
(USDA) recommended diet is 670 times higher than the amount allowed for by the perchlorate
RfD (De Groef 2006). This paper also indicates that the concentration of nitrate allowed in
drinking water would cause an amount of NIS inhibition that is 12 times greater than the
perchlorate RfD. Likewise, Intertox, Inc. also compares the amount of inhibition from
thiocyanate and nitrate in the USD A recommended diet against the perchlorate RfD (Bruce 2003,
Bruce 2005). The OIG agrees with EPA that comparing perchlorate toxicity to the other NIS
inhibitors individually is not a constructive approach to assessing or addressing this public health
issue. We have this opinion because the human body is exposed to all three NIS inhibitors
concurrently and the resulting effect of each on the uptake of iodide by the thyroid is cumulative
and indistinguishable from the others, the perchlorate risk characterization must include exposure
to all three NIS inhibitors. The perchlorate risk characterization performed by EPA and NAS is
incomplete because the exposure from only a single NIS inhibitor was evaluated. The risk from

-------
10-P-0101

the cumulative inhibition effect of all three NIS inhibitors is best assessed through a cumulative
risk assessment.

Another critical aspect of a cumulative risk assessment is to incorporate the risk from the
background exposure to the stressors because the exclusion of a stressor(s) or the background
exposure from a stressor(s) "may seriously distort" the portion of the total estimated risk
attributed to the principal stressor of concern (EPA 2003, appendix C). Cumulative risk
assessment distinguishes between background exposures and source specific exposures to
provide a more complete picture of both total and source-related risks (EPA 2003, appendix C).
For a perchlorate cumulative risk assessment, the background exposure levels and sources of
each of the four stressors need to be determined and clearly distinguished from a specific source
exposure. The Tonacchera Model identifies the lack of iodide in the diet as the dominant
stressor. When the contribution of iodide is not quantitatively factored in to the risk assessment
(i.e., as in a single chemical assessment done for perchlorate), the result is seriously distorted.
Likewise, when the contribution of the other NIS inhibitors is not quantitatively factored in the
perchlorate risk assessment, the result is seriously distorted. When concurrent exposure to all
three NIS inhibitors is considered, perchlorate's contribution to the body's total inhibitor load is
only a small portion of the body's total inhibition load. Knowing the various sources and the
amount of contribution of each of the NIS inhibitors to the body's total inhibition load provides a
clearer, more complete picture of this public health issue for making risk management decisions.

EPA's Framework for Cumulative Risk Assessment stated, "For most exposure situations,
hazard and dose response studies of all of the joint effects from the multiple stressors will not be
available, so that conclusions will have to be based at least partly on the single stressor
information." For perchlorate, a risk characterization cannot be done in humans because adverse
health effects have not been clearly demonstrated in any human population exposed to
perchlorate. However, the Framework for Cumulative Risk Assessment identifies that the risk
characterization of a cumulative risk assessment can be based on the adverse effects observed in
one of the stressors and subsequently applied to the joint effects from all of the multiple
stressors. Therefore, the risk characterization of the four NIS stressors should be done by using a
NIS stressor in which adverse effects in children born to mothers with low maternal TIU during
pregnancy and nursing have been documented and reported. The excessive maternal exposure to
the NIS stressor, the lack of iodide, is the principal NIS stressor in which adverse effects have
been documented and reported in children born to mothers with low maternal TIU during
pregnancy and nursing. Thus, the OIG Analysis characterized the adverse effects from the lack
of iodide stressor and used this information to conduct the initial three steps in a risk
characterization (i.e., hazard characterization, dose-response assessment, and exposure
assessment) of this public health issue. The exposure to the lack of iodide stressor was related
back to the joint exposure level of all four NIS stressors by integrating the joint exposure into a
single parameter or "common metric." The single parameter for measuring the NIS stress level
is the TIU. The Tonacchera Model approach was to mathematically calculate the resulting TIU
level from the concurrent joint exposure to all four NIS stressors.

The cumulative risk assessment "also includes aspects of cumulative risk that may be
outside of EPA's current legislative mandates and where expertise outside of the Agency would
be needed to address certain questions if they should arise" (EPA 2003, p xviii). EPA authority

26

-------
10-P-0101

extends to the regulation of environmental contaminates. However, according to the Tonacchera
Model, the dominant stressor in this public health issue is the iodide nutrition level (i.e., the lack
of iodide in the diet). Overseeing the iodide nutrition level is clearly not in EPA's legislative
mandate, but addressing the needs of iodide-deficient pregnant women in the U.S. population is
necessary to effectively address this public health issue. The NAS Committee was qualitatively
aware of the crucial role of an adequate supply of iodide during pregnancy, so much so that the
NAS Committee "... recommends that consideration be given to adding iodide to all prenatal
vitamins" (NAS 2005, p 18). However, under the 1994 Dietary Supplement Health and
Education Act, the FDA does not approve dietary supplements or have the legislative mandate to
evaluate whether prenatal vitamins should contain iodide (FDA 2007b). Under the Act, the
dietary supplement manufacturers are responsible for determining supplement's safety and the
selection of ingredients for a dietary supplement (i.e., if iodide is to be added to prenatal
vitamins). The medical community, not EPA, is responsible for any practice to identify and treat
iodide deficiency in pregnant women with iodide containing prenatal vitamins or other
treatment(s). Therefore, EPA's potential regulation of perchlorate exposure from the
environment represents only a small portion of both the problem and solution to this public
health issue. Furthermore, addressing this public health issue will require a diverse group of
agencies, organizations, and disciplines to fully characterize and address the issue.

Another key aspect is that cumulative risk assessments can be qualitative as well as
quantitative (EPA 2003, p 7 and section 3.3.4, p 58). This aspect is important because it allows
for the incorporation of risk from other chemical exposures in which there is insufficient
information to quantify accurately the risk, but scientists know the presence and nature of the
risk. With respect to this public health issue, the uptake of iodide is not the only step in the
production and use of thyroid hormones in the body that is disrupted by chemical exposures.
Numerous inorganic and synthetic chemicals have been documented to interfere with almost
every major step in the production, transport, and peripheral tissue metabolism of thyroid
hormones (Howdeshell 2002, p 344, table 1). Furthermore, perchlorate, thiocyanate, and nitrate
are not the only chemicals that disrupt the uptake of iodide by the thyroid. The following table
provides examples of chemicals that interfere with several major thyroid hormone steps:

Thyroid Hormone Step

Examples of Chemicals that Interfere

Iodide uptake by the NIS

2,4-D, Aldrin, Lead, PBBs,

Iodide oxidation by thyroid peroxidase

Lindane, Malathion, Mancozeb

Circulatory transport in blood
(e.g., binding to transthyretin)

Dioxtylphthalate, DDT metabolites,
Dichloroprop, Difocol, Lindane, Malathion,

Cellular metabolism of T4 to T3 by Type I or Type II 5'-
deiodinase in peripheral tissues

Cadmium, Lead, PCB, Dixoin

Increase cellular elimination by glucuronidation of T4/T3 by
the stimulation of the glucuronidase enzyme

Acetochlor, DDT, PBBs, PCBs, Dioxin

Source: Howdeshell 2002 (for full list, see p 344, table 1).

Therefore, any risk assessment and resulting risk management decision(s) addressing this
public health issue must also incorporate the nonquantified risk from the exposure to other
thyroid hormone disrupting chemicals in order to be fully effective and protective of human
health. Section 3.3.4 of EPA's Framework for Cumulative Risk Assessment introduces
qualitative approaches to risk assessment (EPA 2003). Qualitative approaches may be the only

27

-------
10-P-0101

practical means to overcome the problems of complexity and data deficiencies and provide some
insight into the nature and magnitude of cumulative risks (Callahan 2007).

In conclusion, the nature of this public health issue not only meets the requirements to be
addressed by a cumulative risk assessment approach, but the complexity of the interaction among
three NIS inhibitors and the iodide nutritional level requires the use of a cumulative risk
assessment approach to accurately characterize the nature of the problem and to identify
effective and cost-efficient solutions to the problem. Presidential Executive Order 12898, the
intent of Congress under the Food Quality Protection Act of 1996, the 1997 direction given by
former EPA Administrator Browner, and EPA's own risk assessment guidance directs EPA to
consider cumulative risk analyses in all cases for which relevant data are available. The
documented interaction of the three NIS inhibitors - thiocyanate, nitrate, and perchlorate - and
the role of iodide nutrition in the uptake of iodide meet EPA's criteria for using cumulative risk
assessment to address this public health issue. The three NIS inhibitors exceed EPA's mode of
action requirement by sharing the same mechanism of toxicity (i.e., simple competitive
interaction) by inhibiting the uptake of iodide at the NIS. The term "mechanism of activity" is a
more detailed understanding and description of the events at the molecular level meant by "mode
of action" (EPA 2000, p 10). As also required by EPA guidance for dose addition, the dose-
response curves of thiocyanate, nitrate, and perchlorate have the exact same shape (Tonacchera
2004, figure 1). As also required by EPA guidance for dose addition, the three anions act as
concentrations or dilutions of each other (Tonacchera 2004). The relative potency factors for
each of the three NIS inhibitors are known (Tonacchera 2004). Human exposure data are
available on each of the four stressors (i.e., the three NIS inhibitors and iodide nutrition). Since
the EPA requirements for using the dose-addition method are met, there are no significant
technical reasons not to use cumulative risk assessment to characterize the risk to the public.
The only serious technical difficulty in implementing cumulative risk assessment is lack of
experience and familiarity with cumulative risk assessment in the risk assessor community.

2.7 Summary of the OIG Approach

In the opinion of the OIG, the complexity of this public health issue requires the use of a
cumulative risk assessment approach to better understand and characterize all the factors
contributing to this public health issue and to identify an effective and cost-efficient solution to
this problem. Based on the review of the EPA risk assessment guidance, the OIG could simply
just recommend the Agency conduct a cumulative risk assessment on perchlorate. However, due
to the novelty of the cumulative risk assessment approach and the required judgment in applying
the science in a cumulative risk assessment, the OIG has undertaken the unusual step of
providing an example of how a cumulative risk assessment could be conducted on perchlorate.
This example will demonstrate the advantages of using the cumulative risk assessment approach
and provide insight into potential cost-effective remedies to this public health issue. The use of
an in vitro model to evaluate potential remedies of a public health issue is not unfounded; the
NAS Toxicity Testing Committee stated, "In some risk contexts, a dose-response model based
on in vitro results might provide adequate data to support a risk-management decision" (NAS
2007). The application of a cumulative risk assessment approach using the Tonacchera Model
will allow the OIG to make recommendations on needed scientific research, comment on the

28

-------
10-P-0101

current regulatory approach to address the issue, and identify potential cost-effective remedies
for consideration.

29

-------
10-P-0101

3. Sodium (Na+)/lodide Symporter (NIS) Inhibitors

3.1	Background

Sufficient exposure to perchlorate can cause human toxicity by competitively inhibiting
transport of iodide into the thyroid gland by NIS (NAS 2005, p 6). The NAS Committee
identified the inhibition of iodide uptake by the thyroid in humans as the key biochemical event.
The NAS Committee recommended using the inhibition of NIS as the point of departure (POD)
for human risk assessment. The NAS Committee describes NIS inhibition as a nonadverse effect
that precedes all adverse effects from perchlorate exposure. The NAS Committee-recommended
RfD of 0.0007 mg/kg-day was established from a perchlorate exposure level that did not cause a
decreased uptake of iodide in adults.

Since the NAS Committee recommended using NIS inhibition as the POD for human risk
assessment and used the NIS inhibition to establish a perchlorate RfD, a full understanding and a
working quantitative model of this key biochemical event is critical to effectively protect human
health from the adverse effects from perchlorate exposure. The NIS is a trans-phospholipid
membrane protein that actively transports both sodium and iodide ions (i.e., an ion pump) across
the cell's plasmid membrane from the blood serum into the thyroid's follicular cells. The ion
transport is in the same direction. The concentration of iodide ion from the blood serum into the
thyroid's follicular cells is energetically unfavorable. Therefore, the energy from the co-
transport of two sodium ions down sodium's electrochemical gradient allows an iodide ion to
move against its electrochemical gradient and into the cell (Carrasco 2005).

Although NIS has a high affinity for the iodide ion, NIS is also known to transport a wide
variety of monovalent anions with a similar ionic radii to iodide (De Groef 2006; Carrasco
2005). The following monovalent anions are known to be able to concentrate in the thyroid or to
inhibit the NIS: pertechnetate (TCO4"), perchlorate (CIO4), perrhenate (ReO^), thiocyanate
(SCN"), boron tetrafluoride (BF4"), iodide (T), nitrate (N03), bromide (Br"), chloride (CI"),
chlorate (CIO3"), selenocyanate (SeCN"), periodate (IO4"), and bromate (BKV) (Carrasco 2005;
Wolff 1998; Dohan 2007). The relative NIS potencies of some of these anions are as follows
(Wolff 1998):

Tc04- > CIO4" > Re04" > SCN" > BF4" > I" > N03" > Br" > CI"

Notice that the iodide anion does not have the highest affinity for the NIS. Furthermore, both
perchlorate and thiocyanate have a higher affinity for NIS than iodide.

3.2	Relative Potencies of Other NIS Inhibitors

Humans are exposed nutritionally and environmentally to other NIS inhibitors, in
addition to perchlorate, which also competitively inhibit the uptake of iodide. The other
principal NIS inhibitors of concern of human exposure are thiocyanate (SCN") and nitrate
(N03"). As far back as 1953 (Wyngaarden 1953), the relative NIS inhibition potencies of
prevalent nutritional and environmental NIS inhibitors (i.e., perchlorate, thiocyanate, and nitrate

30

-------
10-P-0101

perchlorate) have been studied and quantifed repeatedly with excellent agreement in the results.
The following table provides a summary of these NIS inhibitor potency studies and their results:

Study

Study Description

Relative Potency* of NIS Inhibitor

Perchlorate

Thiocyanate

Nitrate

Tonacchera 2004

In vivo CHO** cells expressing the
human NIS

240

Alexander 1966

In vivo 50% thyroid weight
increase in rats

550

Alexander 1966

In vitro rat thyroid slices

400

Greer 1966

In vitro rat thyroids

240

Wyngaarden 1953

In vivo rat

300

* Relative Potency is based on molar concentrations of each anion.
** Chinese hamster ovary
Source: DeGroef2005.

3.3 Human Thiocyanate Exposure

Vegetables are a significant and natural source of thiocyanates (Tonacchera 2004). At
least 2,500 taxa in the plant kingdom contain cyanogenic glycosides (CGs) (Vetter 2000). Plants
use the CGs as a source to generate hydrogen cyanide, which serves as a feeding deterrent to
grazing animals or insects. Upon human ingestion, the CG-containing plants release cyanide.
The sulfurtransferase enzyme, rhodanese, metabolizes and detoxifies the released cyanide into
thiocyanate (WHO 2004; Hasuike 2004). Therefore, the natural act of eating routinely exposes
humans to the relatively potent NIS inhibitor thiocyanate. As a result, the human body always
has a certain amount of unavoidable exposure to NIS inhibitors and this is referred to as the
body's goitrogen load.

Foods particularly rich in thiocyanate include cabbage, broccoli, Brussels sprouts, corn
(maize), turnips, rapeseed, and mustard seed. Furthermore, cauliflower, cabbage, radishes,
spinach, and tomatoes contain on average 88, 86, 7, 5, and 2 mg/kg (wet weight) of thiocyanate,
respectively. Milk contains 2-10 mg/L thiocyanate. In some parts of the world, the consumption
of cassava is the major source of thiocyanate. Cassava can contain the equivalent of up to 3,400
mg/kg of thiocyanate (dry weight). Where cassava is not a significant part of the diet, cyanide
produced and absorbed by cigarette smoking is the most significant source of thiocyanate.

Thiocyanate serum concentration in nonsmokers is typically in the range of 10-70
(amol/L. By contrast, the thiocyanate serum concentration of smokers is higher, typically in the
range of 80-120 |imol/L.

Thiocyanate may have a regulatory function in normal physiology (Middlesworth 1986).
Thiocyanate is continuously synthesized in the normal rat. In the absence of all dietary intake of
thiocyanate in the rat, the thiocyanate serum concentration actually increased from 500 |ig/dl to
800 |ig/dl during fasting (Middlesworth 1986, figure 1). These data indicate that serum
thiocyanate in rats is maintained in a normal range. These data also imply that humans also
synthesize and maintain thiocyanate serum concentrations in a normal range. A "kidney
threshold" is reported to exist in humans at a thiocyanate serum threshold of 200-300 |imol/L
(Tonnacchera 2004).

-------
10-P-0101

By comparison, the EPA perchlorate RfD of 0.0007 mg/kg-day results in a perchlorate
serum concentration of 0.014 |imol/L. By contrast, the daily thiocyanate exposure in
nonsmokers results in a thiocyanate serum concentration in the range of 10-70 |imol/L.

Therefore, thiocyanate exposure is a significant component of the body's goitrogen load.

3.4 Human Nitrate Exposure

Nitrate is also common in food (Tonacchera 2004). Nitrate occurs in green leafy
vegetables. The consumption of vegetables is the primary source of nitrate exposure (NAS 1995,
p 35). In 1981, the National Research Council (NRC) estimated that 97% of the daily nitrate
intake is from diet (99% for vegetarians) (NAS 1995, p 40, table 4-2). However, nitrate is also
added to processed meats as a preservative. Furthermore, both surface and ground sources of
drinking water commonly contain nitrate due to the agricultural use of inorganic fertilizers,
animal manure, and airborne emissions from utilities, factories, and automobiles. In areas of
intensive agriculture, fertilizer usually is the predominant source of nitrogen. Animal manure
and atmospheric deposition account for smaller amounts of nitrogen contributions nationally
than commercial fertilizer, but are significant secondary sources of nitrogen in certain regions.

In vegetables, the highest nitrate concentrations are found in celery, spinach, lettuce,
beets, radishes, melon, turnip greens, and rhubarb (over 1000 mg/kg of vegetable), (NAS 1995,
p 35). Dairy products contain low concentrations of nitrate and rarely exceed 5 mg/kg in milk
(NAS 1995, p 36). The following table provides a list of common foods that contain nitrate:

Common Foods Containing Nitrates

Milk

Kimchi

Carrots

Bacon

Garlic

Onion

Sausage

Artichoke

Green beans

Pepperoni

Peas

Melon

Beef

Corn

Turnip

Ham

Sweet potatoes

Sweet pepper

Broccoli

Lima beans

Squash

Celery

Cucumber

Cabbage

Lettuce

Tomatoes

Leek

Radish

Parsley

Cauliflower

Spinach

Brussels sprouts

Pumpkin

Beets

White potatoes

Endive

Rhubarb

Eggplant

Kale

Turnip greens

Source: Bruce 2004.

Human exposure to nitrate is both exogenous and endogenous. Exogenous human nitrate
exposure is from diet through the consumption of primarily vegetables. The endogenous human
nitrate exposure is from the body's production of nitric oxide, which is subsequently converted
into nitrate by various types of cells in the body (NAS 1995, p 37). Therefore, nitrate excretion
in urine exceeds the nitrate intake from food and water. Thus, biomonitoring of nitrate in human
urine cannot be directly used to assess exogenous nitrate exposure.

-------
10-P-0101

The amount of human nitrate exposure has been assessed and estimated several times. In
1981, the NRC estimated nitrate intake from food at 40-100 mg/day for males (NAS 1995, p 38)
and that nitrate intake from contaminated water could contribute 22-44 mg/L (NAS 1995, p 38).
In 1998, European investigators report an average dietary nitrate intake of 43-131 mg/day (NAS
1995, p 39). In 1989, Van den Brandt estimated the exogenous nitrate intake at 113 mg/day for
males and 184 mg/day for females (NAS 1995, p 38). In 1992, Jones estimated exogenous
nitrate intake in the United States for omnivores at 76 mg/day and about 260 mg/day for
vegetarians (NAS 1995, p 38). Jones estimated the endogenous production of nitrate to be 62
mg/day (NAS 1995, p 38), which contributes about 45% of a human's total nitrate exposure
(NAS 1995, p 38). Jones estimated the total exogenous and endogenous nitrate exposure to be
138 mg/day (NAS 1995, p 38).

In studies from Denmark and the United Kingdom, the average daily nitrate intake is
estimated to be about 40-50 mg/day for adults (Tonacchera 2004). This nitrate intake value of
40-50 mg/day is on the low end of the nitrate intake estimates reported above. Fortunately, these
data also include the corresponding internal nitrate serum concentrations, which allows for the
subsequent calculation of both the amount of NIS inhibition acting on the thyroid from nitrate
exposure and the relative TIU by the thyroid under various NIS stressor levels. In the Western
world, the typical nitrate serum concentration ranges from 10-140 |imol/L with the mean nitrate
serum concentration being between 30-50 |imol/L (Tonacchera 2004). The serum half-life of
nitrate is 5-8 hours (Tonacchera 2004).

By comparison, the EPA RfD for perchlorate is 0.0007 mg/kg-day, which corresponds to
a daily perchlorate exposure limit of 0.049 mg/day for a 70-kg adult. The daily nitrate intake of
40-50 mg is substantially more (i.e., 1000 times greater on a weight basis) than both the daily
perchlorate exposure limit calculated from the EPA RfD of 0.049 mg/day for a 70 kg adult and
the .0046 mg/day median U.S. exposure to perchlorate estimated by the Centers for Disease
Control and Prevention (CDC) biomonitoring data (Blount 2006). Therefore, nitrate exposure is
also a significant component of the human body's goitrogen load.

3.5 Human Perchlorate Exposure

The U.S. population is ubiquitously exposed to low levels of perchlorate. In 2006,

Blount of the CDC, National Center for Environmental Health (NCEH), assessed the perchlorate
exposure in the U.S. population using urinary biomonitoring data (Blount 2006a). As part of the
2001-2002 NHANES, a nationally representative population of 2,820 U.S. residents (ages 6
years and older) provided urine samples for perchlorate testing. All 2,820 urine samples tested
found a detectable level of perchlorate (i.e., >0.05|ig/L), indicating that perchlorate exposure in
the U.S. population is common. The Blount biomonitoring data indicated median adult
perchlorate exposure was 4.6 |ig/day, and the 95th percentile exposure was calculated to be 16.4
|ig/day in an adult. Furthermore, the FDA TDS estimated lower- and upper-bound average
perchlorate intakes (i.e., range of dietary perchlorate intake) for 2005-2006 from 5.4 to 6.8
|ig/day from food in 25- to 30-year-old women (Murray 2008, table 5). Therefore, both the CDC
biomonitoring and the FDA TDS datasets indicate a low-level background exposure to
perchlorate across the entire U.S. population.

-------
10-P-0101

Potential sources of perchlorate exposure are through the Department of Defense/aerospace
sector (e.g., rocket propellant), commercial products (i.e., safety flares and some fireworks (up to
70% content)), agricultural use (i.e., perchlorate is a contaminant in CNF), and natural
atmospheric production (Dasgupta 2006). The Department of Defense/aerospace sector use
accounts for about 90% of the U.S. annual consumption of perchlorate (Dasgupta 2006).
However, the burning of rocket propellants, safety flares, and perchlorate-containing fireworks
does not result in a significant perchlorate exposure; 0.05% or less of the original perchlorate is
left after use because the fire destroys the perchlorate (Dasgupta 2006). By contrast, the
improper disposal of unused rocket propellants, safety flares, and perchlorate-containing
fireworks, which allows the unused perchlorate to get into the groundwater and surface water,
can result in human perchlorate exposure upon ingestion of contaminated drinking water.
Although this type of improper disposal into water sources would generate "hot spots" of
perchlorate exposure within the population, the improper disposal into water sources would not
generate a uniform background of perchlorate exposure across the entire U.S. population,
because EPA data indicate only 3.16% of the 3,858 Public Water Systems tested in the
Unregulated Contaminant Monitoring Rule (UCMR) had at least one detection of perchlorate
greater than or equal to 5 |ig/L (EPA 2007, table 5). Natural production of perchlorate by the
oxidation of chloride by lightning or ozone is not a significant source of drinking water
contamination because rainfall contains a mean concentration of perchlorate of 0.015 |ig/L (max.
0.2 |ig/L) (Dasgupta 2006).

A potential route of perchlorate exposure that is consistent with generating a uniform low
background exposure level in the entire U.S. population is the contamination of the U.S. food
supply with perchlorate. The FDA TDS identifies that vegetables and dairy foods combined
account for between 46% and 59% of the total estimated intake of perchlorate by teenagers and
adults, respectively (Murray 2008). A potential explanation for the widespread perchlorate
contamination of vegetables and dairy foods is the agricultural use of perchlorate-contaminated
CNF from 1909 to at least 1993. Over the life of CNF use in the United States, the average
perchlorate content of CNF is estimated to be 0.2% (Dasgupta 2006). The agricultural use of
CNF is estimated to have resulted in the environmental release of 1.6 million pounds of
perchlorate per year into U.S. agricultural fields for at least a 63-year period from 1930 to 1993
(Dasgupta 2006). In short, there are few ways to introduce a chemical into the food chain that
are more efficient than using fertilizer as a vector (Dasgupta 2006).

-------
10-P-0101

4. Modeling the NIS Stressors' Influence on the Total Iodide Uptake

4.1	Determining the Total NIS Inhibition Load Acting on the Human Body

The total NIS inhibition load (total goitrogen load) acting on the human body is the
combined effect of all four common NIS inhibitors (i.e., thiocyanate, nitrate, perchlorate, and
iodide). The Tonacchera study investigated the joint effects on the NIS from simultaneous in
vitro exposure to these four NIS inhibitors (Tonacchera 2004). This study identified the
interaction between the NIS inhibitors as being simple competitive interaction and observed no
evidence of synergism or antagonism among the NIS inhibitors. These data indicate that these
NIS inhibitors interact in a simple additive fashion. In other words, the in vitro effect on the NIS
for each NIS inhibitor was indistinguishable from that of the others once adjusted for differences
in concentration and potencies. Therefore, the total amount of in vitro NIS inhibition acting on
the NIS is the combined contribution of each anion on a molar potency adjusted basis (i.e., the
dose-addition method). Thus, the total inhibition acting on the NIS can be expressed as single
PEC representing the same amount of NIS inhibition induced by the joint effect of all four NIS
inhibitors.

The total amount of NIS inhibition (total goitrogen load) acting on the NIS in vitro is
measured as a PEC and is defined in the Tonacchera study by the following equation:

PEC = (Amt of C104" Inhibition) + (Amt ofN03" Inhibition) + (Amt of SCN" Inhibition) + (Amt of I" Inhibition)
PEC = [CKV] + [SCN"] / 15 + [N03"] / 240 + [T] / 30

However, the PEC equation needs to be adjusted into two mathematical forms to
represent in vivo conditions when the amount of total NIS inhibition is being measured in the
blood serum (i.e., serum perchlorate equivalent concentration (SPEC)) or measured from the
weight of ingested NIS inhibitors consumed in the diet or through contaminated drinking water
(i.e., oral perchlorate equivalent concentration (OPEC)).

4.2	Serum Perchlorate Equivalent Concentration (SPEC)

The body's total NIS inhibition is measured by the following SPEC equation (concept
adapted* from Tonacchera 2004):

SPEC = (Amt of CIO4" Inhibition) + (Amt of N03" Inhibition) + (Amt of free SCN" Inhibition)

SPEC = [Perchlorate] + [Nitrate] /240 + ([total serum SCN ] x 0.5) /15
SPEC = [Perchlorate] + [Nitrate] /240 + [free SCN ] /15

Where: [ ] is molarity

Serum inhibition potency of perchlorate = 1
Serum inhibition potency of nitrate relative to perchlorate = 1 / 240
Serum inhibition potency of free thiocyanate relative to perchlorate = 1/15
[free SCN" ]* = [total serum SCN ] x 0.5 [free SCN" /total serum SCN ]

35

-------
10-P-0101

* Note: Tonacchera's PEC equation is derived from in vitro results and had to be
adjusted to reflect in vivo conditions for the binding of thiocyanate to albumin in
the blood stream (see discussion below).

The SPEC equation relates the total NIS inhibition to the molarity of the each of the NIS
inhibitors in the blood serum (i.e., the concentration of each NIS inhibitor in contact with the
NIS protein). However, about half the thiocyanate in blood is bound to albumin (Stoa 1957).
Therefore, only about half of the thiocyanate in human blood serum is free to interact and to
inhibit the NIS protein on the thyroid. The experimental protocol used in Tonacchera's study
(from which each of the anion's potency was determined) incubated the CHO cells expressing
human NIS in Hank's balanced salt solution, which contains 0.5% bovine serum albumin
(Tonacchera 2004). By contrast, human plasma contains 4.8% albumin, by weight (Marieb
1998). This textbook amount of albumin in blood plasma is consistent with the 4.4% and 4.6%
found from direct measurements in nonsmoking and smoking controls, respectively (Hasuike
2004, table 1). Therefore, the in vitro conditions in Tonacchera's experimental protocol had only
about 1/10th the amount of albumin found in blood plasma.

The SPEC equation defined in this document is adapted from Tonacchera's PEC equation
by taking into account the in vivo conditions of having about 48 g/L of albumin in human blood
serum that affect the amount of free thiocyanate. In five human test subjects having a total SCN"
serum concentration ranging from 1.96 mg/L to 2.55 mg/L (i.e., the range is equivalent from
33.7 |imol/L to 43.9 |imol/L, which is in the typical SCN" range identified by Tonacchera), 47.2
±3.4 % of thiocyanate is bound to albumin (Stoa 1957, table 15). Or stated as free thiocyanate,
52.8% of the total thiocyanate in the blood serum is free and unbound to albumin. In rabbit
plasma at a higher SCN" concentration of about 141 & 280 |imol SCN7L, 41.3% and 40.7% of
thiocyanate is bound to the rabbit albumin, respectively (58.7 % and 59.3 % SCN" is free)

(Pollay 1966, table 1). The amount of free SCN" increases with increasing SCN" concentration in
the blood. Since SCN" in blood serum is typically 40 |imol/L, the amount of free SCN" from
Stoa's human data is used. Therefore, the amount of free SCN" at typical human SCN"
concentrations can be determined by multiplying the total SCN" concentration by a factor of
0.528. As the amount of SCN" increases, the amount of free SCN" also increases, but without
having specific data for percent binding of SCN" to human albumin at different concentrations,
the use of 0.5 in the SPEC equation represents a conservative estimate. Therefore, the
thiocyanate term is adjusted by multiplying the total serum SCN" by 0.5 to derive the amount of
available free thiocyanate in the blood able to interact and inhibit the NIS.

Our analysis drops the amount of iodide inhibition term (i.e., [I"]/30] in the original
Tonacchera PEC equation in order to simply the mathematical expression for use in a cumulative
risk assessment. Eliminating a minor term in a polynomial equation is standard mathematical
technique used to simplify a problem. Eliminating the [I"]/30 term in the PEC equation would
not significantly alter the SPEC or OPEC-calculated values. The amount of NIS inhibition from
iodide is a relatively small percentage of the OPEC value (i.e., about 0.1% of the OPEC of a
typical adult exposure level). The typical adult consumes about 150 |ig iodide/day. The oral
inhibition potency of iodide is 1/30. So, a normal adult intake of 150 |ig iodide/day represents
the same amount of NIS inhibition as 5 |ig of perchlorate (i.e., 1/10th the perchlorate RfD of 49
|ig/day). Since the perchlorate RfD has been calculated to represent about 1% of the body's total

-------
10-P-0101

goitrogen load, the iodide inhibition at 150 |ig/day is about 0.1% of the body's total goitrogen
load. Furthermore, the lack of iodide is an NIS stressor, which acts in the opposite to its
inhibition. As the iodide concentration decreases, resulting in more stress on the NIS, the
amount of NIS inhibition from iodide becomes even less significant (i.e., a low intake of iodide
at 50 |ig/day causes the same amount of NIS inhibition as 1.7 |ig of perchlorate (i.e., about
0.03% of the body's total goitrogen load). Conversely, when the iodide concentration increases,
the TIU value increases faster than the decrease in the TIU value from the amount of NIS
inhibition induced by iodide itself. Therefore, elimination of the iodide term from the
Tonacchera PEC equation represents a negligible bias in the calculation of the amount of total
NIS inhibition acting on the body.

The SPEC equation can be used when the serum molarities of each NIS inhibitor is
measured and reported. However, serum molarity data for NIS inhibitors is not commonly
measured or reported in human exposure studies. The oral exposure (i.e., dose) to NIS inhibitors
is more commonly determined and measured as the weight consumed through ingestion.

4.3 Oral Perchlorate Equivalent Concentration (OPEC)

The total NIS inhibitory effect measured upon ingestion of the NIS inhibitors is given by
the following OPEC equation (concept adapted from Tonacchera 2004; De Groef 2006, p 155):

OPEC = (Amt of C104" Inhibition) + (Amt of N03" Inhibition) + (Amt of SCN" Inhibition)

OPEC = grams perchlorate + grams nitrate/150 + 0.5 free SCN" x grams thiocyanate/8.8

OPEC = grams perchlorate + grams nitrate/150 + grams thiocyanate/17.6

Where:

Oral inhibition potency of perchlorate = 1

Oral inhibition potency of nitrate relative to perchlorate = 1/150

Oral inhibition potency of thiocyanate relative to perchlorate = 0.5/8.8 = 1/17.6

The above OPEC equation adjusts the relative NIS inhibitor potencies for the following
three factors:

1. The potency is adjusted to reflect the relationship based on the weights of each NIS
inhibitor ingested instead of the molarity of each NIS inhibitor in the serum.

2. The potency is also adjusted to take into account the different pharmacokinetic properties
of each of the NIS inhibitors upon ingestion (i.e., the different absorption and excretions
characteristics of each NIS inhibitor).

3. The amount of SCN" binding to albumin in human blood serum.

The pharmacokinetic differences are critical in determining the relative NIS inhibition
potencies of each anion upon ingestion. The ingested absorption and serum half-lives of each
anion have to be taken into account. Only about 50% of thiocyanate is absorbed in the
gastrointestinal tract upon ingestion. Meanwhile, 70-90%) of ingested perchlorate and 90-100%
of ingested nitrate is absorbed in the gastrointestinal tract. Furthermore, the half-lives of

-------
10-P-0101

perchlorate, thiocyanate, and nitrate are 8 hours, 6 days, and 5 hours, respectively. Thiocyanate
stays in the serum 18-29 times longer than the other anions. For example, only half of the
ingested amount of thiocyanate is absorbed upon ingestion, but the amount of thiocyanate
absorbed stays in the blood serum much longer, to act on and inhibit the NIS. Furthermore, only
about 0.5 of the total SCN- in human blood serum at typical SCN" concentrations is free to
interact with the NIS protein. Therefore, the NIS inhibition potency of ingested thiocyanate is
about 6% (i.e., 0.5 ^ 8.8 x 100%) - the NIS inhibition potency of ingested perchlorate. The
essential observation from the pharmacokinetics is that ingested thiocyanate has about one-
quarter the potency as ingested perchlorate, but thiocyanate is consumed in mg quantities in the
diet while perchlorate is only consumed in |ig quantities (i.e., median is 4.6 |ig) in the diet. From
this preliminary information, thiocyanate appears to contribute the largest amount of NIS
inhibition in the body. A more detailed analysis of the relative contribution of each NIS inhibitor
is provided in Section 7.2 of this document.

4.4 Tonacchera NIS Model of Competitive Inhibition

The Tonacchera in vitro NIS Model of Competitive Inhibition calculates the TIU by the
NIS from the concentration of the following four anions: iodide, perchlorate, nitrate, and
thiocyanate. The Tonacchera Model is given by the following equation (concept adapted* from
Tonacchera 2004):

TIU = constant x [T] / (1.22 + (SPEC))

Since the constant is not provided, the TIU can be restated as:

TIU is proportional to [T] / (1.22 + ([Perchlorate] + [Nitrate]/240 + [free SCN"]/15)
where:

The [ ] of each anion is express in |imol/L units

The Tonacchera Model is particularly useful because it models the key biological event
in the toxicity of perchlorate - the uptake of iodide by the thyroid by the NIS. The model
predicts the uptake of iodide to the simultaneous exposure to all four variables (i.e., iodide,
perchlorate, thiocyanate, and nitrate) affecting the uptake of iodide. The ability to predict the
amount of uptake of iodide knowing the body's total goitrogen load and the body's iodide
nutritional status is a powerful tool for assessing the risk from perchlorate exposure.

The NAS Committee recommends that the key biochemical event (i.e., the inhibition of
iodide uptake) should be used as the basis of the risk assessment (NAS 2005, p 14, 166, 169).
Therefore, the TIU is key the biochemical event that is perturbed by the NIS stresssors and
should be used as the basis for the risk assessment of this public health issue. The mode of
action identifies that a low TIU during pregnancy and nursing increases the risk of subtle mental
deficits in the children. However, the cumulative effect of all four NIS stressors on the thyroid
(i.e, not just perchlorate exposure) determines the TIU level during the critical time period of
pregnancy and nursing. In other words, the cumulative effect of the four NIS stressors argues
against perchlorate exposure alone as being directly associated with an adverse outcome. By

-------
10-P-0101

contrast, a low TIU during pregnancy and nursing is associated with an adverse outcome.
Therefore, the Tonacchera Model allows the TIU to be calculated from the concurrent exposure
from all four NIS stressors. Furthermore, the Tonacchera Model allows specific adverse effects
to be associated with specific calculated TIU level.

39

-------
10-P-0101

5. Deriving a Reference Dose (RfD)

The NAS Committee emphasized that its approach for deriving a perchlorate RfD differs
from the traditional single chemical risk assessment approach (NAS 2005, p 15). A traditional
single chemical risk assessment derives an RfD from an adverse effect such as a no-observed-
adverse-effect-level (NOAEL) or a lowest-observed-adverse-effect-level (LOAEL). By contrast,
the NAS Committee used the unconventional approach of using a nonadverse effect to derive its
recommended perchlorate RfD. The nonadverse effect was the inhibition of iodide uptake by the
NIS, which is labeled a no-observed-effect-level (NOEL). The NAS Committee used the NOEL
identified in the Greer study (Greer 2002) and set the perchlorate exposure level of 0.007 mg/kg-
day as the POD. The NAS Committee describes its approach of using a NOEL to derive the RfD
as conservative and health protective (NAS 2005, p 15).

The NAS Committee also proposed the following mode-of-action model for perchlorate
toxicity in humans (see below) (NAS 2005, p 167, figure 5-2). The NAS Committee identified
hypothyroidism as the first adverse effect in the continuum of biological effects. A traditional
single chemical risk assessment would use the perchlorate dose that just induces hypothyroidism
(i.e., a LOAEL), but the NAS Committee states that hypothyroidism should not be used as the
basis of a perchlorate risk assessment (NAS 2005, p 166). Overt hypothyroidism is characterized
by an elevated TSH level and a below normal T4 level. Furthermore, subclinical hypothyroidism
is characterized by an elevated TSH level and normal T4 level. Therefore, NAS Committee's
mode-of-action model has identified that transient changes of T4, T3, and TSH levels within a
normal concentration range are nonadverse health effects.

Won adverse
Effect

First

Source: NAS 2005, p 167, figure 5-2.

The final step in deriving an RfD is the application of UFs to the POD. The application
of UFs attempts to account for both uncertainty and the variability in the available data (EPA
2002, p 4-38, section 4.4.5). The seven UFs that can be potentially applied to a POD depending
in the nature of the dataset are: interspecies UF, intraspecies UF, LOAEL to NOAEL UF,

-------
10-P-0101

database UF, subchronic-to-chronic-duration UF, modifying factor, and the Food Quality
Protection Act. The Act allows a lOx safety factor to protect infants and children from pesticide
residues. The value of each UF depends on the quality of the studies available, the extent of the
database, and scientific judgment (EPA 2002, p 4-40, section 4.4.5.1). The NAS Committee
wrote, "No absolute rules exist for application of the [uncertainty] factors, and professional
judgment is a large component of their use." (NAS 2005, p 29). Each one of these UF can have
a value of 1, 3, or 10. If multiple UFs greater than 1 are used, then the UF values are multiplied
together (e.g., 10x 10x3 = 300). Since there is overlap in the coverage of uncertainty in the
UF, the upper limit of total UF applied to a POD should be no more than 3000 (EPA 2002, p 4-
41, section 4.4.5.1).

The NAS Committee used a POD (i.e., labeled as a NOEL) of 0.007 mg/kg-day identified
in the Greer study (Greer 2002) and applied a total UF of 10 (i.e., lOx intraspecies UF) to derive
the NAS Committee's recommended perchlorate RfD of 0.0007 mg/kg-day as the POD. By
contrast, EPA used a POD (i.e., labeled as a LOAEL) of 0.01 mg/kg-day from the Argus rat
study (Argus 2001) and applied a total UF of 300 (i.e., 3x intraspecies UF, lOx LOAEL to
NOAEL UF, 3x duration UF, and 3x database insufficiency, and a lx interspecies UF
(equivalent of omitting the UF)) to derive the EPA draft perchlorate RfD of 0.00003 mg/kg-day
(EPA 2002a, p 7-20-7-23). The following table summaries the differences between how the
NAS Committee and EPA derived a perchlorate RfD:

Perchlorate
Assessment

Biological Effect
Used as the POD

Critical Study

and
Type of POD

POD Value
(mg/kg-day)

Total
UF

Proposed
RFD*
(mg/kg-day)

NAS 2005

Inhibition of iodide uptake
by the NIS

Greer 2002 Human Study
NOEL
(nonadverse effect)

0.007

10

0.0007

EPA 2002

Brain Morphometry,
Thyroid Histopathology,
Thyroid hormone changes

Argus 2001 Rat Study
LOAEL
(adverse effect)

0.01

300

0.00003

* Proposed RfD = POD value Total UF

The scientific disagreement on the manner in which to derive the perchlorate RfD centers
around the following issues:

•	Should animal data or human data be used to determine the POD (NAS 2005, 14; EPA
2002a)?

•	Is the inhibition of iodide uptake an adverse or nonadverse effect (Ginsberg 2005; EPA
2002a, p 7-26, section 7-26)?

•	Which UFs should be applied? And if the UF is applied, at what magnitude (i.e., 1, 3, or
10) (Ginsberg 2005)?

Depending on the manner in which each of these scientific concerns is addressed, the
total UF applied to the POD will be different. Therefore, in a traditional single chemical risk
assessment, the numerical value of the RfD depends significantly on the total UF applied to the

41

-------
10-P-0101

POD. The UFs applied and their size is a matter of scientific judgment in order to account for
uncertainty and variability in the data.

5.1 Issues with the Single Chemical Risk Assessment of Perchlorate

A review of every scientific aspect in the perchlorate single chemical risk assessment is
beyond the capability, resources, and mission of the OIG. The OIG has identified that due to the
interaction of multiple factors, a cumulative risk assessment approach is needed to characterize
and address the risk in this public health issue. However, the OIG review specifically comments
on the following issues relating to a single chemical risk assessment of perchlorate because the
comments provide insight into the science of perchlorate:

•	Is the NAS Committee's unconventional approach of using a NOEL to derive its
recommended perchlorate RfD conservative and health protective? And, if so,
quantify how conservative the NAS recommend perchlorate RfD is as compared to
using the traditional single chemical risk assessment approach to deriving an RfD?

•	Evaluate whether hypothyroidism is the first adverse effect in the continuum of
biological effects from perchlorate exposure.

•	Evaluate the application of UFs to deriving a perchlorate RfD in a single chemical
risk assessment.

5.1.1 Defining an RfD

To evaluate the NAS Committee's unconventional approach of deriving an RfD, the
conventional approach of driving an RfD and a strict regulatory definition of an RfD must be
defined. Since an RfD is part of a risk assessment, the term risk assessment is defined:

Risk Assessment (in the context of human health): The evaluation of scientific
information on the hazardous properties of environmental agents (hazard
characterization), the dose-response relationship (dose-response assessment), and the
extent of human exposure to those agents (exposure assessment). The product of the risk
assessment is a statement regarding the probability that populations or individuals so
exposed will be harmed and to what degree (risk characterization) (EPA 2002; IRIS
2007).

As such, a risk assessment is comprised of four parts: hazard characterization, dose-
response assessment, exposure assessment, and risk characterization. The first portion of a risk
assessment is to determine the hazardous properties of the environmental agent (i.e., perchlorate)
in a hazard characterization, which is defined:

Hazard Characterization: A description of the potential adverse health effects
attributable to a specific environmental agent, the mechanisms by which agents exert
their toxic effects, and the associated dose, route, duration, and timing of exposure (EPA
2002; IRIS 2007).

42

-------
10-P-0101

The hazard characterization identifies the potential adverse health effects in humans
resulting from a specific agent. For the health characterization of perchlorate, the NAS
Committee states that the first adverse health effect in humans is hypothyroidism, which
precedes any potential for mental deficits in offspring.

For a single chemical risk assessment, the hazard characterization of only a specific agent
is a critical limitation that can fatally flaw the risk assessment. Multiple agents exerting their
toxic effects through the same mechanism are not considered together. This limitation can be
without consequence to the single chemical risk assessment if the single chemical is the
dominant factor causing the adverse effect. However, if the single chemical being studied is
only a minor factor among other stronger factors, this limitation in a single chemical risk
assessment can seriously distort the risk characterization. The OIG's use of cumulative risk
assessment will demonstrate the latter to be the situation with perchlorate.

The second part of a single chemical risk assessment is the dose-response assessment,
which is defined:

Dose-Response Assessment: A determination of the relationship between the magnitude
of an administered, applied, or internal dose and a specific biological response. Response
can be expressed as measured or observed incidence or change in a level of response,
percent response in groups of subjects (or populations), or the probability of occurrence
or change in level of response within a population (EPA 2002).

A dose-response assessment identifies that dose that causes an increased occurrence of a
particular biological event in the test subjects. For perchlorate, the NAS Committee identified
that a dose of 0.007 mg/kg-day of perchlorate does not statistically inhibit the uptake of iodide
by the NIS in adult test subjects as compared to controls subjects.

To conduct the third portion of a risk assessment (i.e., the exposure assessment), the RfD
needs to be calculated from the dose-response relationship identified in the dose-response
assessment. RfD is defined:

Reference dose (RfD): An estimate of a daily oral exposure for a given duration to the
human population (including susceptible subgroups) that is likely to be without an
appreciable risk of adverse health effects over a lifetime. It is derived from a benchmark
dose lower confidence limit (BMDL), a NOAEL, a LOAEL, or another suitable point of
departure, with uncertainty/variability factors applied to reflect limitations of the data
used (EPA 2002; IRIS 2007).

Notice that the definition of an RfD is based on avoiding adverse health effects and is
derived from a dose associated with an adverse effect (i.e., BMDL, NOAEL, LOAEL, or POD).
The definition of BMDL, NOAEL, LOAEL, and POD are defined below and all specifically
include the use of adverse effect:

BMDL: A statistical lower confidence limit on the dose or concentration at the
benchmark dose (BMD) or benchmark concentration (BMC) respectively (EPA 2002).

-------
10-P-0101

BMD or BMC: A dose or concentration that produces a predetermined change in
response of an adverse effect. . . compared to the background (EPA 2002).

NOAEL: The highest exposure level at which there are no biologically significant
increases in the frequency or severity of adverse effect between the exposed population
and its appropriate control . . . (EPA 2002; IRIS 2007).

LOAEL: The lowest exposure level at which there are biologically significant increases
in frequency or severity of adverse effects between the exposed population and its
appropriate control group (EPA 2002; IRIS 2007).

POD: The dose-response point that marks the beginning of a low-dose extrapolation.

This point can be the lower-bound on dose for an estimated incidence or a change in
response level from a dose-response model (BMD), or a NOAEL or LOAEL for an
observed incidence, or change in level of response (EPA 2002; IRIS 2007).

Therefore, an RfD is defined by the adverse effect from which it is derived. The RfD is a
dose that is typically one to three orders of magnitude (i.e., 10 to 1000 times) below the NOAEL
or LOAEL from with it was derived by the use of uncertainty/variability factors applied to reflect
limitations of the data used. Thus, the RfD is set at a dose that is below an exposure that is likely
to cause the observed adverse health effects in even the most sensitive group.

The exception to the use of an adverse effect to derive an RfD is that the risk assessment
guidance allows the use of the immediate precursor to the adverse effect. This is seen in the
definition of a critical effect:

Critical Effect: The first adverse effect, or its known precursor, that occurs to the most
sensitive species as the dose rate of an agent increases (EPA 2002; IRIS 2007).

In regard to perchlorate, since the NAS Committee identified hypothyroidism as the first
adverse effect, its immediate precursor in the NAS mode-of-action model (NAS 2005, p 167,
figure 5-2) is listed as thyroid hypertrophy or hyperplasia, but hypertrophy and hyperplasia are
not typically measured in epidemiological studies. Since overt hypothyroidism is characterized
by an elevated TSH level and a below normal T4 level, and subclinical hypothyroidism is
characterized by an elevated TSH level and normal T4 level, the immediate precursor to
hypothyroidism could be considered a perchlorate exposure that induces a statistically significant
change in thyroid hormones levels in the exposed group as compared to the control group.

However, the NAS Committee did not derive its RfD from either hypertrophy,
hyperplasia, or thyroid hormone changes. The NAS Committee derived its perchlorate RfD from
the NOEL observed from the inhibition of iodide uptake by the NIS in the Greer study. A NOEL
is defined as:

NOEL: An exposure level at which there are no statistically or biologically significant
increases in the frequency or severity of any effect between the exposed population and
its appropriate control (EPA 2002; IRIS 2007).

-------
10-P-0101

In the NAS Committee's mode-of-action mode, the inhibition of iodide uptake by the
thyroid is three biological steps before hypothyroidism. Furthermore, the inhibition of iodide
uptake in the thyroid is not an immediate precursor to hypothyroidism. When the biology allows
for the identification of multiple NOAELs, each associated with different biological response,
the highest NOAEL is used for regulatory purposes. For perchlorate, this means a NOAEL
identified for changes in thyroid hormones (i.e., the immediate precursor to hypothyroidism).

This concept is included in the following NOAEL definition used by EPA's Office of
Communications, Education, and Public Affairs (OCEPA):

EPA OCEPA NOAEL: An exposure level at which there are no statistically or
biologically significant increases in the frequency or severity of adverse effects between
the exposed population and its appropriate control; some effects may be produced at this
level, but they are not considered as adverse, or as precursors to adverse effects. In an
experiment with several NOAELs, the regulatory focus is primarily on the highest one,
leading to the common usage of the term NOAEL as the highest exposure without
adverse effects.

Therefore, the NAS Committee's unconventional derivation of an RfD from the
inhibition of iodide uptake (i.e., NOEL) does not meet the strict definition of an RfD in that it is
derived from an adverse effect. Since the NAS Committee's recommended RfD was derived
from a biological response several steps before the adverse effect, the POD value used is at a
dose below a NOAEL or LOAEL that would be conventionally used for deriving an RfD. Thus,
in our opinion, the NAS Committee used a conservative approach of using a NOEL to derived its
RfD, but the degree of conservatism (i.e., quantitatively) is not identified because the value of a
conventionally derived perchlorate RfD is not known since adverse health effects have not been
clearly demonstrated in any human population exposed to perchlorate (NAS 2005, p 177).

The NAS Committee also states its recommended RfD is health protective, but evaluating
this is difficult because the amount of TIU that causes sufficient thyroid stress in pregnant
women to induce mental deficient in their children is not defined in the single chemical risk
assessment approach. Furthermore, following the single chemical risk assessment process does
not guarantee a health protective RfD. By contrast, the application of a cumulative risk
assessment approach expands the understanding of the stressors that affect this public health
issue and provide a clearer, more complete picture of this public health issue for making risk-
management decisions. Therefore, the cumulative risk assessment results can be used to
evaluate whether the NAS Committee-recommended RfD is health protective.

5.1.2 Evaluation of the Application of UFs for Deriving the Perchlorate RfD

The NAS Committee used a nonadverse POD of 0.007 mg/kg-day identified from the
human exposure to perchlorate in the Greer study (Greer 2002) to derive its recommended
perchlorate RfD. By comparison, the EPA used an adverse POD of 0.01 mg/kg-day identified in
the Argus rat study to derive EPA's draft RfD. Note that the respective PODs of 0.007 mg/kg-
day and 0.01 mg/kg-day are essentially the same. Therefore, the difference in the NAS and EPA
RfDs arise from the selection of which UFs are applied (i.e., interspecies UF, intraspecies UF,
LOAEL to NOAEL UF, database UF, subchronic-to-chronic-duration UF, modifying factor) and

45

-------
10-P-0101

the magnitude of each UF applied (i.e., 1, 3, or 10). The NAS Committee used a total UF of 10
while EPA used a total UF of 300.

The EPA risk assessment guidance indicates the preference to use human data over
animal data to set a POD. EPA risk assessment guidance states, "Human or animal data can be
used to derive an RfD, but when adequate and sufficient human data on the most sensitive effect,
the EPA guidance states that the human data should be used" (EPA 2002, p 4-12). Human data
is preferred because it minimizes the uncertainty of extrapolating adverse effects observed in
animal studies to potential effects in humans. This uncertainty is the source and reason an
interspecies UF is often applied when deriving an RfD from animal data (EPA 2002, p 4-42,
section 4.4.5.2). Although neither dataset is ideal because the amount of iodide inhibition that
induces subtle mental deficits in children during gestation is not identified by these studies,
human data are preferred over animal data.

The use of human data is critical for applying a cumulative risk assessment approach to
this public health issue. In animal studies, the ability to test the cognitive abilities of the exposed
animals is very crude and insensitive to subtle mental deficits. For example, how do you
measure a subtle loss of verbal comprehension in a rat? Or, how do you measure a subtle
decrease in the attentiveness of a rat? By comparison, measuring subtle mental deficits such as
verbal comprehension or ADHD in human children is relatively straightforward. The human
exposure to each of the four NIS stressors can be measured or estimated for use in a cumulative
risk assessment (i.e., varying the exposure of the four NIS stressors in an animal study is too
complicated, expensive, and relatively insensitive to identifying the endpoint of subtle mental
deficits). The Tonacchera Model allows the combined effect on the NIS to be determined and
identified into a single variable, TIU. In contrast to the limitations of a single chemical risk
assessment, a cumulative risk assessment can identify the %TIU in pregnant women that is
associated with subtle mental deficits in their children. Therefore, for this public health issue,
animal data are not an adequate substitute for human data (i.e., animal data can supplement, but
not replace, human data for this public health issue).

The NAS Committee used a nonadverse POD to derive its recommended perchlorate
RfD. However, not everyone in the scientific community agrees that the inhibition of iodide
uptake is a nonadverse effect (Ginsberg 2005; EPA 2002a, p 7-26, section 7.1.5.1).

Toxicologists can have different interpretations of what constitutes an adverse effect (Dorato
2005). An approach to classifying effects as adverse or nonadverse are discussed in Dorato's
paper (Dorato 2005, figure 5). If the inhibition of iodide uptake is considered an adverse effect
(e.g., a LOAEL), a LOAEL to NOAEL UF would be applied with a magnitude of 3 or 10
(Ginsberg 2005). So, the issue of whether the inhibition of iodide uptake is adverse or
nonadverse simply affects the amount of total UFs to be applied to the POD to derive the RfD
(see next paragraph - the impact of total UF on this public health issue).

The evaluation of the appropriate selection and magnitude of each UF to be applied to the
derivation of a perchlorate RfD is difficult due to the lack of a clear scientific consensus of their
application to the perchlorate dataset. The value of each UF depends on the quality of the studies
available, the extent of the database, and scientific judgment (EPA 2002, p 4-40, section 4.4.5.1).
The NAS Committee wrote, "No absolute rules exist for application of the [uncertainty] factors,

46

-------
10-P-0101

and professional judgment is a large component of their use." (NAS 2005, p 29). However, the
total UF used directly affects the calculated RfD value and any resulting regulatory exposure
limits set from the RfD. The scientific considerations center around which UF to apply and at
what magnitude. However, there is no scientific discussion if the application of UFs to the POD
accomplishes the intent of an RfD. An RfD sets an oral exposure limit that is likely to prevent
adverse health effects from occurring in the human population over a lifetime.

The perchlorate RfD sets an exposure limit to only one of four known NIS stressors.

Even if the perchlorate RfD is never exceeded in a population, the combined effect from various
exposure level combinations from the other three uncontrolled NIS stressors (i.e., lack of iodide,
thiocyanate, and nitrate) can still generate an unacceptable adverse health effects in humans. All
four NIS stressors conspire to limit the amount of iodide uptake in the thyroid, which initiates all
subsequent adverse health effects, not just perchlorate exposure. The use of the Tonacchera
Model in a cumulative risk assessment approach can identify the %TIU(Loael) and %TIU(noael)
in pregnant women that is associated with subtle mental deficits in their children. The use of
UFs should be applied to the %TIU(loael) or %TIU(noael) to set a %TIU(RfD) exposure limit
(i.e., the integrated effect of all four NIS stressors) and not the POD for perchlorate. The mode
of action identifies that adverse effects are avoided if the amount of TIU in the thyroid is
sufficiently high. Limiting exposure of only perchlorate does not guarantee an acceptable TIU
level. Only by managing the exposure to all four NIS stressors is the TIU level guaranteed to be
a sufficiently high %TIU to avoid adverse effects.

The demonstration of this principle can be seen in the cumulative risk assessment (see
Section 9.3). When the total applied UF value is increased from 10 to 500 and is applied only to
the perchlorate exposure, the corresponding DWEL shifts from 24.5 ppb and 0.5 ppb,
respectively. The calculated increase in %TIU in an adult is estimated to be 0.4% when the
perchlorate DWEL is lowered from 24.5 ppp to 0.5 ppb. In a pregnant woman, our analysis
estimates a 1.0% increase in the %TIU when the perchlorate DWEL is lowered from 24.5 ppb to
6.1 ppb (see Section 9.5). A change of 1.0%TIU in pregnant women is below the statistical
detection in a human radioactive iodide uptake exposure study (e.g, the Greer study). Our
cumulative risk assessment estimates that the %TIU would have to fall 50% from a healthy
iodide uptake level for pregnant women to potentially start inducing subtle mental deficits in
their children. Therefore, the particular total UF (i.e., 10 to 500) applied to the perchlorate POD
in a single chemical risk assessment is not particularly significant in maintaining a sufficient
iodide uptake during pregnancy. By contrast, our analysis indicates that the application of UF
needs to be applied to the %TIU(noael) level, because a low %TIU (and not perchlorate
exposure only) is associated with an adverse outcome. Therefore, the use of a cumulative risk
assessment approach is essential to accurately characterize and to understand the risks affecting
this public health issue and to formulate an effective response(s) to the problem. The cumulative
risk assessment indicates that the dominant stressor in this public health issue is the lack of
iodide stressor, and because of iodide deficiency during pregnancy, about 29% of pregnant
women are below %TIU(noael>

-------
10-P-0101

5.1.3 First Adverse Effect: Hypothyroidism or Hypothyroxinemia

The NAS Committee identified hypothyroidism as the first adverse effect, but
recommended against using hypothyroidism as the basis for a perchlorate risk assessment (NAS
2005, p 166-67). The fetus needs a sufficient supply of T4 during gestation for proper brain
development. The fetus is entirely dependent on the maternal supply of T4 through mid-
gestation (Glinoer 2005, p 1095) and is still sensitive to maternal supply thereafter. Maternal
hypothyroidism is toxic to the fetus because it causes a decreased maternal supply of T4 during
gestation. The adverse effects on the fetus from maternal hypothyroidism are reported in a 1999
New England Journal of Medicine study. This study reported on maternal thyroid deficiency and
the subsequent neuropsychological development of the children (Haddow 1999). The study
identified 47 pregnant women with elevated TSH (i.e., > 6 ul/ml) and another 15 pregnant
women with both elevated TSH and low T4 levels. Of the 48 pregnant women who did not
receive treatment for their thyroid deficiency, the full-scale intelligence quotient (IQ) of their
children averaged 7 points lower (p = 0.005) than the 124 matched control children.

Hypothyroidism is an adverse effect from an extremely low uptake of iodide by the
thyroid, but hypothyroidism is not the first adverse effect from perchlorate exposure.
Hypothyroxinemia is the first adverse effect from a low uptake of iodide by the thyroid. Iodide
deficiency and excess NTS inhibition (i.e., perchlorate exposure) act through the same
mechanism (i.e., each result in a low uptake of iodide by the thyroid). The Tonacchera Model
indicates that the amount of iodide taken up into the thyroid by the NIS is directly related to the
amount of iodide in the blood. If the diet is poor in iodide, the amount of iodide uptake will be
low, regardless of how much or how little NIS inhibition is occurring. Therefore, the lack of
iodide in the diet has the same mechanism of toxicity and outcome as a high NIS inhibition level;
not enough iodide is taken up by the thyroid for the sufficient production of thyroid hormones.
Therefore, the adverse effects caused by iodide deficiency are the same adverse effects that
would be expected from a high NIS inhibition level when not enough iodide is taken up by the
thyroid.

Iodide deficiency is frequently and incorrectly associated with hypothyroidism and
increased TSH (Obregon 2005, p 918). The TSH-independent autoregulation is often
overlooked, or actually not known to younger Western-trained physicians (Obregon 2005,
p 918). Atrophy of the thyroid and overt hypothyroidism occur only when iodide deficiency is
accompanied by a high goitrogen load and/or selenium deficiency (Obregon 2005, p 918).
Clinical or subclinical hypothyroidism is not observed in uncomplicated iodide deficiency
(Obregon 2005, p 918). By contrast, iodide deficiency (i.e., the lack of an adequate uptake of
iodide by the thyroid) induces hypothyroxinemia, a thyroid condition characterized by a
decreased T4 serum level and a normal or slightly elevated T3, without an increase in TSH levels
(i.e., TSH levels are normal) (Obregon 2005, p 918). Mild to moderate iodide deficiency is the
most widespread cause of maternal hypothyroxinemia (Morreale de Escobar 2004, p U25). The
decrease in T4 in pregnant women with hypothyroxinemia in the first 20 weeks of pregnancy is
associated with mental deficits and an increased frequency of ADHD in their children
(Vermiglio 1994). Therefore, maternal hypothyroxinemia induces fetal brain damage through
the same cause as maternal hypothyroidism - a decreased maternal supply of T4.

-------
10-P-0101

5.1.4 Comparison of the NAS Unconventional RfD with a Conventionally Derived
Excess NIS Inhibition RfD

The NAS Committee states that adverse health effects have not been clearly
demonstrated in any human population exposed to perchlorate (NAS 2005, p 177). This lack of
clear demonstration prevents the calculation of a perchlorate RfD using the conventional single
chemical risk assessment approach of deriving an RfD from a NOAEL or LOAEL. However,
perchlorate is only one of three NIS inhibitors continuously acting on the thyroid to prevent the
uptake of iodide. The other two ubiquitously consumed NIS inhibitors are thiocyanate and
nitrate. The Tonacchera Model identifies the relative potency factor for each of the three NIS
inhibitors relative to perchlorate and identifies the interaction of the chemical mixture as simple
competitive interaction (Tonacchera 2004). There is no evidence of synergism or antagonism in
chemical mixtures of these three NIS inhibitors. Therefore, the total amount of NIS inhibition
acting on the thyroid is the combined contribution of each anion on a molar potency adjusted
basis (i.e., a dose-addition method).

The principle of dose addition can be utilized to independently calculate an excess NIS
inhibition RfD and compare it to the NAS-recommended perchlorate RfD. Unlike perchlorate in
which adverse effects have not been documented in humans, adverse effects have been
documented in human populations exposed to excess amounts of the other two NIS inhibitors,
thiocyanate and nitrate. Since the NIS inhibition from thiocyanate and nitrate are
indistinguishable from the NIS inhibition from perchlorate (Tonacchera 2004), the amount of
NIS inhibition in these human populations can be measured and expressed mathematically as a
PEC. Then the conventional single chemical risk assessment approach can use the observed
NOAELs and LOAELs in these human populations to independently derive an excess NIS
inhibition RfD (i.e., a descriptive name for what the perchlorate RfD represents). The excess
NIS inhibition RfD (i.e., which was derived through the conventional approach of using a
NOAEL or LOAEL) can be compared to the NAS-recommended perchlorate RfD (i.e., which
was derived through a unconventional approach of using a NOEL) to quantify how conservative
the NAS-recommended perchlorate RfD might be.

The OIG identified the following three epidemiological studies to calculate an excess NIS
inhibition RfD in which adverse and nonadverse thyroid effects can be observed in human
populations exposed to excess amount of the NIS inhibitors thiocyanate and nitrate:

1. Severe hypothyroxinemia can be observed in the electroplating workers exposed to
cyanide resulting in an excess exposure to the NIS inhibitor thiocyanate (Banerjee 1997).
(See Section 5.1.4.1).

2. Hypothyroxinemia can be observed in women consuming thiocyanate-preserved cow's
milk, which resulted in an excess exposure to the NIS inhibitor thiocyanate (Banerjee
1997b). (See Section 5.1.4.2).

3. Hypertrophy (i.e., an increase in the size of the thyroid) can be observed in children
consuming nitrate-contaminated water, which resulted in an excess exposure to the NIS
inhibitor nitrate (Tajtakova 2006). (See Section 5.1.4.3).

-------
10-P-0101

The NAS Committee states that transient changes in serum thyroid hormones and TSH
concentrations are not adverse health effects (NAS 2005, p 166). Furthermore, the NAS
Committee states that the serum thyroid hormone concentrations should not be used as the POD
for the risk assessment (NAS 2005, p 169). The OIG Analysis of the thyroid function concurs
with this assessment. The thyroid is in a continuous dynamic process (i.e., homeostasis) to
maintain the serum T4 and T3 level within a normal range under ever-changing levels of
exposure to the four NIS stressors. Transient changes in serum thyroid hormones and TSH
concentrations are a reflection of the thyroid working to maintain equilibrium. In contrast to
using transisent changes in serum thyroid hormone and TSH concentrations as the POD, the OIG
Analysis identified two thiocyanate exposure studies (i.e., 1. and 2. from above) in which a long-
term exposure to excess thiocyanate exceeded the thyroid's ability to adapt and maintain a serum
T4 level within a normal range. This is appropriate because the mode of action for this public
health issue identifies that the failure of the thyroid to maintain an adequate serum T4 level
during pregnancy and lactation induces the adverse effects in the offspring. Our analysis
identified a study (i.e., 3. from above) in which a long-term NIS stressor exposure to nitrate
resulted in hypertrophy in the school-aged children.

50

-------
10-P-0101

5.1.4.1 Calculation of an Excess NIS Inhibition RfD Using the Adverse Thyroid
Health Effects Observed in Electroplating Workers Exposed to Cyanide in Adult
Men

Background

Sodium cyanide and potassium cyanide are salts used in the electroplating industry. The
cyanide containing electroplating baths caused occupational exposure of the electroplating
workers to cyanide. Once cyanide is absorbed into the body, about 80% of absorbed cyanide is
metabolized to thiocyanate (SCN") by the sulfurtransferase enzyme, rhodanese (WHO 2004).

Description of Banerjee's Thiocyanate Study in Adult Males

Of 201 male workers in the plant, 35 male workers had worked in the electroplating
process area for at least 5 consecutive years (i.e., the exposed group) (Banerjee 1997). Another
35 male workers who worked outside the manufacturing building were matched for age and
dietary habits (i.e., nonexposed group - control). All subjects were selected randomly from
among employees who did not use tobacco products and who had no history of thyroid disease.

Five mis of blood were collected from each subject. The blood serum was tested for
serum T4, T3, TSH, and thiocyanate.

Iodide Nutritional Status

The iodide nutritional status of the test subjects was not evaluated in this study.
Specifically, no urinary samples were taken to measure iodide excretion levels in this study.

Banerjee's Thiocyanate Study Results

Both the T4 and T3 hormone levels in the exposed workers were all statistically lower
(i.e., p > 0 .05) than the nonexposed workers (i.e., the control group). Both T4 and T3 hormone
levels in the exposed workers are below the normal range for these thyroid hormones. The
exposed workers' TSH value is statistically higher (i.e. ,P>0 .05) than the nonexposed workers.
However, the exposed workers' TSH levels were elevated over the controls, but within the
normal range of 0.2 to 4.0 |iU/ml. The following table summaries the study results:

Study Group

Thiocyanate
(|jmol/L)

t4

(MQ/dL)

t3

(MQ/dL)

TSH
(pU/ml)

35 Nonexposed Workers
(all nonsmokers)

90.8 ±9.02

6.09 ±0.601

111.0 ±9.3

1.20 ±0.301

35 Exposed Workers
(all nonsmokers)

316 ±15.0*

3.81 ± 0.318|

57.2 ±8.1|

2.91 ± 0.20|

Normal Ranges Cited in the
Manufacturers Test Kits



5.5 to 13.5

60 to 200

0.2 to 4.0

* p < 0.01
t p < 0.05

Source: Banerjee 1997, table 1.

51

-------
10-P-0101

Amount of Excess Thiocyanate

The nonexposed workers (all nonsmokers) had a serum thiocyanate concentration of 90.8
± 9.02 (amol/L. The exposed workers had a serum thiocyanate concentration of 316 ± 15.0
(amol/L. Therefore the amount of excess thiocyanate measured in the serum was 225.2 jamol of
SCN7L (i.e., 316 pmol/L - 90.8 pmol/L).

Amount of Excess NIS Inhibition in the Exposed Workers

The amount of excess NIS inhibition from thiocyanate in the exposed workers is
converted to the SPEC (concept taken from Tonacchera 2004):

SPEC = (Amt of CIO4" Inhibition) + (Amt of NO3" Inhibition) + (Amt of SCN" Inhibition)
SPEC = [Perchlorate] + [Nitrate] /240 + ([total serum SCN ] x 0.5) /15
SPEC = [Perchlorate] + [Nitrate] /240 + [free SCN ] /15

Where: [ ] is molarity

Serum inhibition potency of perchlorate = 1
Serum inhibition potency of nitrate relative to perchlorate = 1 / 240
Serum inhibition potency of free thiocyanate relative to perchlorate = 1/15
[free SCN" ] = [total serum SCN ] x 0.5 free SCN" per total serum SCN"

However, the serum molarity concentration for perchlorate and nitrate are neither
measured nor reported in the Baneijee study (i.e., they are potentially confounding factors).
Assuming the perchlorate and nitrate exposure in the Banerjee test subjects is constant between
the exposed workers and the nonexposed workers (i.e., controls). Then, the SPEC equation can
be simplified to express only the excess amount of NIS inhibition from the thiocyanate. Thus:

Excess SPEC(Exposed workers) ([excess SCN ] X 0.5) • 15

= (225.2 pmoles of SCN7L x 0.5) - 15
= 7.51 pmol/L SPEC

Equivalent Amount of Perchlorate Ingestion Corresponding to the Excess SPEC

The Clewell perchlorate physiologically based pharmacokinetic (PBPK) model is used to
calculate the external perchlorate exposure (mg/kg-day) from the excess SPEC observed in the
exposed workers in the Banerjee test subjects (Clewell 2007). The excess SPEC of 7.51 pmol/L
SPEC must be converted into pg/L SPEC unit for use in the Clewell Perchlorate PBPK Model.

pg/L SPEC(ExPosed workers) =7.51 pmol/L SPEC x 99.45 pg perchlorate/pmol

— 747 |Ig/L SPEC(Exposed workers) OI" 0.747 mg/L SPEC(Exposed workers)

The external perchlorate dose required to generate the amount of excess NIS inhibition
observed in the exposed workers is calculated using the Clewell PBPK model as follows:

52

-------
10-P-0101

Ext. C104" Dose = 0.747 mg/L SPEC(Exposed workers) x 1.0 mg/kg-day

1.0 mg/L

= 0.75 mg/kg-day (rounded)

The exposed workers in the Banerjee study were unable to maintain a normal T4 or T3
thyroid hormone level while still maintaining a TSH level in the normal range (refer to table
above). Overt hypothyroidism is characterized by an elevated TSH and below normal T4 while
subclinical hypothyroidism is characterized by an elevated TSH and normal T4. However,
hypothyroxinemia is a less severe thyroid condition and is a common condition in pregnant
women characterized by low maternal fT4 levels with normal TSH levels (Kooistra 2006).
Therefore, the exposed workers' thyroid hormone levels could be described as severe
hypothyroxinemia. Maternal hypothyroxinemia (i.e., during pregnancy) is documented to be
associated with mental deficits (e.g., ADHD) in the children of those mothers (Vermiglio 2004).
The mechanism of fetal brain damage from maternal hypothyroxinemia is the same as maternal
hypothyroidism (i.e., an insufficient maternal supply of T4 to the fetus during gestation).
Therefore, although hypothyroxinemia is not a permanent adverse effect in adult males, maternal
hypothyroxinemia is an adverse effect in the most sensitive populations, pregnant women,
fetuses, and nursing infants. Thus, the OIG has identified the observed severe hypothyroxinemia
in the Baneijee study exposed workers to be an adverse effect (i.e., if severe hypothyroxinemia
were to occur in pregnant woman population) from which an excess NIS inhibition RfD can be
calculated from.

The excess thiocyanate measured in the exposed workers is converted by the use of the
Tonacchera Model and the Clewell perchlorate PBPK model to calculate an external perchlorate
dose that corresponds to the LOAEL of NIS inhibition in an adult male human. In other words,
an external perchlorate exposure of 0.75 mg/kg-day (or about 55 mg/day for a 70 kg adult male)
should generate the same thyroid effects (i.e., severe hypothyroxinemia) observed in the exposed
workers in the Banerjee study.

Calculation of an Excess NIS Inhibition RfD from the Excess Thiocyanate
Exposure

The severe hypothyroxinemia observed in the Baneijee exposed workers is used to
identify the LOAEL for the calculation of the excess NIS inhibition RfD. Therefore, the
observed LOAEL in an adult male is:

LOAEL(Exposed workers) — 0.75 mg/kg-day

The calculation of an NOAEL from a LOAEL is typically accomplished by applying a
UF of 10 to the LOAEL (EPA 2002, p 4-44):

NOAEL(ExPosed workers) = 0.75 mg perchlorate equivalent/kg-day / 10 UF
= 0.075 mg perchlorate equivalent/kg-day

Since the exposed workers are all men, who are less sensitive than women to adverse
thyroid effects, a full UF of 10 is applied to the NOAEL to account for the intraspecies factor to

53

-------
10-P-0101

protect the most sensitive populations. The application of a 10 UF in this calculation is
analogous to the NAS Committee applying a total UF of 10 to the NOEL to protect the most
sensitive population. The NAS Committee used a NOEL to determine the RfD from the excess
perchlorate consumption in the Greer study and a 10 UF was applied to it to derive NAS-
recommended perchlorate RfD (NAS 2005, p 16). The OIG is using a UF of 10 for these
calculations to allow for the direct comparison of the excess NIS inhibition RfD with the NAS-
recommended perchlorate RfD of 0.0007 mg/kg-day. Therefore, the excess NIS inhibition RfD
(from excess SCN" exposure) is calculated as follows:

Excess NIS Inhibition RfD(Aduit males) = 0.075 mg perchlorate equivalent/kg-day / 10 UF

= 0.0075 mg/kg-day

The calculated excess NIS inhibition RfD from excess SCN" exposure in adult males is
0.0075 mg/kg-day. EPA's perchlorate RfD is 0.0007 mg/kg-day obtained from the NAS
Committee. The calculated excess NIS inhibition RfD(Aduit males) is 10.7 times greater than the
NAS-recommended perchlorate RfD.

The advantage of determining the excess NIS inhibition RfD is because it used a
conventional risk assessment approach of deriving an RfD by using a LOAEL observed in a
human population. The NAS Committee describes its own approach of deriving an RfD from a
NOEL as being an unconventional approach, but characterizes it as being a "conservative,
health-protective approach to the perchlorate risk assessment" (NAS 2005, p 15). The OIG's
conventional approach for deriving a perchlorate RfD independently supports the NAS
Committee's statement that its recommended perchlorate RfD is conservative. The excess NIS
inhibition RfD(Aduit males) observed in the Baneijee thiocyanate study of electroplating workers
indicates that the NAS-recommended perchlorate RfD is conservative by a factor of 10.3.

-------
10-P-0101

5.1.4.2 Calculation of an Excess NIS Inhibition RfD Using the Adverse Thyroid
Health Effects from the Ingestion of Thiocyanate-Preserved Cow's Milk by Adult
Women

Background

In Calcutta, India, local dairies add 30-50 mg/L of thiocyanate to preserve cow's milk in
order to activate the lactoperoxidase-thiocyanate-hydrogen peroxide system (LP system). The
LP system forms the hypothiocyanite anion (OSCN), which acts as a bacteriostat to retard the
growth of bacteria. The subsequent ingestion of thiocyanate-preserved cow's milk causes a
relatively high thiocyanate exposure to the consumer.

Description of Banerjee's Thiocyanate Study in Adult Women

Thirty-five women who had consumed 250 ml/day of thiocyanate-preserved milk for
more than 5 consecutive years were designated as the exposed group (Banerjee 1997b). Another
35 women who consumed the same quantity of nonpreserved raw milk were designed the control
group (i.e., the nonexposed group). The control group was matched for age and dietary habits.
All subjects were selected randomly from a population of similar socioeconomic status and
uniform dietary pattern. Furthermore, all test subjects did not use tobacco products and had no
history of thyroid disease.

Five mis of blood were collected from each subject. The blood serum was tested for
serum T4, T3, TSH, and thiocyanate. Iodide levels were measured by urinary iodide
concentration (UIC).

iodide Nutritional Status

The iodide nutritional status of the test subjects was evaluated in this study by UIC. The
UIC of the exposed women was found to be 115 |ig/L (SE 8.5). The UIC of the control group
was found to be 123 |ig/L (SE 7.8). The difference in the UICs between the two groups was not
statistically significant.

Banerjee's Thiocyanate Study Results

Hypothyroxinemia is a thyroid condition characterized by below normal T4 and normal
T3 and TSH levels. Hypothyroxinemia was observed in thiocyanate-exposed female in the
Banerjee adult female thiocyanate study (Banerjee 1997b). The thiocyanate-exposed women
were measured to have below normal T4, and both normal T3 levels and statistically elevated
TSH levels, but the TSH levels were still in the normal range. This level of hypothyroxinemia is
less severe than the hypothyroxinemia observed in the Banerjee adult male thiocyanate study
(Banerjee 1997). The exposed females were measured to have a serum thiocyanate exposure of
230 ± 10.0 (.imol/L. The control women were measured to have a serum thiocyanate exposure of
90.8 ± 9.0 (.imol/L. The observations of the Banerjee study are summarized below:

-------
10-P-0101

Study Group

Thiocyanate
(pmol/L)

t4

(nmol/L)

t3

(nmol/L)

TSH
(uU/ml)

35 Nonexposed Females
(all nonsmokers)

90.8 ±9.0

125.4 ±11.5

1.71 ±0.16

1.09 ±0.28

35 Exposed Females
(drank SCN" preserved milk)

230.0 ± 10.0*

87.8 ±6.6*

2.39 ±0.32

2.49 ±0.20*

Normal Ranges Cited in the
Manufacturers Test Kits



110 to 279

0.93 to 3.12

0.2 to 4.0

* p < 0.01

Source: Banerjee 1997b, table 1.

Amount of Excess Thiocyanate

The control women had a serum thiocyanate concentration of 90.8 pmol/L. The exposed
women had a serum thiocyanate concentration of 230 pmol/L. Therefore, the amount of excess
thiocyanate measured in the serum was 139.2 pmol of SCN7L (i.e., 230 pmol/L - 90.8 pmol/L).

Amount of Excess NIS Inhibition in the Exposed Women

The amount of excess NIS inhibition from thiocyanate in the exposed workers is
converted to the SPEC (concept taken from Tonacchera 2004):

SPEC = (Amt of CIO4" Inhibition) + (Amt of NO3" Inhibition) + (Amt of SCN Inhibition)
SPEC = [Perchlorate] + [Nitrate] /240 + ([total serum SCN ] x 0.5) /15
SPEC = [Perchlorate] + [Nitrate] /240 + [free SCN ] /15

Where: [ ] is molarity

Serum inhibition potency of perchlorate = 1
Serum inhibition potency of nitrate relative to perchlorate = 1 / 240
Serum inhibition potency of free thiocyanate relative to perchlorate = 1/15
[free SCN ] = [total serum SCN] x 0.5 free SCN- per total serum SCN

However, the serum molarity concentration for perchlorate and nitrate are neither measured
nor reported in the Banerjee study (i.e., they are potentially confounding factors). Assuming the
perchlorate and nitrate exposure in the Banerjee test subjects is constant between the exposed
workers and the nonexposed workers (i.e., controls), then the SPEC equation can be simplified to
express only the excess amount of NIS inhibition from the thiocyanate. Thus:

Excess SPEC(EXposed workers) ([excess SC^N ] X 0.5) • 15

= (139.2 pinoles of SCN/L x 0.5) - 15
= 4.64 pmol/L SPEC

Equivalent Amount of Perchlorate Ingestion Corresponding to the Excess SPEC

The Clewell Perchlorate PBPK Model is used to calculate the external perchlorate
exposure (mg/kg-day) from the excess SPEC observed in the exposed women (Clewell 2007).

56

-------
10-P-0101

The excess SPEC of 4.90 |imol/L SPEC must be converted into |ig/L SPEC unit for use in the
Clewell Perchlorate PBPK Model.

|ig/L SPEC(Exposed women) = 4.64 |iinol/L SPEC x 99.45 |ig perchl orate/|iinol

— 461 |ig/L SPEC(Exposed women) Or 0.461 mg/L SPEC(Exposed women)

The external perchlorate dose required to generate the amount of excess NIS inhibition
observed in the exposed workers is calculated using the Clewell PBPK model as follows:

Ext. C104" Dose = 0.461 mg/L SPEC(Exposed women) x 1.0 mg/kg-day

1.0 mg/L

= 0.461 mg/kg-day

The exposed women in the Baneijee study were unable to maintain a normal T4 thyroid
hormone level while still maintaining a TSH level in the normal range (refer to table above).
The SPEC level at which overt hypothyroidism (i.e., elevated TSH and below normal T4) or
subclinical hypothyroidism (i.e., elevated TSH and normal T4) is not known by experimental
data. However, hypothyroxinemia is a less severe thyroid condition and is a common condition
in pregnant women characterized by low maternal fT4 levels with normal TSH levels (Kooistra
2006). Therefore, the exposed women's thyroid condition could be described as
hypothyroxinemia. Maternal hypothyroxinemia (i.e., during pregnancy) is documented to be
associated with mental deficits (e.g., ADHD) in the children of those mothers (Vermiglio 2004).
The mechanism of fetal brain damage from maternal hypothyroxinemia is the same as maternal
hypothyroidism (i.e., an insufficient maternal supply of T4 to the fetus during gestation).
Therefore, although hypothyroxinemia might not have overt adverse symptoms in adult women,
maternal hypothyroxinemia is an adverse effect in the most sensitive populations - pregnant
women, fetuses, and nursing infants. Thus, the OIG has identified the observed
hypothyroxinemia in the Banerjee exposed women to be an adverse effect from which an excess
NIS inhibition RfD can be calculated.

The excess thiocyanate measured in the exposed women is converted by the use of the
Tonacchera Model and the Clewell perchlorate PBPK model to calculate an external perchlorate
dose that corresponds to the LOAEL of NIS inhibition in an adult male human. In other words,
an external perchlorate exposure of 0.461 mg/kg-day (or 32 mg/day for a 70 kg adult female)
should generate the same thyroid effects (i.e., hypothyroxinemia) observed in the exposed
women in the Banerjee study.

Calculation of an Excess NIS Inhibition RfD from the Excess Thiocyanate
Exposure

The severe hypothyroxinemia observed in the Baneijee exposed workers is used to
identify the LOAEL for the calculation of the excess NIS inhibition RfD. Therefore, the
observed LOAEL in an adult male is:

LOAEL(Aduit women) = 0.461 mg/kg-day

57

-------
10-P-0101

The calculation of an NOAEL from a LOAEL is typically accomplished by applying a
UF of 10 to the LOAEL (EPA 2002, p 4-44):

NOAEL(Adult women) = 0.461 mg perchlorate equivalent/kg-day / 10 UF
= 0.0461 mg perchlorate equivalent/kg-day

Since the exposed women are expected to be more sensitive to NIS inhibition and
because the use of hypothyroxinemia is a more subtle adverse thyroid effect than
hypothyroidism, a scientific argument could be made to justify a smaller value of 10 for this UF
(i.e., LOAEL to NOAEL). However, EPA risk assessment guidance states that a reduction of the
intraspecies UF from a default of 10 is to be considered only if the POD is determined from the
data obtained from the susceptible subpopulation (EPA 2002, p 4-43). Since nonpregnant
women are not the sensitive population for this public health issue, this calculation of the excess
NIS inhibition RfD(Women) from the excess exposure to thiocyanate in cow's milk will use a UF of
10. The application of a 10 UF in this calculation is analogous to the NAS Committee applying
a total UF of 10 to the NOEL to protect the most sensitive population. From the excess
perchlorate consumption in the Greer study, the NAS Committee used a NOEL and applied a UF
of 10 to it to derive NAS-recommended perchlorate RfD (NAS 2005, p 16). The use of a UF of
10 in our analysis also allows for the direct comparison of the excess NIS inhibition RfD from
thiocyanate-perserved milk in adult women with the NAS-recommended perchlorate RfD of
0.0007 mg/kg-day. Therefore, the excess NIS inhibition RfD(Women) is calculated as follows:

Excess NIS Inhibition RfD(Aduit women) = 0.0461 mg perchlorate equivalent/kg-day / 10 UF

= 0.0046 mg/kg-day

The calculated excess NIS inhibition RfD from excess SCN" exposure in adult women is
0.0046 mg/kg-day. EPA's perchlorate RfD is 0.0007 mg/kg-day obtained from the NAS
Committee. The calculated excess NIS inhibition RfD(Aduit women) is 6.6 times greater than the
NAS-recommended perchlorate RfD.

The advantage of determining the excess NIS inhibition RfD is it used a conventional risk
assessment approach of deriving an RfD by using a LOAEL observed in a human population.
The NAS Committee describes its own approach of deriving an RfD from a NOEL as being an
unconventional approach, but characterized it as being a "conservative, health-protective
approach to the perchlorate risk assessment" (NAS 2005, p 15). The OIG's conventional
approach for deriving a perchlorate RfD independently supports the NAS Committee's statement
that its recommended perchlorate RfD is conservative. The excess NIS inhibition RfD(Aduit
women) observed in the Baneijee thiocyanate exposure study in milk indicates that the NAS-
recommended perchlorate RfD is conservative by a factor of 6.6.

Discussion

The following findings can be made from the observations in the two Banerjee
thiocyanate exposure studies:

58

-------
10-P-0101

• In both Banerjee thiocyanate exposure studies, the first observed adverse effect from
excess exposure to NIS inhibition was hypothyroxinemia and not hypothyroidism as
stated by the NAS Committee.

• Both Banerjee thiocyanate exposure studies support the concept that the exposure to the
other NIS inhibitors, thiocyanate and nitrate, have the same mechanism of toxicity as
perchlorate, and an excess exposure on a molar potency adjusted basis is just as toxic as
excess exposure to perchlorate. The toxic effects of the three NIS inhibitors are the same
and the evaluation of the risk from three NIS inhibitors should not be separated, but
should be considered together through a cumulative risk assessment approach.

• Both Banerjee thiocyanate exposure studies support that the mathematical description of
the Tonacchera Model for the TIU by the NIS. Hypothyroxinemia is known to be caused
by iodide deficiency, which directly results in a low TIU. The Tonacchera Model
predicts that a low TIU can also be generated when iodide intake is normal, but the total
goitrogen load is excess. In both Banerjee thiocyanate exposure studies, the iodide intake
was normal, but due to the excess exposure to NIS inhibitors, the same adverse effect
observed with iodide deficiency is generated. This indicates that the lack of iodide acts
as an NIS stressor producing the same effect to lower the uptake of iodide in the thyroid
as does the exposure to NIS inhibitors.

• The conventionally derived excess NIS inhibition RfD from excess SCN" exposure in
adult males is calculated to be 0.0075 mg/kg-day. The NAS Committee's recommended
perchlorate RfD is 0.0007 mg/kg-day. The conventionally derived excess NIS inhibition
RfD(Aduit males) is 10.7 times greater than the NAS-recommended perchlorate RfD. The
OIG's conventional approach for deriving a perchlorate RfD independently supports the
NAS Committee's statement that its recommended perchlorate RfD is conservative.

• The conventionally derived excess NIS inhibition RfD from excess SCN" exposure in
adult women is 0.0046 mg/kg-day. The NAS Committee's recommended perchlorate
RfD is 0.0007 mg/kg-day. The conventionally derived excess NIS inhibition RfD(Aduit
women) is 6.6 times greater than the NAS recommended perchlorate RfD. The OIG's
conventional approach for deriving a perchlorate RfD independently supports the NAS
Committee's claim that its recommended perchlorate RfD is conservative.

• Since both the excess NIS inhibition RfD and the NAS Committee's recommended
perchlorate RfD used a single chemical risk assessment approach to derive their
respective RfD, both have the same limitations of a single chemical risk assessment.

The pivotal perchlorate study by Greer used by NAS to derive the perchlorate RfD also
does not identify or account for the varying amounts of NIS inhibition contributed by the other
NIS inhibitors (i.e., nitrate or thiocyanate) among Greer's test subjects. Furthermore, the Greer
study does not identify the baseline amount of perchlorate from dietary exposure in the test
subjects. Therefore, the perchlorate RfD derived from the Greer study is best described as the
amount of excess NIS inhibition from perchlorate (i.e., the additional amount of perchlorate

-------
10-P-0101

given to the Greer's test subjects) that is not expected to reduce the amount of iodide uptake by
the thyroid in sensitive human populations. Thus, the dataset used here from Baneijee's
thiocyanate study is no worse than Greer's study from the perspective of not controlling for the
nonmeasured NIS inhibitors. However, the Banerjee thiocyanate study does identify the baseline
amount of thiocyanate in the controls. Both studies assume the background NIS inhibitor
exposure levels from the nonmeasured NIS inhibitors are the same across the test subjects.
Furthermore, neither the Greer study nor the male Banerjee thiocyanate study identified the
iodide status of the test subjects, which is known to be a factor in the amount of iodide uptake by
the NIS. Therefore, the amount of contributing inhibition from perchlorate and nitrate and the
amount of iodide across the test subjects are potential confounding factors in the Greer and
Banerjee studies. Obviously, a study in which all four variables (i.e., the amount of perchl orate,
thiocyanate, nitrate, and iodide) are measured in the each of the test subjects would be ideal for
applying and evaluating the Tonacchera Model, but this dataset is rare in the published literature.
Our analysis found only one study (Braverman 2005) that measured all four NIS stressors in a
single study population. This study is used to corroborate the calculated %TIU from the
Tonacchera Model with the measured %TIU values from actual radioactive iodide uptake
measurements reported in this human occupational cohort study (see Section 9.1.5).

-------
10-P-0101

5.1.4.3 Calculation of an Excess NIS Inhibition RfD Using the Adverse Effect
Hypertrophy Observed in School-aged Children Drinking Nitrate-Contaminated
Drinking Water

Description of Tajtakova Nitrate Exposure Study

The Tajtakova nitrate exposure study measured thyroid volume and echogenicity by
ultrasound in 342 school-aged children from high-nitrate areas (HNAs) located in the agricultural
lowland who consumed drinking water from shallow wells that were highly contaminated with
nitrate (51-274 mg/L) in the Kosice Region of Slovakia (Tajtakova 2006). The study compared
the HNA children's thyroid volume and echogenicity results against 168 children of the same
age from a low-nitrate area (LNA) who consumed low-nitrate drinking water (<2 mg/L). The
HNA children's results were also compared to 596 children from the City of Kosice who also
consumed low-nitrate drinking water. The ages and genders of the children are as follows:

Study Group

Number of
10-Year-Old Children

Number of
13-Year-Old Children

324 Children from HNA

117 (49 boys / 68 girls)

207 (96 boys/111 girls)

168 Children from LNA

65 (24 boys / 41 girls)

103 (60 boys / 43 girls)

596 Children from Kosice City
(also a low-nitrate area)

171 (68 boys/103 girls)

425 (163 boys / 262 girls)

HNA = high-nitrate area; LNA = low-nitrate area.

Source: Tajtakova 2006.

Blood samples were taking from 315 HNA children and 109 LNA children (no blood
samples were taken from the Kosice City children). The blood was analyzed for levels of TSH,
total thyroxine (tT4), free triiodothyronine (fT3), and thyroperoxidase antibodies (anti-TPO) in
the serum.

iodide Nutritional Status

The urinary iodine concentration in the HNA group (n=53) had a mean ± SD of 154 ±
108 |ig/L, median 150 |ig/L, range 24-320 |ig/L, and 79% of the urinary concentrations were >
100 |ig/L. The urinary iodine concentration in the LNA group (n=32) had a mean ± SD of 125 ±
48 |ig/L, median 126 |ig/L, range 60-173 |ig/L, and 64% of the urinary concentrations were >
100 |ig/L. The recommend urinary iodine concentration for adolescents and children (6-12 yrs)
is 100 to 200 |ig/L (Delange 2005, p 278). Therefore, the urinary iodine concentration in both
HNA and LNA groups indicate satisfactory and healthy iodide intake amounts.

Tajtakova Nitrate Exposure Study Results

In regard to the thyroid volume, both the 10-year-old and 13-year-old HNA children had
statistically significant increases in the size of their thyroid (i.e., hypertrophy) as compared to the
LNA children and the children from Kosice City. The observed thyroid volume in the 10 year
olds increased from 4.58 mis in the LNA children to 5.10 mis in the HNA children (i.e., a 11%
increase). Furthermore, the observed thyroid volume in the 13 year olds increased from 5.23 mis

61

-------
10-P-0101

in the LNA children to 5.97 mis in the HNA children (i.e., a 14% increase). The thyroid volume
results are summarized in the following table:

Study Groups

Thyroid Volume

10-year-old Children

13-year-old Children

324 Children from HNA

5.10 ±0.14 ml

5.97 ±0.11 ml

168 Children from LNA

4.58 ±0.17 ml (p <0.02)

5.23 ±0.15 ml (p < 0.05)

596 Children from Kosice City
(also a low-nitrate area)

4.77 ±0.10 ml (p <0.05)

4.87 ±0.14 ml (p < 0.0001)

HNA = high-nitrate area; LNA = low-nitrate area.

Source: Tajtakova 2006.

In regard to the frequency of hypoechogenicity, both the 10-year-old and 13-year-old
HNA children had statistically higher frequencies of hypoechogenicity as compared to the LNA
children plus the children from Kosice City. The frequency of hypoechogenicity is summarized
in the following table:

Study Groups

Frequency of Hypoechogenicity

10-year-old Children

13-year-old Children

Children from HNA

16/177(13.7%) p<0.01

22/207 (10.6%) p<0.03

Children from LNA & Kosice City

11/236 (4.7%)

30/528 (5.7%)

Source: Tajtakova 2006.

In regard to TSH levels, of the 315 HNA children tested for TSH, 13 were found to be in
the TSH hypothyroid range (i.e., >4.0mU/l). By contrast, of the 109 LNA children tested for
TSH, none were found to be in the TSH hypothyroid range.

In regard to anti-TPO values, of the 315 HNA children tested for anti-TPO, 8 were found
to be anti-TPO positive. By contrast, of the 109 LNA children tested for anti-TPO, none were
found to be anti-TPO positive.

In regard to the blood sample results, the measured tT4 and fT3 levels between the HNA
and LNA schoolchildren were found to have no statistical difference. The blood sample results
for tT4 and fT3 are provided in the following table:

Study Groups

Blood Sample Results

Total Thyroxine (tT4)

Free Triiodothyronine (fT3)

315 Children from HNA

(10- & 13-year-old pooled results)

88.9 ± 26.9 nmol/L

5.9 ± 1.0 pmol/L

109 Children from LNA

(10- & 13-year-old pooled results)

82.1 ±22.6 nmol/L

6.0 ± 0.9 pmol/L

Source: Tajtakova 2006

Estimated Excess Nitrate Exposure from Drinking Water in HLA Children

The total number of nitrate-contaminated wells in the study is 45. The total weighted
nitrate concentration is 6878 well-mg/L. Therefore, the estimated nitrate concentration in the
HLA is 152.8 mg/L (i.e., 6878 well-mg/L divided by 45 wells). Since the LNA children and the
Kosice children drank water containing up to <2 mg/L nitrate, the amount of excess nitrate

62

-------
10-P-0101

concentration in the HNA children's drinking water ingestion is estimated to be 150.8 mg/L (i.e.,
152.8 mg/L-2.0 mg/L).

Number of
Contaminated
Wells

Nitrate Concentration Range
Reported in the Wells
(mg/L)

Mid-Range Nitrate
Concentration
(mg/L)

Weighted Nitrate
Concentration
(well-mg/L)

14

214-264

239

3346

16

101-200

150

2400

15

51-100

75.5

1132

Total = 45





Total = 6878

Source: OIG Analysis.

The amount of excess nitrate exposure per day in the HNA children is estimated to be
150.8 mg/day (i.e., 150.8 mg of excess nitrate/L x 1 liter/person/day). This is calculated from the
amount of excess nitrate concentration in the HNA children's drinking water multiplied by the
median drinking water consumption rate of 1 liter/person/day. Since the water consumption rate
of children in Slovakia is undocumented, the mean daily average ingestion of community water
of 1.0 L/person/day for a consumer in the U.S. population was taken from EPA guidance (EPA
2004, p 4-2). The drinking water consumption rate of Slovakian children may be higher or
lower.

Amount of Excess NIS Inhibition from Excess Nitrate in the School-Aged Children

The amount of excess NIS inhibition from excess nitrate in school-aged children is
measured by the following ingested-weight basis using the OPEC equation (concept adapted
from Tonacchera 2004; De Groef 2006, p 155):

OPEC = (Amt of CIO4" Inhibition) + (Amt of NO3" Inhibition) + (Amt of SCN" Inhibition)
= grams of CIO4" + grams of NO3" /150 + 0.5 free SCN" x grams of SCN" /8.8

Where:

Oral inhibition potency of perchlorate = 1

Oral inhibition potency of nitrate relative to perchlorate =1/150

Oral inhibition potency of thiocyanate relative to perchlorate =1/8.8

Therefore, the amount of Excess NO3" Inhibition is:

Excess OPEC(chiidren) = grams excess Nitrate -M50
= 150.8 mg/day -M50
= 1.005 mg of perchlorate equivalent/day

Calculation of an Excess NIS Inhibition RfD form the Excess Nitrate Exposure

The thyroid hypertrophy observed in the HNA school-aged children is used to identify
the NOAEL for the calculation of the excess NIS inhibition RfD. In a study of 143 Nigerian
adults, the mean thyroid volume was reported to be 9.55 mis ± 1.82 (Ahidjo 2006). This value is
consistent with the thyroid volume range of 8 -12 mis for females and 11-15 mis for males in

63

-------
10-P-0101

Iceland, Sweden, the Netherlands, and the United States (Langer 1989). Therefore, the thyroid
volume in an adult has to approximately double before being considered abnormal. The
recommended upper limit for thyroid volume in an adult population is 18 mis (Ivanac 2004).
The observed thyroid volume in the 10-year-olds increased from 4.58 mis in the LNA children to
5.10 mis in the HNA children (i.e., an 11% increase). Furthermore, the observed thyroid volume
in the 13-year-olds increased from 5.23 mis in the LNA children to 5.97 mis in the HNA children
(i.e., a 14% increase). The recommended upper limit for thyroid volume in children under 15
years old is 16 mis (Ivanac 2004). Although a statistically significant increase in thyroid volume
was observed between the LNA and HNA children in the Tajtakova study from the increase
exposure to nitrate in the drinking water, the increase in thyroid volume was not large enough to
be considered a LOAEL.

No statistical difference was found in tT4 and fT3 thyroid hormone levels between the
HNA and LNA schoolchildren. Since a decreased supply of fT4 during pregnancy is the
proposed mechanism leading to fetal brain damage, the lack of a decrease in the supply of T4
suggests the observed thyroid volume effect is probably occurring below the NOAEL.

Therefore, the excess NIS Inhibition RfD(chiidren) calculated below is probably greater, hence the
use of the greater-than symbol (>).

A NOAEL is expressed on a daily dose-per-kg basis (i.e., mg/kg-day). Therefore, the
average weight of the HNA school-aged children is needed, but this data point is not provided in
the study. Therefore, the average weight of the HNA school-aged children needs to be
estimated. The following table provides weight data on 10- and 13-year-old children to estimate
an average weight for the test population:

Gender

10-Year-Old HNA Children

13-Year-Old HNA Children

Number in
HNA Group

Median
Weight (kg)

Combined
Weight (kg)

Number in
HNA Group

Median
Weight (kg)

Combined
Weight (kg)

Boys

1519

4800

Girls

1972

111

5439

Totals

117

3491

207

10239

Source of weight data: EPA 2004, table 7.2.

Therefore, the estimated median weight of the 324 HNA school-aged children is 42.4 kg
(i.e., (10239 kg + 3491 kg)/( 117 + 207 children)). Therefore, the calculated NOAEL for the
median 42.4 kg HNA school-aged child is:

NOAEL > 1.005 mg perchlorate equivalent/day /42.4 kg
> 0.0237 mg perchlorate equivalent/kg-day

The NAS Committee applied a total UF of 10 (i.e., an intraspecies factor) to the NOEL to
protect the most sensitive population - the fetuses of pregnant women who might have
hypothyroidism or iodide deficiency (NAS 2005, p 16). Although 10- and 13-year-old school-
aged children are expected to be more sensitive to NIS inhibition than an adult, an intraspecies
UF smaller than 10 could be proposed. However, EPA risk assessment guidance states that a
reduction of the intraspecies UF from a default of 10 is to be considered only if the POD is

-------
10-P-0101

determined from the data obtained from the susceptible subpopulation (EPA 2002, p 4-43).

Since children are not the sensitive population for this public health issue, this calculation of the
excess NIS inhibition RfD(Chiidren) from the excess exposure to nitrate in school-aged children
uses a UF of 10 as follows:

Excess NIS Inhibition RfD(Chiidren) > 0.0237 mg perchlorate eq./kg-day / 10 UF

> 0.0024 mg perchlorate eq ,/kg-day

The calculated excess NIS inhibition RfD from excess nitrate exposure in school-aged
children is greater than 0.0024 mg/kg-day. EPA's perchlorate RfD is 0.0007 mg/kg-day,
obtained from the NAS Committee. The calculated excess NIS inhibition RfD(chiidren) is at least
3.4 times greater than the NAS-recommended perchlorate RfD.

Determining the excess NIS inhibition RfD is an advantage because it used a
conventional risk assessment approach of deriving an RfD by using a NOAEL observed in a
human population. The NAS Committee describes its own approach of deriving an RfD from a
NOEL as being an unconventional approach, but characterized it as being a "conservative,
health-protective approach to the perchlorate risk assessment" (NAS 2005, p 15). The OIG's
conventional approach for deriving a perchlorate RfD independently supports the NAS
Committee's statement that its recommended perchlorate RfD is conservative. The excess NIS
inhibition RfD(Chiidren) observed in the Tajtakova nitrate exposure study indicates that the NAS-
recommended perchlorate RfD is conservative by a factor of at least 3.4.

Discussion

The following findings can be made from the observations in the Tajtakova nitrate
exposure study:

• The Tajtakova exposure study supports the concept that the exposure to the other NIS
inhibitors, thiocyanate and nitrate, have the same mechanism of toxicity as perchlorate
and an excess exposure on a molar potency adjusted basis is just as toxic as excess
exposure to perchlorate. The toxic effects of the three NIS inhibitors are the same and
the evaluation of the risk from three NIS inhibitors should not be separated, but should be
considered together through a cumulative risk assessment approach.

• The conventionally derived excess NIS inhibition RfD from excess nitrate exposure in
school-aged children is calculated to be 0.0024 mg/kg-day mg/kg-day. The NAS
Committee's recommended perchlorate RfD is 0.0007 mg/kg-day. The conventionally
derived excess NIS inhibition RfD(Aduitmales) is > 3.4 times greater than the NAS-
recommended perchlorate RfD. The OIG's conventional approach for deriving a
perchlorate RfD independently supports the NAS Committee's claim that its
recommended perchlorate RfD is conservative.

-------
10-P-0101

6. Application of the Tonacchera Model to the Chilean Epidemiological Studies

This section discusses the application of the Tonacchera Model to the two
epidemiological studies conducted on the perchlorate exposed populations of northern Chile.

6.1 Tellez Study

Dr. Rafael Tellez of the Medicine Service in Puente Alto, Chile, led a longitudinal
epidemiological study in northern Chile to evaluate the impact of perchlorate exposure to
drinking water of 113.9 |ig/L on the maternal thyroid during pregnancy, neonatal thyroid
function, the developmental status at birth, breast milk iodide, and perchlorate levels during
lactation (Tellez 2005). The Tellez study was conducted in the same three coastal cities in
northern Chile (i.e., Antofagasta, Chanaral, and Taltal) as the Crump study that evaluated the
impact of perchlorate exposure on school-aged children and newborns (Crump 2000). Since
only preliminary results from the Tellez study were available to the NAS Committee, the Tellez
study was not used in the formulation of the NAS Committee's conclusions (NAS 2005, p 106).

The Tellez study accounted for both drinking water and dietary exposure to perchlorate.
The pregnant mothers from Taltal were found to have perchlorate exposure at about 118.8
|ig/day or 2.42 times the EPA RfD. The following table summarizes the perchlorate exposure in
the pregnant mothers and neonates in the three northern Chilean cities:

Perchlorate Exposure in Chilean Pregnancy and Neonate Study

Parameter

Perchlorate Exposure Groups

Antofagasta

Chanaral

Taltal

Mean Perchlorate in Tap
Water

0.5 jjg/L
< 4.0 |jg/L (all samples)

5.82 ± 0.63 jjg/L

113.9 ± 13.3 jjg/L

Est. Drinking Water
Perchlorate Exposure

0.42 jjg/day

6.1 jjg/day

93.5 jjg/day

Est. Dietary Perchlorate
Exposure

21.7 jjg/day

33.8 jjg/day

25.3 jjg/day

Est. Median Daily
Perchlorate Exposure
(i.e., urinary excretion)

22.1 jjg/day

40.0 jjg/day

118.8 jjg/day

Perchlorate Exposure
Fraction of EPA RfD
of 49 jjg/day

0.45 RfD

0.82 RfD

2.42 RfD

Source: Tellez 2005.

The concept of external dose is a crude technique to measure human exposure. As seen
in Clewell's perchlorate PBPK model, an external dose of 1 |ig/kg-day generates different
internal serum concentrations (i.e., internal indices) in each of the following life stages: fetus,
neonate, child, adult, pregnant women, and lactating women (Clewell 2007, table 4). The use of
an internal indice is critical because the biological impact is directly related to the concentration
of perchlorate in direct contact with the NIS (i.e., bathing the symporter). Therefore, internal
indices of perchlorate serum concentration are a better measure of exposure than the external
perchlorate dose. The same external dose can result in significantly different internal indices.
For example, an external dose of 1 |ig/kg-day (i.e., 70 |ig/day for an adult) would predict an
adult's perchlorate serum concentration to be 2 |ig/L. However, the same external dose of

-------
10-P-0101

1 |ig/kg-day in a pregnant mother is predicted to have a 5 |ig/L perchlorate serum concentration.
Although both the adult and pregnant mother have the same external dose, the pregnant mother
has 2.5 times the perchlorate serum concentration in direct contact with the NIS. This means the
pregnant mother has 2.5 times the amount of NIS inhibition more than the adult for the same
external dose. This is important because the perchlorate RfD is expressed as an external dose,
not an internal indice.

The strategy of the 2007 NAS Toxicity Testing Committee was to define the critical
internal tissue concentration (i.e., internal indices) that perturbs the biological pathway and then
use pharmacokinetic modeling to determine what external dose generates the critical internal
dose (NAS 2007, prepub. summary p 5). To apply the NAS Toxicity Testing Committee's
approach to the perchlorate RfD, the internal perchlorate serum concentration must be
determined at the external RfD dose of 0.7 |ig/kg-day. The Clewell Perchlorate PBPK Model
predicts a perchlorate serum of 2 |ig/L for an adult at an external dose of 70 |ig/day (i.e.,
1 |ig/kg-day or 1.42 times the RfD). If this value is linearly scaled down to the EPA RfD of
49 |ig/day in an adult (i.e., 0.7 |ig/kg-day), the predicted perchlorate serum concentrations at
EPA's RfD in an adult would be 1.4 |ig/L. However, the EPA RfD was derived by applying a
10-fold UF to the NOEL of 7 |ig/kg-day observed in Greer's subjects (NAS 2005, p 15).
Therefore, the perchlorate NOEL corresponds to an external dose of 490 |ig/day. The closest
point provided in Clewell's paper is that an external dose of 700 |ig/day is predicted to generate a
perchlorate serum concentration of 10 |ig/L in an adult (Clewell 2007, table 4). If this value is
linearly scaled down to the perchlorate NOEL of 490 |ig/day in an adult (i.e., 7 |ig/kg-day), the
predicted perchlorate serum concentrations at the NOEL in an adult would be about 7 |ig/L.
Therefore, any perchlorate serum concentrations above 7 |ig/L would be expected to generate
observed effects (i.e., changes in the thyroid hormones and TSH levels).

The following table provides a summary of the internal indices exposures in the pregnant
mothers and neonates in the Tellez study:

Serum Perchlorate Exposures in Tellez Study

Parameter

Perchlorate Exposure Groups

Antofagasta

Chanaral

Taltal

Mothers' Perchlorate Serum
at ~16 Weeks of Gestation

< 4 |jg/L
(n = 3)

< 4 |jg/L
(n = 7)

10.9 ±2.1 jjg/L
(n = 14)

Mothers' Perchlorate Serum
at ~32 Weeks of Gestation

< 4 |jg/L
(n = 5)

< 4 |jg/L
(n = 5)

13.2 ± 1.7 jjg/L

(n = 6)

Neonates' Perchlorate Serum
at Birth (i.e., newborns)

< 4 |jg/L

< 4 |jg/L

19.9 ±5.0 jjg/L
(n = 14)

Source: Tellez 2005.

The observed perchlorate serum concentration in the pregnant Taltal mothers is 10.9 |ig/L
at ~ 16 weeks of gestation and 13.2 |ig/L at -32 weeks of gestation. This agrees with the
predicted perchlorate concentration estimated by the Clewell PBPK model for pregnant women.
An external dose of 70 |ig/day is predicted by the model to generate a perchlorate serum
concentration of 5 |ig/L (Clewell 2007, table 4). A simple linear scaling up of the model's
pregnant mothers' perchlorate serum concentration by a factor of 1.7 (i.e., 118.8 |ig/day
70|ig/day - because of the daily perchlorate exposure in Taltal of 118.8|ig/day) gives a predicted
serum concentration for the pregnant Taltal mothers of 8.5 |ig/L.

67

-------
10-P-0101

In the Tellez study, the pregnant Taltal mothers had perchlorate serum concentrations of
10.9 |ig/L and 13.2 |ig/L. The Taltal neonates had an even higher perchlorate serum
concentrations of 19.9 |ig/L. All of these values are above the perchlorate NOEL observed in
Greer's test subjects. Any perchlorate serum concentrations above 7 |ig/L would be expected to
generate observed effects (i.e., changes in the thyroid hormones and TSH). However, both the
pregnant Taltal mothers and the Taltal newborns had normal health function thyroid performance
(Tellez 2005, table 3).

As an indicator of thyroid stress, the differences in the TSH levels in the pregnant
mothers across the three cities was not statistically significant. However, at each testing interval,
the highest perchlorate exposure group (Taltal) always had the lowest TSH level. The T3, free
T4, and TSH results listed in Table 3 of Tellez's paper for all three testing intervals indicate that
the thyroids in the Taltal pregnant mothers are normal, healthy, and not challenged by the
relatively large perchlorate exposure (Tellez 2005). The following table summaries the TSH
levels in the pregnant mothers:

TSH levels in Pregnant Mothers in the Tellez Study

Parameter

Perchlorate Exposure Groups

Statistical
Significance

Antofagasta

Chanaral

Taltal

TSH level (jjU/ml)

at ~16 Weeks of Gestation

2.63

2.81

2.61

0.91

TSH level (jjU/ml)

at ~32 Weeks of Gestation

3.69

2.55

2.08

0.63

TSH level (jjU/ml)
Postpartum

8.25

2.34

1.95

0.79

Source: Tellez 2005, table 3.

The differences in the free T4 and TSH levels in the neonates across the three cities were
not statistically significant. The T3 level in Chanaral was statistically significantly lower than the
other two cities. These data indicate that the thyroids in the Taltal neonates are normal, healthy,
and not challenged by the relatively large perchlorate exposure. The following table summarizes
the thyroid hormones, TSH, and serum perchlorate levels in neonates across the three cities:

Thyroid Hormones and TSH levels in Neonates in the Tellez Study

Parameter

Perchlorate Exposure Groups

Statistical
Significance

Antofagasta

Chanaral

Taltal

T3 (ng/dL)

79 ± 13.4

73 ± 17.9

82 ±20.6

0.03

Free T4 (ng/dL)

1.07 ± 0.16

1.04 ±0.13

1.03 ±0.14

0.73

TSH Level (jjU/ml)

6.20 ±2.96

6.69 ±4.13

6.31 ±2.91

0.99

Serum Perchlorate (|jg/L)

19.9 ±5.0

0.005

Source: Tellez 2005, table 4.

Given that the perchlorate NOEL determined in the Greer test subjects had an internal
indice of 7 |ig/L, the pregnant mothers and newborns in Taltal had serum perchlorate levels well
above the NOEL. On an internal indices basis, the Taltal pregnant mothers were about 1.6 and
1.9 times the NOEL at -16 and -32 weeks of gestation, respectively. Furthermore, the Taltal
neonates were about 2.8 times above the NOEL. Application of the Greer NOEL requires the
Tellez study to have detected changes in the thyroid hormones and TSH levels in both the Taltal

-------
10-P-0101

pregnant mothers and neonates. This is a striking contradiction between the Greer study and the
Tellez study that must be explained. It is simply not logical that the Taltal pregnant women and
neonates are less sensitive to perchlorate inhibition then the adults in the Greer study. This
contradiction between the Greer study and the Tellez study requires evaluating and considering
potential confounding variable(s) at work to account for this discrepancy.

The Tonacchera Model identifies and simultaneously quantifies the amount of interaction
between the following four variables affecting the uptake of iodide into the thyroid by the NIS:
iodide, perchlorate, nitrate, and thiocyanate. The Tonacchera Model evaluates the total NIS
inhibition load (i.e., total goitrogen load) acting on the body. The Tellez study does not provide
nitrate exposure data in the pregnant mothers, but the study did measure perchlorate serum
concentrations at -16 and -32 weeks of gestation and thiocyanate serum concentrations at
postpartum. The following table summarizes the known NIS inhibitor exposures in the pregnant
mothers (note that the NIS inhibitors are also provided in |imol/L to facilitate their use in
determining the total goitrogen load acting on the pregnant mothers):

NIS Inhibitor Exposures in Pregnant Mothers in Tellez Study

Parameter

Perchlorate Exposure Groups

Antofagasta

Chanaral

Taltal

Perchlorate Serum
at ~ 6 Weeks of Gestation

< 4 |jg/L
(n = 3)
< 0.04 jjmol/L

< 4 |jg/L
(n = 7)
< 0.04 jjmol/L

10.9 ±2.1 jjg/L
(n = 14)
0.11 |jmol/L

Perchlorate Serum
at ~32 Weeks of Gestation

< 4 |jg/L
(n = 5)
< 0.04 jjmol/L

< 4 |jg/L
(n = 5)
< 0.04 jjmol/L

13.2 ±1.7 jjg/L
(n = 6)
0.13 jjmol/L

Thiocyanate Serum
Postpartum

20.0 ±10.9 jjmol/L
(n = 9)

17.4 ±16.1 jjmol/L
(n = 21)

12.6 ±6.4 (jmol/L (n =
10)

Source: Tellez 2005.

The total NIS inhibition load (i.e., total goitrogen load) acting on the body is determined
by calculating the SPEC as follows (concept taken from Tonacchera 2004):

SPEC = (Amt of CIO4" Inhibition) + (Amt of NO3" Inhibition) + (Amt of free SCN

Inhibition)

SPEC = [CIO4] + [NO3"]/240+ [free SCN] /15

Where: [ ] is molarity

Serum inhibition potency of perchlorate = 1
Serum inhibition potency of nitrate relative to perchlorate = 1 / 240
Serum inhibition potency of thiocyanate relative to perchlorate =1/15
[free SCN ] = [total serum SCN] x 0.5 free SCN" /total serum SCN

Using the measured perchlorate serum at -32 weeks of gestation and thiocyanate serum
concentrations at postpartum, the total NIS inhibitor load can be estimated in the pregnant
mothers from each of the three cites. However, the Tellez study did not measure the nitrate
exposure, and as another NIS inhibitor, nitrate could serve as an uncontrolled confounding
variable. However, the amount of NIS inhibition typically contributed from nitrate is only about
11% (i.e., nitrate SPEC of 0.167 |imol/L out of a total SPEC for the adult of 1.501 |imol/L) of

69

-------
10-P-0101

the body's total goitrogen load. In the Western world, the typical nitrate serum concentration
ranges from 10-140 |imol/L with the mean nitrate serum concentration being 30-50 |imol/L
(Tonacchera 2004). Therefore, an estimated nitrate exposure of 40 (amol/L will be used for all
three cities in this calculation of the pregnant mothers' total NIS inhibition load (i.e., since each
value is the same, this will not change the relative ranking of total goitrogen loads between the
cities). The nitrate exposure in Taltal would have to increase by at least a factor of 2 (i.e., to 80
(amol/L) over nitrate exposure in Antofagasta and Chanaral to cause Taltal to have the largest
total goitrogen load. The following are calculations for the total NIS inhibition in the pregnant
mothers in each of the three cities from the Tellez study (expressed as SPECs):

Antofagasta SPEC = 0.04 (imol/L + (40.0 (imol/L 240) + (20.0 (imol/L x 0.5 free SCN" ^ 15)
= 0.04 (imol/L + 0.17 (imol/L + 0.67 (imol/L
= 0.88 (imol/L

Chanaral SPEC = 0.04 (imol/L + (40.0 (imol/L 240) + (17.4 (imol/L x 0.5 free SCN" ^ 15)
= 0.04 (imol/L + 0.17 (imol/L + 0.58 (imol/L
= 0.79 (imol/L

Taltal SPEC = 0.13 junol/L + (40.0 junol/L - 240) + (12.6 junol/L x 0.5 free SCN" - 15)
= 0.13 (imol/L + 0.17 (imol/L + 0.42 (imol/L
= 0.72 (imol/L

The total NIS inhibition load (i.e., total goitrogen load) acting on the neonate can also be
determined by calculating the SPEC. The Tellez study did not measure thiocyanate
concentrations in neonates. However, the OIG identified four papers that document the transfer
of thiocyanate from the mother's blood to the neonate's blood through the placenta (Hauth 1984;
Laurberg 2004; Nafstad 1995; Vanderpas 1984). We provided this information above in a table
under the section Total NIS Inhibitor Exposure in Fetuses. These four papers clearly report
thiocyanate serum concentration ratios of the cord serum to the mother's serum in the range of
68-111%. The observed thiocyanate concentrations in the pregnant mother at the postpartum
examination was 20.0 |imol, 17.4 |imol, and 12.6 |imol for Antofagasta, Chanaral, and Taltal,
respectively.

The closest match in the data from the four papers is in Hauth's study. The Hauth study
reported a thiocyanate serum concentration of 32.2 |imol and 23 |imol in the mother's serum and
the neonate's serum, respectively. In the Hauth study, the observed thiocyanate serum
concentration ratio of the cord serum to the mother's serum in nonsmokers was 71%. The
amount of thiocyanate transfer from the mother to the neonate will be used to estimate the
expected thiocyanate concentrations in the neonates in the Tellez study. The amount of nitrate in
the neonate's serum is not reported in the Tellez study. Therefore, nitrate is unknown and
represents a potential confounding variable. However, the amount of NIS inhibition typically
contributed by nitrate in an adult is only about 11% (i.e., nitrate SPEC of 0.167 |imol/L out of a
total SPEC for the adult of 1.501 (amol/L) of the body's total goitrogen load. Furthermore, if the
nitrate exposure is the same across the three cities, the relative ranking of total NIS inhibition
load would not change. The following table summarizes the neonate exposure to NIS inhibitors
(note that the NIS inhibitors are also provided in |imol/L to facilitate their use in determining the
total goitrogen load acting on the neonates):

70

-------
10-P-0101

NIS Inhibitor Exposures in Neonates in Tellez Study

Parameter

Exposure Groups

Antofagasta

Chanaral

Taltal

Perchlorate Serum in
Neonates at Birth

< 4 |jg/L
(n = 4)
< 0.04 jjmol/L

< 4 |jg/L
(n = 1)
< 0.04 jjmol/L

19.9 ±5.0 jjg/L
(n = 14)
0.20 jjmol/L

Thiocyanate Serum in
Pregnant Mother at
Postpartum

20.0 ±10.9 jjmol/L

(n = 9)

17.4 ±16.1 jjmol/L
(n = 21)

12.6 ±6.4 (jmol/L (n =
10)

Estimated Neonate
Thiocyanate Serum Using a
70% Equilibrium

14.2 jjmol/L

12.4 jjmol/L

8.9 jjmol/L

Source: Tellez 2005.

The following are calculations for the total NIS inhibition in neonates in each of the three
cities in the Tellez study (expressed as SPECs):

Antofagasta SPEC = 0.04 (imol/L + (nitrate ? (imol/L 240) + (14.2 (imol/L x 0.5 free SCN" ^ 15)
= 0.04 (imol/L + unknown nitrate amount + 0.47 (imol/L
= 0.51 (imol/L

Chanaral SPEC = 0.04 (imol/L + (nitrate ? (imol/L 240) + (12.4 (imol/L x 0.5 free SCN" ^ 15)
= 0.04 (imol/L + unknown nitrate amount + 0.41 (imol/L
= 0.45 (imol/L

Taltal SPEC = 0.13 (imol/L + (nitrate ? (imol/L 240) + (8.9 (imol/L x 0.5 free SCN" ^ 15)
= 0.20 (imol/L + unknown nitrate amount + 0.30 (imol/L
= 0.50 (imol/L

The Tellez epidemiological study was designed to compare the thyroid health effects in
the high perchlorate exposure city of Taltal against two cities, Antofagasta and Chanaral, which
have much lower perchlorate exposure. The authors concluded that "... the perchlorate dose in
Taltal was not sufficiently high to stress the maternal or neonatal thyroid." However, nitrate and
thiocyanate are also known NIS inhibitors and act as potential confounding variables. Since
thiocyanate levels were measured in the Tellez study, this analysis shows that the level of
thiocyanate exposure in both the pregnant mothers and neonates causes a much greater amount
of NIS inhibition than even the amount of NIS inhibition from consuming 118 ppb of perchlorate
in the tap water. The relatively low thiocyanate exposure in Taltal causes the Taltal group to
actually have the least amount of total NIS inhibition and the least "stressed" or challenged
thyroids in the group (i.e., excluding the impact of iodide, to be discussed later). In other words,
when considered from a total goitrogen load, the experimental design of this epidemiological
study is flipped: Taltal is the control group (i.e., no longer the exposed group), and Antofagasta
is the exposed group (i.e., no longer the control group). Actually, since there is only a relatively
small difference in the Total NIS Inhibitor Load in the pregnant mothers between Taltal and
Antofagasta, the study becomes the equivalent of comparing three control groups together. From
the view point of total NIS inhibition load, there is no surprise then that the data show little
difference in the thyroid health between the three cities. The following table summarized the
relative total NIS inhibition load in the Tellez study:

71

-------
10-P-0101

Relative Total NIS Inhibition Load in the Tellez Study

Parameter

Cities

Antofagasta

Chanaral

Taltal

Mean Perchlorate in Tap Water

0.5 jjg/L

5.82 jjg/L

113.9 jjg/L

Estimated Total Perchlorate Exposure

22.1 jjg/day

40.0 jjg/day

118.8 jjg/day

Relative Perchlorate Exposure

Low

medium

high

Estimated Total NIS Inhibition Load
in Pregnant Mothers

0.88 jjmol/L

0.79 jjmol/L

0.72 jjmol/L

Est. NIS Inhibition Load in Neonate
(excluding nitrate due to a lack of data)

0.51 jjmol/L

0.45 jjmol/L

0.50 jjmol/L

Total NIS Inhibition Load Observed in

Pregnant Mothers

(i.e., total goitrogen load)

Highest
(becomes the
experimental group)

In the middle

Lowest
(becomes the
control group)

Difference in Total NIS Inhibition Load
Observed in Neonates
(i.e., total goitrogen load)

Not much
difference

Not much
difference

Not much
difference

Source: OIG Analysis; summary of Tellez study results.

The Tellez study can be related back to the conditions in the United States using the
Tonacchera Model. In the United States, the typical total NIS inhibition load is 1.500 |imol/L
(which, calculated previously in this document, assumes a median serum concentration of
40|amol/L for both nitrate and thiocyanate). In the Tellez study in Chile, the pregnant mothers'
total NIS inhibition load was 0.88 |imol/L, 0.79 |imol/L, and 0.72 |imol/L in Antofagasta,
Chanaral, and Taltal, respectively. Therefore, the typical U.S. adult has a total NIS inhibition
load that is 170%, 190%, and 208% greater than those in Antofagasta, Chanaral, and Taltal,
respectively. Furthermore, the median UIC is 140 |ig/L in U.S. pregnant women (NAP 2000).
In the Tellez study, the entire Chilean study group had a reported median UIC of 269 |ig/L which
is 92% greater than the United States. The recommended UIC for pregnant and lactating women
is reported at 200-300 |ig/L (Delange 2005, p 278, table 1 IE.8). The median UIC indicating
optimal iodide nutrition during pregnancy and lactating women could be in the range of 150-230
|ig/L (Delange 2004).

The difference in TIU between pregnant women in the United States and Chile can be
calculated using the Tonacchera Model TIU is calculated by using the following equation:

TIU = constant x [T] / (1.22 + [SPEC])

where: SPEC = [C104 ] + [N03"]/240 + [freeSCN]/15

Note: The Tonacchera Model uses concentration expressed in units of |imol/L.

Restated:

TIU is proportional to [T] / (1.22 + [SPEC])

Estimated TIU in U.S. Pregnant Women

The TIU in the U.S. pregnant women is determined by substituting values into the
Tonacchera Model. The [T] in the serum is unknown but is directly related to the UIC. The [T]
is assigned a value of "x". In the United States, the typical total NIS inhibition load is 1.501

72

-------
10-P-0101

(amol/L (which, calculated previously in this document, assumes a median serum concentration
of 40|amol/L for both nitrate and thiocyanate).

Therefore, the TIU in the U.S. pregnant women is expressed by:

TIU (U.S. pregnant women) || [T] / (1.22 + [SPEC])

where: the symbol || means "proportional to"

Substitution gives:

TIU (U.S. pregnant women) || X / (1.22 + (1.501 (^mol/L)

Estimated TIU in Chilean Pregnant Women

The TIU in the Chilean pregnant women is determined by substituting values into the
Tonacchera Model. The [T] is unknown but is directly related to the UIC. The median UIC of
269 |ig/L is 92% greater than in the U.S. pregnant women. The [T] in the Chilean pregnant
women is assigned a value of "1.92x". For simplicity of comparison to the United States, the
Chilean pregnant mothers' total NIS inhibition load from all three cities is averaged to 0.80
|imol/L (i.e., (0.88 |imol/L + 0.79 |imol/L + 0.72 |imol/L) ^ 3).

Therefore, the TIU in the Chilean pregnant women is expressed by:

Percent Change in TIU of U.S. Pregnant Women as Compared to Chilean Pregnant
Women

The percent change in TIU of U.S. pregnant women as compared to Chilean pregnant
women from the Tellez study is given by the following equation:

% Change in TIU of U.S. Pregnant Women = U.S. TIU - Chilean TIU x 100%

Chilean TIU

Substitution gives:

% Change in TIU of U.S. Pregnant Women = ((0.368x - 0.950x)/ 0.950x) x 100%
% Change in TIU of U.S. Pregnant Women = - 61%

Therefore, the median U.S. pregnant woman has a 61% less TIU as compared to the
Chilean pregnant women in the Tellez study. In other words, the U.S. median pregnant woman
has only 39% of the amount of iodide uptake as her Chilean counterpart. In the Tellez study, the
entire Chilean cohort had a reported median UIC of 269 |ig/L, which is in the middle of the
recommended UIC for pregnant and lactating women of 200-300 |ig/L (Delange 2005, p 278,
table 11E.8). The 61%TIU deficit is driven by the difference in the goitrogen load and iodide
nutritional status between the United States and Chile. In other words, U.S. pregnant women

TIU (U.S. pregnant women)
TIU (U.S. pregnant women)

TIU (Chilean pregnant women) || [T] / (1.22 + [SPEC])

73

-------
10-P-0101

have about twice the goitrogen load as their Chilean counterparts, while having only about half
the dietary intake of iodide. Both factors contribute to the decreased uptake of iodide by U.S.
pregnant women.

Relative Contribution of the TIU Deficit from Goitrogen Load and from iodide
Nutrition

The Tonacchera Model can be used to assess the amount of TIU deficit attributed to the
difference in the goitrogen load and to assess the amount of TIU deficit attributed to the
difference in iodide nutrition. This is calculated by analyzing the contribution of each to the TIU
separately.

Change in TIU from the change in goitrogen load = ((0.368x - 0.495x)/ 0.495x) x 100%

= -25.7%

Change in TIU from the change in iodide nutrition = ((x - 1.92x)/ 1.92x) x 100%

= -47.9%

Note: The 26% and 48% are not added directly together to get the total 61% decrease TIU deficit
value. They are combined mathematically in the following way to give 61%:

Total Decrease = l-((l-0.257)(l-0.479))

= 1-((0.743)(0.521)

= l-(0.387)

= 0.612 or 61%

To further explain, of the original Chilean TIU, 26% is lost due to the increased U.S. goitrogen
load leaving 0.74 TIU. Of the remaining 0.74 TIU, 48% of this amount is lost due to the
decreased U.S. iodide nutrition leaving only 0.385 TIU (i.e., 0.74 - (0.74 x 0.48). This leaves
39% of the original TIU, or a 61% decrease from the Chilean TIU (defined as 100% for this
comparison).

Therefore, the larger goitrogen load in the United States, by itself, causes the TIU to
decrease by 61%. By contrast, the lower iodide nutritional status in the United States, by itself,
causes the TIU to decrease by 48%. When the two factors (i.e., increased goitrogen load and
decrease availability of iodide) act together, a 61% decrease in the TIU in the United States is
predicted by the Tonacchera Model.

74

-------
10-P-0101

6.2 Crump Study

The Crump study compared the thyroid hormone and TSH levels in 9,784 newborns and
in 162 school-aged children in three Chilean cities (Crump 2000). The mean urinary iodide,
TSH, and T3 levels were not statistically different in the schoolchildren among the three cities.
However, the NAS Committee specifically commented, "Serum free T4 levels were significantly
higher in children in Chanaral and Taltal then Antofagasta ... the difference was opposite the
direction predicted on the basis of competitive perchlorate inhibition of iodide uptake" (NAS
2005, p 100). This contradiction in observed direction of the effect on free T4 levels should have
raised an inquiry into reviewing the mechanism (i.e., because competitive inhibition from
perchlorate only does not appear to explain the observed effect). Ideally, this should have
triggered the possibility that a confounding variable was at work in the data (i.e., other NIS
inhibitors).

The potential confounding variable is the different levels of thiocyanate exposure
observed in the Tellez study between Taltal and the other two cities. Based on the trend in
amount of total NIS inhibition load observed across the three cities in the Tellez study (identified
in the previous section), the trend in the serum free T4 levels in the Crump schoolchildren is in
the direction predicted on the basis of total competitive inhibition of iodide uptake. Using the
total goitrogen load better explains the trend in observations seen in the serum free T4 levels.
The following table summarizes the study results in the schoolchildren:

Summary of Crump Study Results of All Schoolchildren by City

Parameter

Antofagasta
(n = 60)

Chanaral
(n = 60)

Taltal
(n = 60)

Perchlorate Concentration in Drinking

ND





Water (jjg/L)

All samples < 4.0 |jg/L

5.5 ±1.6

111.6 ±6.7

Perchlorate Exposure

very low group

low group

high group

TSH foil/ml)

3.3 ±1.8

2.9 ±1.3

3.0 ±1.4

Free T4 (ng/dL)

1.2 ± 0.1

1.3 ±0.1

1.4 ±0.2

T3 (ng/dL)

208.6 ±29.7

211.0 ±20.9

212.0 ±21.9

Urinary Iodide (|jg/L)

756 ± 404

614 ±357

766 ± 474

Est. Total NIS Inhibition Load







in Pregnant Mothers in Tellez Study

1.54 jjmol/L

1.37 jjmol/L

1.14 jjmol/L

Relative Total NIS Inhibition Load







Observed in Pregnant Mothers and







Neonates in Tellez Study

high

medium

low

Source: Crump 2000, table 2.

For the newborns in the Crump study, the newborn TSH levels were slightly lower in
Taltal (highest perchlorate exposure group) as compared with the two low perchlorate exposure
cities (p<0.001) (Crump 2000). Furthermore, the log serum TSH levels in the Taltal newborns
were statistically lower than the other two cities (p<0.001); the log TSH levels did not vary
significantly between the newborns in Antofagasta and Chanaral (Crump 2000; NAS 2005).
Again, the NAS Committee specifically commented, "The lower mean in Taltal is, again,
opposite in direction of what would be expected in association with increased exposure to
perchlorate" (NAS 2005, p 101). Again, this contradiction in observed direction of the predicted
inhibition effect of perchlorate on TSH levels should have raised an inquiry into reviewing the
mechanism (i.e., competitive inhibition of perchlorate only does not appear to explain the

75

-------
10-P-0101

observed effect). Ideally, this should have triggered the possibility that a confounding variable is
at work in the data (i.e., the other NIS inhibitors).

The potentially confounding variable is the different levels of thiocyanate exposure
observed in the Tellez study between Taltal and the other two cities. Based on the trend in
amount of total NIS inhibition load of pregnant mothers observed across the same three cities in
the Tellez study, the trend in the TSH levels in the Crump neonates is in the direction predicted
on the basis of the total competitive inhibition of iodide uptake in the pregnant mothers.
Furthermore, by day 7, when the neonate's thyroid is the sole influence on TSH level (i.e,
mother's residual influence is gone), the neonate TSH levels are essentially the same across the
three cites (see table below). This mirrors the observation that the total goitrogen load estimated
in the neonates from the 2005 Tellez study is also essentially the same across the three cities.
Using the total goitrogen load better explains the observations seen in the serum TSH levels in
the Taltal newborns than the perchlorate exposure level only. The following table summaries the
study results in the neonates:

Summary of Crump Study Results of Neonates by City

Parameter

Antofagasta
(n = 8888)

Chanaral
(n = 468)

Taltal
(n = 428)

Perchlorate Concentration in Drinking

ND





Water (jjg/L)

All samples < 4.0 |jg/L

5.5 ±1.6

111.6 ±6.7

Perchlorate Exposure

Very low group

low group

high group

TSH (jjU/ml) Overall Both Genders

3.4 ±7.8

3.0 ±2.5

2.4 ±1.9

Estimated Total NIS Inhibition Load in







Pregnant Mothers from the Tellez Study

0.88 jjmol/L

0.79 jjmol/L

0.72 jjmol/L

Relative Total NIS Inhibition Load







Observed in Pregnant Mothers and
Neonates in Tellez Study

high

medium

low

TSH (uU/ml) Day 7 Both Genders

2.4 ±2.0

2.1 ±1.4

2.1 ±1.7

Est. NIS Inhibition Load in Neonate from







the Tellez Study

(excluding nitrate - lack of data)

0.51 jjmol/L

0.45 jjmol/L

0.50 jjmol/L

Source: Crump 2000, table 8.

According to the NAS Committee, the major criticisms of the Crump study indicated that
the school-aged children had an iodide status different than those in the United States and a
higher prevalence of goiter (NAS 2005, p 103). The mean urinary iodide excretion in the
children was about 3 times higher than in the United States (NAS 2005, p 103). The NAS
Committee finally concluded that the Crump study could be considered in evaluating perchlorate
exposure in the United States (NAS 2005, p 105).

The Crump study did not determine the amounts of nitrate or thiocyanate in any of the
subjects. The NAS Committee notes the possibility of "uncontrolled confounders" in the Crump
study as in any epidemiologic study (NAS 2005, p 103). Since the OIG has shown that the
thiocyanate NIS inhibitor was an uncontrolled confounder in the Tellez study, this same
confounder probably applies to the Crump study because it involves the same population from
the same cites with the same potential difference in dietary habits and explains the observations
better.

76

-------
10-P-0101

6.3 Conclusions

The OIG's analysis of the Tellez study shows the advantages of applying cumulative risk
assessment to the environmental problem of perchlorate exposure. The biological step perturbed
by perchlorate exposure is the uptake of iodide by the NIS. However, the uptake of iodide by the
NIS is not just a function of exposure to perchlorate. The amount of iodide uptake by the NIS is
the result of the combined effects of simultaneous exposure to all four anions: iodide,
perchlorate, nitrate, and thiocyanate. The effects of each anion are defined by the Tonacchera
Model, where TIU is proportional to [T] / (1.22 + [SPEC]). Measuring the internal indices of
exposure (i.e., |imol/L concentration in serum) for iodide, perchlorate, nitrate, and thiocyanate is
needed to more accurately characterize the amount of stress on the thyroid.

The Chilean pregnant women and neonates have healthy thyroids because their iodide
nutritional status is sufficient to tolerate their total goitrogen load. The fact that the relatively
high exposure to 118 |ig/day of perchlorate from water and food in Taltal represents only a small
portion (11%) of their total goitrogen load, and to say that this level of perchlorate exposure did
not generate thyroid health problems, is to miss the larger point. The potential impact of
perchlorate exposure cannot be assessed without knowing the concurrent exposure to the other
NIS inhibitors (nitrate and thiocyanate) and the iodide status of the population/individual. The
essential finding from the Tellez study is that the total goitrogen load in Taltal (including the
perchlorate) is small compared to the healthy iodide nutritional status.

The critical information needed to characterize this public health problem is determining
when the iodide nutritional level is insufficient to tolerate the total goitrogen load. The
relationship between the iodide nutritional level and total goitrogen load is dynamic. As the
availability of iodide increases, the tolerance for larger total goitrogen loads increases. In other
words, iodide-deficient individuals cannot tolerate as much goitrogens as an individual with a
healthy iodide intake. Defining when the iodide nutritional level is insufficient for the total
goitrogen load and its impact on the regulation of perchlorate is the subject of a later section of
this document.

-------
10-P-0101

7. Assessment of Total NIS Inhibitor Exposure
7.1 Perchlorate Exposure

To assess the role of perchlorate exposure on this public health issue, the amount of
perchlorate exposure must be documented in adults, fetuses, and nursing neonates.

7.1.1 Adults

In 2006, Blount of the CDC, National Center for Environmental Health (NCEH),
assessed the perchlorate exposure in the U.S. population using urinary biomonitoring data
(Blount 2006a). As part of the 2001-2002 NHANES, a nationally representative population of
2,820 U.S. residents (ages 6 years and older) provided urine samples for perchlorate testing. All
2,820 urine samples tested found detectable levels of perchlorate (i.e., >0.05|ig/L), indicating
that perchlorate exposure in the U.S. population is common. Blount's biomonitoring results are
summarized below:

Population

Blount's U.S. Perchlorate Exposure Assessment
from 2001-2002 NHANES Using Urinary Biomonitoring Data

Median

95th Percentile

Adults
(20 years and older)

0.066 jjg/kg-day
(4.6 jjg/day for a 70 kg adult)

0.234 jjg/kg-day
(16.4 jjg/day for a 70 kg adult)

Reproductive Females
(15-44 years old)

0.057 jjg/kg-day
(4.0 jjg/day for a 70 kg adult)

0.214 jjg/kg-day
(15.0 jjg/day for a 70 kg adult)

Source: Blount 2006a.

The perchlorate RfD is 0.7 |ig/kg-day (equal to 0.0007 mg/kg-day). For a 70 kg adult,
the perchlorate RfD allows the "safe" exposure to 49 |ig of perchlorate per day for an adult.
Therefore, the median perchlorate exposure is 9.4% and 8.1% of the perchlorate RfD in adults
and females of reproductive age, respectively. Furthermore, the perchlorate exposure at the 95th
percentile is 33% and 31% of the perchlorate RfD in adults and females of reproductive age,
respectively. Therefore, the Blount biomonitoring study indicates that the perchlorate exposure
in the majority of adults in the United States is well below the RfD. In fact, Blount's
biomonitoring study found that only 11 of the 1,618 adults (i.e., 20 years or older) tested had an
estimated perchlorate exposure in excess of the perchlorate RfD of 0.7 |ig/kg (i.e., 49 |ig per
day). This corresponds to about 0.7% of the adult U.S. population that might have perchlorate
exposures greater than the RfD.

In May 2007, the FDA's Center for Food Safety and Applied Nutrition estimated the
human dietary exposure to perchlorate from the consumption of 27 foods and beverages from
samples collected in Fiscal Years 2004 and 2005. The preliminary human perchlorate dietary
exposure assessment, "Preliminary Estimation of Perchlorate Dietary Exposure Based on FDA
2004/2005 Exploratory Data," was posted on FDA's Website on May 31, 2007 (FDA 2007a).
FDA's preliminary testing represents only 32% of the food and 42% of the beverages in the diet.
Since these samples targeted foods grown in areas known to have perchlorate present, the results
from this study are not representative of total perchlorate exposures. The remaining 68% of the
foods in the diet are expected to contain little to no perchlorate, although this assumption cannot

-------
10-P-0101

be verified. Using FDA's stated assumption, FDA's preliminary perchlorate dietary exposure
results represent the bulk of the perchlorate exposure expected from the U.S. diet:

Population

FDA's Preliminary Perchlorate Dietary Exposure Estimate

Mean

90th Percentile

All Ages (2 and up)

0.053 jjg/kg-day
(3.7 jjg/day for a 70 kg adult)

0.12 jjg/kg-day
(8.4 jjg/day for a 70 kg adult)

Females (15-45 years)

0.037 jjg/kg-day
(2.6 jjg/day for a 70 kg adult)

0.074 jjg/kg-day
(5.2 jjg/day for a 70 kg adult)

Source: FDA 2007a.

On January 2, 2008, FDA published their TDS on perchlorate and iodide (Murray 2008).
The TDS is an ongoing market basket survey in which 285 core foods (TDS foods) in the U.S.
food supply are collected and analyzed to determine levels of various contaminants and nutrients
in those foods. The foods collected in the TDS (referred to as the TDS food list) represent the
major components of the diet of the U.S. population. The food list is based on results of national
food consumption surveys and is updated from time to time to reflect changes in food
consumption patterns. The FDA TDS reported the upper and lower bounds for the average
dietary intakes of perchlorate for each of the 14 age/sex groups from FDA's TDS samples
collected in Fiscal Years 2005 and 2006. The upper and lower bounds for the average dietary
intakes of perchlorate for infants 6 to 11 months and nonpregnant women 25 to 30 years are:

Age/Sex Group Label

FDA's Total Dietary Study: Dietary Intake of Perchlorate

Lower and Upper Bound
AverageTotal Dietary Intake

90tn Percentile
Total Dietary Intake

Infants

(6 to 11 months)

0.26-0.29 jjg/kg-day
2.4 to 2.7 ijg/day

Not provided for
in the TDS Design*

Women

(25 to 30 years)

0.09 to 0.11 jjg/kg-day
5.4 to 6.8 jjg/day

Not provided for
In the TDS Design*

* FDA TDS is not designed to estimate intakes at the extremes (e.g., 10th or 90th percentiles). Murray 2008, p 8.
Source: Murray 2008, table 5.

The FDA TDS study has two significant limitations for characterizing the perchlorate
exposure in a risk assessment. Perchlorate exposure during pregnancy and nursing is the critical
period in which NIS stressor exposure could induce a low TIU resulting in an increased risk for
adverse effects in the developing child. However, the FDA TDS does not provided estimates for
dietary intakes for subgroups with specific nutritional needs (e.g., pregnant or lactating women).
The second limitation is that the FDA TDS does not provide estimates for the distribution of
perchlorate exposure within the age/sex group populations (e.g., 10th and 90th percentiles).
Ideally, a risk assessment would benefit from an exposure estimate of the 90th percentile, but the
FDA TDS does not provide this type of information.

FDA's human perchlorate preliminary dietary exposure assessment and the FDA TDS are
in general agreement with the perchlorate exposure measured in Blount's biomonitoring data.
Specifically, FDA's preliminary mean perchlorate exposure calculated for an adult is 3.7 |ig/day.
The FDA TDS estimated a lower and upper bound average total dietary intake for all age/sex
groups > 25 years old at 0.08 to 0.12 |jg/kg-day (i.e., 5.6 to 8.4 jjg/day for 70 kg adult) (Murray
2008, table 5). These values agree with the median perchlorate exposure of 4.6 |ig/day estimated
from the Blount biomonitoring data. The average perchlorate exposures from the FDA

-------
10-P-0101

preliminary data and Blount biomonitoring data are less than 10% the perchlorate RfD of 49
|ig/day for an adult. Furthermore, FDA's preliminary dietary exposure assessment at the 90th
percentile is calculated at 8.4 |ig/day or 17% of the RfD. The estimated perchlorate exposure at
the 95th percentile using the Blount biomonitoring data is calculated at 16.4 |ig/day, or 33% of
the RfD. Both upper percentile values are significantly less than the perchlorate RfD of 49
|ig/day for an adult.

7.1.2 Fetuses

Assessing the perchlorate exposure in fetuses is difficult; direct measurements are not
possible due to the risk to the child. Therefore, the use of PBPK modeling is required to predict
the amount of exposure in the fetus. The Clewell Perchlorate PBPK Model has accurately
predicted the serum perchlorate concentrations in sensitive populations including fetuses
(Clewell 2007). The Clewell Perchlorate PBPK Model predictions for perchlorate serum
concentrations were corroborated against pregnant women at the -16 and -32 week of gestation,
fetuses at birth observed in the Tellez study, and 7.4-year-old schoolchildren observed in the
Crump study (Clewell 2007) (see figure below). This use of the Clewell Perchlorate PBPK
Model provides the ability to estimate the fetal perchlorate exposure based on the mother's
perchlorate exposure.

Clewell Perchlorate PBPK Model Predicted Perchlorate Serum Concentrations

GW 15 GW 33 Birth 7.4 yrs

The PBPK model predicted values (light grey) are the average ± SD (standard deviation). The
measured perchlorate serum concentrations (dark grey) are the mean ± SD. The pregnant
woman value is from the Tellez Chilean study (Tellez 2005). The 7.4-year-old child value is
reported by Gibbs (Gibbs 2004) of the Chilean schoolchildren from the Crump study (Crump
2000).

Source: Clewell 2007, figure 10.

The Clewell Perchlorate PBPK Model predicts that the internal perchlorate serum
concentration in a fetus will be 5 times greater then the internal perchlorate serum concentration
in an adult from an external perchlorate dose of 0.001 mg/kg-day (Clewell 2007, table 4). Using
the Blount biomonitoring data to characterize the perchlorate exposure in U.S. population, if a
pregnant woman is exposed to the median adult perchlorate exposure of 4.6 |ig/day (i.e., 9.4% of

-------
10-P-0101

the RfD), the fetus would be expected have an internal perchlorate serum concentration that
corresponds to 47% that of the RfD. Likewise, if a pregnant woman is exposed to the 95th
percentile perchlorate exposure for an adult of 16.4 |ig/day (i.e., 33% of the RfD), the fetus
would be expected have an internal perchlorate serum concentration that corresponds to 1.67
times the perchlorate RfD. Finally, if a pregnant woman is exposed to perchlorate at the RfD,
the fetus would be expected have an internal perchlorate serum concentration that corresponds to
5 times the perchlorate RfD.

7.1.3 Nursing Neonates

If exclusively breastfed, nursing infants are dependent on their mothers' milk for all their
nutrition, including the essential element, iodide. Both thyroidal and extrathyroidal NIS
concentrate iodide from the serum into the thyroid by roughly the same magnitude (i.e., thyroid
to serum ration (T/S) is 20- to 40-fold under steady-state conditions) (De La Vieja 2000, p 1093).
In cases of iodide deficiency, the TSH stimulates the production of cyclic adenosine
monophosphe (cAMP) in the thyroid to enhance the rate of active transport of iodide into the
thyroid by the NIS (McMurray 1983, p 170). The T/S ratio may increase up to 300:1. TSH
regulates the expression and activity of the NIS in the thyroid gland only (Merrill 2003; Clewell
2003). TSH does not regulate NIS expression in the mammary gland or any other extrathyroidal
tissue (Merrill 2003). The NIS expression and regulation appear to be controlled by prolactin
and/or oxytocin (Merrill 2003).

Nursing neonates are a sensitive population because an adequate supply of thyroid
hormones is required for proper brain development. However, nursing neonates may be at a
significant risk due to a relatively large perchlorate exposure (i.e., on a per-body-weight basis)
due to the active concentration of perchlorate in breast milk by the NIS. Neonate perchlorate
exposure is a function of the weight of the infant, the amount of milk consumption, and the
perchlorate concentration of the milk. A 1-month-old infant's median weight is 4.14 kg (EPA
1997a) and these infants consume a mean intake of 673 ml of milk (Dewey 1983). A 6-month-
old infant's median weight is 7.53 kg (EPA 1997a), and those infants consume a mean intake of
896 ml of milk (Dewey 1983). Therefore, the mean perchlorate concentration in breast milk that
induces a perchlorate RfD exposure of 0.0007 mg/kg-day in a median sized neonate is:

For a 1-month-old infant:

(0.7 |ig/kg-day x 4.14 kg body weight) ^ 0.673 L of milk/day = 4.3 |ig/L

For a 6-month-old infant:

(0.7 |ig/kg-day x 7.53 kg body weight) ^ 0.896 L of milk/day = 5.9 |ig/L

In fall 2003 and spring 2004, Kirk evaluated the concentrations of perchlorate and iodide
in the breast milk of 36 healthy women across 18 States and 23 healthy women across 14 States,
respectively (Kirk 2007). The following table summaries the study results for the concentration
of perchlorate and iodide found in human breast milk:

-------
10-P-0101



Perchlorate and Iodide Concentrations in Human Breast Milk



Number of

Range

Mean *

Median

Analyte

Subjects (n)

(mq/l)

(mq/l)

(mq/l)

Perchlorate

36

1.4 to 92.2

10.5 ± n/a

3.3

Iodide

23

4.5 to 184.5

63.3 ± n/a

33.5

* No mean statistics provided in the paper. The paper provided an average for each test subject comprising 4 to 6
samples per subject.

Source: Kirk 2005.

Published in February 2007, Kirk evaluated the concentrations of perchlorate,
thiocyanate, and iodide in the breast milk of 10 nonsmoking lactating women with five subjects
from the Texas Panhandle and one subject from each of the following States: Colorado, Florida,
Missouri, New Mexico, and North Carolina (Kirk 2007). Using an average milk consumption of
100 ml/kg-day, the authors found that 3 of 10 subjects averaged breast milk perchlorate
concentrations above 7 |ig/L, which would cause these 3 infants to exceed the perchlorate RfD of
0.7 |ig/kg-day. As the infant grows, the infant's exposure per unit weight decreases. The
authors project that an average weight infant consuming an average amount of milk would
exceed the perchlorate RfD for the first 2 months of life. Furthermore, the authors found that the
median iodide concentration of 55.2 |ig/L is substantially below the recommended level of 110-
130 |ig/day. The following table summarizes the study's results for the concentration of
perchlorate and iodide found in human breast milk:

Analyte

Perchlorate and Iodide Concentrations in Human Breast Milk
in 10 Nonsmoking Women

Number of
sample (n)

Range

(mq/l)

Mean ± SD

(mq/l)

Median

(mq/l)

Perchlorate

147

0.5 to 39.5

5.8 ±6.2

4.0

Iodide

108

3.1 to 334

87.9 ±80.9

55.2

Source: Kirk 2007.

Between July 2002 and April 2006, breast milk and urine of 57 healthy lactating women
in the Boston area were analyzed for perchlorate and iodide concentrations (Pearce 2007). The
following table summaries the study's results for the concentration of perchlorate and iodide
found in human breast milk:



Perchlorate and Iodide Concentrations in Breast Milk and Urine Samples





from 57 Healthy Boston Women





Number of

Range

Mean ± SD

Median

Analyte

sample (n)

(mq/l)

(mq/l)

(mq/l)

Perchlorate

49

1.3 to 411

33 ±77

9.1

Urinary Perchlorate

56

0.37 to 127

8.2 ± 19

3.0

Iodide

57

2.7 to 1968

205 ±271

155

Urinary Iodide

57

25 to 920

155 ±142

114

Source: Pearce 2007.

82

-------
10-P-0101

Estimate Using the Clewell Perchlorate PBPK Model

The Clewell Perchlorate PBPK Model predicts that the internal perchlorate serum
concentration in a 1.5-month-old nursing neonate will be 4 times greater then the internal
perchlorate serum concentration in an adult from an external perchlorate dose of 0.001 mg/kg-
day (Clewell 2007, table 4). Using the Blount biomonitoring data to characterize the perchlorate
exposure in the U.S. population, if a pregnant woman is exposed to the median adult perchlorate
exposure of 4.6 |ig/day (i.e., 9.4% of the RfD), the nursing neonate would be expected to have an
internal perchlorate serum concentration that corresponds to 37% that of the RfD. Likewise, if a
pregnant woman is exposed to the 95th percentile perchlorate exposure for an adult of 16.4
|ig/day (i.e., 33% of the RfD), the nursing neonate would be expected have an internal
perchlorate serum concentration that corresponds to 1.34 times the perchlorate RfD. Finally, if a
pregnant woman is exposed to perchlorate at the RfD, the nursing neonate would be expected
have an internal perchlorate serum concentration that corresponds to 4 times the perchlorate RfD.

7.2 Total NIS Inhibitor Exposure

Since perchlorate is not the only NIS inhibitor to which these three groups are exposed,
the perchlorate exposure must be evaluated in context with the total amount of NIS inhibition on
the body (i.e., the goitrogen load). To determine the total goitrogen load, the concurrent amount
of exposure to thiocyanate and nitrate must be estimated in each of these groups. Knowing the
exposure to each NIS inhibitor in each group allows the total goitrogen load to be determined for
each group to assess the contribution of perchlorate to the body's total goitrogen load.
Furthermore, the Tonacchera Model also quantifies the impact of the availability of iodide on the
thyroid's ability to uptake iodide. Therefore, the iodide nutrition status of each group must be
determined and evaluated (see Section 8 of this document).

7.2.1 Total NIS Inhibition Load Acting on Adults

The total NIS inhibition load (i.e., total goitrogen load) acting on the body is determined
by knowing the serum concentration of the three environmentally prominent NIS inhibitors:
perchlorate, nitrate, and thiocyanate. The serum concentration of each is then used to calculate
the SPEC.

The perchlorate serum concentration in an adult was predicted using the Clewell
Perchlorate PBPK Model (Clewell 2007). The perchlorate RfD was dervived by applying a 10-
fold UF to the NOEL of 7 |ig/kg-day measured in Greer's subjects (NAS 2005, p 15). Therefore,
the perchlorate NOEL corresponds to an external dose of 490 |ig/day. The closest point
provided in Clewell's paper is that an external dose of 700 |ig/day is predicted to generate a
perchlorate serum concentration of 10 |ig/L in an adult (Clewell 2007, table 4). If this value is
linearly scaled down to the perchlorate NOEL of 490 |ig/day in an adult (i.e., 7 |ig/kg-day), the
predicted perchlorate serum concentrations at the NOEL in an adult would be about 7 |ig/L (i.e.,
0.070 |imol/L), Likewise, the Clewell Perchlorate PBPK Model predicts a perchlorate serum of
2 |ig/L for an adult at an external dose of 70 |ig/day (i.e., 1 |ig/kg-day or 1.42 times the RfD)
(Clewell 2007). If this value is linearly scaled down to the EPA RfD of 49 |ig/day in an adult
(i.e., 0.7 |ig/kg-day), the predicted perchlorate serum concentrations at EPA's RfD in an adult

83

-------
10-P-0101

would be 1.4 |ig/L (i.e., 0.014 |imol/L). If this Clewell data is scaled down to the 95th percentile
of the U.S. perchlorate exposure level of 16.4 |ig/day in an adult measured in the Blount
biomonitoring study, the predicted perchlorate serum concentrations at 95th percentile of the
U.S. perchlorate exposure level in an adult would be 0.47 |ig/L (i.e., 0.0047 |imol/L)(Blount
2006a). If the Clewell data are scaled down to the median U.S. perchlorate exposure level of 4.6
|ig/day in an adult measured in the Blount biomonitoring study, the predicted perchlorate serum
concentrations at median U.S. perchlorate exposure level in an adult would be 0.13 |ig/L (i.e.,
0.0013 |imol/L)) (Blount 2006a). The following table summarizes the predicted perchlorate
serum concentrations in an adult using the Clewell Perchlorate PBPK Model:



External Perchlorate Dose





in an a 70 kg Adult

Predicted Perchlorate Serum

Perchlorate Exposure Level

(Mg/day)

Concentration in an Adult

NOEL (7 jjg/kg-day)

490

7.0 |jg/L or 0.070 jjmol/L

RfD (0.7 jjg/kg-day)

49

1.4 |jg/L or 0.014 jjmol/L

95th Percentile

16.4

0.47 |jg/L or 0.0047 jjmol/L

50th Percentile

4.6

0.13 |jg/L or 0.0013 jjmol/L

Source: OIG Analysis.

In the Western world, the typical nitrate serum concentration ranges from 10-140
|imol/L, with the mean nitrate serum concentration being between 30-50 |imol/L (Tonacchera
2004). Thiocyanate serum concentration in nonsmokers is typically in the range of 10-70
(amol/L. By contrast, the thiocyanate serum concentration of smokers is higher and is typically
in the range of 80-120 |imol/L (Tonacchera 2004). A thiocyanate serum concentration of 40.5
(amol/L was observed in the controls in Braverman's study of workers at the ammonium
perchlorate plant in Cedar City, Utah (Bravermen 2005). Therefore, although variable, the
typical nitrate and thiocyanate serum concentrations in the U.S. population are 40 |imol/L and 40
|imol/L, respectively. These values will be used for the calculation of the total NIS inhibition
load in a U.S. adult.

The total NIS inhibition load (i.e., total goitrogen load) acting on the body is determined
by calculating the SPEC. The SPEC is calculated as follows (concept taken from Tonacchera
2004):

SPEC = (Amt of CIO4" Inhibition) + (Amt of NO3" Inhibition) + (Amt of free SCN Inhibition)
SPEC = [CIO4] + [N03"] /240 + [free SCN"] /15

Where: [ ] is molarity

Serum inhibition potency of perchlorate = 1

Serum inhibition potency of nitrate relative to perchlorate = 1/240

Serum inhibition potency of thiocyanate relative to perchlorate = 1/15

The following are calculations for the total NIS inhibition in a U.S. adult at the NOEL,
RfD, 95th percentile, and median perchlorate exposure levels expressed as SPECs:

84

-------
10-P-0101

SPEC (at NOEL)

SPEC (at RfD)

= 0.070 (imol/L + (40 (imol/L 240) + (40 (imol/L x 0.5 free SCN" 15)
= 0.070 (imol/L + 0.167 (imol/L + 1.333 (imol/L
= 1.567 junol/L

= 0.014 (imol/L + (40 (imol/L 240) + (40 (imol/L x 0.5 free SCN" 15)
= 0.014 (imol/L + 0.167 (imol/L + 1.333 (imol/L
= 1.511 (imol/L

SPEC (at 95th CL04") = 0.0047 junol/L + (40 junol/L - 240) + (40 junol/L x 0.5 free SCN - 15)
= 0.0047 (imol/L + 0.167(imol/L + 1.333 (imol/L
= 1.505 (imol/L

SPEC (median C104)

SPEC (w/no CIO4)

= 0.0013 (imol/L + (40 (imol/L 240) + (40 (imol/L x 0.5 free SCN 15)
= 0.0013 (imol/L + 0.167 (imol/L + 1.333 (imol/L
= 1.501 (imol/L

= 0.0 (imol/L + (40.0 (imol/L 240) + (40.0 (imol/L x 0.5 free SCN ^ 15)
= 0.0000 (imol/L + 0.1667 (imol/L + 1.333 (imol/L
= 1.500 (imol/L

The following table summarizes the total NIS inhibition in a typical U.S. adult having 40
|imoles/L exposures to both nitrate and thiocyanate at various levels of perchlorate exposure.
Furthermore, the table identifies the relative contribution of each NIS inhibitor to the total NIS
inhibition:

Relative Contribution of Each NIS Inhibitor
to the Total Amount of NIS Inhibition in an Adult

Perchlorate
Exposure Level

Total Amount of
NIS Inhibition in
an Adult
(SPEC)

Perchlorate

Nitrate
at

(40 |jmoles/L)

Thiocyanate
at

(40 |jmoles/L)

At the Perchlorate NOEL
(7 jjg/kg-day)

1.567 jjmoles/L

4.5%

10.7%

85.1%

At the Perchlorate RfD of
(0.7 jjg/kg-day)

1.511 jjmoles/L

0.93%

11.1%

88.2%

95th percentile of U.S.
population (Blount 2006)

1.505 jjmoles/L

0.31%

11.1%

88.6%

50th percentile of U.S.
population (Blount 2006)

1.501 jjmoles/L

0.09%

11.1%

88.8%

Zero Perchlorate

1.500 jjmoles/L

0.00%

11.1%

88.9%

Source: OIG Analysis.

The Tonacchera Model establishes a mathematical equation to calculate the TIU of the
NIS from the serum concentrations of the three NIS inhibitors: perchlorate, nitrate, and
thiocyanate. The iodide concentration in the serum (i.e., [T]) is unknown but should be directly
related to the UIC. Therefore, [T] is assigned a value of "x" in these calculations. The following
calculates the TIU at the perchlorate exposure of zero (theoretical) and at the RfD and the
amount of change in TIU from the exposure of perchlorate at the RfD and at the NOEL:

85

-------
10-P-0101

The TIU with no perchlorate exposure is expressed by:

TIU (w/no Perchlorate Exposure) || [I ] / (1-22 + [SPEC])

where: the symbol || means "proportional to"

Substitution gives:

TIU (w/no Perchlorate Exposure) || x/ (1.22 + 1,500|^mol/L)

TIU (w/no Perchlorate Exposure) || X / 2.720
TIU (w/no Perchlorate Exposure) || 0.3676X

The TIU at a perchlorate exposure level of the RfD is expressed by:

TIU (Perchlorate Exposure at RfD) || [I ] / (1.22 + [SPEC])

Substitution gives:

TIU (Perchlorate Exposure at RfD) || X / (1.22 + 1.51 1 (^mol/L)

TIU (Perchlorate Exposure at RfD) || X / 2.73 1
TIU (Perchlorate Exposure at RfD) || 0.3662X

The change in TIU with perchlorate exposure at RfD = TIU (RfD) - TIU (no CIO/) x 100%

TIU (no C104")

Substitution gives:

Change in TIU at perchlorate RfD = ((0.3662x - 0.3676x)/ 0.3676x) x 100%
Change in TIU at perchlorate RfD = - 0.38% (i.e., a decrease of 0.4 %)

The TIU at a perchlorate exposure level of the NOEL is expressed by:

TIU (Perchlorate Exposure at NOEL)	|| [I ] / (1.22 + [SPEC])

Substitution gives:

TIU (Perchlorate Exposure at NOEL)	|| x/ (1.22 + 1.567 (^mol/L)

TIU (Perchlorate Exposure at NOEL)	|| x/ 2.787

TIU (Perchlorate Exposure at NOEL)	|| 0.3588X

Change in TIU with perchlorate exposure at RfD = TIU (NOEL) - TIU (no CIO4") x 100%

TIU (no CIO4")

Substitution gives:

Change in TIU at perchlorate RfD = ((0.3588x - 0.3676x)/ 0.3676x) x 100%
Change in TIU at perchlorate RfD = - 2.4% (i.e., a decrease of 2.4 %)

In a typical adult with nitrate and thiocyanate exposures of 40 |imoles/L exposures, the
above calculations predict that the exposure to perchlorate at the RfD or at the NOEL will
decrease the uptake of iodide by the thyroid by 0.4% and 1.9%, respectively. However, the
Tonacchera Model predicts TIU is proportional to [T] for all levels of iodide nutrition for any
fixed underlying goitrogen load between 0.1 |imol/L to about 12 |imol/L PEC. The TIU vs log
PEC relationship becomes nonlinear above 12 |imol/L PEC (Tonacchera 2004, figure 2).
Therefore, the TIU could remain unchanged at the perchlorate exposure level of the RfD and at
the NOEL in an adult if "x" (iodide concentration in the serum) increases by 0.4% and 2.4%,
respectively.

86

-------
10-P-0101

Seasonal and Dietary Variation in NIS inhibitor Load

By comparison, consumption of common vegetables is a significant and natural source of
thiocyanates. In a Norwegian health study of 25,300 people, the mean thiocyanate serum levels
(± SD) were 33.9 ± 14.0 |imol/L and 33.4 ± 14.0 |imol/L in nonsmoking males and females,
respectively (Foss 1986). The thiocyanate serum levels in 95% of the nonsmokers ranged
fromlO |imol/L to 60|imol/L. Foods particularly rich in thiocyanate include cabbage, broccoli,
Brussels sprouts, corn (maize), turnips, rapeseed, and mustard seed. Foods are known to contain
varying quantities of thiocyanate and the content varies considerably from sample to sample. An
individual's thiocyanate serum level will vary depending on the frequency and amount of
consumption of thiocyanate-containing food types.

For example, a nonsmoking individual consumed 100 grams of raw Brussels sprouts each
day for a week to document a rise in thiocyanate serum levels from 31 to 80 |imol/L. Dietary
thiocyanate exposure also exhibits a change of about 10 |imol/L, presumably due to seasonal
changes in the composition of the diet (Foss 1986). Cabbage is a high thiocyanate content food
with consumption higher during autumn than any other time of year. The following table
demonstrates the seasonality of thiocyanate serum levels in nonsmokers:

Serum Thiocyanate (|jmol/L)

1st Quarter

2nd Quarter

3rd Quarter

4th Quarter

Study Group

of the Year

Nonsmoking Males (n = 6212)

29.5

30.2

39.4

Nonsmoking Females (n = 7908)

29.9

30.1

39.9

39.4

Source: Foss 1986.

The purpose of reviewing natural thiocyanate exposure from the diet is to show that a 10
(amol/L change in human thiocyanate serum levels is common. Both the nonsmoking males and
females in the Norwegian health study had a ± 14.0 |imol/L SD with regard to the mean (Foss
1986). Furthermore, the seasonal change in thiocyanate serum level in the Norwegian
nonsmokers was about 10.0 |imol/L.

In regard to the amount of NIS inhibition, a change of 10.0 |imol/L in human thiocyanate
serum levels represents a much greater amount of NIS inhibition than exposure to perchlorate at
the RfD in an adult. The amount of NIS inhibition of 10.0 |imol/L in human thiocyanate serum
is measured by the Tonacchera Model as SPEC and has value of 0.333 |imol/L (i.e., (10 |imol/L
x 0.5 free SCN) ^ 15). By comparison, perchlorate exposure at the RfD has an estimated SPEC
of 0.014 |imol/L. Therefore, the fluctuation in the human thiocyanate serum of 10.0 |imol/L
causes a change in the amount of NIS inhibition that is about 24 times greater than the amount of
NIS inhibition caused by perchlorate exposure at the RfD.

7.2.2 Total NIS Inhibition Load Acting on Fetuses

The total NIS inhibition load (i.e., total goitrogen load) acting on the fetus is determined
by estimating the serum concentration in the fetus of the three environmentally prominent NIS
inhibitors: perchlorate, nitrate, and thiocyanate. The serum concentration of each is then used to
calculate the SPEC in the fetus. Direct measurements of fetal exposures to each of the NIS

-------
10-P-0101

inhibitors are not possible due to the unacceptable risk to the unborn child. However, a direct
measurement of serum concentration is possible at birth without risk to the fetus by measuring
the serum concentration in umbilical cord blood.

The fetal thiocyanate serum concentration can be estimated by direct measurements of
thiocyanate in both maternal and cord serums that have been reported in at least four studies (see
table below). The thiocyanate ratio of cord serum to maternal serum observed in nonsmokers is
71%, 88%, and 92% observed in the Hauth, Laurberg, and Vanderpass studies, respectively
(Hauth 1984; Laurberg 2004; Vanderpass 1984). For estimating the fetal thiocyanate serum
concentration, the thiocyanate ratio of cord serum to maternal serum of 71% observed in the
nonsmokers from the Hauth study is used (Hauth 1984). Using the thiocyanate ratio of cord
serum to maternal serum of 71% provides the lowest expected value in the fetal thiocyanate
serum concentration (i.e., use of either 88% or 92% to estimate the fetal thiocyanate serum
concentration would decrease the relative contribution of NIS inhibition from perchlorate). The
fetal thiocyanate serum concentration is estimated by multiplying the typical thiocyanate serum
concentration found in the Western world of 40 |imol/L by the thiocyanate ratio of cord serum to
maternal serum of 71%. Therefore, the typical fetal thiocyanate serum concentration in the
Western world is estimated to be at least 28.4 |imol/L. This estimated fetal thiocyanate serum
concentration is a conservative value and is based on the direct thiocyanate measures observed in
cord blood.



Mother's

Cord SCN"





SCN" Serum

Serum



Study

Concentration

Concentration

SCN Ratio

(Exposure Type/Description)

(pmol/L)

(pmol/L)

Cord Serum/Mother Serum

Hauth 1984







Smokers

95

72

76%

Passive Smokers

35.9

26

72%

Nonsmokers

32.3

23

71%

Laurberg 2004
Smokers

84.9 ±25.4

94.6 ±31.9

111%

Nonsmokers

54.7 ±18.2

48.3 ± 15.5

88%

Nafstad 1995







24 Smokers & 4 Nonsmokers

69

47

68%

Vanderpass 1984

Central Africa (dietary SCN)

133 ±7

134 ±2

101%

Belgian Controls

36 ±7

33 ±7

92%

Sources: Hauth 1984; Laurberg 2004; Nafstad 1995; Vanderpass 1984.

The Clewell PBPK model predicts a perchlorate serum concentration (i.e., an internal
indice) of 1.4 |ig/L (i.e., 0.014 |imol/L) in an adult at an external exposure dose at the RfD. The
Clewell PBPK model predicts a pregnant woman's perchlorate serum concentration as being 2.5
times greater than an adult's perchlorate serum concentration at an external exposure dose at the
RfD which corresponds to 3.5 |ig/L or 0.035 |imol/L. Likewise, the Clewell PBPK model
predicts a fetus's perchlorate serum concentration as being 5 times greater than an adult's
perchlorate serum concentration at an external exposure dose at the RfD which corresponds to 7
|ig/L or 0.070 |imol/L.

88

-------
10-P-0101

Although NIS is present in the placenta, the transport of nitrate from the mother's blood
to the fetal blood would be expected (i.e., analogous to the known transport of nitrate from
maternal blood into breast milk). However, the OIG could not find a study that reports the
amount of nitrate found in cord blood. Although nitrate contributes to fetal NIS inhibition, the
amount of NIS inhibition from nitrate is problematic to estimate and is not used in the following
calculation of the total NIS inhibition load on the fetal thyroid with a perchlorate exposure at the
RfD:

SPEC(RfD) = 0.070 |imol/L + (|imol/L of NO3" unknown) + ((40 |imol/L x 0.71 ratio) x 0.5 -M5)
= 0.070 |imol/L + (|imol/L of NO3" unknown) + ((28.4 |imol/L) x 0.5 15)
= 0.070 |imol/L + (NO3" contribution unknown) + 0.947 |imol/L
= 1.017 |imol/L

In a fetus, the perchlorate at the RfD contributes about 6.9% of the total NIS inhibition
load acting on the thyroid. (Note: With the contribution from nitrate or the use of a higher SCN"
maternal/cord serum ratio, this estimated value will be smaller.)

The TIU of the fetus with the mother having an external exposure dose at the RfD is
expressed by:

TIU (FetuS) || [T]/(1.22 +[SPEC])

Substitution gives:

TIU (Fetus) || x/ (1.22 + 1.017 |imol/L)

TIU (Fetus) II x / 2.237

TIU (Fetus) || 0.4470x

The TIU of the fetus with the mother having no perchlorate exposure is expressed by:

TIU (Fetus with no perchlorate exposure) || [I ] / (1.22 ~l~ [SPEC])

Substitution gives:

TIU (Fetus with no perchlorate exposure) || x/ (1.22 + 0.947 (^mol/L)

TIU (Fetus with no perchlorate exposure) 11 X / 2.167

TIU (Fetus with no perchlorate exposure) || 0.4615x

The change in %TIU(Fetus) with mother exposed at RfD versus no perchlorate exposure is
given by:

= TIU (Fetus with exposure) - TIU (Fetus with no CIO4") x 100%
TIU(fetus with no CIO4")

89

-------
10-P-0101

Substitution gives:

%TIU(Fetus) w/ mother exposed at RfD = ((0.4470x - 0.4615x)/ 0.4615x) x 100%

= - 3.1% (i.e., a decrease of 3.1 %)

(Note: With the contribution from nitrate, this estimated value will be < 3.1%.)
7.2.3 Total NIS Inhibition Load Acting on Nursing Infants

The total NIS inhibition load acting on nursing infant's thyroid at a perchlorate RfD
exposure to the nursing mother can be estimated by adding the amount of NIS inhibition
contributed from perchlorate, nitrate, and thiocyanate together. All three of these NIS inhibitors
are found in human breast milk.

In regard to thiocyanate in breast milk, a Norwegian study reported the mean thiocyanate
concentration in breast milk as 9.0 mg/L in smokers and 10.5 mg/L in nonsmokers (Dorea 2004).
By contrast, the mean thiocyanate concentration in breast milk in the United States is reported to
be 4.2 mg/L in smokers and 0.92 mg/L in nonsmokers (Dorea 2004). The reported thiocyanate
concentrations in breast milk in nonsmokers, which range from a low of 0.92 mg/L to a high of
10.5 mg/L, are large (i.e., an order of magnitude), which introduces a lot of uncertainty into the
estimation. For purposes of estimating (i.e., a conservative estimate) the total NIS inhibition
load acting on nursing infant's thyroid at a perchlorate RfD exposure to the nursing mother, the
mean thiocyanate concentration in breast milk in the United States of 0.92 mg/L in nonsmokers
will be used.

In regard to nitrate concentrations in breast milk, 20 healthy, lactating women in the State
of Iowa had their breast milk sampled and tested for nitrate (Dusdieker 1996). The study found
4.4 ±3.6 mg nitrate/L in human milk (Dusdieker 1996).

The concentration of perchlorate in breast milk when the lactating mother is exposed to
perchlorate at the RfD is estimated from the perchlorate PBPK podel in rats. The perchlorate
PBPK model in rats predicts that about 50% of the daily maternal perchlorate dose at 0.01
mg/kg-day is transferred to the nursing pup through breast milk (Clewell 2003, 416, (language in
the abstract)). If the same relationship holds in humans, lactating women exposed at the
perchlorate RfD (i.e., 49 |ig/day) would be predicted to transfer 24.5 |ig perchlorate/day to the
nursing infant. Neonate perchlorate exposure is a function of both the weight of the infant, the
amount of milk consumption, and the perchlorate concentration of the milk. A 1-month-old
infant's median weight is 4.14 kg (EPA 1997a) and that infant consumes a mean intake of 673
ml of milk (Dewey 1983). A 6-month-old infant's median weight is 7.53 kg (EPA 1997a) and
that infant consumes a mean intake of 673 ml of milk (Dewey 1983). Since a 1-month-old infant
consumes 0.673 L of milk/day, a breast milk concentration of 36.4 |ig perchlorate/L transfers
24.5 |ig perchlorate/day to the nursing infant (i.e., the 50% transfer the mother's RfD perchlorate
exposure to the nursing infant predicted by the PBPK model). Likewise, a 6-month-old infant
consumes 0.896 L of milk/day, a breast milk concentration of 27.3 |ig perchlorate/L transfers
24.5 |ig perchlorate/day to the nursing infant.

-------
10-P-0101

The total NIS inhibition load acting on a nursing infant's thyroid at a perchlorate RfD
exposure to the nursing mother estimates perchlorate breast milk concentrations of 36.4 |ig/L for
a 1-month-old and 27.3 |ig/L for a 6-month-old (i.e., perchlorate concentration in breast milk is
estimated to be higher because less milk volume is generated to meet the child's needs). These
PBPK model estimates of 36.4 |ig/L and 27.3 |ig/L for perchlorate concentrations in human
breast milk are substantially above the measured median perchlorate concentrations in human
breast milk of 3.3 |ig/L, 4.0 |ig/L and 9.1 |ig/L (Kirk 2005; Kirk 2007; Pearce 2007). The
exposure to each of the NIS inhibitors found in human breast milk is estimated as follows:

For a 1-month-old infant:

36.4 |ig C1047L x 0.673 L of milk/day = 24.5 |ig C1047day
920 |ig SCN7L x 0.673 L of milk/day = 619 |ig SCN" /day
4400 |ig MV/L x 0.673 L of milk/day = 2961 |ig nitrate/day

For a 6-month-old infant:

27.3 |ig C1047L x 0.896 L of milk/day = 24.5 |ig C1047day
920 |ig SCN7L x 0.896 L of milk/day = 824 |ig SCN" /day
4400 |ig N037L x 0.896 L of milk/day = 3942.4 |ig nitrate/day

The total NIS inhibitory effect measured upon ingestion of the NIS inhibitors is given by
the following OPEC equation (concept adapted from Tonacchera 2004; De Groef 2006, 155):

OPEC = (Amt of C104- Inhibition) + (Amt of N03- Inhibition) + (Amt of SCN- Inhibition)
OPEC = grams perchlorate + grams nitrate/150 + 0.5 free SCN- x grams thiocyanate/8.8
OPEC = grams perchlorate + grams nitrate/150 + grams thiocyanate/17.6

Where:

Oral inhibition potency of perchlorate = 1

Oral inhibition potency of nitrate relative to perchlorate = 1/150

Oral inhibition potency of thiocyanate relative to perchlorate = 0.5/8.8 = 1/17.6

The total NIS inhibition load acting on nursing infant's thyroid at a perchlorate RfD
exposure to the nursing mother is estimated as follows:

For a 1-month-old infant:

OPEC(i month) = (Amt C104" Inhibition) + (Amt of N03" Inhibition) + (Amt of SCN"
Inhibition)

OPEC(i month) = |ig perchlorate + |ig nitrate/150 + |ig thiocyanate/17.6
OPEC(l month) — 24.5 |ig C104"/day + 2961 |ig nitrate/day ^ 150 + 619 |ig SCN" /day ^ 17.6
OPEC(i month) = 24.5 |ig C104"/day + 19.74 |ig C104"equiv./day + 35.2 |ig C104"equiv./day
OPEC(i month) = 79.4 |ig C104"equivalent/day

For a 6-month-old infant:

OPEC(6 month) = (Amt C104" Inhibition) + (Amt of N03" Inhibition) + (Amt of SCN"
Inhibition)

OP EC (6 month) = Hg perchlorate + |ig nitrate/150 + |ig thiocyanate/17.6

91

-------
10-P-0101

OPEC(6 month) 24.5 |ig C104"/day + 3942.4 |ig nitrate/day ^ 150 + 824 |ig SCN" /day ^ 17.6
OPEC(6 month) = 24.5 |ig C104Vday + 26.3 |ig C104"equiv./day + 46.8 |ig C104"equiv./day
OPEC(6 month) = 97.6 |ig C104"equivalent/day

At both 1 month and 6 months, thiocyanate exposure in breast milk contributes the
largest portion of the NIS inhibition. These OPEC values were calculated using the lower
estimated thiocyanate concentration in breast milk of 0.92 mg/L in nonsmokers. The OPEC
values would be much larger if the higher estimated thiocyanate concentration in breast milk of
10.5 mg/L in nonsmokers was used. Using the lowest thiocyanate exposure at 1 month, maternal
perchlorate exposure at the RfD is estimated to result in perchlorate contributing about 30% of
the nursing infant's total NIS inhibition load. Using the lowest thiocyanate exposure at 6
months, maternal perchlorate exposure at the RfD is estimated to result in perchlorate
contributing about 25% of the nursing infant's total NIS inhibition load.

This characterization of a nursing infant's total NIS inhibition load clearly indicates that
perchlorate is not acting alone and the use of a single chemical risk assessment of perchlorate to
characterize the risk of NIS inhibition acting on a nursing infant's thyroid is incomplete. A
cumulative risk assessment approach is needed to characterize and understand the sources and
total risk of NIS inhibition acting on a nursing infant's thyroid. Furthermore, a cumulative risk
assessment approach is needed to quantify the effect of all four NIS stressor on the TIU by the
nursing infant.

7.2.4 Total NIS Inhibition Load Acting on Non-nursing Infants

To calculate or estimate the total NIS inhibition load (i.e., total goitrogen load) acting on
the thyroid of non-nursing infants, the non-nursing infant exposure to each of the three
environmentally prominent NIS inhibitors (i.e., thiocyanate, nitrate, and perchlorate) must be
known. Unfortunately, neither the serum concentrations nor the amount of oral consumption to
all three NIS inhibitors are available to calculate or estimate the total NIS inhibition load (i.e.,
total goitrogen load) acting on the thyroid of non-nursing infants.

The perchlorate exposure in non-nursing infants can be estimated using the results from
the 2008 FDA Food Dietary Study. The 2008 FDA Food Dietary Study reports the total
perchlorate intake from food for 6- to 11-month-old infants to be 0.26-0.29 |ig/kg-day (i.e., not
including potential perchlorate exposure from water) (Murray 2008, Table 5). Since the
perchlorate RfD is 0.6 |ig/kg-day, the perchlorate exposure from food for 6- to 11-month-old
infants of 0.26-0.29 |ig/kg-day represents 37% to 41% of the perchlorate RfD. This suggests a
Relative Source Contribution (RSC) of about 60% for non-nursing infants.

Unfortunately, this estimated perchlorate RSC is derived using a single chemical risk
assessment process that is characterized as being outdated. In other words, limiting only
perchlorate to protect public health does not insure that the total NIS inhibition load acting on the
non-nursing infants is "safe" because the NIS inhibition exposure from thiocyanate and nitrate in
the food and water of the non-nursing infant is not considered. The public health risk to non-
nursing infants from a potentially excessive amount of NIS inhibitors is not known without
knowing the non-nursing infants' exposure to thiocyanate and nitrate. However, the relative

92

-------
10-P-0101

contribution of each NIS inhibitor to the total amount of NIS inhibition load acting on the body
is known for adults. In typical nonsmoking U.S. adults, thiocyanate and nitrate from the
consumption of food and water contributes the vast majority (i.e., > 99.9%) of the body's total
NIS inhibition load (see Section 7.2.1). Since a non-nursing infant's diet contains the same food
sources (e.g., vegetables (see Sections 3.3 and 3.4) that contribute thiocyanate and nitrate in an
adult's diet, the vast majority of the total NIS inhibition load acting on the thyroid of the non-
nursing infant is still expected to be from thiocyanate and nitrate and not from perchlorate.
Therefore, from a cumulative risk assessment standpoint, the majority of risk to non-nursing
infants from exposure to NIS inhibitors is from thiocyanate and nitrate. Furthermore, to
effectively characterize and manage the risk, the exposure to all three NIS inhibitors needs to be
determined and the combined exposure needs to be managed to prevent an excessive total NIS
inhibition load acting on the non-nursing infant.

Another aspect of a cumulative risk assessment is that the body's tolerance for NIS
inhibition is a function of the amount of iodide in the diet. The lack of iodide is also an NIS
stressor on the thyroid and also limits the amount of TIU by the thyroid. Therefore, within
certain biological limits, the more iodide in the diet, the body can tolerate a larger total NIS
inhibition load without adverse effects. The converse of this relationship is also true, within
certain biological limits: the less iodide in the diet, the body cannot tolerate as much total NIS
inhibition load before adverse effects are observed. In short, there is no single limit to the
amount of total NIS inhibition the body can tolerate; the body's toleration of NIS inhibition is on
a sliding scale depending on the amount of iodide in the diet.

In regard to the iodide nutrition observed in non-nursing infants, the iodide nutrition level
can be estimated using the results from the 2008 FDA Food Dietary Study. The 2008 FDA Food
Dietary Study identifies that baby foods and dairy products account for nearly all (90%) of the
estimated iodine intake by infants. The 2008 FDA Food Dietary Study reports the range of
estimated lower- and upper-bound average iodide intakes for 2003-2004 for 6- to 11-month-old
infants to be 144-155 |ig/day (Murray 2008, table 7). The 2008 FDA Food Dietary Study
indicates that an adequate intake of iodide for 6- to 11-month-old infants is 130 |ig/day (Murray
2008, table 7). Therefore, the 2008 FDA Food Dietary Study indicates that the majority of 6- to
11-month-old U.S. infants have an adequate intake of iodide, which allows these infants to
tolerate a larger amount of total NIS inhibition load compared to an infant with an iodide-
deficient diet. Unfortunately, the 2008 FDA Food Dietary Study was not designed to provide the
iodide intake distribution within each population subgroup. This iodide intake distribution
within the U.S. population subgroups would help to better characterize the public health risk
because the risk to the public from NIS inhibitors increases with the decreasing intake of iodide.

-------
10-P-0101

8. Risk Characterization of NIS Stress

In 2003, the EPA Framework for Cumulative Risk Assessment identifies, "For most
exposure situations, hazard and dose response studies of all of the joint effects from the multiple
stressors will not be available, so that conclusions will have to be based at least partly on the
single stressor information" (EPA 2003, p 44). For perchlorate, a risk characterization cannot be
done in humans because adverse health effects have not been clearly demonstrated in any human
population exposed to perchlorate (NAS 2005, p 177). However, the EPA Framework for
Cumulative Risk Assessment allows the risk characterization to be based on the adverse effects
observed in one of the stressors, and then be applied to the joint effects from the multiple
stressors. Therefore, the risk characterization for NIS stressors must be done by using an NIS
stressor in which adverse effects in children born to mothers with low maternal TIU during
pregnancy and nursing have been documented and reported. The excessive maternal exposure to
the NIS stressor, the lack of iodide, is the principal NIS stressor in which adverse effects have
been documented and reported in children born to mothers with low maternal TIU during
pregnancy and nursing. Thus, the adverse effects from the lack of iodide stressor can be used to
conduct the initial three steps in a risk characterization (i.e., hazard characterization, dose-
response assessment, and exposure assessment) to this public health issue. The exposure to the
lack of iodide stressor that induces an adverse effect can be related back to the joint exposure
level of all four NIS stressors by integrating the joint exposure into a single parameter or
"common metric" (EPA 2003, p 55, section 3.3.3.1). The single parameter for measuring the
NIS stress level is the TIU. The Tonacchera Model can be used to mathematically calculate the
resulting TIU level from the concurrent joint exposure to all four NIS stressors.

The inclusion of the lack of iodide stressor is essential in characterizing and
understanding this public health issue. An insufficient uptake of iodide in both the pregnant
mother's thyroid and her child's thyroid during gestation and lactation is the initiating biological
event that triggers the subsequent potential for permanent brain damage in the developing fetus
and nursing infant. An insufficient uptake of iodide can be caused by the exposure to an
excessive amount of NIS inhibitors, but an insufficient uptake of iodide can also be caused by
the lack of iodide in the diet (i.e., iodide deficiency). The excessive ingestion of NIS inhibitors
or an iodide-deficient diet (or a combination of both) can lead to an insufficient uptake of iodide;
therefore, excessive NIS inhibitors or the lack of iodide both initiate the same subsequent
adverse effects. Thus, critical insight into this public health issue can be learned by studying the
adverse effects observed in the children of iodide-deficient pregnant women. The adverse effects
observed in the offspring of iodide-deficient pregnant women are the same adverse effects
expected to be observed in the offspring of pregnant women with an excessive exposure to NIS
inhibitors, because the two groups share the same mechanism of toxicity (i.e., a low TIU).

The 1991 EPA Guidelines for Developmental Toxicity Risk Assessment clearly states that
human studies are the most appropriate data for conducting developmental toxicity risk
assessments (EPA 1991, p 1). Human data avoid the uncertainties of extrapolating from animals
studies to humans. Human data also provide the ability to identify subtle mental deficits (e.g.,
ADHD or verbal comprehension) that are not possible to detect or identify in animal studies.
Unfortunately, subtle adverse effects can be observed in the children born to iodide-deficient
pregnant women. The type and severity of these adverse effects that occur in the children born

-------
10-P-0101

to iodide-deficient pregnant women can be studied and identified. Furthermore, through the use
of the Tonacchera Model, the amount of TIU that results in adverse effects to the offspring can
be calculated from the iodide exposure levels in the pregnant women. To accomplish this task,
the lack of iodide stressor will be evaluated using the traditional risk assessment techniques of a
hazard characterization, dose-response assessment, and an exposure assessment.

95

-------
10-P-0101

8.1 Hazard Characterization

EPA's Framework for Cumulative Risk Assessment states that a stressor does not have to
be the exposure to a chemical, but the "absence of a necessity" (EPA 2003). In terms of the
necessity for the thyroid to uptake enough iodide to make a sufficient supply thyroid hormones
for proper brain development, the potential absence of iodide in the diet of pregnant women and
neonates is the dominant stressor in this public health issue, according to the Tonacchera Model.
To characterize the risk from this nonchemical NIS stressor, the same process used for chemical
agents is appropriate, except this time the adverse health effects are observed by the lack of
exposure. The hazard characterization is the description of the potential adverse health effects
attributed to a specific environmental stressor; the mechanisms by which a stressor exerts its
toxic effect; and the associated dose, route, duration, and timing of exposure (EPA 2002).

The absence of iodide in the diet (i.e., poor iodide nutrition) acts through the same
mechanism of toxicity as perchlorate and the other NIS inhibitors. Although the lack of iodide
does not act as an NIS inhibitor, the lack of iodide has the same biological result by preventing
an adequate uptake of iodide by the NIS. Therefore, the lack of iodide would have the same
adverse health effects as proposed by an excess exposure to perchlorate.

The EPA mode of action for perchlorate proposed a continuum of possible health effects
from perchlorate exposure that included the potential for permanent neurodevelopmental deficits
in infants (EPA 2002a). The NAS Committee mode of action for perchlorate also identifies a
continuum of biological effects that includes the possibility for abnormal fetal and child growth
and development (NAS 2005). However, the NAS Committee "emphasizes" that the inhibition
of iodide uptake by the thyroid has been the only consistently documented effect of perchlorate
exposure in humans (NAS 2005, p 165). The NAS Committee also states that the continuum of
possible health effects is only proposed and has not been clearly demonstrated in any human
population exposed to perchlorate (NAS 2005, p 165). Therefore, the potential perchlorate
hazard for neurodevelomental deficits in infants is only theoretical.

By contrast, the neurodevelopmental hazard from this nonchemical NIS stressor (i.e., the
lack of iodide) is not theoretical. The neurodevelopmental hazard from this nonchemical NIS
stressor is known, characterized, and observed in children born in iodide-deficient populations.
Iodide deficiency is fundamental in the cause of endemic cretinism (Delange 2005a). Cretinism
is the extreme, irreversible mental retardation and impairment of physical development of
children. In severely iodide-deficient populations, the prevalence of cretinism can reach 5% to
15% of the population. However, the combined effect of the four NIS stressors on the uptake of
iodide by the thyroid does not reach this level of severity in the United States.

The neurodevelopmental hazard from the lack of iodide is a not threshold effect, but is
better characterized by a dose-response relationship. In other words, the severity of mental
deficits and the frequency of occurrence of mental deficits increase with the severity of the
iodide deficiency. One cross-sectional epidemiological study in children observed a dose-
response relationship between UIC and IQ, (i.e, IQ decreases with decreasing UIC) (Santiago-
Fernandez 2004). Children born and raised in mild iodide-deficient populations are not reported
to have intellectual or cognitive deficits (Delange 2001), but are observed to have slower

-------
10-P-0101

reaction times (Lombardi 1995; Vitti 1992). However, children born and raised in moderate
iodide-deficient populations have minor, subtle, or overt neurological, psychological, and
intellectual deficits (Delange 2005, p 278; Lombardi 1995). These mental deficits are of the
same nature but less prominent than the mental deficits found in children born in severe iodide-
deficient populations. Brief thyroid failure during fetal development or infancy, when the brain
is still developing, is thought to cause these mental deficits.

In 1994, Bleichrodt conducted a meta-analysis of 18 studies comparing the mental
development in children between either iodide-sufficient and iodide-deficient regions or iodide
prophylaxis (iodide supplement) and a placebo (Bleichrodt 1994). Meta-analysis relies on the
fact that statistical significance is highly dependent on sample size. By combining the results
from 18 smaller studies, the meta-analysis generated a larger dataset - a total of 2,214 subjects -
to identify a statistically significant trend. A meta-analysis generates a Cohen lvalue where d is
the difference between the two group means (i.e., control and experimental groups) divided by
the standard deviation. A Cohen lvalue of a d= 0.2 is classified as a small effect size, a d= 0.5
is classified as a medium effect size, and a d= 0.8 is classified as a large effect size. The
Bleichrodt meta-analysis of the effects of iodide deficiency on cognitive development found a
large effect size with a lvalue = 0.9. Therefore, the mean IQ scores between iodide-deficient
group and the iodide-sufficient group were 0.9 SD or 13.5 IQ points apart. In other words, 82%
of the children from the iodide-deficient group score below the average IQ score of a iodide-
sufficient child.

The specific adverse mental deficits are documented to be observed in children born to
mothers with mild iodide deficiency, moderate iodide deficiency, and hypothyroidism are
provided in the following three sections.

Adverse Effects Observed in Mild iodide-Deficient Populations

Lombardi 1995 Study: The Lombardi epidemiological study evaluating the
neuropsychological impairment of mild iodide deficiency during fetal and neonatal life reported
delayed reaction times in children aged 6-10 in Tuscany, Italy (Lombardi 1995). The study
measured the reaction time between visual stimulus and the response for two child populations:
a mildly iodide-deficient area (IDA) and an iodide-sufficient area (ISA). The measurement of
simple reaction time is a sensitive test to detect subtle neurological damage from mild iodide-
deficiency during fetal and neonatal brain development. The following table summaries the
study:

Village

Number of
Children

Urinary Iodide Concentration
(mean ± SD)

Description of Iodide Level

Borgo a Mozzano

719

80.1 ±57

Iodide-Deficient Area (IDA)

Marina di Pisa

106

173 ±95

Iodide-Sufficient Area (ISA)

Source: Lombardi 1995.

Reaction time is known to decrease with age (i.e., explains the observed decrease in
reaction time with age in this dataset, but the significance of the study is that reaction time of
ISA children is statistically significantly shorter than the reaction time of the IDA children at

-------
10-P-0101

each age (p<0.05)). The following figure provides the observed reaction times between iodide-
deficient and -sufficient children by year (note: the x-axis is age of the child in years):

Fig. 1 - Comparison between RTs (msec) of children born in io-
dine deficient area (IDA) and in iodine sufficient area (ISA). RTs
were expressed as median. *p <0.05.

Vitti 1992 Study: The Vitti study also evaluated the neuropsychological impairment of
mild iodide deficiency during fetal and neonatal life by measuring the reaction times in two
groups of schoolchildren in Vagli, Italy (Vitti 1992). Group 1 contained 30 children (mean age
11.5 years) who were born before iodide prophylaxis. Group 2 contained 27 children (mean age
7.3 years) who were born after iodide prophylaxis. This study measured the reaction time
between a visual stimulus and the response observed in the two groups and compared them
against their age-matched controls living in an ISA. The results are summarized in the following
table:

Study Group

Vagli Children
Mean RT

Aged-Matched
Controls
Mean RT*

Statistical
Significance

Group 1 (mean age 11.5 yrs)
(born before iodide prophylaxis)

355 msec

322 msec

p < 0.05

Group 2 (mean age 7.3 yrs)
(born after iodide prophylaxis)

473 msec

442 msec

Not statistically
significant

*RT - Retention Time
Source: Vitti 1992, table 4.

The Vitti paper discusses that the slower retention times observed in children born in
mild ID As are potentially not related to cognitive impairment, but might be related to slightly
altered efficiency of the peripheral nerves to transmit a signal to and from the brain.

98

-------
10-P-0101

Santiago-Fernandez 2004 Study: The Santiago-Fernandez study evaluated the possible
relationship between UIC (i.e., a surrogate for iodide intake) and the intellectual capacity
observed in 1,221 schoolchildren in the province of Jaen in southeast Spain (Santiago-Fernandez
2004). The mean ages of the children were 10.8 ± 2.9 yrs (range 6-16 years) and were from 1st,
5th, and 8th grades. The IQ was determined by the Cattell's g factor test. IQ is statistically
lower in children with UIC <100 |ig/L than children with UICs >100 |ig/L. This observation is
summarized in the following table:

Urinary Iodide Concentration

Intelligence Quotient

Statistical Significance

> 100 |ig/L

99.03 ± 15.81

p = 0.01

< 100 |ig/L

96.40 ± 17.46

Source: Santiago-Fernandez 2004, table 2.

The IQ level of children is associated with the type of salt consumed. In the United
States, iodized salt contains about 100 parts per million (ppm) potassium iodide, which
corresponds to 77 |ig iodide per gram of salt (Dunn 1998). By contrast, both common salt and
sea salt contain negligible amounts of iodide. As a result, iodized-salt-consuming children had
the highest level of UIC of about 118 |ig/L. The common-salt- and sea-salt-consuming children
had a statistically lower level of UIC of about 99 |ig/L and 94 |ig/L, respectively. The IQ of the
iodized-salt-consuming children with the higher iodide intake was about 101, while the IQs of
the common-salt- and sea-salt-consuming children with lower iodide intake was statistically
lower at about 97 and 96, respectively. These results are summarized in the following table:

Type of Salt Consumed

Iodide Content
(|jg iodide/g of salt)

Urinary Iodide Cone.

(mq/l)

Intelligence
Quotient

Iodized [rock salt]

77 jjg iodide/g of salt

118.22 ±77.39a

100.63 ± 15.44°

Common

[noniodized rock salt]

0.7 jjg iodide/g of salt*

99.48 ±68.85b

96.64 ± 16.85d

Marine

[sea salt contains very little
iodide]

1.4 jjg iodide/g of sea salt**

94.44 ± 70.40b

95.50 ± 18.22d

Statistical significance between a & b is p = 0.0004
Statistical significance between c & d is p = 0.001

Source: Santiago-Fernandez 2004, table 5; * Aquaron 2000; ** Fisher 1980.

The IQ level of children is associated with the amount of milk consumed. Milk is known
to be a significant source of iodide in the diet. As a result, children consuming milk 3 or more
times a day have highest levels of UICs of about 119 |ig/L, while the children consuming milk 2,
1, or < 1 time(s) per day had a statistically lower level of UIC of about 99, 87, and 81 |ig/L,
respectively (see table below). The children consuming milk 3 or more times per day or 2 times
per day have statistically higher IQs of about 99 and 100, respectively, than children consuming
milk 1 or < 1 time per day, who had IQs of 94 and 94, respectively. These results are
summarized in the following table:

99

-------
10-P-0101

Frequency of milk intake

Urinary Iodide Cone.

(mq/l)

Intelligence Quotient

3 times per day

118.60 ±79.80a

98.01 ± 15.96°

2 times per day

98.68 ±68.70°

99.90 ± 16.23°

1 time per day

87.10 ±58.40°

93.64 ± 19.41°

< 1 time per day

80.82 ±48.60°

93.75 ± 16.93°

Statistical significance between a & b is p = 0.0001
Statistical significance between c & d is p = 0.0008
Source: Santiago-Fernandez 2004, table 5.

The UICs in the children of the Santiago-Fernandez study do not reflect the iodide
nutritional status of the pregnant mother and do not attempt to measure any potential brain
development issues during gestation. Furthermore, the Santiago-Fernandez study is not an
interventional study to evaluate whether the observed effects on IQ are reversible if the UIC is
increased in the low IQ children.

The major contribution of the Santiago-Fernandez study was to document the sensitivity
of the child's thyroid to iodide intake and the resulting cognitive performance of the brain.
Another major significant contribution of the Santiago-Fernandez study was to document that
iodized salt and milk are significant sources of iodide. Milk is a significant source of iodide in
children which is confirmed by the FDA TDS (Murray 2008). Likewise, while iodized salt
contains 77 |ig iodide/g of salt (Dunn 1998), noniodized salt and sea salt contain only about 0.7
|ig iodide/g of noniodized salt (Aquaron 2000) and about 1.4 |ig iodide/g of sea salt (Fisher
1980). Therefore, noniodized salt and sea salt do not provide a significant source of iodide in the
diet, which results in a corresponding decrease in cognitive performance (Santiago-Fernandez
2004, table 5).

100

-------
10-P-0101

Adverse Effects Observed in Moderate Iodide Deficient Populations

Bleichrodt 1989 Study: A study evaluated the mental and psychomotor development
effects on children of moderate iodide deficiency (Bleichrodt 1989). The study evaluated 162
children aged 2 months to 12 years in an IDA in Spain having a mean urinary iodine of 45 ± 41
jag/L. The control group consisted of 193 children aged 2 months to 12 years old in a non-ID A
having a mean urinary iodine of 77 ± 46 jig/L. Both T4 and iodide urinary excretion levels were
markedly lower in the iodide-deficient group. The mental development scores are significantly
lower (p < 0.01) across all three age groups (i.e., 0-2.5 years, 2.5-6 years, and 6-12 years). The
mental development scores are summarized in the following table:

Age Group

Mental Development Scores

Control Children

Moderately Iodide Deficient Children

0-2.5 yrs

108.1 ± 16.0

95.9 ± 16.3

2.5-6 yrs

105.1 ±11.6

88.1 + 15.0

6-12 yrs

101.6 ± 13.1

88.0 ± 12.7

Source; Bleichrodt 1989, table 3.

The moderately iodide-deficient group was subsequently treated twice with 2 ml of
Lipiodol orally to correct its iodide deficiency and was retested 32 months later. However, the
average mental developmental scores and distribution between the moderately iodide-deficient
group and the group after having been treated are nearly similar (i.e., mean scores of 90.2 and
88 .0 respectively). This observation is consistent with the belief that the mental deficit was
irreversible and occurred during fetal and neonatal development and is not a temporary side
effect of iodide deficiency. The distribution of mental development scores for the three age
groups (0-2.5 years, 2.5-6 years, and 6-12 years) are provided below:

Source: Bleichrodt 1989, figure 2.

101

-------
10-P-0101

Figure 3 Mental Development
age 2§-6

nwm Control

Iodine Deficient
Normal Distribution

so 60 70 80 90 100 110 120 130 140 ISO
Source: Bleichrodt 1989, figure 3.

20

OP

W '5

>
u
z

111
J
0

111

K 5

10 -

JLJL

Figure 4 Mental Development
age 6-12

KWM Control
T rea ted

Iodine Deficient
Normal Distribution

SO

60

70

80

90

100

110

120 130 140 ISO

•Source: Bleichrodt 1989, figure 4.

The Bleichrodt study also evaluated the psychomotor development of the 6- to 12-year-
olds. The study found the coordination of body movements to be similar between the two
groups, but the iodide-deficient children scored significantly lower (p <0.01) in manual dexterity
and speed of reaction. The moderately iodide-deficient group was subsequently treated twice
with 2 ml of Lipiodol orally to correct their iodide deficiency and retested 32 months later.
However, the average scores and distribution of scores between the moderately iodide-deficient
group and the group after having been treated are nearly identical. This observation is consistent
belief that the mental deficit was irreversible and occurred during fetal and neonatal development
and is not a temporary side effect of iodide deficiency. The distribution of scores for manual
dexterity, speed of reaction, and coordination of movements are provided below:

102

-------
10-P-0101

Figure 7 Manual Dexterity
age 6-12

SO 60 70 SO 90

Source: Bleichrodt 1989, figure 7.

100

110

120

130

140

150

Figure 8 Speed of reaction
age 6-12



[\\\vi Control
T reated

Iodine Deficient
Normal Distribution

50 60 70 80 90 100 110

Bleichrodt 1989, figure 8.

120

130 140 ISO

Figure 9 Coordination of movements

age 6-12

fcSNNJ Control
¦H Iodine Deficient
	 Normal Distribution

50 60 70 80 90

Source: Bleichrodt 1989, figure 9.

fQvwt Control
T reated
¦¦ Iodine Deficient
	 Normal Distribution

103

-------
10-P-0101

Vermiglio 2004 Study: The Vermiglio 2004 prospective study evaluated the
neuropsychological development of 16 children born to 16 healthy mothers living in a
moderately IDA (Area A) and compared them against 11 control children born to 11 age-
matched women in a marginally sufficient iodide area (Area B). The urinary iodide excretion
level in Area A during pregnancy was 48.1 ± 38.2 |ig/day. The urinary iodide excretion level in
Area B during pregnancy was 95.2 ± 55.8 |ig/day. All children from both Areas A and B were
euthyroid at neonatal screening, at the first examination (18-36 months), and at the second
examination (8-10 years).

In 2001-2002, at the age of 8-10 years, the children were evaluated for ADHD and their
intelligence was tested. In the moderately IDA A, 11 of 16 (68.7%) children had ADHD. Of the
11 ADHD children, 5 had combined-type ADHD subtype, 5 had the predominantly hyperactive-
impulsive-type ADHD subtype, and 1 child had predominantly inattentive-type ADHD subtype.
None of the control children from Area B had ADHD.

The Wechsler Intelligence Scale for Children (WISC)-III full-scale IQ scores, and
subscale scores for ADHD-positive children from Area A, the ADHD-negative children from
Area A, and the control children in Area B are summarized below:

TABLE S, Neuropsychological test scores to ADHD+ve aid. ADHD-ire childraa tram. iodiie-ddMeot. area (area At as compared with,
iodliie-euHieient matched children (area B)

test

Ar.Mi A ,vr>nt>--..

¦ hll'liktli ui !"<*

A API IP

'Jlii'lu il Mi -

Ar> n
¦ litl-ii-'ji in n*

PvaJua

toMligBn.ee WISC-III full-scale IQ score (t-IQj

8iS.1 ' J ii '1

m.u ±

2/i

Jin

± in

¦ 0 H00 f
il .M.V

WISC-III fteedam-from-distractibUity score

8.8 ± 2.5

K.r» ±

2-5

If

± ' 3

•• J i.UI

•¦-U.U51

WISC-III verbal Kj| scare

88.6 ± 11.4

m ±

2.rt

tlx

± 1J i

<0.0016











<0.QS*

WISC-III verbal section













General information

Mm £ J I,

«v$ ±

^ n



± 1! ^

NS

General comprehension

~ ± J 4

J'l±

2/f

Jff.V"

± :l.l

»-".0.05

Arithmetic



l'U> ±

o .1

ln.7

± L'.l

•-.0.0056













 5

<0 .OS*

Coding

a.7 ± :i,4

hli±

Hi

II."

±

MS

Mazes

H,( ± H

*. ±



11"

± l!,c>

<0.06*













<0.01*

Nfi N'li s-iRnififflnL

"RiMill-t rt'f'T 1" IR11 Al'UD+ve children due to the fact that in 1/11,

extreme iii:tlt"ntii vi fnn t ml
r jUiIIT'-'-Tn ; s i'.fliitmlF.

J AIUiri+v..;'s API II I-v.

Source: Vermiglio 2004, table 2.

The intelligence WISC-III full scale IQ scores of 88.0 ± 6.9 for the ADHD-positive
children from Area A were statistically lower than the corresponding IQ scores of 110 ± 10
control children from Area B. Of the 15 subscale scores, the following 9 subscale scores of the
ADHD-positive children from Area A were found to be statistically lower than the

104

-------
10-P-0101

corresponding scores in the control children from Area B: WISC-III freedom-from-
distractibility score, WISC-III verbal IQ score, WISC-III verbal section tests (general
comprehension, arithmetic, similarities, vocabulary), WISC-III performance IQ score, and
WISC-III performance section tests (picture completion and mazes).

The free T4 level and TSH level of the pregnant women at weeks 8, 13, and 20 for the
three groups (ADHD positive from Area A, ADHD negative from Area A, and controls from
Area B) are provided below:

i.H

1.5
||J
rftfl J

Oii J

(LI,

Sr.li week

1 .

¦« 5 6 ;
I Ml
-------
10-P-0101

Vermiglio stated that ADHD is a developmental disorder involving difficulties with
sustained attention, distractibility, poor impulse control, and hyperactivity or inability to regulate
activity level to situational demands (Vermiglio 2004). The disorder is believed to arise early in
childhood (3-7 years) and is considered organic in pathology. The current view is that
dysfunctions of both the frontal and prefrontal lobes, involved in impulsivity and motor activity
control, and of cortical and subcortical activity areas, involved in inhibition of irrelevant
responses and executive function, are responsible for the disorder.

ADHD is a common behavioral disorder among children. The estimated occurrence of
ADHD at 19 years of age is 7.5% (6.5%-8.4% at a 95% confidence interval (CI)) (Barbaresi
2004). A high prevalence ADHD {10%) is reported in children with generalized resistance to
thyroid hormone (GRTH). GRTH is a disease caused by mutations in the thyroid receptor-P
gene, which reduces the response of tissues to thyroid hormone (i.e., the developing fetal nerve
cells do not get enough signal from the thyroid hormone to develop properly). Since the ADHD
prevalence in the Vermiglio study was 69% in the moderately iodide-deficient and 87.5% in
hypothyroxinemic, it suggests that a decreased signal through the thyroid receptor-P (i.e., by a
mutated thyroid receptor-P gene, decreased maternal supply of T4 to the fetus during the first 20
weeks caused by iodide deficiency, or a combination of both) is associated with ADHD. The
Vermiglio study suggests that the rates of ADHD could be significantly reduced by preventing
iodide deficiency during pregnancy. Furthermore, the Vermiglio study suggests the possibility
that providing a normal thyroid hormone level during gestation may allow a child with GRTH to
have normal brain development by not challenging the mutated thyroid receptor-P gene.

The Vermiglio 2004 study is one of four studies reporting that maternal hypothyroxinemia
during early pregnancy results in neurodevelopmental deficits in children (Kooistra 2006).
Hypothyroxinemia is a common condition in pregnant women characterized by low maternal fT4
levels with normal TSH levels (Kooistra 2006). Hypothyroxinemia is regarded to be without
consequences for the mother and fetus (Kooistra 2006). Hypothyroxinemia reflects a condition in
which the pregnant mother has difficulty meeting her own T4 needs and is unable to meet the fetal
demand for T4 for proper brain development (Kooistra 2006). Fetal hypothyroxinemia is induced
by iodide deficiency because the immature fetal thyroid is unable to increase its avidity for iodide
despite up-regulation of NIS expression in the fetal thyroid and placenta during iodide deficiency
(Delange 2005a). The increased iodide clearance rate will further decrease the iodine stores of the
fetal thyroid and decrease fetal T4 synthesis (Delange 2005a). The transfer of maternal T3 does not
protect the fetal brain. The pregnant woman transfers maternal T4 until birth, at which time
maternal T4 represents 20% to 50% of the cord serum T4. Therefore, the fetal thyroid is required
to provide the major T4 during the last half of pregnancy when the fetal thyroid is unable to adapt
to a decreased supply of iodide caused by its mother's iodide deficiency. In rats, the fetus can
compensate for hypothyroxinemia by increasing the fetal brain deiodinase type 2 activity, which
helps protects the fetal brain from T3 deficiency (Delange 2005a).

106

-------
10-P-0101

Fenzi 1990 Study: A neuropsychological assessment in school-aged children in an area
of moderate iodide deficiency found minor impairment of perception, motor, and attentive
functions (Fenzi 1990). The epidemiological study evaluated 384 schoolchildren ages 6-14 years
in eastern Tuscany having a mean urinary iodine excretion of 39 ± 38 |ig/g creatinine. The
control group consisted of 352 sex- and age-matched schoolchildren from an ISA with a mean
urinary iodine excretion of 88 ± 15 |ig/g creatinine. The tT4, tT3, and TSH levels were within the
normal range for both groups. The study failed to show major differences in the global
neuropsychological performance and cognitive performance between the two groups. However,
significant differences were observed in subtest for information, verbal, and coding tests that
were administered to the third graders. The lower scores in the information and verbal subtests
resulted in a lower verbal IQ for the third grader in the moderately IDA. Furthermore, the lower
score on the coding test, which is less correlated with global cognitive abilities, suggests minor
impairment in perception, motor, and attentive functions.

Vermiglio 1990 Study: In the Vermiglio 1990 study, the visual perceptual integrative
motor ability was investigated in 719 6- to 12-year-olds living in two moderately ID As in Sicily,
Italy. A modified Bender-Gestalt test was administered to all 6- to 12-year-old children. The
test consists of copying as accurately as possible a number of geometrical figures, explores
visual perceptive and neuromotor manual ability, and is affected by specific segments of
intellectual function (i.e., memory, spatial concepts, and ability to organize and represent). The
children in the moderately deficient areas scored statistically worse on the Bender-Gestalt test
than the children in the iodide-sufficient control area. The results are summarized in the
following table:

Performance on
Bender-Gestalt Test

Moderately Iodide-
Deficient Areas A + B
(number of children)

Iodide-Sufficient
Control Area C
(number of children)

Statistical
Significance
(X method)

Defective

99 (13.8%)

13 (3.5%)

P < 0.000001

Borderline

124 (17.2%)

14 (3.8%)

P < 0.000001

Nondefective

496 (69.0%)

343 (92.7%)

P < 0.0000005

Source: Vermiglio 1990, table 3.

A Terman-Merrill test was administered to 96 of the 99 defective children and 62 of the
124 borderline children found on the Bender-Gestalt test in the moderately ID As A + B. The
Terman-Merrill test evaluates general intellectual aptitude. Intellectual aptitude deficiencies
appeared in the following segments listed in decreasing percentage order of involvement:
vocabulary (40%), logical and critical capacities (32%), reading abilities (30%), calculating
abilities (28%), handwriting (25%), memory (20%), and comprehension (15%). Of the 96
defective children tested, 91 (i.e., 94.8%) had IQ scores less then 90% on the Terman-Merrill
test. Of the 62 borderline children tested, 35 (i.e., 56.4%) had IQ scores less then 90% on the
Terman-Merrill test. The results from the Terman-Merrill test are summarized below:

Performance on the
Bender-Gastalt Test

Intelligence Quotient Score on the Terman-Merrill Test

<90

90-95

96-100

Defective (n = 96)

91 (94.8%)

Borderline (n = 62)

35 (56.4%)

Nondefective (n = 12)

Source: Vermiglio 1990, table 4.

107

-------
10-P-0101

Adverse Health Effects Observed in Children of Hypothyroid Pregnant Mothers

A 1999 maternal thyroid deficiency study reported on the subsequent neuropsychological
development of the children born to hypothyroid mothers (Haddow 1999). The study identified
47 pregnant women with subclinical hypothyroidism having elevated TSH (i.e., > 6 ul/ml) and
normal T4 levels. The study also identified another 15 pregnant women with overt
hypothyroidism having both elevated TSH and low T4 levels. A survey of pregnant women in
the United States identified that 2.5% of pregnant U.S. women have hypothyroidism (Utiger
1999). Likewise, North American and European evaluations identify that up to 0.5% of pregnant
women (1 in 200) may have overt hypothyroidism and up to 2.5% of women (1 in 40) have
undetected subclinical hypothyroidism before pregnancy (Morreale de Escobar 2004, p U33).
Since hypothyroidism is difficult to diagnose, the median time to a hypothyroid diagnosis in
women is 5 years (Haddow 1999).

Of the 48 pregnant women that did not receive treatment for their thyroid deficiency, the
full-scale IQ of their children averaged 7 points lower (p = 0.005) than the 124 matched control
children. The children from the 48 untreated pregnant women had statistically poorer
performance on the following nine neuropsychological tests:

WISC-III full-scale IQ score (Intelligence Test)

Number of children with IQ score < 85 (Intelligence Test)

WISC-III freedom-from-distractibility score (Attention Test)

Continuous Performance Test score > 8% (Attention Test)

Word Discrimination (Language Test)

WISC-III verbal IQ score (Language Test)

PIAT-R* reading-recognition score (School Performance Test)

WISC-III performance IQ score (Visual-motor performance)

Pegboard-test score (Nondominant hand) (Visual-motor performance)

*(PIAT-R) stands for Peabody Individual Achievement Score - Revised

The performance results on all 15 neuropsychological tests were observed to be lower for
the untreated children than the control children. However, the statistical significant of this trend
was not evaluated.

The children of the 14 treated pregnant women had performance results on 13 of the 15
neuropsychological tests that were comparable to the control children. However, the study did
not conduct a statistical comparison between the children of the treated pregnant women with
hypothyroidism and the control children (i.e., to determine how effective the treatment was).
Even after treatment, the results of the children of the 14 treated pregnant women were worse
than both the control and untreated children on the following two tests: Continuous Performance
Test (Attention Test) and School difficulties and learning problems (School Performance Test).
The performance results for all 15 neuropsychological tests for the children of both the treated
and untreated pregnant women are provided below:

108

-------
10-P-0101

Table 4. \"kvr->i--yi.h*i.ix.irAi. Ti/.i Si • i-.i.-. imon., ihk < hii.hw:': or Womiii; wrin
HiTi:air.T.oincM iiitjn'i, Ppjics'ANi i as i ViMrArxi* '.mu inn (. musriN nr M.vrt iir.r
< riVTK'M. 'VojiK-i, SlTAlirtF.D A<"< nRDJNV, Whcthkh. nit- Hwim'TrnntsM

Was Bhio Tkboid.*

CMILDREN OF

Treated

WOMEN WITH

H¥f>OTH¥BOB!SM P
TtST (N" 14) VALUE t

Inicllit»eiKe

'.VIM -III lulUc-ile H.j -core 111 0.20

'.VIM III lull-^cjle k,> M.ore 0 0.90

- ¦ X 5, I.

Anemi-ui

^VIm -III lived.im-tr-mi 103 0.80

dKirjuibiliry m.. i iv*
i in

V" i d ivcri mi rwi ii >n 11.4 1'« 411

'•VI • i HI verbal Iij v,-i're 111 ".Sii

schn.il pL-rh.rnuiKC

rirtl" R raJinft-rev:..ynition score 101 >«.sn

riAL R rejdiny o ¦mprehc-nsi.Qn Its ">.4U

%o ire

Vlii.i'1 dillkullics and Ic.trnlng: 29 0.0S

pn .hleni1. i"«'1

Kepe.ued j yrjde >if 7 0.S0

VKiuJ-nvm -r perh 'rm.uice

Scire mi Do'.clfpnieni.irieM 182 0.80

m| V Nil j I-M. ii>ir Iniear.iiii.n
"•VIM -III peril>rm,nice IiJ wi.a- 189 0.30

I'eyb. unl-iesr MMit
l"ii mi in .ii ii li.nn.il "4h

Ni .iidi 'ill ill am kind! N7 i' "ii

CMILDREN OF
UNTBEATEB

Women with

HfPOTHmOIDISBt
(N-«8)

108
If

VALUES (N-1241

9?
33

10.0
10.3

1:01

21
8
94
99

(I i'

H "2

ll.i il if,

n.ii^
U .1IM

II .1 iu

0.3
0.1
0.01

II nr.
Ii 14

107

102
If

Hl.2

i: 4

100

101
II

105

§3
S#

~Wis< -in detunes VVechder Intelligence Scale for Children, third edition, and I'IAi-R Peabody
InJr.idu.il Achievement Tom, revised

fl'he P % aliic. -ire t. ¦ r i lie >.i.mp.irW<>n of the children of the treated women with the children of
tile IIIII realed wmic-li

}•; ine uihiii received ire.uiiieiii K-1>ire. Inn nm during;, the prenn.mo. under Miid>.
§Hie 1" -.jluec .nv I. -r die ci .inp.iriM 'ii "I (lie children nl I he umrejted H-uneii iih the children of
ill*, ci'iimol women.

|A higher score or percental imitates more problems.

Source: Haddow 1999, p 554, table 4.

109

-------
10-P-0101

8.2 Dose-Response Assessment

A dose-response assessment is defined as "a determination of the relationship between
the magnitude of an administered, applied, or internal dose and a specific biological response.
Response can be expressed as measured or observed incidence, or change in level of response,
percent response in groups of subjects (or population), or the probability of occurrence or change
in level of response within a population" (IRIS 2006, EPA 2002).

Ideally, one would like to know the internal dose of free iodide concentration in the blood
serum in contact interacting with the NIS to establish the dose. However, determining an
internal dose of free iodide in the blood serum is problematic. Healthy levels of inorganic iodide
concentrations in serum are very minute and are about 1 pg/dL (Demers 2003, p 77). Due to the
difficulty in measuring plasma inorganic iodide (PII) concentration, the measurement of PII is
limited to research studies in pregnancy. Since about 90% of ingested iodide is excreted in the
urine, the median urinary concentration is the most practical measurement of iodide intake in a
population (Dunn 1998). Therefore, in most epidemiological studies, the UIC is used as a
surrogate for iodide intake.

The observed biological responses for iodide is inversely proportional to dose (i.e., the
more you have in your diet, the less adverse effects that are observed). The dose-response curve
for iodide is summarized in the following table:

Mild

Moderate

Severe

Iodide

Observed Biological Response

Sufficient

Deficiency

Median UIC (jjg/L) in school-aged
children and adults

(UIC is a surrogate measurement for
iodide intake (i.e., dose))

100 to 200

99 to 50

49 to 20

<20

Prevalence of goiter (%)
in school-aged children

(Goiter Grades > 0)

< 5.0

5.0-19.9

20.0-29.9

>30

Frequency of thyroid volume (%) in
school-aged children

(>97th percentile by ultrasound)

< 5.0

5.0-19.9

20.0-29.9

> 30

Frequency of elevated serum TSH in
newborns (%)

(> 5 mU/L whole blood)

< 3.0

3.0-19.9

20.0-39.9

>40

Source: Delange 1998.

The frequency of the following biological responses are observed to increase with the the
lack of iodide (i.e., decreasing iodide dose): goiter in school-aged children, frequency of thyroid
volume >97th in school-aged children, and elevated TSH levels in newborns. This pattern of
response to the lack of iodide shows that a segment of the population is more sensitive (i.e.,
thyroid-sensitive population) than the rest of the population. In IS As (i.e., little to no thyroid
stress), the baseline frequency is that up to 5% of school-aged children have goiter and up to 3%
of newborns have elevated TSH. By comparison, in the minimal thyroid stress population
induced by mild iodide deficiency, the frequency goiter in school-aged children increases to up
to 19.9% and the frequency of elevated TSH in newborns increases to up to 19.9%. In other

110

-------
10-P-0101

words, at least 80% of the newborns are unaffected by mild iodide deficiency, while up to 20%
are adversely impacted (i.e., a heterogenous response is observesed in the population). Likewise,
in the considerable thyroid stress population induced by moderate iodide deficiency, the
frequency goiter in school-aged children increases to up to 29.9% and the frequency of elevated
TSH in newborns increases to up to 39.9%. In other words, at least 60% of the newborns are
unaffected by moderate iodide deficiency, while up to 40% are adversely impacted (i.e., a
heterogenous response is observesed in the population). Furthermore, even in severe IDA, not
everyone gets goiter nor does everyone have cretins. For example, in seven severe iodide-
deficient cities, the percentage of elevated TSH in newborns did not reach 100% but ranged from
47%) in Osh, Kyrgyzstan, to 80% in Lahore, Paskistan, in relation to the severity of the iodide
deficiency (Sullivan 1996). In summary, the biological response to iodide deficiency is not
uniform across the human population and has to be taken into account in a cumulative risk
assessment.

This observation (i.e., not all individuals develop goiters or elevated TSH levels in ID As)
indicates a heterogeneous level of adaptability to iodide deficiency within the human population.
A genetic modeling study supports this observation. In euthyroid subjects, quantitative genetic
modeling calculated that genetic factors account for 71% (61-78%, CI 95%) of the individual
differences in thyroid volume, where as the estimate for unique environmental effects account
for only 29% (22-39%, CI 95%) (Hansen 2004). Therefore, the iodide-sensitive thyroid
population should be added to the list of sensitive groups for this public health issue. The iodide-
sensitive thyroid population is not a trivial concern (e.g., such as an unexplained rare
idiosyncratic reaction), because elevated TSH in neonates occurs up to 17% in the mild iodide-
deficient population (i.e., 19.9% - 3% background) and up to 37% in the moderately iodide-
deficient population (i.e., 39.9% - 3% background). Therefore, neonates have elevated serum
TSH more frequently than adults for the same level of iodide deficiency (Delange 1998). Thus,
neonates appear hypersensitive to iodide deficiency. Thus, the most sensitive population for
cumulative risk assessment purposes is the iodide-sensitive fetus and neonate during gestation
and lactation whose mothers are iodide deficient.

Since the early 1970s, newborns are screened for permanent congenital hypothyroidism
(CH) (incidence of CH is about 1 in 4,000 births) by measuring the serum TSH (Delange 1998).
However, CH screening also detects transient primary hypothyroidism (incidence can be as high
as 1 in 10 neonates). Neonatal TSH screening is one of the indicators recommended by the
World Health Organization (WHO), the United Nations International Children's Emergency
Fund (UNICEF), and the International Council for the Control of Iodine Deficiency Disorders
(ICCIDD) for assessing iodide deficiency disorders and their control (Delange 1998). Neonatal
TSH is the single indicator that focuses on potential brain damage, which is the major impact of
iodide deficiency (Delange 1998). Neonates with elevated TSH levels are at a high risk of
developing subclinical hypothyroidism in infancy and early childhood (Delange 1998).

The Calaciura 2002 study evaluated 56 children with transient neonatal subclinical
hypothyroidism at the neonatal TSH screening (Calaciura 2002). All 56 children had a TSH >
20 mU/L at the initial neonatal screening, but during the recall/confirmation examination at 22.2
± 6.2 days, 33 children (Group I) had a normal fT4 level and fully normal TSH level (i.e., < 5.0
mU/L) and the remaining 23 children (Group II) had a normal fT4 level with high normal/slightly

111

-------
10-P-0101

elevated TSH level (i.e., 5-17 mU/L) (Calaciura 2002). Upon follow-up examination in early
childhood, 70% of Group I children and 36% of Group II children had mild thyroid dysfunction.
Calaciura determined that all infants with elevated TSH at neonatal screening are at risk for
developing subclinical hypothyroidism in early childhood (odds ratio 44.6%; 95% CI 38.8-
50.4%). Furthermore, for the first time, several different thyroid abnormalities were identified in
neonates with elevated TSH levels at birth. Twenty-four of the 56 children had one or more of
the following thyroid abnormalities: genetic, immunological, and morphological. This
observation documents a partially dysfunctional or underperforming thyroid in a subset of the
human population that makes them more sensitive to NIS stressors (e.g., lack of iodide). The
adaptability of the thyroid to respond to stress is not uniform in the human population and,
therefore, should be thought of as being heterogeneous in the human population. The cumulative
risk assessment has to incorporate this difference in the thyroid's adaptability within the human
population.

Cognitive deficits in the children of mildly iodide-deficient pregnant women have not
been documented (Delange 2001). Unfortunately, these epidemiological studies only compare
children from a mild IDA against control children in an ISA. However, the human thyroid's
ability to compensate for mild iodide deficiency is not uniform across the population. Up to 17%
of humans who may have partially dysfunctional thyroids with limited adaptability have
difficulty compensating for mild iodide deficiency. Since neonatal TSH level is the single
indicator that focuses on potential brain damage, the partially dysfunctional thyroid subgroup can
be tentatively identified by grouping the neonates with an elevated TSH level. The current
epidemiological studies evaluating the cognitive performance of children of mildly iodide-
deficient pregnant mothers do not segregate the neonates with elevated TSH who are at greater
risk of developing mental deficits for evaluation. The potential for lower cognitive performance
in these elevated TSH neonates, averaging 8.5% (i.e., 17% ^ 2) of the population in mild IDAs,
is potentially lost in the average cognitive score of the school-aged children (i.e., the cognitive
scores from the 91.5% of the normal TSH neonates statistically overwhelms the scores of the
8.5%) neonates with elevated TSH). Therefore, in order to confirm that a mild iodide deficiency
in pregnant women does not cause any cognitive deficits in partially dysfunction thyroid
fetuses/neonates (i.e., a potential sensitive group), the cognitive scores of children who had an
elevated TSH level at birth need to be compared to the cognitive scores of appropriate controls
(i.e., cognitive scores of children from ISAs or cognitive scores of children from mildly IDAs
with normal TSH levels at birth). Therefore, in short, current epidemiological studies have not
ruled out the possibility that cognitive deficits could be observed in fetuses/neonates with
partially dyfunctional thyroids.

The adverse effect of concern for this public health issue is the potential for mental
deficits in fetuses and neonates. The dose-response relationship for mental deficits from the lack
of iodide is characterized by the epidemiological studies identified in the hazard characterization
assessment. These studies are primarily designed to identify and detect mental deficits. They
were not specifically designed to establish a well-documented dose-response curve showing an
increased occurrence of mental defects with decreasing dose of iodide. However, these studies
show, as required by the definition of a dose-response assessment, a "change in level of
response" with decreasing dose of iodide (i.e., a decreasing maternal TIU). The severity of the
mental deficits clearly increases as the lack of iodide increases (i.e., as the maternal TIU

112

-------
10-P-0101

becomes lower). The following table summarizes the observed mental deficits in the children of
mothers living in ID As:

Iodide Exposure Levels
in Pregnant Women

Observed Mental Deficits in the
Children of Mothers in Iodide-Deficient Areas

Iodide sufficient

No increased occurrence of mental or physical effects

Mild iodide deficiency

Delayed reaction time

Potential cognitive effects in partially dysfunctional thyroid

individuals (see discussion above)

Increased frequency of mild thyroid dysfunction in childhood

Moderate iodide deficiency

Subtle, minor, or overt neurological, psychological, and intellectual
deficits

Severe iodide deficiency

Cretinism 5-15% prevalence

(severe, permanent mental and physical defects)

Source: OIG Analysis.

The NAS Committee concluded the first adverse effect in its mode-of-action model is
hypothyroidism and any effects downstream (e.g., neurodevelopmental deficits in infants) of
hypothyroidism clearly would be adverse (NAS 2005, p 166). The NAS Committee stated that
"developmental deficits" are clinical manifestations of hypothyroidism (NAS 2005, p 29). The
NAS Committee identifies hypothyroidism as a deficiency in thyroid hormone production (NAS
2005, p 35). Overt (i.e., primary) hypothyroidism is indicated by high serum TSH
concentrations and low T4 (Ladenson 2005). Mild hypothyroidism is indicated by elevated TSH
and normal T4 (Ladenson 2005). The NAS Committee stated that:

Iodide can be obtained only by ingestion of food or water that contains it.

Therefore, iodide deficiency and reduction in thyroid hormone production
can occur if iodide intake is very low. Because the body maintains the
serum concentrations of thyroid hormones with narrow limits through
feedback control mechanisms, there is remarkable compensation for
iodide deficiency. Generally, thyroid production is normal even when
iodide intake is quite low. Hypothyroidism occurs only if daily iodide
intake is below about 10 to 20 |ig (about one-fifth to one-tenth of the
average intake in the United States) [10-20 |ig of iodide intake equates to
severe iodide deficiency]. However, in pregnant women, iodide
deficiency of that severity can result in major neurodevelopmental deficits
and goiter in their offspring. Lesser degrees of iodide deficiency may also
cause important neurodevelopmental deficits in infants and children (NAS
2005, p 6).

The NAS Committee also states:

Furthermore, among people 15-44 years old, including pregnant women,
there are no differences in serum TSH and T4 concentrations between
those with urinary iodide values less than 50 |ig/L and those with higher
values ..." (NAS 2005, p 46-47).

113

-------
10-P-0101

These statements represent the NAS Committee's position that neurodevelopmental
deficits in infants would not be expected to occur until severe iodide deficiency during
pregnancy induces overt hypothyroidism in the mother, resulting in an elevated TSH
concentration accompanied by a low T4.

Unfortunately, the NAS Committee's proposed mode-of-action model for perchlorate is
not completely correct. The NAS Committee identifies that adults with moderate iodide
deficiency have normal TSH and T4 levels (NAS 2005, p 48). The NAS adverse effect
continuum presumes that pregnant women have to become hypothyroid before
neurodevelopmental deficits are observed. However, this statement is not true. For example,
due to the added stress of pregnancy, moderately iodide-deficient pregnant women are observed
to develop hypothyroxinemia (i.e., normal TSH and low fT4). The hypothyroxinemia in
moderately iodide-deficient pregnant women resulted in mental deficits of their children
(Vermiglio 2004, figure 2). Furthermore, under mild iodide deficiency during pregnancy, the
maternal thyroid function at delivery is characterized by "relative hypothyroxinemia," which is
an increased T3/T4 ratio (i.e., indicating a preference for T3 secretion) and slightly increased TSH
levels (i.e., 97% of the pregnant women were still in the normal range) (Glinoer 1992).

Likewise, the less-stressed state of mildly iodide-deficient pregnant women are be observed to be
euthyroid (Glinoer 1995, figure 4), but they are documented to have children with permanent
delayed reaction times (Lombardi 1995; Vitti 1992) and are documented to have up to a 17%
increased occurrence of newborns with elevated TSH levels (i.e., subclinical hypothyroidism)
over background levels observed in iodide-sufficient populations (Delange 1998). Therefore,
pregnant women do not have to develop overt hypothyroidism, as proposed by the NAS
Committee, to increase the risk of their offspring developing mental deficits.

Although the maternal hypothalamic-pituitary-thyroid axis has a large ability to adapt to
the stress from iodide deficiency without showing obvious changes in TSH and T4 levels (Soldin
2005, p 184, figures 2 and 3), the fetal thyroid is immature and less able than the mother's
thyroid to adapt to the stress induced by mild or moderate iodide deficiency. Fetuses and
neonates do not have a thyroidal store of hormones and have a proportionally higher demand for
thyroid hormones because they use them for both development and homeostasis, while adults use
thyroid hormones primarily for homeostasis only. During the third trimester, the fetal thyroid
provides the majority of T4 (i.e., maternal T4 represents 20% to 50% of the cord serum T4)
(Delange 2005a). Under iodide-deficient conditions, the immature fetal thyroid is unable to
increase its avidity for iodide despite up-regulation of NIS expression in the fetal thyroid and
placenta during iodide deficiency (Delange 2005a). The increased iodide clearance rate will
further decrease the iodine stores of the fetal thyroid and decrease fetal T4 synthesis (Delange
2005a). Fetuses and neonates appear to be more sensitive to iodide deficiency than pregnant
women (Delange 1998). For mild to moderate iodide deficiency, maternal measurements of
TSH, fT4, and T4 are not appropriate markers to identify thyroid stress from iodide deficiency
(Soldin 2005, p 184). Thyroid function tests are not a measure of iodide sufficiency (Soldin
2005, p 184). Therefore, maternal urinary iodide is the only marker known that can serve as an
indirect measure of the amount of fetal thyroid stress induced by maternal mild to moderate
iodide deficiency (Soldin 2005, p 184).

114

-------
10-P-0101

During mild and moderate iodide deficiency, the fetus has some ability to adapt to a decreased T4
hormone supply by increasing the fetal brain type 2 deiodinase (D2) activity, which protects the
fetal brain from T3 deficiency (Delange 2005a, p 737). The compensation mechanism is
observed in tracer studies of neonatal rat brains (Bianco 2005, p 123). The increased production
of T3 from T4 by the increased activity of fetal rat brain type 2 deiodinase and the prolonged
residence time of T4 by the increased activity of fetal rat brain type 3 deiodinase (D3) mitigates
the effects from severe iodide deficiency and mild to moderate hypothyroidism (Bianco 2005,
p 123). The activation of this fetal rat brain D2-D3 deiodinase compensatory mechanism clearly
indicates that the fetal rat brain is not getting an adequate supply of T4 during severe iodide
deficiency and mild to moderate hypothyroidism.

In summary, although the pregnant woman is often euthyroid during iodide deficiency,
the fetal thyroid is less able to adapt to this iodide deficiency, resulting in the increased risk for
fetal brain damage even though the mother's thyroid hormone level remains normal. Thyroid
function tests are not a measure of iodide sufficiency (Soldin 2005, p 184). Therefore, maternal
urinary iodide is the only marker known that can serve as an indirect measure of the amount of
fetal thyroid stress induced by maternal mild to moderate iodide deficiency (Soldin 2005, p 184).
The lowest UIC level in pregnant women in which an adverse effect (e.g., delayed reaction time)
is observed in the fetus is about 100 |ig/L, which corresponds to a mild iodide deficiency.

115

-------
10-P-0101

8.3 Exposure Assessment

An exposure assessment is defined as "an identification and evaluation of the human
population exposed to a toxic agent [i.e., stressor in cumulative risk assessment terms],
describing its composition and size, as well as the type, magnitude, frequency, route, and
duration of exposure" (EPA 2002).

As the background exposure levels to perchlorate, thiocyanate, and nitrate must be
determined to conduct a cumulative risk assessment, the background exposure level to the lack
of iodide (i.e., the iodide nutritional status) also needs to be known. As the dose-exposure
assessment has identified the iodide levels at which various adverse effects occur, the exposure
assessment determines the portion of the population that is exposed at these levels. Since
neurological, psychological, and intellectual deficits are known to occur in moderately iodide-
deficient populations, the exposure assessment needs to determine the size of the U.S. pregnant
women subpopulation that might be moderately iodide deficient. Furthermore, since delayed
reaction times and the possibly of cognitive deficits in partially dysfunctional thyroid individuals
occur in mildly iodide-deficient populations, the exposure assessment must determine the size of
the U.S. pregnant women subpopulation that might be mildly iodide deficient.

U.S. Iodide Nutritional Status

CDC's National Center for Health Statistics (NCHS) surveyed the iodide levels in the
U.S. population through NHANES. NHANES is designed to collect data on the health and
nutritional status of a statistically representative sample of the U.S. population. According to
WHO, iodide-sufficient populations have median UIC greater than 100 |ig/L, and no more than
20% of the population should have a UIC below 50 |ig/L (Hollowell 1998). Although the
median urinary levels in the U.S. population are adequate by the WHO definition, the median
UIC has decreased more than 50% from NHANES I (1971-1974) to NHANES III (1988-1994),
and remained at this lower value in NHANES 2000 (see table below). The decrease in iodide
consumption since 1984 is thought to be due to the reduction in the amount of iodide in milk and
the replacement of iodide salts by bromine salts as the dough conditioner in commercial bread
production (Hollowell 1998).

Population Group

Urinary Iodide Levels (Hg/L)

NHANESI
(1971-1974)

NHANES III
(1988-1991)

NHANES 2000

U.S. Population
(6-74 years of age)

320 ± 6 jjg/L

145 ± 3 |jg/L

161 ± 7 (jg/L

Source: CDC/NCHS 2007.

The NHANES 2000 median urinary iodide level of 161 |ig/L (95% CI 14.7 - 17.6) agrees
with the NHANES III median of 145 |ig/L. However, the NHANES 2000 urinary iodide data
represent only 1433 samples, which does not allow for detailed analysis of the iodide status of
subpopulations (e.g., pregnant or lactating women). Since the NHANES 2000 median agrees
with the NHANES III median, the NHANES III distribution data (i.e., the best available dataset)
are used to assess the amount of the U.S. population, and the amount of the sensitive pregnant

116

-------
10-P-0101

and lactating population, with inadequate iodide. The following table summarizes the
distribution of urinary iodide values determined in the NHANES III survey:

Population Group

Urinary Iodide
Concentration
Mean (SEM)

Percentiles for Urinary Iodide (pg/L) from NHANES III
(1988-1994)

5th

10th

25th

50th (Median)

U.S Population
(ages 6-71+)
(n = 21,298)

275 ijg/L (27)

30 jjg/L

45 jjg/L

81 jjg/L

145 jjg/L

Pregnant (n = 343)

196 jjg/L (11)

43 |jg/L*

58 jjg/L

92 jjg/L

140 jjg/L

Lactating (n = 95)

161 |jg/L (22)

25 |jg/L*

29 |jg/L*

81 |jg/L*

109 |jg/L*

SEM = Standard Error of the Mean

* These values are potentially unreliable in a statistical sense based on an insufficient sample size.

Source: NAP 2000.

The severity of iodide deficiency can be determined by the amount of iodide in the urine.
The following table identifies the urinary iodide levels for each of the three categories of iodide
deficiency:

Parameter

Classification of Iodine Deficiency

Mild

Moderate

Severe

Median Urinary Iodide Levels

99 to 50 |jg/L

49 to 20 |jg/L

< 20 jjg/L

Source: Delange 2005, p 265.

The NHANES III survey documents that 8.1% of males and 15.1% of females are
moderately iodide deficient (i.e., UICs < 50 |ig/L) (Hollowell 1998). The NHANES III survey
identifies that 14.9 % ± 1.2 of the U.S. women of childbearing age are moderately iodide
deficient and that 6.9 % ± 1.9 of pregnant U.S. women are moderately iodide deficient
(Hollowell 1998). The number of lactating U.S. women with moderate iodide deficiency cannot
be precisely determined due to the small sample size in the NHANES III for this category (see *
footnote above), but appears to be about 16%. Furthermore, the NHANES III survey indicates
that about an additional 22% (i.e., corresponding to about the 29th percentile) of U.S. pregnant
women are mildly iodide deficient. The NHANES survey also indicates that about an additional
26% (i.e., corresponds to about the 42nd percentile) of the U.S. lactating women are mildly
iodide deficient. The demands of pregnancy on the thyroid cause the UIC to steady decrease
from the first to the third trimester (Glinoer 2005). The burdens of lactation continue the trend.
Therefore, to have a greater percentage of lactating women than pregnant women both
moderately and mildly iodide deficient is consistent with the biology.

The NHANES III survey also documents that the proportion of the U.S. population that
was moderately iodide deficient was 4.5 times higher in 1988-1991 than in 1971-1974
(Hollowell 1998). The percentage of women of childbearing age (15-44 years) who were
moderately iodide deficient increased 3.8 from the 1971-1974 period to the 1988-1991 period
(Hollowell 1998). Furthermore, the percentage of pregnant women who were moderately iodide
deficient increased 6.9% from the 1971-1974 period to the 1988-1991 period (Hollowell 1998).

Since the NHANES 2000 data do not provide iodide nutritional information for pregnant
women, the OIG checked the current validity of the NHANES III data by evaluating the results

117

-------
10-P-0101

from a 2004 urinary iodide sampling of 100 consecutive healthy, pregnant women in an inner-
city obstetric clinic in Boston, Massachusetts (Pearce 2004). The Pearce study shows a similar
distribution of mildly and moderately iodide-deficient pregnant women as was observed in the
NHANES III survey (Pearce 2004). The Boston sample found 9% of the pregnant women were
moderately deficient (i.e., < 50 |ig/L UIC). The Boston sample found about 21% of the pregnant
women (i.e., the 30th percentile) were mildly iodide deficient (i.e., UIC was greater than 50 |ig/L
but less than 100 |ig/L). Finally, 49% of the Boston sample of pregnant women had UICs below
the U.S. recommended dietary allowance (RDA) of 150 |ig/L (i.e., corresponds to a daily intake
of - 220 |ig/day for a pregnant women). Therefore, the Boston sampling confirms that the
percentage of iodide-deficient pregnant women has remained stable over the last decade since
the 1988-1994 NHANES IV.

Criticism of the NHANES III Data to Accurately Assess Iodide Deficiency

The NAS Committee directly commented on the NHANES III survey results that 15% of
women of childbearing age and 7% of pregnant women had UIC less than 50 |ig/L (NAS 2005,
p 46). NHANES measured UIC through a single spot (untimed) analysis. The NAS Committee
noted, "... distribution of iodide values measured in a spot urine samples is broader than values
measured repeatedly in individual subjects (Anderson et al. 2001), this leads to overestimation of
the number of subjects with both low and high values" (NAS 2005, p 46). In short, the NAS
Committee's opinion is that NHANES overestimates the percentage of U.S. women of
childbearing age and pregnant women who are moderately iodide deficient due to the limitations
of a single urine spot test.

An Associate Clinical Professor of Medicine and Epidemiology of the Yale School of
Medicine provided public comments on the same topic to California's Draft Public Health Goal
for Perchlorate in Drinking Water on behalf of the Lockheed Martin Corporation (Borak 2002).
The Associate Clinical Professor indicated that the NHANES III survey "... exaggerates and
perhaps misrepresents the iodine sufficiency of the U.S. population generally and that of
pregnant women specifically" (Borak 2002). The Associate Clinical Professor challenged the
meaningfulness of the NHANES III survey results that indicate that 15% of women of
childbearing age and 7% of pregnant women may be moderately iodide deficient on the
following two perspectives:

1. A single spot urine sample reflects very recent dietary intake rather than diet
generally. The Associate Clinical Professor also cites the Anderson field study
(Anderson 2001), which measured urinary iodide samples in 15 men monthly for 12
months. The variability of urine iodide levels within a month across the individual
samples was much greater than the variation of individual means considered across the
population (i.e., 6.7% of individual samples were < 25 |ig/L but no individual had a
yearly average level < 25 |ig/L). Therefore, in the Associate Clinical Professor's opinion,
the NHANES III survey results "... overstates the proportion of subjects with very low
(and very high) urinary iodide levels." The Associate Clinical Professor suggested that if
multiple urine samples were obtained instead of single spot samples from pregnant
women, few (if any) would have levels of < 50 |ig/L.

118

-------
10-P-0101

2. In a follow-up study by Hollowell (Hollowell 2002), TSH and total T4 levels were
evaluated in the serum samples collected for the NHANES III survey. This study found
that 0.3% had overt hypothyroidism (i.e., elevated TSH and below normal T4) and 4.3%
had subclinical hypothyroidism (i.e., elevated TSH and normal T4). The Associate
Clinical Professor indicated that this study found that moderate iodide deficiency does
not predispose an individual to hypothyroidism.

The OIG addresses these criticisms by the NAS Committee and the Associate Clinical
Professor regarding the utility of using the NHANES III survey to estimate the frequency of
iodide deficiency with in the U.S. population with the following comments:

• The OIG's biggest concern is that both the NAS Committee and the Associate Clinical
Professor's comments cite a 12-month "male" urine study. The variation in a spot urine
sample is expected to vary because of several factors: variation in urine dilution due to
the body's level of hydration, diurnal iodide variation, time since last meal, and seasonal
iodide intake variation (Demers 2003). However, the essential point is to know how the
atypical biology of the most sensitive group (i.e., pregnant women) affects the results of
the urinalysis. Pregnancy produces pronounced changes on the body. For the spot urine
test, the relevant change is that renal blood flow and glomerular filtration increase, which
leads to the increased clearance of iodide from the plasma (Glinoer 2005). This change
in the kidneys results in a lower plasma-iodide concentration and the elevated excretion
of iodide in the urine. Since pregnant women are both using more iodide (i.e., to provide
for the fetus) and are losing more iodide through the kidneys, the recommended iodide
for pregnant women is much higher (e.g., RDA of 230 |ig iodide intake/day
(WHO/UNICEF/ICCIDD RDA of 250-300 |ig iodide intake/day) versus 150 |ig iodide
intake/day for an adult). Since pregnancy causes an increase in the iodide excretion rate,
the UIC will appear higher than normal for the given level of iodide intake, thus giving a
false reading (i.e., bias high) of the actual iodide nutrition status of the pregnant women
(Demers 2003, p 78). Therefore, the increased excretion from the kidneys during
pregnancy biases the UIC high, while the spot test can bias the UIC low; thus, the two
effects probably cancel each other out. Thus, the NHANES III survey results on the rate
of iodide deficiency in pregnant women are probably reasonable, but these results do
represent a significant uncertainty in the science concerning this public health issue (i.e.,
what portion of the U.S. pregnant women population is moderately or mildly iodide
deficient?).

• A closely related concern is how to interrupt, for example, a UIC of 50 ug/L in pregnant
women. A UIC of 50 ug/L in nonpregnant adults is considered to be moderately iodide
deficient. The recommended UIC for healthy, nonpregnant adults is 100 to 200 |ig/L
(Delange 2005, p 278, table 1 IE.8). Therefore, the median healthy UIC value is 150
|ig/L. Thus, a UIC of 50 ug/L in nonpregnant adults is one-third of normal. However,
the recommended healthy UIC in pregnant women is higher. Several credible sources
have opined on the recommended UIC for pregnant women. The average recommended
UIC for pregnant women from these opinions was 204 |ig/L. The following table
summarizes the sources and recommended UICs for pregnant women:

119

-------
10-P-0101

Recommended Iodide

Recommended Urinary

Intake in Pregnant Women

Iodide Concentration

Source

(Mg/day)

(Mg/L)

WHO/UNICEF/ICCIDD*

200

200-300

Dr. Francois Delange, MD**

225-300

150-230

WHO Technical Consultant Group***

225-375

150-249

Recommend Dietary Allowances (RDA)|

220|

150|t

Average Value =>

258

204

* Source: Delange 2005, p 278, table 11E.8.

** Source: Delange 2004.

*** Source: ATA 2006.
f Source: NAP 2000.
ft Source: Pearce 2004.

Therefore, a UIC of 50 ug/L in pregnant women is a lower fraction of the healthly
recommended UIC for pregnant women. A UIC of 50 ug/L in pregnant women is less
than one-fourth the average recommended UIC level for pregnant women. A UIC of 50
ug/L in pregnant women could be as much as one-sixth the recommend UIC level if the
upper limit of 300 ug/L from the WHO/UNICEF/ICCIDD is used. Therefore, a UIC 50
ug/L in pregnant women represents a greater level of iodide deficiency than the same
corresponding UIC of 50 ug/L in nonpregnant adults.

• To assess whether the NHANES III survey data overestimate the percentage of U.S.
pregnant women with iodide deficiency, the OIG searched for an independent source of
data that could be used to corroborate the NHANES III survey results. Iodide levels in
human breast milk (i.e., a different excretion route) are an independent source of data that
can be used to independently assess the iodide deficiency status of U.S. lactating women.
The observed iodide content of human breast milk can be used to document the frequency
of iodide deficiency in U.S. pregnant women. The WHO/UNICEF/ICCIDD uses the
following iodide breast milk values to classify the severity of iodide deficiency in goiter
endemic areas:

Parameter

Classification of Iodide Deficiency

Mild

Moderate

Severe

Iodide Levels
In Human Breast Milk

35 to 50 |jg/L

20 to 35 |jg/L

< 20 jjg/L

Source: Delange 1994, p 109.

The urinary iodide results from the NHANES III survey indicate that about 42% of the
U.S. lactating women population is iodide deficient. By comparison, three studies
measuring the iodide content of breast milk had the following findings:

> Of the 23 lactating women measured for iodide in the Kirk 2005 study, the iodide
content of the breast milk in 14 of 23 lactating women (60%) was measured to be
iodide deficient (i.e., < 50 |ig/L) (Kirk 2005).

120

-------
10-P-0101

> Of the nine lactating women in the Kirk 2007 study, the iodide content of the
breast milk in four of nine lactating women (44%) was found to be iodide
deficient (i.e., < 50 |ig/L) (Kirk 2007).

> Of the 57 lactating women measured for iodide in the Pearce 2007 study, the
median iodide breast milk was reported to be 155 |ig/L (range 2.7 - 1968 |ig/L)
(Pearce 2007). Since only summary statistics were provided (i.e., not individual
results for each women), the percentage of iodide-deficient women could not be
identified. Pearce points out that differences in analytical methods for measuring
total iodide may explain some of the differences between the Pearce 2007 study
and the Kirk 2005 and 2007 studies.

Although the sampling size in these three studies is too small to adequately identify the
distribution of iodide deficiency in the U.S. lactating population as being mild, moderate,
or severe, the percentage of iodide-deficient lactating women in the two Kirk studies
(60% and 44%) do independently support the conclusion from the NHANES III survey
that about 42% of the U.S. lactating women population is iodide deficient.

• Iodide surveys using the spot test (e.g., NHANES III survey, NHANES 2000, and Boston
area study) are appropriate to identify the average UIC of a population. The inherent
differences in urine dilution in a spot urine survey are partially compensated for by using
a large number (-50) of subjects in each study population (Demers 2003, p 77). Due to
these limitations, the spot test urine surveys are not well suited to accurately characterize
the size of the high or low UICs in the population. Characterizing the high and low
UICs in the population is critical, because the population with high and low UICs is the
population susceptible to adverse health effects from iodide.

• Additional iodide survey(s) should be designed to accurately characterize the population
with high and low iodide intakes. Iodide intake is best determined by a 24-hour urine
sample; however, logistics make this method impractical for epidemiological studies
(Demers 2003, p 77). The difference in level of dilution of a spot urine sample can be
partially compensated for using results normalized to urine creatinine (i.e., |ig iodide
excreted/gram creatinine (Ul/Cr ratio)), but the diurnal and seasonal cycles of iodine and
creatinine urinary excretion are different along with seasonal variations in iodide intake
(i.e., higher UIC in winter) (Demers 2003). Furthermore, urinary iodide excretion can
vary even in healthy, well-nourished individuals. The Associate Clinical Professor
recommends taking more samples to better characterize the UIC. However, this is
unlikely to happen because the CDC does not consider iodide nutrition to be a problem or
priority (the NHANES 2000 survey only took 1,433 samples to measure iodide instead of
the 21,298 samples in the NHANES III survey). Therefore, a practical substitute of the
ideal but impractical 24-hour urine sample is the age- and sex-adjusted (Ul/Cr) ratio in a
fasting morning specimen, which comes close to the true 24-hour iodide excretion if
nutrition is generally adequate (Demmers 2003, p 77). Since urinary iodide is diurnal and
reaches a median early in the morning or 8-12 hours after the last meal, another practical
substitute could be to take spot urine samples at these times (Demers 2003).

121

-------
10-P-0101

• In pregnant women with iodide deficiency, a steady decrease in UIC is seen from the first
trimester through the third trimester, hence revealing the underlying tendency toward
iodide deficiency associated with the pregnancy (Glinoer 2005). This steady decline in
UIC continues through lactation. This trend is seen in the NHANES III survey data with
the UIC of lactating women being consistently lower than the UIC of pregnant women
for the same percentile. The point is the NHANES III data do not indicate when the spot
urine sample was taken during the pregnancy (i.e., which trimester(s)). If the majority of
the data points are early on in the pregnancy, the UIC might be biased high and not
representative of the last trimester of the pregnancy.

• The Associate Clinical Professor questions the association between iodide deficiency and
hypothyroidism. The NAS Committee's mode-of-action model assumes that pregnant
women have to become hypothyroid before mental deficits in their children occur.
However, the epidemiological studies of iodide deficiency in euthyroid pregnant women
show mental deficits in their children (see hazard characterization section above for
details). These studies indicate that the mother does not have to be hypothyroid before
mental deficits can occur in her children. Although the mother can be euthyroid during
iodide deficiency, the fetal thyroid is less able to adapt to the iodide deficiency, resulting
in the possibility of permanent lifelong mental deficits. Therefore, the Associate Clinical
Professor's issue that the Hollowell's study (Hollowell 2002) does not associate iodide
deficiency with hypothyroidism is not critical to assessing the risk. By contrast, iodide
deficiency clearly stresses the thyroid of pregnant women; this situation is only partially
reversible during late postpartum (Glinoer 1992).

• The Associate Clinical Professor points out that the spot urine samples reflect only recent
dietary intake and not yearly averages of UIC. However, due to the speed at which the
fetal brain develops, a 2- or 3-week period of insufficient thyroid hormones may be
adequate to irreversibly affect the brain's development and result in permanent, lifelong
mental deficits in the child. This short time frame is observed in a study of the treatment
of congenital hypothyroidism, where neonates who took longer than 2 weeks to have
their thyroid hormone levels normalize had significantly lower cognitive, attention, and
achievement scores than those infants who had normal thyroid function within 1 or 2
weeks of therapy (Selva 2005).

Assessing Iodide Exposure through the FDA Total Dietary Study

Exposure can also be assessed by estimating the amount of dietary intake. On January 2,
2008, FDA published a TDS measuring the dietary intake of iodide (FDA 2008). The TDS
involved the purchase and iodide testing of 285 foods and beverages that represents the "total
diet" of the average U.S. population. Iodide has been analyzed for in all TDS foods since 2003.
The TDS was designed to provide a reasonable estimate of total dietary exposure to iodide from
all foods in the diet. The contribution of iodide from each food group was determined by
summing the estimated intakes from all TDS foods within each of the 12 food groups. The
percentage of total intake of iodide is calculated for each food group for 14 age/sex groups. The
percentage contribution of estimated daily intake of iodide for each food group for three age/sex
groups is provided in the following table:

122

-------
10-P-0101

Total Dietary Study
(TDS) Food Group

Percentage Contribution (%) of Each TDS Food Group
to the Estimated Daily Intake of Iodide for 2003-2004

Non-Nursing Infants
(6-11 months old)

Children
(6 years old)

Women
(25-30 years old)

Baby Food

Beverage

Dairy

Egg

Fat/oil

Fruit

Grain

Legume

Mixture

Meat, poultry, fish

Sweets

Vegetable

Source: Murray 2008, table 6.

The FDA TDS identifies that for 6- to 11-month-old non-nursing infants, the baby food
group is the dominant source of iodide in the diet, contributing 56% of the total intake of iodide
followed by the dairy food group at 34%. Likewise, the FDA TDS identifies that for 6-year-old
children, the dairy food group is the dominant source of iodide in the diet, contributing 70% of
the total intake of iodide. Furthermore, the FDA TDS also identifies that for 25- to 30-year-old
women (i.e., the closest FDA age/sex group corresponding to women of childbearing age), the
dairy food group is the dominant source of iodide in the diet, contributing 49% of the total intake
of iodide. The FDA TDS clearly identifies that dairy and grain are the most significant sources
of iodide in the adult diet (Murray 2008, p 7).

The FDA TDS for iodide also determines the estimated upper and lower average iodide
intake for each food group for 14 age/sex groups. The estimated total iodide intake for each
age/sex group can then be compared against the estimated average requirement (EAR) and the
RDA for each age/sex group. EARs are used to both assess the adequacy of population intakes
and as the basis for calculating RDAs for individuals (NAP 2000). RDA is a goal for individual
intake. RDAs are set to meet the needs of almost all (i.e., 97% to 98%) individuals in a group
(NAP 2000). The estimated upper- and lower-bound average iodide intakes for three age/sex
groups are provided in the following table:

123

-------
10-P-0101

Total Dietary Study
(TDS) Food Group

Range of Estimated Lower and Upper Bound
Average Iodide Intake
for 2003-2004 for Each Food Group (pg/person/day)

Infants
(6-11 months)

Children
(6 years old)

Women
(25-30 years old)

Baby Food

82.8 -88.3

0.0-0.0

Beverage

0.0-1.8

0.1 - 11.3

0.2-32.1

Dairy

50.8-50.8

187.9-188.0

83.2 - 83.2

Egg

2.5-2.5

5.1 - 5.1

6.0-6.0

Fat/oil

0.0-0.0

0.2-0.3

0.6-0.7

Fruit

1.6-3.1

5.8-9.7

4.2-7.4

Grain

3.6-4.1

37.1 - 39.5

32.3- 34.8

Legume

0.0-0.1

0.2-0.5

0.2-0.6

Mixture

1.9-2.5

11.2-13.3

12.6-15.8

Meat, poultry, fish

0.5-0.6

4.0-5.3

4.6-6.4

Sweets

0.0-0.1

2.7-3.4

2.8-3.2

Vegetable

0.6-1.1

1.2-3.9

1.4-5.8

Total Iodide Intake

144-155

255 - 280

148-196

Estimated Average Requirement (EAR)

130

160 (pregnancy)*
209 (lactation)*

Recommended Daily Allowance (RDA)*

130

150

220 (pregnancy)
290 (lactation)

Source: Murray 2008, table 7.
* Source: NAP 2000.

The range of average iodide intakes for the 6- to 11-month-old non-nursing infants
subpopulation is above both the EAR and RDA for that subpopulation. Likewise, the range of
average iodide intakes for the 6-year-old children subpopulation is also above EAR and RDA for
that subpopulation.

Similarly, the range of estimated average iodide intakes for the 25- to 30-year-old women
subpopulation is above the EAR for nonpregnant women of this subpopulation. However, the
FDA TDS does not estimate intakes for subgroups with specific nutritional needs (e.g., pregnant
or lactating women). However, the iodide intake of pregnant and lactating women is the critical
information needed to assess the risk from this public health issue. Pregnant and lactating
women have increased caloric needs, which would likely result in "somewhat higher" iodide
intake values (Murray 2008, p 8). From a population standpoint, and considering the added food
intake of pregnant and lactating women, the range of estimated average iodide intakes for the 25-
to 30-year-old women subpopulation of 148-196 |j,g/person/day generally meets the EARs of 95,
160, and 209 |ig/person/day for women, pregnant women, and lactating women of this age
group. However, even when considering the added food intake of pregnant and lactating women
(e.g., estimated at 20% more), the increased estimated average iodide intakes for pregnant and
lactating women has difficulty meeting the RDAs of 220 and 290 |ig/person/day for pregnant
and lactating women. In other words, if the average intake of iodide for a population just meets

124

-------
10-P-0101

the RDA, this means that half of the pregnant and lactating women (i.e., the individuals below
the average intake of iodide) are below the RDA for pregnant and lactating women. This finding
from the FDA TDS for iodide is fully consistent with the observation that 49% of the Boston
sample of pregnant women in the Pearce study had UICs below the U.S. RDA of 150 |ig/L
pregnant women (i.e., corresponds to a daily intake of ~ 220 |ig/day for a pregnant woman)
(Pearce 2004). Furthermore, this finding from the TDS for iodide is in agreement with the
NHANES III results indicating the 50th percentile of pregnant women (i.e., UIC of 140 |ig/L) is
below the U.S. RDA of 150 |ig/L for pregnant women. Therefore, if the iodide exposure
assessment is done by measuring iodide excretion or by measuring iodide intake by food
consumption, the iodide exposure assessment comes to the same conclusion that about half (i.e.,
50%) of the pregnant women are below the RDA for iodide.

The FDA TDS does not provide the required information needed to evaluate and quantify
the risk in the most sensitive group, pregnant women. The perchlorate and iodide intakes in
pregnant women are needed assess the risk to pregnant women. The perchlorate intake is
essential to quantify the total amount of NIS inhibition occurring in pregnant women.
Furthermore, iodide intake is also essential in quantifying the amount of TIU occurring in
pregnant women. However, the design of the FDA TDS did not identify pregnant women as one
of the 14 age/sex population groups in the study (Murray 2008, p 8).

The perchlorate and iodide intake in pregnant women is needed to characterize the
perchlorate and iodide exposure occurring in nursing infants. A nursing mother will pass her
perchlorate exposure level (as well as her thiocyanate and nitrate exposoure level) and iodide
nutrition level onto her nursing infant. In a cumulative risk assessment, the exposure level to all
four NIS stressors in pregnant women must been known and is essential to assessing the risk to
nursing infants. However, the FDS TDS does not provide this sought after exposure information
in pregnant women. The perchlorate and iodide intakes for pregnant women were not
incoporated into the design of the FDS TDS (Murray 2008).

Another critical limitation of the FDA TDS study is that the distribution of iodide intake
is not determined in pregnant women. The FDA TDS only determined the statistical range of the
average iodide intake. This type of information is useful to evaluate whether the median iodide
dietary intake in a population is reasonable. However, adverse health effects from iodide intake
in pregnant women occur at the extremes (i.e., the upper and lower 10%). For this public health
issue, subtle mental deficits in children occur in either pregnant women who are mildly or
moderately iodide deficient. The percentage of pregnant women with low iodide intake (i.e.,
poor iodide nutrition) is a critical piece of information for the cumulative risk assessment of this
public health issue. Unfortunately, the FDA TDS was neither designed for nor provides
nutritional information on the proportion of the U.S. pregnant women subpopulation that is either
mildly or moderately iodide deficient.

125

-------
10-P-0101

Iodide Exposure Assessment of the Fetus and Nursing Infant is Directly Related
to the Mother's Iodide Status

Since the mother is the sole provider of both thyroid hormones and iodide to the
developing fetus during pregnancy, the mother's iodide status can directly affect the developing
fetus. The fetal thyroid becomes operational only after midgestation (Glinoer 2005, p 1095).
During the first and second trimester, the developing fetus is almost exclusively dependent on
the mother to provide an adequate supply of thyroid hormones for proper brain development
(Glinoer 2005). However, by the third trimester, the supply of thyroid hormones is of
essentially fetal origin, but the fetus is still dependent on the mother for an adequate supply of
iodide for proper brain development (Glinoer 2005).

During lactation, the mother provides the nursing infant with iodide. The mother's iodide
status can directly affect the developing infant. The regulation of iodide into breast milk is only
partially known (Laurberg 2002). However, study results observed in Danish mothers indicate
that iodide content of breast milk seems to vary in parallel with the iodide intake of the mother
(Nohr 1994; Laurberg 2002, figure 1). The results of the iodide study of Danish mothers and
infants are summarized below:

Iodide Ratio Evaluated

Ratio
(median)

Percentiles
(25% - 75%)

Number
Tested

Ratio of iodide concentration in breast milk to the
mother's urinary iodide concentration

0.96

0.50-1.96

n = 142

Ratio of infant's urinary iodide concentration to
mother's iodide breast milk concentration

1.00

0.60-1.64

n = 137

Ratio of infant's urinary iodide concentration to
the mother's urinary iodide concentration

1.00

0.65-1.75

n = 141

In other words, this study indicates that, on average in a population, the iodide status of
the mother is directly reflected in the iodide concentration of her breast milk. As the breast milk
is subsequently consumed by the infant, on average in a population, the infant's iodide status is
directly related to the mother's iodide concentration in her breast milk. As such, on average in a
population, the mother's iodide status directly corresponds to the iodide status of the nursing
infant. Therefore, during lactation, the iodide nutritional status of the lactating mother is a good
measure of the iodide nutritional status of the nursing child (Laurberg 2002). For example, if the
lactating mother is moderately iodide deficient, then it is reasonable to expect the nursing infant
to be also moderately iodide deficient.

The Size of the U.S. Sensitive Population Potentially Affected by the Lack of
Iodide

The final task in the iodide exposure assessment is to determine the size of the U.S.
sensitive population potentially affected by the lack of iodide. Although about half of pregnant
women are iodide sufficient (i.e., RDA for pregnancy of 150 |ig/L (Pearce 2004)), a significant
portion of the sensitive groups (i.e., pregnant women, fetuses, neonates born with elevated TSH
levels, and nursing infants) have either mild or moderate iodide deficiency. The following table
provides an estimate of the size of each of the sensitive groups:

126

-------
10-P-0101

Severity
of Iodide
Deficiency

Sensitive Population

Percentage of
4 Million Annual
U.S. Births

Estimated U.S.
Population Potentially
Affected Per Year

Moderately

Pregnant Women/Fetuses

6.9%

276,000

Elevated TSH Neonates (Occurrence
rate of 17.0% to 36.9% rate above
iodide sufficient rate up to 3%)

1.2% to 2.5%
(i.e., 0.069x0.17;
and 0.069 x 0.369)

48,000 to 100,000

Nursing Infants

3.7%
(i.e., 0.069 x 0.53
nursing frequency)

148,000

Mildly

Pregnant Women/Fetuses

22%

880,000

Elevated TSH Neonates
(Occurrence rate of 0% to 16.9% rate
above iodide sufficient rate up to 3%)

0% to 3.7%
(i.e., 0.22x0.0;
and 0.22x0.169)

Oto 148,000

Nursing Infants

11.7%
(i.e., 0.22x0.53
nursing frequency)

468,000

The NHANES III survey identifies that about 6.9% of the U.S. pregnant women are
moderately iodide deficient. Since the United States has about 4 million births per year (Martin
2003), up to approximately 276,000 children are born every year to moderately iodide-deficient
mothers in the United States (4,000,000 births x 0.069). Therefore, up to 276,000 children are
potentially affected every year by the more severe adverse effects associated with low maternal
TIU during pregnancy (i.e., maternal UIC less than 50 ug/L).

In addition, the NHANES III survey can be used to identify the percenation of U.S.
pregnant women who are mildly iodide deficient. The NHANES III survey identifies
approximately the 29th percentile of U.S. pregnant woman population as having a UIC less than
100 |ig/L and therefore, as being iodide deficient. The percentage of U.S. pregnant women who
are mildly iodide deficient can be estimated by taking the total number of iodide-deficient
pregnant women (i.e., 29%) and subtracting the number of moderately iodide deficient pregnant
women (i.e., 6.9%). Therefore, approximately 22% (i.e., 29% - 6.9%) of the U.S. pregnant
woman population is estimated to be mildly iodide deficient. Since the United States has about 4
million births per year (Martin 2003), up to approximately 880,000 children are born every year
to moderately iodide-deficient mothers in the United States (i.e., 4,000,000 births x 0.22).
Therefore, up to 880,000 children are potentially affected every year by the minimal adverse
effects associated with low maternal TIU during pregnancy (i.e., maternal UIC between 50 and
100 ug/L).

The neonatal TSH level at birth is the principal indicator that identifies fetal thyroid stress
and focuses on the potential for neonatal brain damage (Delange 1998). Due to the importance
of this indicator, the iodide exposure assessment estimates the number of elevated TSH neonates
above the baseline rate observed in iodide-sufficient populations to be 0 to 148,000 and 48,000
to 100,000 neonates per year in mildly and moderately iodide-deficient populations, respectively.
This is the portion of the population that may have partially impaired thyroid function, which
makes it particularly susceptible to iodide deficiency.

127

-------
10-P-0101

The nursing infant's brain is still developing and still vulnerable to an insufficient uptake
of iodide by the thyroid. The nursing infant's iodide nutritional status is directly related (i.e., a
ratio of 1.00) to the iodide nutritional status of the mother (Nohr 1994; Laurberg 2002, figure 1).
From a population standpoint (i.e., not an individual standpoint), a moderately iodide-deficient
nursing mother is likely to have a moderately iodide-deficient nursing infant. Therefore, the
number and level of iodide deficiency in nursing infants can be estimated from the number and
level of iodide-deficient mothers. Furthermore, the fact that not all mothers breastfeed has to be
factored into the estimate. At least 53.6% ±1.7 of U.S. infants are breastfed at least once, while
22.4% ±1.2 of U.S. infants are breastfed at 6 months (Li 2002). Therefore, the iodide exposure
assessment estimates the number of U.S. nursing infants with mildly and moderately iodide
deficiency to be up to 468,000 and 148,000, respectively.

128

-------
10-P-0101

9. Till Cumulative Risk Assessment

Unlike a conventional single chemical exposure assessment, a cumulative risk
assessment accounts for all the known stressors associated with a public health issue.

Cumulative risk assessment expands our understanding of the public health impacts of
environmental exposures (Fox 2004) and provides a clearer, more complete picture of a public
health issue for making risk management decisions (EPA 2003, appendix A). Cumulative risk
assessment will reduce risks to the extent that it can be integrated into prevention strategies to
track and protect public health (Fox 2004). Therefore, this cumulative risk assessment must
account for the following known stressors affecting this public health issue:

1. The amount of NIS inhibition from thiocyanate.

2. The amount of NIS inhibition from nitrate.

3. The amount of NIS inhibition from perchlorate.

4. The lack of iodide in the diet.

5. The heterogeneous thyroid performance within the human population.

6. The interference of the other major steps in the use of thyroid hormones (e.g., production,

transport, and peripheral tissue metabolism).

The OIG's approach to conducting a cumulative risk assessment is to apply the NAS
Toxicity Testing Committee's vision for toxicity testing to this public health issue. The NAS
Toxicity Testing Committee's vision of toxicity testing is to develop a quantitative, mechanistic,
dose-response model of the cellular pathway that is perturbed by the stressor(s) (NAS 2007).
Subsequent pharmacokinetic modeling would identify a safe human exposure level that prevents
the environmental chemical from reaching a toxic tissue concentration. The NAS Toxicity
Testing Committee's vision takes advantage of advances in science (e.g., proteomics and
genomics) in which the increased understanding of the biological mechanisms allows in vitro
toxicity testing to be applied to environmental risk assessments. As the complexity of the risk
assessment increases (i.e., more factors evaluated), animal testing is neither practical nor
sensitive enough to observe and define the relationship between an increasing number of
stressors. For example, four stressors each administered at three different doses (i.e., high,
medium, and low) would require 81 separate animal studies to evaluate all the possible
combinations (i.e., 34). To conduct 81 Argus-equivalent perchlorate rat studies is impractical.
Furthermore, no scientific techniques are available to measure subtle cognitive deficits in rat
offspring. These limitations in conventional animal testing require the implementation of NAS
Toxicity Testing Committee's vision of toxicity testing in order to characterize the risk from
exposure to the four NIS stressors.

A quantitative, mechanistic, dose-response model of the NIS cellular pathway is available
in the published literature. The Tonacchera Model is a multifactorial mathematical equation that
characterizes the biological performance of the NIS to uptake iodide to the simultaneous
exposure to all four NIS stressors (Tonacchera 2004). The Tonacchera Model defines the
relationship between the four NIS stressors (1-4 listed above) and the TIU by the NIS.

Therefore, the use of the Tonachera Model results in a single variable, the TIU, that measures the
resulting cumulative effect (i.e., integrated effect) of the simultaneous exposure to all four NIS
stressors. Since a low TIU result triggers the subsequent adverse effects, knowing the relative

129

-------
10-P-0101

combinations of the four NIS stressors that result in a low TIU is critical to addressing this public
health issue. In order to define the combination of the four NIS stressors that result in an
unacceptably low TIU requires the development and implementation of a cumulative risk
assessment.

The OIG is using of the Tonacchera Model to conduct a cumulative risk assessment
that incorporates the four NIS stressors. This is in contrast to the outdated approach of a single
chemical risk assessment (EPA 1992; NAS 1994; EPA 1997b; EPA 2000; Callahan 2007).
Nevertheless, with any environmental model, any results need to be corroborated with other
independent sources of information (e.g., predicted level of stressor exposure is observed to
result in adverse effects in humans) (EPA 2003 a).

In order to conduct a cumulative risk assessment, the exposure to each of the stressors
must be characterized. The human exposure to perchlorate and lack of iodide is relatively well
characterized. By contrast, the human exposure to thiocyanate and nitrate is less well
characterized. Therefore, this cumulative risk assessment uses only the typical thiocyanate and
nitrate exposure levels in the U.S. population to demonstrate the application of cumulative risk
assessment to this public health issue. Ideally, the exposure distribution of thiocyanate and
nitrate in the U.S. population would be a useful piece of information. The 90th percentile
thiocyanate exposure of nonsmokers in the U.S. population would be useful in the subsequent
modeling. Likewise, the 90th percentile of nitrate exposure in the U.S. population would also be
useful in the subsequent modeling.

9.1 Corroboration of Tonacchera Model with Effects Observed in Humans

The Tonacchera Model must be corroborated with effects observed in humans before
being used for risk characterization and before being used to define an RfD based on a %TIU
level. Section 3.1.3 ofEPA's Draft Guidance on the Development, Evaluation, and
Application of Regulatory Environmental Models (November 2003) establishes the need to
corroborate all environmental models before risk assessors use the models as the basis of rule
making or regulation (EPA 2003a). The Tonacchera Model represents an environmental model
that calculates the TIU by the thyroid based on the environmental exposure to all four NIS
stressors. As such, the Tonacchera Model results must be corroborated with effects observed
in humans. This corroboration is performed in this section of the OIG Analysis.

"Robustness" is an aspect of model corroboration. Section 3.1.3.2 of EPA's Draft
Guidance on the Development, Evaluation, and Application of Regulatory Environmental
Models defines robustness as the capacity of a model to perform equally well across the full
range of environmental conditions (EPA 2003a). A model cannot be corroborated against the
data or similar data used to generate or calibrate the model. An independent measure of the
model's performance is that the model agrees with results from other sources of data. Since
the Tonacchera Model is derived from in vivo data, all human results represent potential
corroboration from a different data source. The OIG Analysis corroborates the Tonacchera
Model in Sections 9.1.1 through 9.1.6. The corroboration of the Tonacchera Model is robust
because of the multiple points of agreement and the ability to explain the basis of the observed

130

-------
10-P-0101

effect across a variety of human data sources and types. The major points of corroboration are
(see the individual sections in 9.1 for a more detailed explanation):

• The calculated %TIU levels for adverse thyroid effects from chronic excess NIS
inhibitor exposure in both men and women is consistent with the NAS
Committee's statement that a 75% reduction in TIU would be needed to induce
adverse effects in adults (which is not the sensitive group) (see Section 9.1.1).

• The Tonacherra Model calculates lower %TIU levels for the more severe
thyroid condition, hypothyroidism, than for the less-severe thyroid condition
hypothyroxinemia (see Sections 9.1.2 and 9.1.3).

• The calculated %TIU levels for adverse thyroid effects from chronic excess NIS
inhibitor exposure show that women are more prone than men to adverse
thyroid effects induced by a low TIU, which is consistent with clinical
observations (see Sections 9.1.2 and 9.1.3).

• The Tonacchera Models calcluates a decrease in %TIU that agrees with the
measured radioactive iodide uptake (RAIU) levels in both the Greer Perchlorate
Exposure Study and the Braverman Perchlorate Occupational Exposure Study
(see Sections 9.1.4 and 9.1.5).

• The findings from the Tonacchera Model on the role and contribution of each of
the NIS stressors to this public health issue are consistent with the role and
contribution of iodide and NIS inhibition on the etiology of endemic cretinism,
which occurs through the same mechanism of toxicity (see Section 9.1.6).

• The Tonacchera Models provides a quantified, scientific explanation as to why
there are no observed health effects from the exposure to 119 ug of perchlorate
per day in the Chilean epidemiological studies (see Section 6).

A major uncertainty in the science concerning this public health issue is how much of a
reduction in the TIU causes an adverse effect in a pregnant woman, a fetus, or a nursing
neonate. In other words, the TIU level associated with adverse effects has not been specifically
identified. However, the NAS Committee estimated that a 75% reduction in TIU would be
needed to induce adverse effects. Specifically, the NAS Committee states, "To cause declines
in thyroid hormone production that would have adverse health effects, iodide uptake would
most likely have to be reduced by at least 75% for months or longer" (NAS 2005, p 8).
Scientific criticism of the NAS perchlorate assessment states that this NAS statement was not
supported by any data, analysis, or references (Ginsberg 2005). Furthermore, a member of the
NAS Committee stated to the Massachusetts Department of Environmental Protection's
Advisory Committee on Health Effects that the 75% value was not derived by any quantitative
analysis but was based on a clinical observation that a loss of a large fraction of the thyroid can
be tolerated in adults (MassDEP 2006). The essential point is the 75% reduction in TIU was
not determined experimentally from exposure to any of the four stressors and was not
determined for the sensitive group: pregnant women and their fetuses (i.e., NAS

131

-------
10-P-0101

characterization of the tolerance of nonpregnant adults is potentially not protective of the
sensitive group). To emphasize the limitations of this NAS statement, the concern for the
sensitive group can be restated as a question: Can a pregnant woman with three-quarters of
her thyroid removed still have enough thyroid hormone production during the added stress
occurring during gestation and lactation to have a healthy child without any mental deficits?

The TIU level associated with adverse effects must be identified or estimated. However,
determing TIU levels in humans is impractial because of the need to conduct radioactive 123I
uptake measurements. By contrast, the Tonacchera Model is able to calculate a TIU level for a
given set of exposures to all four NIS stressors in both the human population in general and in
the sensitive groups of concern (i.e., pregnant women, fetuses, and nursing neonates). This
allows the TIU levels associated with adverse effects to be identified and compared to normal
TIU level observed in a healthy human adult. However, the accuracy of the Tonacchera Model
to estimate TIU levels must be corroborated with observed human data and information. The
corroboration of the Tonacchera Model is performed in remainder of Section 9.1.

9.1.1 NAS Hypothyroidism Statement

The Tonacchera Model indicates that the TIU is directly proportional to the concentration
of free iodide in the serum when the total goitrogen load is held constant (see below).

The Tonacchera Model states that
TIU || [T]/(1.22 +(SPEC))
where the symbol || means "proportional to."

If the SPEC is held constant (i.e., the level of NIS inhibitors remains the same), then
TIU || [T]

Thus,

TIU(Hypothyroidism) || [I j(Hypothyroidism)

The [T] in Tonacchera Model is the internal concentration of free iodide in the blood
serum that is able to directly interact with the NIS. Unfortunately, free iodide concentrations are
rarely reported. Healthy levels of inorganic iodide concentrations in serum are very minute and
are about 1 pg/dL (Demers, 2003, p 77). Because of the difficulty in measuring PII, the
measurement of PII is limited to research studies in pregnancy. Likewise, determining the
iodide intake from the iodide content of food is extremely difficult for methodological reasons
(Delange 2005, p 267). Therefore, the median urinary concentration is the most practical
measurement of iodide intake in a population (Dunn 1998). Iodide balance studies have shown
that adults are in balance with their iodide environment (Delange 2005, p 267). About 90% of
ingested iodide is excreted in the urine (Dunn 1998), while usually a negligible amount (i.e., 5
|ig/day) is excreted in fecal matter. Therefore, in general terms, iodide intake is proportional to
urinary elimination. Thus, the UIC is used as a surrogate measure for the iodide concentration in
the serum.

132

-------
10-P-0101

The NAS Committee states, "Hypothyroidism occurs only if daily iodide intake is below
about 10 to 20 |ig (about one-fifth to one-tenth of the average intake in the United States) [an
intake of only 10 to20 |ig of iodide equates to severe iodide deficiency]" (NAS 2005, p 6). The
recommended iodide intake is 150 |ig/day for an adult (Delange 2005, p 278, table 1 IE.8).
Therefore, the estimated TIU at which hypothyroidism occurs can be estimated by the ratio of
iodide intake associated with hypothyroidism (i.e., range of 10 to 20 |ig/day) to a healthy iodide
intake of 150 |ig/day for an adult. Therefore, the TIU(HyPothyroidism) = (range of 10 to 20 |ig/day ^
150 |ig/day) x 100%) = range of 6.7% to 13.3%. However, the NAS Committee statements that
10 - 20 ug of iodide per day is "about one-fifth to one-tenth of the average intake in the United
States." This description of a daily iodide intake of l/5th to 1/10th corresponds to a
TIU(Hypothyroidism) = 10% tO 20%.

The recommended UIC for a healthy adult is 100 to 200 |ig/L (Delange 2005, p 278,
table 1 IE.8). Therefore, the median healthy UIC value is 150 |ig/L. Severe iodide deficiency is
characterized by a UIC of < 20 |ig/L (Delange 2005, p 265, table 11E.3). Severe iodide
deficiency can generate symptoms of hypothyroidism (i.e., low T4; high TSH) (Delange 2005,
p 272, table 1 IE.6). Therefore, the estimated TIU at which hypothyroidism occurs can be
estimated by the ratio of UIC of severe iodide deficiency to median healthy UIC value.

Therefore, the TIU(Hypothyroidism) = (20 |ig/L h- 150 |ig/L) x 100%) = 13.3%.

In summary, from the estimates made above from the NAS Committee's hypothyroidism
statement, the TIU(HyPothyroidism) is expected to occur in range from 6.7% to 20% with the low
teens being most probable (i.e., 13.3%). This is consistent with the NAS Committee statement:
"To cause declines in thyroid hormone production that would have adverse health effects, iodide
uptake would most likely have to be reduced by at least 75% for months or longer" (NAS 2005,
p 8). In other words, the TIU(HyPothyroidism) is expected to occur at 25% or less.

9.1.2 Hypothyroxinemia in Men Exposed to Excess Thiocyanate

The TIU level at which overt hypothyroidism (i.e., elevated TSH and below-normal T4)
or subclinical hypothyroidism (i.e., elevated TSH and normal T4) is not known by experimental
data. However, hypothyroxinemia is a less-severe thyroid condition and is a common condition
in pregnant women, characterized by low maternal fT4 levels with normal TSH levels (Kooistra
2006). Maternal hypothyroxinemia is documented to be associated with mental deficits (e.g.,
ADHD) in children (Vermiglio 2004). Iodide deficiency induces thyroid changes, characterized
by low serum T4, normal or slightly elevated T3, and a normal TSH level (i.e., as a consequence
of the normal or elevated T3) (Obregon 2005). Therefore, iodide deficiency induces thyroid
changes consistent with hypothyroxinemia. Since both the lack of iodide and NIS inhibitors act
through the same mechanism of action, an excess NIS inhibitor exposure will also induce thyroid
changes consistent with hypothyroxinemia, observed during iodide deficiency (i.e., a low TIU).

Severe hypothyroxinemia is observed in thiocyanate-exposed electroplating workers in
the Banerjee adult male thiocyanate study (Banerjee 1997). The thiocyanate-exposed workers
were measured with below normal T4, just slightly below normal T3 levels, and statistically
elevated TSH levels, but the TSH levels are still in the normal range. The exposed workers were

133

-------
10-P-0101

measured to have a serum thiocyanate exposure of 316 ± 15.0 (unol/L. The observations of the
Banerjee study are summarized below:

Banerjee Study Group

Thiocyanate
(|jmol/L)

(MQ/dL)

TSH
(pU/ml)

35 Nonexposed Workers
(all nonsmokers)

90.8 ±9.02

6.09 ±0.601

111.0 ±9.3

1.20 ±0.301

35 Exposed Workers
(all nonsmokers)

316 ± 15.0*

3.81 ± 0.318|

57.2 ± 8.1|

2.91 ± 0.20|

Normal Ranges Cited in the
Manufacturers Test Kits

5.5 to 13.5

60 to 200

0.2 to 4.0

* p < 0.01
t p < 0.05

Source: Banerjee 1997, table 1.

The Tonacchera Model can be used to calculate the TIU level that corresponds to the
hypothyroxinemia observed in the exposed workers in the Banerjee thiocyanate adult male study.
The TIU calculations are as follows:

The Tonacchera Model states:

TIU || [T]/(1.22 +(SPEC))

where the symbol || means "proportional to".

If the [T] is held constant, a normal [T] level is assigned the variable "x." (Note: the
variable x corresponds to the normal [T] level in the U.S. population of 145 |ig/L), then
the TIU for normal U.S. exposure (i.e., to be used as the normal TIU reference point) to
the four NIS variables is given by the following:

TIU(N0rmai) 11 x

(1.22 + SPEC(Normal))

Since SPEC(nomiai) was previously calculated in this paper to be 1.501 (amol/L from the
median exposures to thiocyanate (SCN), nitrate, and perchlorate in the U.S. population,

TIU(normal) is.

TIU(N0rmal) 11 X

(1.22+ 1.501)

TIU(Normal) 11 X

(2.721)

TIU(Normal) || 0.3675X

In the Banerjee thiocyanate adult male study, the iodide nutritional level was not
determined (which is typical in the epidemiological studies in that they measure only a
subset of the four NIS stressors). However, the iodide nutritional level of the

134

-------
10-P-0101

thiocyanate-exposed workers is believed to be sufficient due to the year and location of
the study (Gibbs 2006). However, a different Baneijee study did measure the iodide
nutritional level in 70 test subjects in India in the same year (Baneijee 1997a). A random
sampling of UIC in 50 to 100 people can be used to define the iodide nutritional status of
a population (Delange 2005, p 267). In the Banerjee thiocyanate women study, the
average UIC of 70 subjects was 119 |ig/L (i.e., avg. 115 |ig/L and 123 |ig/L). Therefore,
a UIC of 119 |ig/L (i.e., 0.82x) is a reasonable value to use for the TIU(HyPothyroxinemia)
calculation in the exposed workers. The TIU(HyPothyroxinemia) calculation will also be been
done using the UIC found in the U.S. population of 145 |ig/L in order to define an upper
limit for the TIU(HyPothyroxinemia) range. Thus, assuming a UIC of 119 |ig/L, then the
TIU(Hypothyroidism) observed in the workers is given by:

TIU(Hypothyroxinemia) 11 0.82 X

(1.22 + SPEC(SCN Exposed Workers))

Since SPEC(scn exposed workers) is given by.

SPEC(scN exPosed workers) = (CIO4" Inhibition) + (NO3" Inhibition) + (free SCN" Inhibition)

SPEC(scNexposedworkers) = [CIO4"] + [N03"] /240 + ([total serum SCN"] X 0.528) /15

The amount of perchlorate and nitrate exposure is not measured in the Banerjee
thiocyanate study, but the default values observed in the U.S. population for perchl orate
and nitrate are used for the calculation:

SPEC(scn" exposed workers) = 0.0013 |imol/L + 0.167 |imol/L + (316 |imol/L SCN" x 0.528) /15

SPEC(scn" exposed workers) = 0.0013 |imol/L + 0.167 |imol/L + 11.12 |imol/L

SPEC(SCN exposed workers) — 11.29 [^mol/L

Inserting 11.29 |imol/L for SPEC(Scn" exposed workers)

TIU(Hypothyroxinemia) 11 0.82x

(1.22+ 11.29)

TIU(Hypothyroxinemia) || 0.066x

The %TIU(Hypothyroxinemia) in the thiocyanate-exposed adult male workers is given by the
following equation:

%TIU(Hypothyroxinemia) — (TIU(SCN exposed workers) ~ TIU(Normal)) X 100%

= (0.066x -h 0.3675x) x 100%

= 18.0%

135

-------
10-P-0101

Assuming the UIC of the exposed workers was 119 |ig/L, the amount of NIS inhibition
from the thiocyanate in the exposed workers resulted in a calculated TIU(HyPothyroxinemia) of 18.0%
as compared to the normal TIU observed in the U.S. population. Assuming the UIC of the
exposed workers was the same as the U.S. population (i.e., 145 |ig/L), the amount of NIS
inhibition from the thiocyanate in the exposed workers results in a calculated TIU(Hypothyroxinemia)
(calculation not shown, but identical to the above expect "0.82x" is replaced with just "x") of
21.8% as compared to the normal TIU observed in the U.S. population.

In summary, as calculated from measurements from the Banerjee adult male thiocyanate
study, a %TIU(HyPothyroxinemia) in the range of 18.0% to 21.8% is shown to induce a T3 and T4
thyroid hormone deficiency in adult males consistent with severe hypothyroxinemia. Since
hypothyroxinemia is a less-extreme thyroid condition than hypothyroidism, %TIU(HyPothyroidism)
should be less than the %TIU(HyPothyroxinemia). From the estimates made above from the NAS
Committee's hypothyroidism statement, the %TIU(HyPothyroidism) is estimated to occur in a range
from 6.7% to 20%, with the low teens typical (i.e., 13.3%). The %TIU(HyPothyroxinemia) in males is
in the range of 18.0% to 21.8%, which is above the TIU(HyPothyroidism) (i.e., low teens). The
calculation of %TIU(HyPothyroxinemia) for severe hypothyroxinemia in men was performed to
corroborate that the results from Tonacchera Model is consistent with information from other
sources (e.g., the clinical observations of the NAS experts).

9.1.3 Hypothyroxinemia in Women Exposed to Excess Thiocyanate

Hypothyroxinemia (i.e., below normal T4 and normal T3 and TSH levels) is observed in
thiocyanate-exposed females in the Banerjee adult female thiocyanate study (Banerjee 1997a).
In Calcutta, India, 30-50 mg/L of thiocyanate is added to preserve milk by activating the
lactoperoxidase-thiocyanate-hydrogen peroxide system (LP system) found in milk in order to
retard the growth of bacteria. The thiocyanate-exposed females are measured to have below
normal T4, and both normal T3 levels and statistically elevated TSH levels, but the TSH levels
are still in the normal range. The exposed females were measured to have a serum thiocyanate
exposure of 230 ± 10.0 (imol/L. The observations of the Banerjee study are summarized below:

Study Group

Thiocyanate
(|jmol/L)

(nmol/L)

TSH
(pU/ml)

35 Nonexposed Females
(all nonsmokers)

90.8 ±9.0

125.4 ±11.5

1.71 ±0.16

1.09 ±0.28

35 Exposed Females
(drank SCN" preserved milk)

230.0 ± 10.0*

87.8 ±6.6*

2.39 ±0.32

2.49 ±0.20*

Normal Ranges Cited in the
Manufacturers Test Kits

110 to 279

0.93 to 3.12

0.2 to 4.0

.* p < 0.01

Source: Banerjee 1997a, table 1.

The Tonacchera Model can be used to calculate the TIU level that corresponds to the
hypothyroxinemia observed in the exposed women in the Banerjee thiocyanate adult female
study. The TIU calculations are as follows:

136

-------
10-P-0101

The Tonacchera Model states:

TIU|| [I"]/(1.22 +(SPEC))

where the symbol || means "proportional to"

If the [I"] is held constant, a normal [I"] level is assigned the variable "x," then the TIU for
normal U.S. exposure (i.e., to be used as the normal TIU reference point) to the four NIS
variables is given by the following:

TIU(U.S. normal) || X

(1.22 + SPEC(U.S. normal))

Since SPEC(normai) was previously calculated in this paper to be 1.501 from median
exposures to thiocyanate, nitrate, and perchlorate in the U.S. population, TIU(normai) is:

TIU(U.S. normal) || X

(1.22 + 1.501 |imol/L)

TIU(U.S..normal) || X

(2.721)

TIU(U.S. normal) || 0.3675X

In the Baneijee thiocyanate adult female study, the iodide nutritional level was measured.
The average UIC in the expose females was determined to be 115 |ig/L (SE 8.5). Since
the iodide nutritional level in the U.S. population is 145 |ig/L, the [T] level in the exposed
workers is assigned a value of 0.793x (i.e., 115 |ig/L ^ 145 |ig/L). Thus, then the
TIU(Hypothyroidism) observed in the workers is given by:

TIU(Hypothyroxinemia) 11 0.793x

(1.22 + SPEC(SCN Exposed Females))

Since SPEC(scn exposed females) is given by.

SPEC(scn" exposed females) = (CIO4" Inhibition) + (NO3" Inhibition) + (free SCN" Inhibition)

SPEC(scn exposed females) = [CIO4"] + [N03"] /240 + ([total serum SCN"] X 0.528) /15

The amount of perchl orate and nitrate exposure is not measured in the Banerjee
thiocyanate study, but the default values observed in the U.S. population for perchl orate
and nitrate are used for the calculation:

SPEC(scn" exposed females) = 0.0013 |imol/L + 0.167 |imol/L + (230 |imol/L SCN" x 0.528) /15

SPEC(scn" exposed females) = 0.0013 |imol/L + 0.167 |imol/L + 8.096 |imol/L

137

-------
10-P-0101

SPEC(SCN exposed females) 8.26 |^mol/L

Inserting 8.26 |imol/L for SPEC(SCN Exposed Females)

TIU(Hypothyroxinemia) || 0.793x

(1.22 + 8.26)

TIU(Hypothyroxinemia) || 0.0836X

The %TIU in the thiocyanate exposed females is given by the following equation:

%TIU(Hypothyroxinemia) — TIU(SCN exposed females) ~ TIU(U.S. normal) X 100%
%TIU(Hypothyroxinemia) = (0.0836X - 0.3675x) X 100%

%TIU(Hypothyroxinemia) — 22.7%

The amount of NIS inhibition from the thiocyanate in the exposed females resulted in a
calculated %)TIU(Hypothyroxinemia) of 22.7% as compared to the normal TIU observed in the U.S.
population. The Baneijee thiocyanate adult female study, a %TIU(HyPothyroxinemia) of 22.1% is
observed to induce hypothyroxinemia in females. Since hypothyroxinemia is less extreme
thyroid condition than hypothyroidism, %TIU(Hypothyroidism) should be less than the
%TIU(HyPothyroxinemia). From the estimates made above from the NAS Committee's
hypothyroidism statement, the %TIU(Hypothyroidism) is estimated to occur in range from 6.7% to
20%), with the low teens being most probable (i.e., 13.3%>). The %TIU(HyPothyroxinemia) in women is
22.1% and is above the NAS Committee's TIU(HyPothyroidism) of the low teens. The calculation of
%TIU(HyPothyroxinemia) for hypothyroxinemia in women was performed to corroborate the results
from Tonacchera Model; the Tonacchera Model results are consistent with information from
other sources (e.g., the clinical observations from the NAS experts).

Furthermore, the %TIU(HyPothyroxinemia) in women is 22.1% and is above the
%TIU(HyPothyroxinemia) found in men of 18.0%) to 21.8%). This finding is consistent with the
observation that women are more sensitive to adverse thyroid effects than men. In the original
Whickman survey, the rate of high serum TSH concentrations was higher in women than men in
each decade of age (Vanderpump 2005, p 401, figure 19.2). Furthermore, both the original
Whickman survey and the subsequent 20-year follow-up survey found the rate of goiter is higher
in women than men (Vanderpump 2005, p 403).

9.1.4 Perchlorate PBPK Model and the Greer Perchlorate Exposure Study

The Tonacchera Model, as with any environmental model, must be corroborated with
other independent sources of information (EPA 2003a). The Tonacchera Model can be used to
calculate the TIU at various external doses of perchlorate. The Tonacchera Model uses the
Clewell Perchlorate PBBK Model to convert external perchlorate doses into internal perchlorate
doses in the blood serum. As such, the Tonacchera Model is an example of the NAS Toxicity
Testing Committee's (NAS 2007) quantitative, mechanistic, dose-response model of the cellular

138

-------
10-P-0101

pathway that is perturbed by the environmental agent. The Clewell perchlorate PBPK model is
the subsequent pharmacokinetic model called on by the NAS Toxicity Testing Committee to
convert the internal exposure dose at the cell (i.e., in this case, the perchlorate serum
concentration in |ig/L) to an external exposure dose (i.e., how much is consumed - mg/kg-day).
As seen in the Clewell perchlorate PBPK model, the conversion factor from internal to external
perchlorate dose is not constant, but changes due to biological differences across life stages (i.e.,
adult, fetus, neonate, child, pregnant woman, lactating woman) and changes with external
exposure levels (Clewell 2007, p 423, table 4). Therefore, the calculated TIUs will be slightly
different for each life stage and external dose level.

The %TIUs calculated by the Tonacchera Model will be compared against the published
predicted %TIUs from the Clewell Perchlorate PBBK Model and the %TIUs observed in the
Greer study. Since, the Greer study was of adults, this comparison of %TIUs is only for adults
and not for the other life stages. The Clewell perchlorate PBPK model predicted the percentage
inhibition of TIU across various life stages (Clewell 2007, p 423, table 5). The Greer study
calculated a 24-hour TIU from the difference in the amount of RAIU from before and after 14
days of a known external perchlorate exposure at four different levels (Greer 2002, p 931, table
1, 24-hr uptake).

The TIUs calculated by the Tonacchera Model at various external perchlorate doses are
provided in the next two tables (see below). Below is an example of the TIU calculations used to
populate the values in each of the columns in the next two tables. The resulting internal
perchlorate dose (i.e., perchlorate serum concentration) in an adult is calculated for an external
perchlorate dose of 0.001 mg/kg-day using the Clewell perchlorate PBPK model (Clewell 2007,
p 423, table 4) in the following calculation:

Int. CIO4" Dose(Aduit) = 0.001 mg x 0.002 mg/L Int. CIO4" Dose = 0.002 mg/L

kg-day 0.001 mg/kg-day Ext. CIO4" Dose

The units of the internal perchlorate dose are converted from mg/L to |imol/L using the
following calculation:

Internal CIO4" Dose(Aduit) = 0.002 mg/L x 1000 ^g x 1 f^mole = 0.020 |imol/L

1 mg 99.45 |ig

The SPEC of an internal perchlorate dose of 0.020 (amol/L is calculated using the following
SPEC equation adapted from the Tonacchera Model (typical U.S. exposures for nitrate and
thiocyanate are used in for this calculation):

139

-------
10-P-0101

The TIU for this external perchlorate dose is calculated using the following Tonacchera Model
equation:

TIU(Ext. perchlorate dose) 11 X

(1.22 + SPEC(Ext. perchlorate dose))

|| X

(1.22+ 1.52 |imol/L)

II x

2.74
|| 0.3650x

The %TIU for this external perchlorate dose is calculated using the following equation:

%TIU(Ext. perchlorate dose) — TIU(Ext. perchlorate dose) ~ THJ(U.S. normal) X 100%

= (0.3650x - 0.3675x) x 100%

= 99.3%

The following table provides the TIUs calculated by the Tonacchera Model at various
external perchlorate doses (see sample calculations above to see how the values were determined
for each column):

External CI04"

Dose
(mg/kg-day)

Internal CI04"
Dose In an
Adult
(mg/L serum)

Internal CI04"
Dose In an
Adult
(|jmol/L)

Serum Perchlorate
Equivalent Cone.
(SPEC)
(|jmol/L)

Calculated Total
Iodide Uptake
(TIU)

0.00013*

0.0013

1.501|

0.3675X

0.000175 (% RfD)

0.00035

0.0035

1.505

0.3670X

0.0007 (RfD)

0.0014

0.014

1.511

0.3662X

0.001

0.002

0.020

1.52

0.3650X

0.007 (NOEL)

0.0088**

0.088

1.589

0.3560X

0.01

0.1

1.60

0.3546X

0.02

0.2

1.70

0.3325X

0.1

1.0

2.50

0.2688X

0.5

5.0

6.50

0.1295X

1.0

10.0

11.50

0.0786X

* Median perchlorate exposure in U.S. population (Blount 2006a).

f From median NIS inhibitor exposures in the U.S. population (calculated in a previous section).

** Used an estimated conversion factor of 0.009375 mg/L per .0075 mg/kg-day using a linear interpolation from the

Clewell perchlorate PBPK model.

The following table provides the %TIUs calculated by the Tonacchera Model at various
external perchlorate doses and compares them against the published predicted %TIUs from the
Clewell Perchlorate PBBK Model and the %TIUs observed in the Greer study:

140

-------
10-P-0101

External
CI04" Dose
(mg/kg-day)

Calculated Total
Iodide Uptake
(TIU)

%TIU Calculated

from the
Tonacchera Model

%TIU Predicted by
the Clewell
Perchlorate
PBPK Model*

%TIU
Measured in
the

Greer Studyf

0.3675X

100

0.000175

0.3670

99.9

0.0007

0.3662X

99.6

CIO4" RfD

0.001

0.3650X

99.3

99.4

0.007

0.3560X

96.9

98.3 ±8.3
(NOEL)

0.01

0.3546X

96.5

0.02

0.3325X

93.2

83.6 ±4.1

0.1

0.2688X

73.1

55.3 ±3.9

0.5

0.1295X

35.2

32.9 ±3.8

1.0

0.0786X

21.3

* Source: Clewell 2007, p 423, table 5.
f Source: Greer 2002, p 931, table 1, 24-hr uptake.

In an adult, the Tonacchera Model calculated a 99.6 %TIU at the perchlorate RfD of
0.0007 mg/kg-day, which corresponds to a 24.5 ppb DWEL. The Tonacchera Model also
calculated a 99.9 %TIU at ]A the perchlorate RfD of 0.0000175 mg/kg-day in an adult which
corresponds to a 6.1 ppb DWEL. Therefore, the Tonacchera Model estimated a 0.3% reduction
in the TIU when the DWEL is lowered from 24.5 ppb to 6.1 ppb.

The %TIUs calculated from the Tonacchera Model for an external perchlorate dose for
0.001, 0.01, 0.1, and 1.0 mg/kg-day are in excellent agreement with the %TIU predicted by the
Clewell perchlorate PBPK model (see table above). This is expected because the %TIU
calculation from the Tonacchera Model uses the Clewell perchlorate PBPK model to convert an
external perchlorate dose to an internal perchlorate dose. Although the %TIUs calculated from
the Tonacchera Model agree with the %TIUs predicted by the Clewell perchlorate PBPK model,
this corroboration of results is not a robust test of the Tonacchera Model because only one
stressor (i.e., perchlorate) is varied, while the other three stressors (i.e., lack of iodide,
thiocyanate, and nitrate) are held constant.

In order to understand the potential sources of differences between the two sets of %TIU
values, a reexamination of the Greer study data with consideration to all four NIS stressors is
appropriate. The Greer study calculated a 24-hour TIU from the difference in the amount of
RAIU from before and after 14 days of a known external perchlorate exposure level. However,
the Greer study did not attempt to measure or control for the other three known NIS stressors
(i.e., lack of iodide, thiocyanate, and nitrate). Therefore, the other three known NIS stressors can
act as confounding variables in the Greer study.

141

-------
10-P-0101

Although the exposures to the other three NIS stressors were not directly measured in the
Greer study, the amount of change in an individual's TIU from the other three known NIS
stressors can be observed in the 24-hour RAIU results (Greer 2002, Figure 2, provided below).
If the contribution from the other three NIS stressors was negligible, the measured 24-hour
RAIU should be the same as before exposure (i.e., baseline visit (BV)) and as 15 days post
exposure (i.e., PI 5) when there was no additional external perchlorate dose. At each of the four
external perchlorate doses, the average raw 24-hour % RAIUs before exposure (i.e., BV) and 15
days post exposure (i.e., P15) was statistically similar (Greer 2002, table 1, 24-hour uptake
column). However, the raw 24-hour RAIUs before exposure (i.e., BV) and 15 days post
exposure (i.e., PI 5) are far from similar for some of the individual test subjects (Greer 2002,
figure 2). In figure 2 in the Greer study, if the contribution from the other three NIS stressors
was negligible to an individual's TIU, the graphs should be mirror images around exposure day
14 (E14) and the graphs should not have any crossing lines. However, this is not shown in figure
2 of the Greer study.

¦—

CT3

T3
£

f-J

PIS

P15

<—

<=s-

CNI

E14

P15

T3
O)

E14

Figure 2. The 24-hr RAIU at the baseline visit and on E14 and PI5 for each subject in the
(CI 0.02-, and (£J) 0.007-mg/kg-day dose groups.

Source: Greer 2002, figure 2.

P15

0.5-, {f?) Q.1-,

The following table summarizes some of the larger %TIU changes observed in several of
the individual test subjects that occurred in the 29 days between the BV and the post 15 days
exposure (PI5):

142

-------
10-P-0101

External CI04"

Dose
(mg/kg-day)

Example
Test Subject

Estimated Raw
RAIU at Baseline
Visit*

Estimated Raw
RAIU at Post 15

Days*

Individual %TIU
change over the
29 days

0.007 (fig. 2 d)

Purple

12% decrease

0.007 (fig. 2 d)

Blue '0'

40% increase

0.02 (fig. 2 c)

Purple '0'

22% decrease

0.02 (fig. 2 c)

Blue '0'

29% increase

0.1 (fig. 2 b)

Black

70% increase

0.1 (fig. 2 b)

Red 'o'

26% decrease

0.5 (fig. 2 a)

Red 'o'

36% decrease

0.5 (fig. 2 a)

Purple'0'

67% increase

* The raw numerical RAIU data for individuals were not provided in the published Greer study, but were estimated
from Greer figure 2 (provide above).

The point of identifying these large changes in %TIU of individual subjects in the Greer
study is to highlight that a single chemical risk assessment (i.e., perchlorate only) can seriously
distort the perception of risk from perchlorate when the other NIS stressors are not considered.
The Greer study is designed to quantify the amount of perchlorate exposure that causes no
statistically significant change in the RAIU between before perchlorate exposure and during
perchlorate exposure, to identify the NOEL for perchlorate. Then, a UF of 10 was applied by the
NAS Committee to derive the perchlorate RfD to ensure perchlorate exposure at the RfD does
not cause a significant change (i.e., < 0.2 %TIU) in an individual's TIU to be protective.
However, these large changes in %TIU of individual subjects in the Greer study are occurring at
each exposure dose without any apparent adverse thyroid effects. A perchlorate-only chemical
risk assessment does not attempt to explain how these large changes in %TIU are occurring in
the population and does not attempt to identify the source(s) of the effect. If the perchlorate-only
chemical risk assessment considers a 1.7 %TIU to be potentially risky, why should the potential
risk inherent in double-digit percentage changes in the %TIU observed in the Greer study
population be ignored, overlooked, or discounted?

Fortunately, a cumulative risk assessment approach incorporates the risk from the
background exposure to all four NIS stressors because the exclusion of one or more stressor(s) or
the background exposure from a stressor(s) will seriously distort the portion of the total
estimated risk attributed to perchlorate (i.e., the principal stressor of concern) (EPA 2003,
appendix C). Furthermore, a cumulative risk assessment distinguishes between background
exposures and source-specific exposures to provide a more complete picture of both total and
source-related risks (EPA 2003, appendix C). A cumulative risk assessment identifies the
amount of risk from the exposure levels observed for each stressor. In regard to this public
health issue, the Tonacchera Model identifies the lack of iodide in the diet as the dominant
stressor. When the contribution of iodide is not quantitatively factored into the risk assessment
(i.e., as in a single chemical assessments done for perchlorate), the result is seriously distorted.

The Greer study measured the RAIU in 37 test subjects at the start of the study and
named this RAIU measurement the BV. The observed raw 24-hour RAIU at the BV was 9.8%
to 33.7%. Therefore, prior to perchlorate exposure, the 24-hour RAIU varied by a factor of
about 3.4 times. The BV results capture the range in %TIUs found in the study's population.

143

-------
10-P-0101

The median 24-hour RAIU at the BV is estimated to be 21.75%. Assigning this as the typical
RAIU in the population, the range in %TIUs in the study's population is then estimated to range
from 45% to 155% of normal (i.e., 100% is nominal). The Greer BV data show that the TIU is
varying ± 55% in the study population prior to perchlorate exposure without adversely effecting
thyroid hormone levels. The fT4, tT4, tT3, and TSH levels were measured at BV and were all
found to be in the normal range prior to and after the experiment (Greer 2002, table 4).

To manage the risk affecting this public health issue, the 45-155% range in %TIUs of the
Greer study's population needs to be understood and explained. The Greer study does not
explain this observation. However, the Tonacchera Model provides insight into this observation
by quantifying the TIU contribution from each of the stressors. For example, if all the variations
in %TIUs in the Greer study population were from variations in the iodide nutritional status, the
Greer study population would be expected to have a UIC in the range of 65 |ig/L and 225 |ig/L.
Since the NHANE III survey identified the 10th to the 90th percentile UIC in the U.S. population
to be from 45 |ig/L to 386 |ig/L, the %TIU range observed in the Greer study population can be
easily accounted for by the expected variation in iodide nutritional status found in the U.S.
population. Furthermore, this range in TIU indicates that variation in an individual's iodide
exposure during the experiment represents a large uncontrolled confounding variable in the
measurement of the amount of NIS inhibition from perchlorate in the Greer study.

The %TIUs calculated from the Tonacchera Model for an external perchlorate dose for
0.007 and 0.5 mg/kg-day are in excellent agreement with the %TIU observed in the Greer study
(see table above). However, the %TIUs calculated from the Tonacchera Model for an external
perchlorate dose for 0.02 and 0.1 mg/kg-day are not in particularly good agreement with the
%TIU observed in the Greer study. Agreement in the 0.5 mg/kg-day dose makes sense because
the administered perchlorate dose represents a 333% increase in the NIS inhibition load on the
body (see table below). Any unbalanced variation in the background exposure levels from the
other NIS inhibitions (nitrate and thiocyanate) will have a smaller effect on the observed %TIU.
By contrast, in the other three perchlorate dose levels (i.e., 0.007, 0.02, 0.1 mg/kg-day), the
administered perchlorate only represents a minor or small fraction of the total NIS inhibition load
on the body. Any unbalance variation in the background exposure levels from the other NIS
inhibitions, nitrate and thiocyanate, among the test subjects will have a much larger effect on the
observed %TIU. And, at all perchlorate exposure levels, the potential unbalanced variation in
the iodide intake (i.e., the dominant NIS stressor) among the test subjects has largest potential to
confound the observed %TIU.

External
CI04" Dose
(mg/kg-day)

Internal CI04~
Dose in an Adult
(|jmol/L serum)

Estimated Total SPEC
in the Greer Study
(|jmol/L)*

% of Estimated

Total SPEC
Contributed by
CIO4" Dose

Estimated Total
SPEC as a % of
Normal SPEC

0.007

0.014

1.511

0.9%

101%

0.02

0.2

1.70

11.7%

113%

0.1

1.0

2.50

40%

166%

0.5

5.0

6.50

76%

433%
(i.e., a 333% inc. over
normal SPEC)

* Background NIS inhibitor exposure in the U.S. population is estimated at 1.501 pmol/L SPEC (see Section 7.2.1).

144

-------
10-P-0101

According to the Tonacchera Model, these potential confounding factors in the Greer
study do not invalidate the identification of the NOEL. This is because the Tonacchera Model
agrees with the Greer study that a perchlorate dose of 0.007 mg/kg-day produces a minimal
change in the %TIU. The Tonacchera Models estimates a 96.9 %TIU at a perchlorate dose of
0.007 mg/kg-day. The Greer study measured a 98.3 ± 8.3 %TIU at a perchlorate dose of 0.007
mg/kg-day. The measured 1.7% reduction in the TIU in the Greer study (Greer 2002, table 1) is
not statistically different than a 0.0% reduction in the TIU.

In order to isolate the contribution of perchlorate NIS inhibition on the %TIU, the other
NIS stressors must also be measured through the course of the experiment. In this experimental
design, any unbalance variation in the background exposure to the other NIS stressor could be
mathematically factored out to isolate the only the contribution of perchlorate NIS inhibition on
the %TIU. Fortunately, the Braverman perchlorate occupational cohort study took
measurements of all four NIS stressors before and during exposure. This will allow the variation
in the other NIS stressors to be identified and factored out the confounding variables, if
necessary. This will be demonstrated in the next section.

9.1.5 Braverman Perchlorate Occupational Exposure Study

In the Braverman perchlorate occupational exposure study, workers at an ammonium
perchlorate production facility in Cedar City, Utah, had the RAIU and all four NIS stressors
measured preexposure and during exposure (Braverman 2005). The study also measured the
following thyroid parameters: T4, fT4, T3, TSH, and thyroglobulin (Tg). Measurement of all
four NIS stressors allows the full application of the Tonacchera Model to the study results.

In the study, the exposed workers had their serum measured for thiocyanate, nitrate, and
perchlorate concentrations preexposure and during exposure. During exposure, the exposed
workers' mean serum concentrations were 3487.8 |ig/L of SCN", 7926.8 |ig/L of NCV, and 838.4
|ig/L of CIO4". Since the Tonacchera Model uses units of |imol/L, the corresponding serum
concentrations are 60.0 |imol/L of SCN", 127.8 |imol/L of NO3", and 8.43 |imol/L of CIO4".

Using these values, the SPEC(Dudng exposure) of the exposed workers can be calculated as follows:

SPEC(During exposure) [CKV] + [N03"] /240 + [free SCN"] /15

SPEC(During exposure) = 8.43 |imol/L + (127.8 |imol/L ^ 240) + (60.0 |imol/L x 0.5 free SCN" 15)

SPEC(Duringexposure) = 8.43 |imol/L + 0.53 |imol/L + 2.00 |imol/L

SPEC(During exposure) — 10.96 |ilTlol/L

The exposed workers were measured to have a median urinary iodide to creatinine ratio
of 174 |ig I/g creatinine and an average age of 33.6 years (Braverman 2005, table 3). The
NHANES III survey identified the median urinary iodide to creatinine ratio of 30- to 39-year-
olds to be 108.3 |ig I/g creatinine (i.e., the normal iodide excretion level for this age group)

(Hollowell 1998, table 2). Therefore, the perchlorate-exposed workers had a 1.61 times greater
urinary excretion than the median urinary excretion of the 30- to 39-year-olds in the NHANES
III survey. The following calculation determines the exposed workers %TIU(During exposure) 36%
using the TIU(normai) of 0.3675x as the baseline:

145

-------
10-P-0101

Inserting 10.96 |imol/L for SPEC(During exposure) and 1.61x for the [I"](During exposure) into the
Tonacchera Model gives:

TIU(During exposure) II rrifF. xposed workers)

(1-22 SPEC(Exposed workers))

TIU(During exposure) II 1.61x

(1.22+ 10.96 |imol/L)

TIU(During exposure) 11 0.1322x

The %TIU(During exposure) in the exposed is calculated using the following equation:

/OTIU(During exposure)

- (TIU(During exposure) '

TIU(Normal)) X 100%

%TIU(During exposure) = (0.1322X - 0.3675x) X 100%

%TIU(During exposure) — 36.0 %

The Braverman study reports a mean RAIU(Duringexposure) of 13.5 % ± 7.1% and a mean
RAIU(Preexposure) of 21.5 % ± 8.2%). The study determined that the perchlorate exposure in the
workers resulted in an average 38%> decrease in RAIU (i.e., 38%> NIS inhibition) during exposure
as compared to the workers' preexposure baseline. The corresponding %>TIU for comparison is
calculated as: 1 - (13.5%> 21.5%) x 100% = 62.8%. At first glance, the Tonacchera Model
calculated %oTIU(During exposure) of 36.0 % does not agree with the measured %TIU in the study of
62.8%). However, these two %>TIU values cannot be compared directly because of the following
two reasons:

• The %>TIU values are calculated from different TIU baselines. The Tonacchera
Model used a TIU(Normai) of 0.3675x as the baseline, which was calculated from the
typical U.S. exposure to the four NIS stressors. The %>TIU from the Braverman study
used the TIU from the workers' preexposure as the baseline. The TIU(preexPosure) is
calculated as 0.3370x, which represents a higher NIS stress level than the typical NIS
stress level of 0.3675x. Since all four NIS stressors were measured, the TIU(preexPosure)
can be calculated and the raw RAIU(preexposure) value can be normalized to the NIS
stressor condition at the TIU(Normai) for the comparison.

• Since the TIU(Dudng exposure) is the integrated effect from combined exposure to all four
NIS stressors, to measure only the NIS inhibition from perchlorate requires the
exposure from the other three NIS stressors to remain constant. However, the
TIU(During exposure) is confounded by a substantial change in the median urinary iodide
to creatinine ratio between during exposure and during preexposure (i.e., the iodide
stressor will be shown to be the dominant stressor in this public health issue). The
median urinary iodide to creatinine ratio was 174 |ig I/g creatinine during exposure
and 133 jug I/g creatinine at preexposure. By contrast, the mean thiocyanate serum
concentrations during exposure and preexposure are essentially the same at 7926.8

146

-------
10-P-0101

|ig/L and 7638.6 |ig/L, respectively (i.e., 2.00 |imol/L SPEC vs 1.90 |imol/L SPEC).
Likewise, the mean nitrate serum concentrations during exposure and preexposure are
essentially the same at 3487.8 |ig/L and 3304.0 |ig/L, respectively (i.e., 0.53 |imol/L
SPEC vs 0.51 |imol/L SPEC). Since all four NIS stressors were measured, the
difference in urinary iodide between preexposure and during exposure can be factored
out of the TIU(During exposure)- Since all four NIS stressors were measured, the
TlU(Preexposure) can be calculated and the raw RAIU(preexPosure) value can be normalized
to the TIU(Normai) for the comparison.

The TIU(preexposure) can be calculated, but the SPEC(preexposure) must be calculated first.

During exposure, the exposed workers' mean serum concentrations during the preexposure were
measured to be 3304.0 |ig/L of SCN", 7638.6 |ig/L of N03", and 2.0 |ig/L of C104". Since the
Tonacchera Model uses units of |imol/L, the corresponding serum concentrations are 56.9
|imol/L of SCN", 123.2 |imol/L of N03", and 0.02 |imol/L of C104". Using these values, the
SPEC(Preexposure) of the exposed workers can be calculated as follows:

SPEC(Preexposure) = [C104"] + [N03"] /240 + [free SCN"] /15

SPEC(preexposure) = 0.02 |imol/L + (123.2 |imol/L ^ 240) + (56.9 |imol/L x 0.5 free SCN" -M5)

SPEC(Preexposure) = 0.02 |imol/L + 0.51 |imol/L + 1.90 |imol/L

SPEC(preeXposure)— 2.43 |imol/L

The exposed workers at preexposure were measured to have a median urinary iodide to
creatinine ratio of 133 |ig I/g creatinine and an average age of 33.6 years old (Braverman 2005,
table 3). The NHANES III survey identified the median urinary iodide to creatinine ratio of 30-
to 39-year-olds to be 108.3 |ig I/g creatinine (i.e., the normal iodide excretion level for this age
group) (Hollowell 1998, table 2). Therefore, the exposed workers at preexposure had a 1.23
times greater urinary excretion than the median urinary excretion of the 30- to 39-year-olds in
the NHANES III survey. The following calculation determines the exposed workers'

/oTIU(preeXposure) •

Inserting 2.43 |amol/L for SPEC(preexposure) and 1.23x for the [^preexposure) into the

Tonacchera Model gives:

TlU(Preexposure) 11 [I 1 ("Preexposure'I

(1.22 + SP EC(Preexposure))

TlU(Preexposure) 11 1 -23x

(1.22 + 2.43 (imol/L)

TlU(Preexposure) || 0.3370X

The TlU(Preexposure) of 0.3370x is less than the TIU(Normai) of 0.3675x. Therefore, the
TlU(Preexposure) value is only 0.917 of TIU(Normai) (i.e., 0.3370x 0.3675x). In other words, the
four NIS stressors at preexposure is a slightly more stressful NIS condition than the typical
exposure to the four NIS stressors found in the U.S. population. The raw RAIU(preexPosure) of
21.5% is multiplied by a factor of 1.09 (i.e., 1/0.917) to estimate what the RAIU(preexPosure) would

147

-------
10-P-0101

have been if preexposure was under the typical NIS stressor exposure conditions found in the
U.S. population (i.e., a UIC of 145 |ig/L and a SPEC(Normai) of 1.501 (amol/L SPEC). Thus, to
normalize the raw RAIU(preexPosure) for a common baseline of comparison to the Tonacchera
Model, the normalized RAIU(preexposure) is calculated as follows:

normalized RArU(preeXp0Sure) raw RAIU(preeXp0Sure) x (1/0.917)
normalized RAIU(preexPosure) = 21.5% x (1/0.917)
normalized RAIU(preexposure) = 23.4 %

The difference in urinary iodide between preexposure and during exposure can be
factored out of the raw RAIU(During exposure)- The exposed workers were measured to have a
median urinary iodide to creatinine ratio of 174 |ig I/g creatinine. The median urinary iodide to
creatinine ratio was 133 jug I/g creatinine at the preexposure. Therefore, the median urinary
iodide to creatinine ratio during exposure was 1.30 times greater than at preexposure. The raw
RAIU(During exposure) must be reduced by a factor of 0.764 to factor out the increased TIU
contributed by the higher intake of iodide during exposure. Therefore, the adjusted RAIU(During
exposure) represents the RAIU caused by only the increase in perchlorate exposure, while the other
three NIS stressors are either essentially the same (i.e., thiocyanate and nitrate) or factored out
(i.e., as in iodide). The adjusted RAIU(During exposure) is calculated as follows:

adjusted RAIU(During exposure) — raw RAIU(preexposure) X (1/1.30)
adjusted RAIU(DUringexposure) 13.5/o x 0.764
adjusted RAIU(During exposure) — 10.3%

Therefore, the normalized %TIU(During exposure) for comparison to the Tonacchera Model
calculated %>TIU(DuringExposure) of 36.0 % is calculated as follows:

Normalized %TIU(During exposure) — adj. RAIU(During exposure) ~ normalized RAIU(preeXp0Sure)x 100%
Normalized %TIU(Duringexposure) = (10.3%) 23.4%) x 100%)

Normalized %oTIU(Duringexposure) = 44%) (i.e., a 56%> NIS inhibition)

After factoring out the differences in baselines and iodide exposures, the normalized
%oTIU(Dudng exposure) of 44% compares favorable with the Tonacchera Model calculated
% TIU (During exposure) of 36.0 %>. The standard deviations in the raw RAIU measurements in the
Braverman study were greater than 38%> of the measured RAIU value for both during exposure
and preexposure. This means the difference in the normalized %>TIU(During exposure) and the
Tonacchera Model calculated %>TIU(During exposure) is well within the statistical variation of the
RAIU measurement.

9.1.6 Qualitative Risk Assessment of NIS Stressors

In 2003, EPA's l'ramework for Cumulative Risk Assessment identified that a cumulative
risk assessment considers the joint impact of multiple stressors. In Section 3.2.2.1, the
Framework for Cumulative Risk Assessment states, "Studies on individual stressors can,
however, provide informative qualitative information for multistressor assessments, particularly
regarding hazard identification" (EPA 2003, p 44). Since adverse health effects have not been

148

-------
10-P-0101

clearly demonstrated in any human population exposed to perchlorate (NAS 2005, p 177), the
potential human response to NIS inhibition from perchlorate exposure can be studied by
qualitatively evaluating the human response to exposure to the other NIS inhibitors, such as
thiocyanate. Therefore, the human biological response to the NIS inhibition from thiocyanate
provides insight into the potential human response to other NIS inhibitors, such as perchlorate.

Insight into the Toxicity of NIS Inhibition through the Qualitative Review of
Endemic Cretinism

Endemic cretinism is the irreversible severe impairment of intellectual and physical
development (Delange 2005a, p 731). In both neurologic cretinism and myxedematous
cretinism, cretins are severely mentally retarded. Endemic cretinism occurs in 5-15% of severe
goiter populations. Endemic cretinism constitutes the extreme expression of the spectrum of
abnormalities in physical and intellectual development from iodide deficiency (i.e., a low
maternal %TIU) during gestation. Iodide deficiency is the principal factor in the etiology of
endemic cretinism. The degree of iodide deficiency correlates with the frequency of cretinism.
Since the level of iodide deficiency is an indirect measure of TIU level, the degree of maternal
TIU correlates with the frequency of cretinism. A low TIU during pregnancy is the same
mechanism of toxicity as perchlorate. Thus, endemic cretinism occurs through the same
mechanism of toxicity as perchlorate and represents a case study with which to qualitatively
evaluate the factors influencing the toxicity of NIS inhibition. The main difference between
endemic cretinism and the public risk from excess NIS stressor exposure in the United States is
that both the severity of the maternal %TIU and the severity of the resulting mental deficits are
less in the United States.

Iodide deficiency is the main factor in the etiology of endemic cretinism (Delange 2005a,
p 734). Delange makes this conclusion based on the following:

1. The degree of iodide deficiency correlates with the frequency of cretinism

2. The prophylactic action of iodide in preventing endemic cretinism (i.e., correcting iodide
deficiency).

3. Cretinism has emerged in previously unaffected populations in the Jimi Valley of New
Guinea, due to the recent onset of iodide deficiency when the local, iodide-rich rock salt
was replaced with low-iodide industrial salt.

Naturally occurring goitrogens in the diet are also a factor in the etiology of endemic
cretinism (Delange 2005a, p 735). Cassava is a staple food in tropical areas. Cassava contains
cyanogenic glucosides. Upon ingestion, the cyanogenic glucosides in cassava are hydrolyzed
into cyanide, which is subsequently metabolized by the body into thiocyanate. Thiocyanate is a
potent NIS inhibitor. However, dietary intake of thiocyanate from properly prepared cassava is
not sufficient to induce endemic cretinism alone (i.e., NIS inhibition from thiocyanate is not as
strong a factor as iodide deficiency), but thiocyanate NIS inhibition works in conjunction with
iodide deficiency to induce endemic cretinism. In other words, the role of thiocyanate on
endemic cretinism is entirely due to the aggravation of iodide deficiency (Delange 2005a, p 735).
The toxicity of thiocyanate is not fixed, but the critical threshold of inducing endemic cretinism
from thiocyanate NIS inhibition is related to the dietary intake of iodide (i.e., as measured by

149

-------
10-P-0101

UIC). Goiter develops when the UlC-to-thiocyanate ratio falls below the critical threshold of
about 3 |ig of iodide per mg of dietary thiocyanate (Delange 2005, p 267). Therefore, the
observed toxicity of thiocyanate NIS inhibition is not fixed, but is related to the dietary intake
level of iodide.

The lack of selenium is an additional factor in the etiology of endemic cretinism.
However, selenium deficiency works through a different mechanism of toxicity than goitrogens.
Since selenium deficiency is believed to be extremely rare in the U.S. population and works
through a different mechanism of toxicity, this factor is not directly relevant to the perchlorate
public health issue and, therefore, is not discussed here.

The role of the NIS inhibition from thiocyanate on endemic cretinism is consistent with
the risk characterization observed in the OIG's cumulative risk assessment of NIS inhibition on
this public health issue. Endemic cretinism is the extreme example of abnormalities in physical
and intellectual development from extremely low maternal TIU. However, since the mechanism
of toxicity of cretinism is the same as that of NIS inhibition from perchlorate, the observed role
of the risk factors in endemic cretinism can be qualitatively applied to the perchlorate public
health issue. Specifically, the following observations from the OIG Analysis are consistent with
the observed role of the risk factors in endemic cretinism:

• The OIG Analysis identifies the lack of iodide stressor (i.e., iodide deficiency) as the
dominant NIS stressor in inducing a low maternal TIU resulting in increased risk of
subtle mental deficits in the children. This risk characterization is consistent with the
dominant role that iodide deficiency has in endemic cretinism.

• The OIG Analysis identifies that the thiocyanate NIS stressor has a smaller impact on
lowering TIU than does the lack of iodide stressor. This risk characterization is
consistent with the observation that thiocyanate NIS inhibition from cassava
consumption is not sufficient by itself to induce endemic cretinism, which makes
thiocyanate NIS inhibition only a contributing factor on the etiology of endemic
cretinism.

• The OIG Analysis identifies that the amount of NIS inhibition from the typical range of
thiocyanate exposure in the United States in iodide-sufficient individuals (i.e., UIC
greater or equal to 145 |ig/L) is not sufficient to lower maternal TIU enough to increase
the risk of subtle mental deficits in her child. However, the amount of NIS inhibition
from the typical range of thiocyanate exposure in the United States, in conjunction with
iodide deficiency (i.e., UIC less than or equal to 100 |ig/L), is sufficient to lower
maternal TIU enough to increase the risk of subtle mental deficits in her child. This risk
characterization is consistent with and parallels the observation that thiocyanate NIS
inhibition from cassava consumption is not sufficient by itself to induce endemic
cretinism, but works in conjunction with iodide deficiency to induce endemic cretinism.

• The OIG Analysis identifies that the toxicity of NIS inhibition is not fixed, but the
body's ability to tolerate NIS inhibition is directly related to the iodide nutrition level.
This risk characterization is consistent with the clinical observations in endemic goiter

150

-------
10-P-0101

regions where the toxicity of thiocyanate NIS inhibition from cassava consumption is not
fixed, but has a critical threshold ratio related to the dietary iodide intake level (i.e., 3 |ig
of dietary iodide per mg of thiocyanate).

• The OIG Analysis identifies that the lack of iodide (i.e., iodide deficiency) during
pregnancy (i.e., UIC less than or equal to 100 |ig/L) is the major factor increasing the
risk of permanent subtle mental deficits in children. The OIG Analysis identified that
the most effective and efficient approach to correcting this public health issue of low
maternal TIU during pregnancy and nursing is to add iodide to all prenatal vitamins and
to ingest them before and during pregnancy and nursing. This risk characterization is
consistent with the clinical observations in endemic goiter populations where endemic
cretinism is prevented by correcting iodide deficiency through the injection of iodized oil
(Delange 2005a, p 741-42), (i.e., correcting low maternal TIU).

Therefore, the qualitative evaluation of the roles of iodide deficiency and NIS inhibition in the etiology
of endemic cretinism are consistent with the cumulative risk characterization of this public health issue
posed by the potential for inducing low maternal TIU from excessive NIS inhibitor exposure.

151

-------
10-P-0101

9.2 Assessing the Contribution of Each NIS Stressor on the Till

9.2.1 Evaluation of the Amount of NIS Inhibition Contributed by Each NIS
Inhibitor

A critical aspect of a cumulative risk assessment is to incorporate the risk from the
background exposure to the stressors because the exclusion of a stressor(s) or the background
exposure from a stressor(s) "may seriously distort" the portion of the total estimated risk
attributed to the principal stressor of concern (EPA 2003, appendix C). Cumulative risk
assessment distinguishes between background exposures and source-specific exposures to
provide a more complete picture of both total and source-related risks (EPA 2003, appendix C).
As shown by the Tonacchera experiment used to derive the Tonacchera Model, the three NIS
inhibitors, thiocyanate, nitrate, and perchlorate, act together to inhibit the uptake of iodide by the
NIS (Tonacchera 2004). The inhibition of the NIS by thiocyanate, nitrate, and perchlorate is
"indistinguishable" from a concentration or dilution of each other (Tonacchera 2004). Therefore,
the NIS inhibition contribution by perchlorate exposure has to be evaluated in context to the total
body's burden of NIS inhibition.

To address this aspect of a cumulative risk assessment, the typical background exposure
to thiocyanate and nitrate was estimated in the U.S. population to evaluate the amount of the
inhibition of the NIS of each NIS inhibitor at various perchlorate exposures. As calculated
previously in this document, the following table summarizes the results:

Relative Contribution of Each NIS Inhibitor
To the Total Amount of NIS Inhibition in an Adult

Perchlorate
Exposure Level

Total Amount of NIS
Inhibition in an Adult
(SPEC)

Perchlorate

Nitrate
at

(40 |jmoles/L)

Thiocyanate
at

(40 |jmoles/L)

At the perchlorate NOEL
(0.007 mg/kg-day)

1.567 jjmoles/L

4.5 %

10.7%

85.1%

At the perchlorate RfD of
(0.0007 mg/kg-day)

1.511 jjmoles/L

0.93%

11.1%

88.2%

95th percentile of U.S.
population (Blount 2006)

1.505 jjmoles/L

0.31%

11.1%

88.6%

50th percentile of U.S.
population (Blount 2006)

1.501 jjmoles/L

0.09%

11.1%

88.8%

Zero perchlorate

1.500 jjmoles/L

0.00%

11.1%

88.9%

This table clearly shows that when the amount of NIS inhibition from perchlorate
exposure is evaluated in context to the typical amount of NIS inhibition contributed from the
routine exposure to thiocyanate and nitrate, the amount of NIS inhibition from perchlorate
exposure is relatively small. At the perchlorate RfD of 0.007 mg/kg-day in an adult, the amount
of NIS inhibition from perchlorate exposure is less than 1%. Therefore, as identified in EPA
guidance, the exclusion of the background exposure to thiocyanate and nitrate in single chemical
risk assessment "seriously distorts" the portion of the total estimated risk attributed to the
principal stressor of concern, perchlorate (EPA 2003, appendix C).

152

-------
10-P-0101

As indicated by the Tonacchera Model, perchlorate is only one of four stressors that
contribute to the amount of iodide taken up by the NIS. A thorough cumulative risk assessment
identifies the exposure levels of the stressors that result in adverse effects. For this public health
issue, this means that this cumulative risk assessment must identify the level of TIU in pregnant
women that leads to fetal brain damage. Once the critical level of TIU is identified, then each of
the four NIS stressors must be managed to prevent the critical TIU level from being exceeded.

9.2.2 Evaluation of %TIU as a Function of Total NIS Inhibition Load

The amount of iodide taken up by the thyroid is the integrated effect from the exposure of
all four NIS stressors. However, the three NIS stressors (thiocyanate, nitrate, and perchlorate)
are "indistinguishable" from a concentration or dilution of each other (Tonacchera 2004).
Therefore, their combined NIS inhibitor exposure can be expressed as a single variable reflecting
the total NIS inhibition load on the body (i.e., SPEC), which is measured in |imol/L SPEC.

Thus, the Tonacchera Model is simplified into a two-variable equation. This simplification
allows the TIU to be expressed as a function of the body's total NIS inhibition load (i.e., SPEC)
at a given level of iodide nutrition (e.g., normal iodide intake level). Therefore, to examine this
impact of SPEC on TIU, the iodide is held constant, but the SPEC is allowed to change. The
following is an example calculation:

The Tonacchera Model stated as a two-variable equation:

TIU || [T]/(1.22 +(SPEC))

If the [T] is held constant (i.e., at a stable iodide nutritional level - "x"), the total NIS
inhibition exposure level (i.e., SPEC) can be calculated for any given TIU level by using the
following equation:

TIU(Normai) x fraction of TIU 11 X

(1.22 + SPEC(%tiu))

Given TIU(Normai) in the U.S. population for an adult is 0.3675x, then solving for

SPEC(°/oTiu) becomes

SPEC(% tiu) 11 x - 1.22

0.3675x x (fraction of TIU)

For a TIU level of 25% of normal, the SPEC(25%tiu) becomes

SPEC(25%tiu) || x - 1.22

0.3675x x (0.25)

SPEC(25%tiu) || x - 1.22

0.091875x

SPEC(25%tiu) || 10.88 - 1.22

153

-------
10-P-0101

SPEC(25%tiu) || 9.66

This calculated value estimates that a total goitrogen load of 9.66 |imol/L SPEC at a
typical U.S. iodide nutritional level will induce the TIU to be lowered to 25% of the normal TIU.
Since the SPEC(Median) is 1.501 |imol/L, an increased NIS inhibitor exposure amount of 8.16
|imol/L SPEC, or 544% increase of normal TIU, is needed to lower normal TIU to 25%. Using
the Tonacchera Model to calculate the values, the following table provides the SPEC(ooTiu) for an
adult at several %TIU levels:

%TIU Level
of Normal

Iodide
Nutrition

(%)
of Normal

SPEC^-fiu)

(|jmol/L)

Additional SPEC
Exposure Above U.S.
Normal* to Achieve TIU
Level
(|jmol/L)

Percent Increase
in Additional NIS Inhibitor
Load Needed to Generate
Specified TIU Level
(%)

100

16.92

15.41

1027

100

12.38

10.88

725

100

9.66

8.16

544

100

6.55

5.05

337

100

4.22

2.72

181

100

3.32

1.81

121

100

2.42

0.92

100

1.80

0.30

100

1.64

0.14

9.3

100

1.528

0.027

1.8

100
(normal TIU)

100

1.501

110

100

1.25

-0.25

-16.7

* U.S. normal SPEC is 1.501 pmol/L (see calculation in Section 7.2.1).
Source: OIG Analysis; Tonacchera Model calculations.

This modeling of %TIU as a function of the body's total NIS inhibition load at normal
iodide intake shows that the thyroid can tolerate a lot of exposure to NIS inhibitors before
changes in thyroid hormone production. As observed in the adult males and females in the
Banerjee thiocyanate studies, hypothyroxinemia is not expected to be observed until %TIU falls
at or below 22.7% (Banerjee 1997; Banerjee 1997b). Therefore, in an adult with a normal iodide
intake, the thyroid can tolerate up to an additional 8.16 |imol/L SPEC of NIS inhibition above
the background NIS inhibition load of 1.501 |imol/L SPEC before adult thyroid is unable to
maintain T4 production. This represents a 544% increase in the body's normal total goitrogen
load. Only under rare exposure conditions to NIS inhibitors does the total goitrogen load
increase to this point in the Western world (e.g., from occupational exposures or from
medicines). By comparison, the amount of NIS inhibition contributed by perchlorate at the RfD
in an adult is only 0.014 |imol/L SPEC (i.e., a 0.9% increase of the normal goitrogen load on the
body).

The modeling of %TIU as a function of Total NIS Inhibition Load at a normal iodide
intake is graphically presented below:

154

-------
10-P-0101

20.0-

Tonacchera Model's Calculation of
%TIU as a Function of Total NIS Inhibiton Load
At a Normal Iodide Nutritional Level

C
O

15.0-

J-I o
5 E

10.0-

TO
"o

LU
Cl

5.0-

0.0-

%TIU Associated with Adverse Effects in Non-pregnant Adult Women

Upper SPEC Range for Smokers

Upper SPEC Range for Non-smokers

—-—•

* —~ • .

20% 40% 60% 80%

%TIU- (Normal is 100%)

100%

120%

Source: OIG Analysis - graphic representation of the table immediately above.

Since the range of thiocyanate and nitrate exposure in the Western world is known, the
upper range for the total NIS inhibition load in the Western world can be calculated for both
smokers and nonsmokers using the Tonacchera Model. The total NIS inhibition load is
separated into smokers and nonsmokers because the act of smoking is a significant source of
cyanide, which is subsequently metabolized into the NIS inhibitor thiocyanate. For nonsmokers,
the thiocyanate serum concentration is typically in the range of 10-70 |imol/L. By contrast, for
smokers, the thiocyanate serum concentration is higher and is typically in the range of 80-120
(amol/L. In the Western world, the typical nitrate serum concentration ranges from 10-140
|imol/L (Tonacchera 2004). Therefore, the upper range of the total NIS inhibitor load in a
nonsmoker is then calculated usingl40 |imol/L for serum nitrate, the70 |imol/L for serum
thiocyanate, and 0.0047 |imol/L for perchlorate (i.e., the 95th percentile observed in the Blount
Biomonitoring Study (Blount 2006a (see Section 7.2.1)). Therefore, the SPEC(Upper range nonsmoker)
is given by the following calculation:

SPEC (Upper range nonsmoker) [C1CV] + [NCV] /240 + [free SCN] /15

SPEC (Upper range nonsmoker) = 0.0047 |ilTlol/L + (140 |ilTlol/L ^ 240) + (70 |ilTlol/L X 0.5 free SCN

SPEC (Upper range nonsmoker) = 0.0047 |ilTlol/L + 0.583 |ilTlol/L + 2.333 |ilTlol/L

SPEC (Upper range nonsmoker) — 2.92 |J.lTlol/L

A nonsmoker with a normal iodide intake exposed to NIS inhibitors at an SPEC (uPPer range
nonsmoker) of 2.92 |iinol/L (purple line in the graph above) has a calculated %TIU of 66%. This is
well above the %TIU associated with hypothyroxinemia or hypothyroidism. Therefore, at a
normal iodide nutritional level, the upper observed range for total NIS inhibition load in the
nonsmoking population does not sufficiently decrease the %TIU level to prevent the thyroid
from adapting and maintaining thyroid hormone production in the normal range.

15)

Likewise, the upper range of the total NIS inhibitor load in a smoker is then calculated
using 140 |imol/L for serum nitrate, 120 |imol/L for serum thiocyanate, and 0.0047 |imol/L for

155

-------
10-P-0101

perchlorate (i.e., the 95th percentile observed in the Blount biomonitoring study (Blount 2006a
(see Section 7.2.1)). Therefore, the SPEC(Upper range smoker) is given by the following calculation:

SPEC

(Upper range smoker) [C104"] + [N03"] /240 + [free SCN"] /15

SPEC (Upper range smoker) = 0.0047 |imol/L + (140 |imol/L ^ 240) + (120 |imol/L x 0.5 free SCN" ^ 15)

SPEC (Upper range smoker) = 0.0047 |imol/L + 0.583 |imol/L + 4.000 |imol/L

SPEC (Upper range smoker)

4.59 |imol/L

A smoker with a normal iodide intake exposed to NIS inhibitors at an SPEC (uPPer range smoker) of
4.58 (amol/L (brown line in the graph above) has a calculated %TIU of 47%. This is well above
the %TIU associated with hypothyroxinemia or hypothyroidism. Therefore, at a normal iodide
nutritional level, the upper observed range for total NIS inhibition load in the smoking
population does not sufficiently decrease the %TIU level to prevent the thyroid from adapting
and maintaining thyroid hormone production in the normal range.

To summarize, the graph of %TIU as a function of total NIS inhibition load at a normal
iodide nutrition level identifies that the cumulative effect of all three NIS inhibitors (even at the
upper range of exposure to each NIS inhibitor found in the U.S. population) does not exceed the
thyroid's ability to compensate for the stress. In other words, the largest total NIS inhibition
loads observed in the U.S. population are insufficient to lower the %TIU below the critical level
of 22.7% to result in hypothyroxemia. Therefore, this analysis identifies that the essential role
and contribution of the lack of iodide NIS stressor (i.e., the fourth NIS stressor) is to induce a
sufficiently low %TIU (i.e., 22.1% in nonpregnant women) to decrease serum T4 levels below
the normal range. The added stress from the lack of iodide NIS stressor is the critical factor
affecting this public health issue. An iodide-deficient diet is required to induce a low enough
%TIU level to result in T3 or T4 serum level below the normal range.

Perchlorate Occupational Cohort Study

In the Braverman perchlorate occupational cohort study, the ammonium perchlorate
workers (i.e., all were males) at Cedar City, Utah, facility had their serum measured for
thiocyanate, nitrate, and perchlorate (Braverman 2005). The mean serum concentrations during
exposure were 3487.8 |ig/L of SCN", 7926.8 |ig/L of NO3", and 838.4 |ig/L of CIO4". Since the
Tonacchera Model uses units of |imol/L, the corresponding serum concentrations are 60.0
|imol/L of SCN", 127.8 |imol/L of NO3", and 8.43 |imol/L of CIO4". Using these values, the
SPEC of the exposed workers can be calculated as follows:

SPEC(Exposed workers) = [C104"] + [N03"] /240 + [free SCN"] /15

SPEC(ExPosed workers) = 8.43 |imol/L + (127.8 |imol/L ^ 240) + (60.0 |imol/L x 0.5 free SCN" 15)

SPEC(ExPosedworkers) = 8.43 |imol/L + 0.53 |imol/L + 2.00 |imol/L

SPEC(Exposed workers) — 10.96 [^mol/L

In reviewing the table above, an adult with a SPEC of 10.96 and normal iodide intake
would be expected to have %TIU of about 22.3%. This %TIU is barely above the
%TIU(HyPothyroxinemia) range of 18.0%) to 21.8%) in the Banerjee adult male thiocyanate study that
induced a T3 and T4 thyroid hormone deficiency in adult males consistent with severe
hypothyroxinemia. The calcuated %>TIU of 22.3%> from a total NIS inhibition load of 10.96

156

-------
10-P-0101

(amol/L at a normal iodide nutrition level is in the range in which hypothyroxinemia would be
expected, but the thyroid hormones in the perchlorate-exposed workers are well within the
normal range. This is because the TIU integrates the exposure to all four NIS stressors (i.e., not
just the NIS inhibitors). The iodide stressor is not included in the SPEC value (i.e., the
calculation held iodide nutritional level at a normal level of 100%).

If the exposed workers had a normal or below-normal iodide intake, their %TIU would
be low enough to cause adverse thyroid hormone effects (e.g., hypothyroxinemia). However, the
exposed workers in the Braverman study were measured to have an median urinary iodide to
creatinine ratio of 174 |ig I/g creatinine and an average age of 33.6 years (Braverman 2005, table
3). The NHANES III survey identified the median urinary iodide to creatinine ratio of 30- to 39-
year-olds to be 108.3 |ig I/g creatinine (i.e., the normal iodide excretion level for this age group)
(Hollowell 1998, table 2). Therefore, the perchlorate-exposed workers had a 1.61 times greater
urinary excretion than the median urinary excretion of the 30- to 39-year-olds in the NHANES
III survey. The following calculation shows that the above-median intake of iodide by the
exposed workers increased their %TIU from a potentially adverse TIU level of 22.3% to an
asymptomatic TIU level of 36% (i.e., the increased iodide intake in the exposed workers
protected them from adverse thyroid effects):

Inserting 10.96 umoll* for SPEC(Exposed workers) £ind 1.61x for the [I ](Exposed workers) into the

Tonacchera Model gives:

TIU(Exposed workers) 11 IT Ifexposed workers'!

(1.22 + SPEC(Exposed Workers))

TIU(Exposed workers) 11 1 -61x

(1.22 + 10.96 (imol/L)

TIU(Exposed workers) || 0.1322X

The %TIU(ExPosed workers) in the exposed is calculated using the following equation:

%TIU(Exposed workers) — (TIU(Exposed Workers) ~ TIU(Normal)) X 100%

%TIU(Exposed workers) = (0.1322X H- 0.3675x) X 100%

%TIU(Exposed workers) — 36.0 %

An exposed worker with an exposure to NIS inhibitors of SPEC(EXposed worker) of 10.96 |imol/L
and a iodide intake 1.61 times greater than normal has a calculated %TIU of 36%. This is well
above the %TIU associated with hypothyroxinemia or hypothyroidism. Therefore, the exposed
workers in the Braverman study would be expected to have thyroid hormones in the normal range,
and they were found to have thyroid hormones in the normal range. In other words, the exposed
workers' increased iodide intake during exposure protected them from adverse thyroid effects.

157

-------
10-P-0101

9.2.3 Evaluation of the Role of Iodide Nutrition on the Ability to Tolerate NIS
Inhibition

The Tonacchera Model can also be used to evaluate the impact of iodide nutrition on the
amount of iodide taken up by the NIS. The three iodide nutrition levels modeled are the
fractions 0.34, 0.50, and 1.00 of normal which correspond to a UIC of 50 |ig/L, 72.5 |ig/L, and
145 |ig/L at a SPEC(normai) of 1.501 |imol/L SPEC. The UICs of 50 |ig/L, 72.5 |ig/L, and 145
|ig/L correspond to borderline moderate iodide deficiency, mild iodide deficiency, and sufficient
iodide levels. The following table calculates the SPEC(%tiu) for each of the three iodide nutrition
levels for a given %TIU level (sample calculation not shown, but similar to previous
calculations):

%TIU Level
of Normal

Iodide
Nutrition
Fraction
of Normal

SPEC^-fiu)

(|jmol/L)

Additional SPEC
Exposure
above U.S. Normalft
to Achieve TIU Level
(Mmol/L SPEC)

Increase in Additional NIS
Inhibitor Load Needed to
Generate Specified TIU Level
(%)

0.34*

8.03

6.53

435

0.34*

4.94

3.45

230

0.34*

3.41

1.90

127

22.7

0.34*

2.85

1.35

0.34*

2.48

0.98

0.34*

1.86

0.36

0.34*

1.501

0.0

0.50|

12.38

10.88

725

0.50|

7.85

6.35

423

0.50|

5.58

4.08

272

22.7

0.50|

4.77

3.27

218

0.50|

4.22

2.72

181

0.50|

3.32

1.81

121

0.50|

2.78

1.28

0.50|

1.501

0.0

1.00

16.92

15.41

1027

1.00

12.38

10.88

725

22.7

1.00

10.77

9.27

618

1.00

9.66

8.16

544

1.00

6.78

5.28

352

1.00

4.22

2.72

181

100
(normal TIU)

1.00

1.501

0.0

* A 0.34 fraction of iodide nutrition corresponds to borderline moderate iodide deficiency

(i.e., UIC of 50 |jg/L normal UIC of 145 |jg/L in an adult),
f A 0.50 fraction of iodide nutrition corresponds to mild iodide deficiency

(i.e., UIC of 72.5 |jg/L normal UIC of 145 |jg/L in an adult),
ft U.S. normal SPEC is 1.501 pmol/L (see calculation in previous section).

158

-------
10-P-0101

The table shows that as iodide nutrition decreases, the thyroid is less able to tolerate a
large NIS inhibition load. For example, to maintain a %TIU above the asymptomatic %TIU of
22.7% in an non-pregnant adult, at an iodide nutrition fraction of 1.0, 0.5, and 0.34, the body can
tolerate up to a total NIS inhibition load on the body of 6.78, 2.78, and 1.501 |imol/L SPEC,
respectively. An iodide nutrition fraction of 0.34 corresponds to borderline moderate iodide
deficiency. To summarize, the body's ability to cope with NIS inhibitors decreases rapidly as
the iodide intake decreases.

The modeling of %TIU as a function of total NIS inhibition load at three iodide intake
levels of 100%, 50%, and 34% is graphically presented below:

Tonacchera Model's Calculation of
%TIU as a Function of Total NIS Inhibition Load

At Three Iodide Nutrition Levels {i.e., 100%, 50%, and 34%)

14.0

12.0-

10.0-

8.0-

6.0-

4.0-

2.0-

0.0

20%

%TIU Associated with Adverse Effects in Non-pregnant Women
%TIU at an iodide Nutrition Level of 100%

%TIU at an Iodide Nutrition Level of 50%

%TIU at an Iodide Nutrition Level of 34%

40%

60%
%TIU

80% 100% 120%

Source: OIG Analysis - graphic representation of the table immediately above.

This graph demonstrates that the lack of iodide NIS stressor dramatically lowers the
%TIU level for any given total NIS inhibition load (i.e., shifts the curve to the left). For
example, decreasing the normal iodide nutrition level in half (i.e., from 100% to 50%) at a
constant typical total NIS inhibition load proportionally decreases the %TIU in half (i.e., from
100% to 50%). By contrast, doubling the typical total NIS inhibition load from 1.501 |imol/L
SPEC to 3.0 |imol/L SPEC at a constant normal iodide nutrition level results in a %TIU of
64.5% (i.e., a decrease of 35.5%). Therefore, for the NIS stressor exposure conditions in this
example, the lack of iodide NIS stressor has about 1.4 times the effect as the same amount of
change in the total NIS inhibition load. This clearly demonstrates that the lack of iodide NIS
stressor has a larger effect on the %TIU than all the three other NIS stressors combined.
Therefore, the lack of iodide NIS stressor is the dominant stressor affecting the public health
issue.

159

-------
10-P-0101

As seen in the Banerjee thiocyanate study in male electroplating workers (Banerjee
1997), a %TIU of about 20% is symptomatic and presents as severe hypothyroxinemia. In this
study, the low %TIU was brought on by a large total goitrogen load on the thyroid. However,
since the %TIU is a function of all four NIS stressors, a low %TIU causing the symptoms of
severe hypothyroxinemia can also be brought on by a borderline moderate iodide deficiency with
a total goitrogen load found in the upper range of the U.S. population. For example, adult males
with a 20 %TIU, at iodide nutrition levels of 1.0, 0.5, and 0.34, can tolerate total NIS inhibition
loads of up to 12.38, 5.58, and 3.41 |imol/L SPEC, respectively. Since the upper range of the
total NIS inhibition load found in the U.S. population is 4.59 |imol/L SPEC for smokers and 2.92
(amol/L SPEC for nonsmokers, a portion of only the male smoker subpopulation with borderline
moderate iodide deficiency (i.e., iodide nutrition fraction of 0.34) with the highest observed
levels of the total NIS inhibition load would exceed the tolerable limit of 3.41 (amol/L SPEC.
Therefore, a portion of the borderline moderate iodide-deficient male smokers would be
expected to show symptoms of hypothyroxinemia. By comparison, borderline moderate iodide-
deficient male nonsmokers would not be expected to show symptoms of hypothyroxinemia.
Furthermore, all the male nonsmokers and the smokers at iodide nutrition factions of 1.0 and 0.5
would be below the maximum tolerable SPEC values for their iodide nutritional status (i.e., their
%TIU > 20%).

As seen in the Banerjee thiocyanate exposure study on females (Banerjee 1997b), a
%TIU of 22.7% induced hypothyroxinemia in nonpregnant females. In this study, the low %TIU
was brought on by a large total goitrogen load on the body. However, since the %TIU is a
function of all four NIS stressors, a low %TIU causing the symptoms of hypothyroxinemia can
also be brought on by a borderline moderate iodide deficiency with a total NIS inhibition load
found in the upper range of the U.S. female population. For example, adult females with a 22.7
%TIU, at iodide nutrition levels of 1.0, 0.5, and 0.34, can tolerate total NIS inhibition loads of up
to 10.77, 4.77, and 2.85 |imol/L SPEC, respectively. Since the upper range of the total NIS
inhibition load found in the U.S. population is 4.59 |imol/L SPEC for smokers and 2.92 |imol/L
SPEC for nonsmokers, both female smokers and nonsmokers with borderline moderate iodide
deficiency (i.e., iodide nutrition fraction of 0.34) with the highest observed levels of the total NIS
inhibition load would exceed the tolerable limit of 2.85 (amol/L SPEC. Therefore, a portion of
the borderline moderate iodide-deficient female smoker and nonsmoker subpopulations would be
expected to show symptoms of hypothyroxinemia. By comparison, all the female nonsmokers
and the smokers at iodide nutrition factions of 1.0 and 0.5 would be below the maximum
tolerable SPEC values for their iodide nutritional status (i.e., their %TIU > 22.7%).

The clinical observations in areas of endemic cretinism show that the consumption of
cassava is not the cause of this health problem, but serves to aggravate the underlying problem of
severe iodide deficiency in these areas (Delange 2005a, p 735). The consumption of cassava
causes a large NIS inhibition load on the body, but is not large enough to cause health problems
on its own. Only when the consumption of cassava occurs in conjunction with a low iodide
intake do health problems occur. The above table and graph express the same finding as that
observed in the clinical observation in areas of endemic cretinism. In other words, the table and
graph state that with a healthy iodide intake, the consumption of NIS inhibitors is not large
enough to induce health problems. However, if iodide intake is low enough, the consumption of

160

-------
10-P-0101

moderate amounts of NIS inhibitors can induce health problems; but once again, the root
problem of the health issue is iodide deficiency, not the moderate consumption of NIS inhibitors.
The next section also identifies iodide intake as the dominant stressor and quantifies the impact
on this public health issue.

In conclusion, the NIS inhibitor exposure levels found in the U.S. population are not
individually sufficient to induce adverse health effects. However, combined with the added NIS
stress caused by iodide defiency, the cumulative effect of all four NIS stressors on the TIU by the
thyroid is sufficient to induce adverse effects in the U.S. population. Therefore, NIS inhibitors
are not the source of the problem, but act to modulate the toxicity of iodide deficiency.

9.2.4 Evaluation of the Relative Influence of Each NIS Stressor

A critical aspect of a cumulative risk assessment is that a stressor does not have to be the
exposure to a chemical, but the "absence of a necessity" (EPA 2003, p 2). The role of iodide
nutrition is critical to this public health issue. The Tonacchera Model can be simplified into a
two-variable equation that establishes the relationship between the amount of iodide and the
body's total NIS inhibition load, which is measured in |imol/L SPEC. The following table shows
the relationship between iodide nutrition and the body's total NIS inhibition load (a sample
calculation is not shown, but is similar to previous examples):

%TIU Level
of Normal

Iodide
Nutrition

(%)
of Normal

SPEC
(|jmol/L)

Reduction in SPEC Exposure
From U.S. Normal* to Achieve
TIU Level
(|jmol/L)

Percent Change in
Total NIS Inhibitor Load
(%)

100

1.528

0.027

1.8

1.501

0.0

100
(normal TIU)

100

1.501

0.0

100

99.4

1.486

0.015

- 1.0

100

1.474

0.027

- 1.8

101

100

1.474

0.027

- 1.8

* U.S. normal SPEC is 1.501 pmol/L (see calculation in previous section).

This table shows that the lack of iodide stressor is the dominant NIS stressor of the four
NIS stressors. A 1% decrease in the iodide nutrition level has the same effect as lowering the
body's total NIS inhibition load by 1.8%. In short, the lack of iodide stressor has about twice the
impact on the %TIU as does the impact on the %TIU of all three NIS stressors combined at their
typical exposure level. Therefore, by not being able to include the nonchemical stressor, lack of
iodide, the conventional single chemical risk assessments for perchlorate "seriously distort" the
portion of the total estimated risk attributed to the principal stressor of concern (i.e., perchlorate).
By implementing a cumulative risk assessment approach, the contribution of the lack of iodide
stressor is quantitatively factored into the total risk, and the portion of total risk attributed to
perchlorate and each of the other three NIS stressors can be measured.

The Tonacchera Model can also be used to quantify the impact on the TIU from each of
the four NIS stressors. The following table calculates the percent reduction in TIU as a result of

161

-------
10-P-0101

a 25% change in the typical concentration of each NIS stressor (a sample calculation is not
shown, but is similar to previous examples):

Modeled Condition

SPEC
(|jmol/L)

Raw TIU
Value

%TIU of Baseline
at SPEC(RfD) &
Normal [ij

Reduction in TIU as
a Result of Modeled
Condition

SPEC(RfD) + normal [I ]

1.511|

0.36617x

100 (baseline)

SPEC(RfD) + 25% I [1]

1.511

0.27463X

75.0

25%

SPEC(RfD) (25%t [CIO4 ])*

1.5175

0.36530X

99.8

0.2%

SPEC(RfD) (25%| [NO:/])*

1.5558

0.36026X

98.4

1.6%

SPEC(RfD) (25%| [SCN ])*

1.8473

0.32602X

89.0

11.0%

f means "increase in"

| means "decrease in"

* Modeled serum condition is at normal [I ]
f Estimated typical total NIS inhibition load at the perchlorate RfD

Note: The table uses a 25% change in each of the NIS stressors so that the perchlorate effect would be on scale
(i.e., a fraction of a % change in TIU).

The table shows the change in %TIU as the result of a 25% change in serum
concentration of each of the NIS stressors. At the estimated total NIS inhibition load on the body
at the perchlorate RfD of 1.511 |imol/L SPEC, an isolated 1% change in the serum concentration
of each of the stressors (i.e., lack of iodide, thiocyanate, nitrate, and perchlorate) induces changes
in %TIU of 1%, 0.44%, 0.06%, and 0.008% respectively. Therefore, the relative impact of the
stressors on this public health issue can be described by the following mathematical expression:

Lack of Iodide > SCN » N03" » C104"

Where: " > " means "greater than"

" » " means "much greater than"

The relative impact of each of the NIS stressors can be described in the following words:
the lack of iodide (i.e., iodide nutrition) is the dominant stressor (i.e., strongest NIS stressor),
thiocyanate exposure is a medium-strength stressor, nitrate exposure is a weak NIS stressor, and
perchlorate exposure is a very weak NIS stressor. The relative impact of each of the NIS
stressors is useful information when evaluating the course of action to take to effectively address
this public health issue.

162

-------
10-P-0101

9.2.5 Assessing the Effectiveness of Lowering the Total NIS Inhibition Load to
Compensate for Poor Iodide Nutrition

The Tonacchera Model can be used to assess the effectiveness of potentially reducing the
body's total NIS inhibition load to compensate for poor iodide nutrition to maintain a normal
TIU level. The following table summaries the modeled results (a sample calculation is not
shown, but is similar to previous examples):

TIU Level

(%)
of Normal

Iodide
Nutritional

Level
Fraction
of Normal

Calculated SPEC
Exposure
Required to Maintain
Normal TIU level
(|jmol/L)

Reduction in SPEC
Exposure Needed to
Maintain Normal TIU Level
(pmol/L)

Percentage
Reduction in
SPEC Exposure
(%)

100

1.00

1.501*

100

0.99

1.474

0.027

1.8

100

0.98

1.447

0.054

3.6

100

0.95

1.365

0.136

9.1

100

0.935

1.324

0.176

11.8

100

0.90

1.229

0.272

18.1

100

0.75

0.821

0.680

45.3

100

0.69

0.657

0.843

56.2

100

0.50

0.140

1.361

90.7

* 1.501 pmol/L SPEC is the baseline amount of total NIS inhibition acting on the body.

The table shows that reducing the total NIS inhibition load acting on the body is neither
practical nor biologically achievable approach for compensating for poor iodide nutrition. A 1%
decrease in the iodide nutritional level (i.e., corresponds to a mean UIC of 143.6 |ig/L) can be
offset by decreasing the body's total NIS inhibition load by 0.027 |imol/L SPEC. By
comparison, the amount of NIS inhibition in an adult contributed by perchlorate at the RfD to the
body's total NIS inhibition load is only 0.014 |imol/L SPEC.

Likewise, the table shows that a 7.5% decrease in the iodide nutritional level (i.e.,
corresponds to a mean UIC of 136 |ig/L) can be offset by decreasing the body's total NIS
inhibition load by 0.176 |imol/L SPEC. By comparison, the amount of NIS inhibition
contributed by both perchlorate exposure at the RfD and the typical nitrate exposure in an adult
is only 0.181 |imol/L SPEC. So, all exposure to perchlorate at the RfD and nitrate would have to
be eliminated to offset a 7.5% decrease in the body's intake of iodide.

Furthermore, the table shows that a 31% decrease in the iodide nutritional level to
borderline mild iodide deficiency (i.e., corresponds to a mean UIC of 100 |ig/L) can be offset by
decreasing the body's total NIS inhibition load by 0.843 |imol/L SPEC. To achieve this, all of
the exposure to perchlorate at the RfD (i.e., 0.014 (amol/L SPEC), all of the typical nitrate
exposure (i.e., 0.167 |imol/L SPEC), and half of the typical thiocyanate exposure (i.e., 0.667
|imol/L SPEC) would have to be eliminated to offset the decrease in %TIU from borderline mild
iodide deficiency. Such a reduction in the total NIS inhibition load is not practical. All nitrate-
containing foods would have to be eliminated from the diet, to include bacon, beef, sausage,
potatoes, and carrots, and most of the foods rich in thiocyanate would have to be eliminated from

163

-------
10-P-0101

the diet, to include cabbage, broccoli, Brussels sprouts, corn (maize), cauliflower, cabbage,
radishes, spinach, tomatoes, and milk.

Furthermore, thiocyanate may have a regulatory function in normal physiology
(Middlesworth 1986). Thiocyanate is continually synthesized in the normal rat. In the absence
of all dietary intake of thiocyanate in the rat, the thiocyanate serum concentration actually
increased, from 500 |ig/dl to 800 |ig/dl, during the fasting (Middlesworth 1986, figure 1). These
data indicate that serum thiocyanate in rats is maintained in a normal range. This suggests that
significantly reducing serum thiocyanate in humans from a normal range is not biologically
achievable.

In conclusion, reducing the average exposure to NIS inhibitors is not an effective solution
to compensate for iodide deficiency. The only effective manner to address the adverse health
effects of a low %TIU resulting from iodide deficiency is to increase the intake of iodide in the
diet.

9.3 Application of UFs in a Multifactorial Cumulative Risk Assessment

The conventional approach of deriving an RfD in a single contaminant characterization is
to apply one or more UFs to a LOAEL, NOAEL, or BMD, to address uncertainties in the data to
ensure the RfD is below an exposure that might induce an adverse effect in a sensitive
population. The NAS Committee applied a UF of 10 to the NOEL to derive the recommend
perchlorate RfD of 0.0007 mg/kg-day. The number, type, and magnitude of UFs to apply in the
derivation of an RfD is a matter of scientific judgment. This is a major source of contention in
the derivation of a perchlorate RfD. The following table calculates the estimated total goitrogen
load in an adult for various applications to UFs to the perchlorate NOEL of 0.007 mg/kg-day,
which correspond to several DWELs (sample calculation not shown, but similar to previous
calculations):

Description of
Dose Level

Drinking Water

Equivalent
Level (DWEL)

External Dose
(mg/kg-day)

Corresponding
Uncertainty
Factor (UF)

Calculated Total
Goitrogen Load
(pmol/L SPEC)

Greer NOEL

245 ppb

0.007

1.567

CI04" RfD

24.5 ppb

0.0007

1.511

California

Drinking Water Limit

6.0 ppb

0.000171

40.8|

1.5034

Massachusetts
Drinking Water Limit

2.0 ppb

0.0000571

122.5|

1.5012

EPA draft RfD

1.0 ppb

0.00003

233.3*

1.5006

Provided to assess the
impact of a large UF on
the TIU

0.5 ppb

0.000014

500

1.5003

* EPA's draft perchlorate RfD was 0.00003 mg/kg-day external dose, but was derived from a different POD using a
UF of 300.

f For this evaluation of the impact of applying different UFs to the setting of a perchlorate RfD on the TIU, the
external perchlorate dose and UF listed were back calculated from the Greer NOEL to arrive at the State's Drinking
Water Limits (this not the manner in which the States set their limits).

164

-------
10-P-0101

The following table summaries the calculated TIU, %TIU, and percentage increase in
TIU in an adult at various applications to UFs to the perchlorate NOEL of 0.007 mg/kg-day,
which corresponds to several DWELs (sample calculation not shown, but similar to previous
calculations):

Description of
Dose Level

Drinking Water
Equivalent Level
(DWEL)

Calculated
Total Iodide
Uptake in an
Adult
(TIU)

Percentage TIU
of Baseline

(%)

Change in
%TIU from
Baseline

Greer NOEL

245 ppb

0.35881X

100 baseline

CI04" RfD

24.5 ppb

0.36617x

102.05

2.05

California

Drinking Water Limit

6.0 ppb

0.36718x

102.33

2.33

Massachusetts
Drinking Water Limit

2.0 ppb

0.36748X

102.42

2.42

EPA draft RfD

1.0 ppb

0.36756X

102.44

2.44

Provided to assess the impact
of a large UF on the TIU

0.5 ppb

0.36761X

102.45

2.45

Applying a UF of 10 to the NOEL reduces the perchl orate DWEL from 245 ppb to 24.5
ppb, which would increase an adult's TIU by 2.05%. Applying a UF of 40.8 to the NOEL
reduces the perchlorate DWEL to 6.0 ppb; this further reduction in the perchlorate DWEL from
24.5 ppb to 6.0 ppb increases an adult's TIU by only 0.28%. Applying a UF of 122.5 to the
NOEL reduces the perchlorate DWEL to 2.0 ppb; this further reduction in the perchlorate DWEL
from 6.0 ppb to 2.0 ppb increases an adult's TIU by only 0.09%. Applying a UF of 233.3 to the
NOEL reduces the perchlorate DWEL to 1.0 ppb; this further reduction in the perchlorate DWEL
from 2.0 ppb to 1.0 ppb increases an adult's TIU by only 0.02%. Applying a UF of 500 to the
NOEL reduces the perchlorate DWEL to 0.5 ppb; this further reduction in the perchlorate DWEL
from 1.0 ppb to 0.5 ppb increases an adult's TIU by only 0.01%.

In short, this calculation shows that the use of an increasing UF value has the surprising
result of having only a minimal impact on the TIU. This is contrary to the intent of the use of
UFs in deriving an RfD. The explanation of this result is that a single chemical risk assessment
is based on the assumption that the chemical is the principal cause (i.e., dominant stressor) of the
adverse effect. So, if the exposure to the single chemical is kept sufficiently below the POD, the
adverse effect will be avoided in even the most sensitive human subpopulation. However,
perchlorate is only one of four stressors that affect the TIU of the NIS. Perchlorate also
contributes the least amount of stress of the four NIS stressors and explains why the application
of a UF to perchlorate exposure has a minimal effect on the TIU. Since the mechanism of
toxicity results from the lack of a sufficient uptake of iodide to make an adequate supply of
thyroid hormones for proper brain development during gestation and lactation, the concept of
UFs should be applied to the TIU, which addresses the integrated result from the simultaneous
exposure to all four stressors. The magnitude of the TIU is the single variable that determines a
favorable or unfavorable outcome in this public health issue. In other words, since a low TIU
leads to the adverse effect of brain damage, the minimum TIUs observed in the human
population need to be kept above the TIU level that results in adverse effects. To apply UFs to

165

-------
10-P-0101

the TIU, the equivalent LOAEL or NOAEL TIU value must be determined; this will be
discussed in the next section.

166

-------
10-P-0101

9.4 Determining %TIU Levels of Concern

9.4.1 Identifying a %TIU(loael) in Pregnant Women to Prevent Subtle Mental
Deficits in Children

The Tonacchera Model can be used to estimate the LOAEL %TIU in pregnant women
when fetal brain damage occurs. The Tonacchera Model states:

TIU|| [T]/(1.22 +(SPEC))

where: the symbol || means "proportional to"

If the SPEC is held constant (i.e., the level of NIS inhibitors remains the same), then,
TIU || [T]

Therefore, when the SPEC(Normai), then %TIU(loael) is given by:

%TIU(LOAEL) — ([I ](Fetal brain damage) ~ [I"] (Normal pregnant women) ) X 100 /o

Since subtle fetal brain damage is known to occur in moderately iodide-deficient
pregnant women (Bleichrodt 1989; Vermiglio 2004; Vermiglio 1990), as before, using UIC as a
surrogate, the [T](Fetai brain damage) is 50 |ig/L. The other information need is the recommended
healthy UIC in pregnant women (i.e., [T] (Normal pregnant women))- Several credible sources have
opined on the recommended UIC for pregnant women. Instead of relying on a single source, the
average recommended UIC for pregnant women was determined and found to be 204 |ig/L. The
following table summarizes the sources and recommended UICs for pregnant women:

Recommended Iodide

Recommended Urinary

Intake in Pregnant Women

Iodide Concentration

Source

(Mg/day)

(mq/l)

WHO/UNICEF/ICCIDD*

200

200-300

Dr. Francois Delange, MD**

225-300

150-230

WHO Technical Consultant Group***

225-375

150-249

Recommend Dietary Allowances (RDA)|

220|

150|t

Average Value =>

258

204

* Source: Delange 2005, p 278, table 11E.8.

** Source: Delange 2004.

*** Source: ATA 2006.

f Source: NAP 2000.

ft Source: Pearce 2004.

Therefore, the %TIU(loael) in pregnant women is calculated as follows:

%TIU(LOAEL) ([I ](Fetal brain damage) ' [I"] (Normal pregnant women) ) X 100 /o

%TIU(loael) = (50 |ig/L - 204 |ig/L) x 100%

%TIU(loael) = 24.5%

167

-------
10-P-0101

From the observed subtle brain damage in the offspring of moderately iodide-deficient
pregnant women (Bleichrodt 1989; Vermiglio 2004; Vermiglio 1990), the LOAEL in pregnant
women is estimated to be 24.5 %TIU. This agrees with 22.7%TIU(HyPothyoxinemia) observed to
generate hypothyroxinemia in women exposed to excess amounts of thiocyanate (Banerjee
1997b). Furthermore, inducing hypothyroxinemia in pregnant women is consistent with the
proposed mechanism that an insufficient supply of maternal fT4 during gestation starves the
developing fetal brain of an adequate supply of thyroid hormones, resulting in fetal brain
damage. Also of note, 50% of the moderately iodide-deficient pregnant women in the Vermiglio
2004 study were observed to have hypothyroxinemia. Eleven of the 16 children born to
moderately iodide-deficient pregnant women in the Vermiglio 2004 study had ADHD.

The 24.5 %TIU(Loael) integrates the combined effects of all four NIS stressors into a
single variable. In a cumulative risk assessment, this is the numerical value that the concept of
UFs shall be applied to. In a single chemical risk assessment, UFs are applied to the LOAEL or
NOAEL to derive an RfD to address the uncertainties inherent in the data. If this concept is
applied to the %TIU(Loael), typically a UF of 10 is applied to a LOAEL to estimate a NOAEL.
And as in the case of the NAS Committee, an intraspecies UF of 10 was applied to derive the
perchlorate RfD. Using this simplified template for the estimation of UFs to apply to the
%TIU(loael), a total UF of 100 would be used. The application of a UF of 100 to 24.5
%TIU(loael) (i e., to back off the adverse effect level by two orders of magnitude) would derive
a %TIU(RfD equivalent) = 2450%. This value corresponds to a UIC of about 5000 |ig/L. This is
biologically unrealistic. Obviously, the conventional rules of applying UFs during a single
chemical risk assessment do not apply to this cumulative risk assessment.

9.4.2 Identifying a %TIU(Noael) in Pregnant Women to Prevent Subtle Mental
Deficits in Children

The cumulative risk assessment of this public health issue recognizes that the dose-
response curve for iodide exposure to the thyroid is a u-shape curve (Delange 2005, p 281).

When the UIC of pregnant women is below 50 |ig/L, adverse effects are observed in their
children, but when the UIC of pregnant women goes above 500 |ig/L (ATA 2006), other adverse
clinical effects can occur. By analogy, since %TIU represents the same mechanism (i.e., the
amount of iodide available to the thyroid), the %TIU dose-response curve will also be u-shaped.
A UIC of 50 |ig/L and 500 |ig/L result in a %TIU in a pregnant woman of 24.5 %TIU(loael) and
245 %TIU(Excess limit), respectively.

EPA's Integrated Risk Information System defines a NOAEL as:

The highest exposure level at which there are no biologically significant increases
in the frequency or severity of adverse effect between the exposed population and
its appropriate control; some effects may be produced at this level, but they are
not considered adverse or precursors of adverse effects (IRIS 2007).

The traditional use of a NOAEL in a single chemical exposure is as the lowest chemical
exposure level without adverse effects. The typical use of a NOAEL assumes that increasing
chemical exposure produces an increasing the severity or frequency of adverse effects.

168

-------
10-P-0101

However, the %TIU dose-response curve is u-shaped. An adverse effect occurs when the %TIU
is either too high or too low. Therefore, the application of the NOAEL concept to a u-shape
dose-response curve generates the need for two NOEALs to prevent insufficient uptake of iodide
and to prevent an excessive uptake of iodide. For the purposes of identifying each NOAEL, the
following labels will be used: %TIU(noael) and %TIU(noael for excess)- A UIC above 500 |ig/L in
pregnant women (ATA 2006) is excessive and results in a %TIU(Noael for excess) of 245%.

The key is to define a %TIU(NOael) that is " . . . [the lowest %TIU] exposure level at
which there are no biologically significant increases in the frequency or severity of adverse effect
between the exposed population and its appropriate control." A %TIU(inSufficient noael) can be
derived from the following adverse effects occurring in children of mildly iodide-deficient
pregnant women:

• Mildly iodide-deficient pregnant women have children with statistically significant
delayed reaction times to visual stimuli (Lombardi 1995; Vitti 1992). Although not a
particularly severe adverse effect, delayed reaction times to visual stimuli is a sensitive
test for a subtle change that is reproducible between studies and is undesirable.

• The heterogeneous thyroid performance within the human population (i.e., the 5th item to
be addressed by this cumulative risk assessment) can be observed in the children of
mildly iodide-deficient pregnant women. Newborns of mildly iodide-deficient women
have an increased frequency of 0 to 16.9% of elevated TSH above the < 3% background
levels observed in iodide-sufficient areas. In other words, at least 83% of the newborns
are unaffected by mild iodide deficiency, while up to 16.9% are potentially adversely
affected. Although a mildly iodide-deficient pregnant woman is able to adapt to mild
iodide deficiency and will probably remain euthyroid throughout pregnancy, the fetus's
thyroid is less able to adapt to iodide deficiency. Fetal hypothyroxinemia is induced by
iodide deficiency because the immature fetal thyroid is unable to increase its avidity for
iodide despite up-regulation of NIS expression in the fetal thyroid and placenta during
iodide deficiency (Delange 2005a). The increased iodide clearance rate will further
decrease the iodine stores of the fetal thyroid and decrease fetal T4 synthesis (Delange
2005a). Tracer studies in mild hypothyroid neonatal rats (i.e., elevated TSH without
below normal fT4) have observed increased D2 deiodinase activity of the central nervous
system (CNS) in order to compensate for a decreased supply of T3 in the CNS. This
implies that newborns with elevated TSH (analogous to mild hypothyroidism in the
neonatal rats) may have had to activate their D2 deiodinase compensatory mechanism in
the CNS (Bianco 2005, p 123). This indicates that the fetal thyroid may not be making
enough T4 during mild iodide deficiency and increases the possibility that subtle brain
damage may have occurred. However, epidemiology studies have neither identified nor
evaluated this subgroup (i.e., elevated TSH newborns) for the possibility of subtle mental
deficits. Neonatal TSH is the single indicator that focuses on potential brain damage,
which is the major impact of iodide deficiency (Delange 1998). An additional adverse
effect in neonates with elevated TSH is observed in a study by Calacirua, in which infants
with elevated TSH at neonatal screening are at risk for developing subclinical
hypothyroidism in early childhood (odds ratio 44.6%; 95% CI 38.8-50.4%) (Calaciura

169

-------
10-P-0101

2002). Delange also points out that neonates with elevated TSH levels are at a high risk

of developing subclinical hypothyroidism in infancy and early childhood (Delange 1998).

Clearly, the observed delayed reaction time, the increased occurrence of elevated TSH in
newborns, and the increased risk of developing subclinical hypothyroidism meet the NOAEL
definitional requirement that the NOAEL exposure level must be above the level of "increases in
the frequency or severity of adverse effect(s)." Since mild iodide deficiency in pregnant women
is characterized by a UIC between 100 |ig/L to 50 |ig/L, using a UIC of 100 |ig/L prevents a
%TIU exposure level that induces an increased frequency of elevated TSH levels in the iodide-
sensitive subpopulation. Therefore, the %TIU(Noael) in pregnant women is calculated as
follows:

%TIU(NOAEL) — ([I ](inc. freq. of elevated TSH in newborns) ~ [I ] (Normal pregnant women) ) X 100 /o

%TIU(Noael) = (100 |ig/L - 204 |ig/L) x 100%

%TIU(Noael) = 49%

9.4.3 Identifying a %TIU(RfD) in Pregnant Women to Prevent Subtle Mental Deficits
in Children

EPA's Framework for Cumulative Risk Assessment (EPA 2003) defines an (RfD) as, "An
estimate (with uncertainty spanning perhaps an order of magnitude) of a daily exposure to the
human population (including sensitive subgroups) that is likely to be without an appreciable risk
of deleterious non-cancer effects during a lifetime (EPA 2003)."

EPA's risk assessment guidance states that a reduction of the intraspecies UF from a
default of 10 is to be considered only if the POD is determined from the data obtained from the
susceptible subpopulation (EPA 2002, p 4-43). Since the POD (i.e., %TIU(noael) = 49%) was
determined in the susceptible subpopulation, an intraspecies UF of 1 should be used. However,
the %TIU(noael) should not be used as the %TIU(RfD), because the uptake of iodide is not the
only step in the production and use of thyroid hormones in the body that is disrupted by chemical
exposures. Numerous inorganic and synthetic chemicals have been documented to interfere with
almost every major step in the production, transport, and peripheral tissue metabolism of thyroid
hormones (Howdeshell 2002, p 344, table 1). The following table provides examples of
chemicals that interfere with several major thyroid hormone steps:

Thyroid Hormone Step

Examples of Chemicals that Interfere

Iodide uptake by the NIS

2,4-D, Aldrin, Lead, PBBs,

Iodide oxidation by thyroid peroxidase

Lindane, Malathion, Mancozeb

Circulatory transport in blood
(e.g., binding to transthyretin)

Dioxtylphthalate, DDT metabolites,
Dichloroprop, Difocol, Lindane, Malathion,

Cellular metabolism of T4 to T3 by Type I or Type II 5'-
deiodinase in peripheral tissues

Cadmium, Lead, PCB, Dixoin

Increase cellular elimination by glucuronidation of T4/T3 by
the stimulation of the glucuronidase enzyme

Acetochlor, DDT, PBBs, PCBs, Dioxin

Source: Howdeshell 2002 (for full list - see p 344, table 1).

As stated in the EPA's /•'ramework for Cumulative Risk Assessment, an aspect of a
cumulative risk assessment is that it can be qualitative as well as quantitative (EPA 2003). This

170

-------
10-P-0101

aspect is necessary to incorporate risk from other chemical exposures in which there is
insufficient information to accurately quantify the amount of risk, although the presence and
nature of the risk is known. The rational for using a nonquantitative approach for setting the
%TIU(RfD) is also based on the understanding of the mechanism of action for thyroid hormones
(Yen 2005). Proper fetal/neonate brain development occurs when a sufficient supply of
intracellular T3 binds to the thyroid hormone receptor TRP-1. The T3/TRP-1 is a ligand-
regluated transcription factor that along with other transcriptional cofactors causes the expression
of genes containing thyroid hormone response elements in the promoter region (Yen 2005).

In short, thyroid hormones represent a "chemical signal" in the body that regulates the
transcription of genes during CNS development. Although the %TIU(Noael) provides enough
uptake of iodide by the thyroid to make a sufficient supply of T4, subsequent loses of T4in
subsequent thyroid hormone steps due to the potential interference by other industrial chemicals
(e.g., higher than normal cadmium exposure during gestation) could result in a insufficient suppy
of T4 at the target brain tissue. In other words, the "chemical signal" started out strong enough at
the thyroid, but was weakened by interference from other industrial chemicals along the way that
resulted in an insufficient "chemical signal" in the developing brain cell. The amount of
interference by these other industrial chemicals is qualitatively known to occur, but is not easily
quantified for use in a conventional risk assessment. However, a cumulative risk assessment
allows this interference effect to be qualitatively factored into the assessment. The concept is
that if the original "chemical signal" is made strong enough, it may tolerate some interference
along the way and still be strong enough at the end to be correctly interrupted and allow for the
appropriate level of gene transcription to occur for proper brain development.

The qualitative factor proposed is the equivalent of using a 1.5 UF. Therefore, the
%TIU(RfD) in pregnant women is calculated as follows:

%TIU(RfD) = %TIU(noael) x 1.5 UF

%TIU(RfD) = 49% x 1.5 UF
%TIU(RfD) = 74%

At a typical 1.501 |imol/L SPEC level, a %TIU of 74% corresponds to a UIC in pregnant
women of 151 |ig/L. Since the median UIC in pregnant women in the NHANES III survey was 140
Hg/L, this means about half of U.S. pregnancies each year occur below the %TIU(RfD> This indicates a
clear need to increase the %TIU levels in this half of the U.S. pregnant women population with these
low %TIUs.

171

-------
10-P-0101

9.5 Calculating the %TIU in Pregnant Women at Various Perchlorate Exposures

The Clewell perchlorate PBPK model identifies that the internal perchlorate exposure
(i.e., internal indices) is not the same for each life stage for the same external perchlorate dose
(Clewell 2007, p 423, table 4). Specifically, the conversion factor from an external to internal
perchlorate dose is 2.5 times higher at 0.001 mg/kg-day for a pregnant woman than for an adult.
In other words, a pregnant woman experiences 2.5 times more NIS inhibition from higher serum
perchlorate level than an adult for the same external perchlorate dose of 0.001 mg/kg-day.
Therefore, the calculated TIUs in pregnant women will be slightly lower then the calculated
TIUs in adults (see previous table in Section 9.1.4) at the same external perchlorate dose. The
following table summaries the calculated TIUs for pregnant women with various external
perchlorate exposures levels (sample calculation not shown, but similar to previous calculations):

External CI04"

Dose
(mg/kg-day)

Internal
CI04"
Dose in
Pregnant
Women
(|jmol/L)

Serum
Perchlorate
Equivalent
Cone. (SPEC)
in Pregnant
Women
(|jmol/L)

Iodide
Nutritional
Level in
Pregnant
Women
(Fraction of
Normal)

Calculated
Total
Iodide
Uptake
(T'U)

%TIU Calculated
from the
Tonacchera
Model

0.000066*
(in

nonpregnant
adults)

0.0013*

1.501|

1.00

0.3675X

100
(nonpregnant
adult %TIU
reference level)

0.000066*

0.00325

1.503

1.00

0.3672X

99.9

0.000175
(% RfD)

0.0088

1.510

1.00

0.3663X

99.7

0.0007
(RfD)

0.035

1.536

1.00

0.3628X

98.7

0.007 (NOEL)

0.399**

1.900

1.00

0.3205X

87.2

0.000066*

0.00325

1.503

0.49***

0.1799X

49.0

0.000175
(% RfD)

0.0088

1.510

0.49***

0.1795X

48.8

0.0007
(RfD)

0.035

1.536

0.49***

0.1778X

48.4

0.007 (NOEL)

0.399**

1.900

0.49***

0.1570X

42.7

* Median perchlorate exposure in U.S. population (Blount 2006a).

f From median NIS inhibitor exposures in the U.S. population (calculated in Section 7.2.1).

** Used an estimated conversion factor of 0.0425 mg/L per .0075 mg/kg-day using a linear interpolation from the

Clewell perchlorate PBPK model.

*** Average recommended urinary iodide concentration in pregnant women = 204 |jg/L (see Section 9.4.1).

Therefore, an iodide nutritional level of 0.49 in pregnant women corresponds to 100 |jg/L, which is the urinary iodide
limit for mild iodide deficiency.

In a pregnant woman at the average recommended iodide nutrition level, the Tonacchera
Model calculates a 98.7 %TIU at the perchlorate RfD of 0.0007 mg/kg-day, which corresponds
to a 24.5 ppb DWEL. By comparison, the calculated %TIU in adults at the perchlorate RfD is
99.6% (see table in Section 9.1.4). Therefore, the Tonacchera Model estimates that a pregnant
woman's %TIU is affected more by the same external perchlorate exposure than is an adult's
%TIU (i.e., a decrease of 0.9 %TIU). In a pregnant woman, the Tonacchera Model also
calculates a 99.7 %TIU at a perchlorate exposure level of 0.0000175 mg/kg-day (i.e., a quarter of

172

-------
10-P-0101

the perchlorate RfD level), which corresponds to a 6.1 ppb DWEL. Therefore, in a pregnant
woman, the Tonacchera Model estimates a 1.0% increase in the %TIU when the perchlorate
drinking water exposure is lowered from 24.5 ppb to 6.1 ppb.

By contrast, in a pregnant woman with mild iodide deficiency (i.e., with a UIC of 100
|ig/L), the Tonacchera Model calculates a 48.4 %TIU at the perchlorate RfD of 0.0007 mg/kg-
day, which corresponds to a 24.5 ppb DWEL. This is a decrease of 50.3%TIU as compared to a
pregnant woman at the average recommended iodide nutrition level having a same perchlorate
exposure at the perchlorate RfD (i.e., 98.7% minus 48.4%). In other words, for a pregnant
woman with perchlorate exposure at the RfD, decreasing her iodide intake from the average
recommended iodide nutrition level (i.e., a UIC of 204 |ig/L) to a mildly iodide-deficient level
(i.e., a UIC of 100 |ig/l) lowers her TIU level by 50.3%TIU. By comparison, a pregnant woman
with mild iodide deficiency and a perchlorate exposure level of 0.0000175 mg/kg-day (i.e., a
quarter of the perchlorate RfD), which corresponds to a 6.1 ppb in drinking water, would have a
calculated %TIU of 48.8%. Therefore, in a pregnant woman with mild iodide deficiency, the
Tonacchera Model estimates only a 0.4 % increase in the %TIU when the perchlorate drinking
water exposure is lowered from 24.5 ppb to 6.1 ppb. Clearly, an increase of 0.4% in %TIU in
mildly iodide-deficient pregnant women estimated from the potential act of lowering the
perchlorate drinking water exposure from 24.5 ppb to 6.1 ppb is not sufficient to offset the
50.3% decrease in %TIU from the loss of iodide in the diet during mild iodide deficiency during
pregnancy. This example in mildly iodide-deficient pregnant women shows that the lack of
iodide is the dominant stressor affecting this public health issue and the risk to the public from a
low %TIU during pregnancy and lactation cannot be managed without addressing iodide
deficiency within this sensitive subpopulation.

Evaluating these results from another perspective, the Greer study observed a %TIU of
98.2%) ± 8.3%) at the perchlorate NOEL of 0.007 mg/kg-day. Therefore, the predicted 1.0%
reduction in the TIU in pregnant women when the DWEL is lowered from 24.5 ppb to 6.1 ppb is
not statistically detectable in a human experiment. From the Greer data, a 1.0% change in the
%TIU is only 1/12th the random variation observed in the %TIU at the perchlorate NOEL. By
contrast, the Greer study measures a %TIU range of 45% to 155% in the population before the
exposure to perchlorate (i.e., in the 37 test subjects at the BV). Therefore, %TIU varied ± 55%
in Greer's study population without measuring any thyroid hormones outside the normal ranges
for TSH, fT4, tT4, or tT3. Potentially regulating perchlorate at a DWEL of 6.0 ppb instead of
24.5 ppb prevents about a 1% change in TIU in pregnant women; a 1%TIU change is only a
small fraction of the ± 55% normal variation observed in the %TIU at baseline in the Greer study
population. Therefore, decreasing the perchlorate drinking water concentration from 24.5 ppb to
6.0 ppb will not have a significant effect on the %TIU observed in people. A single chemical
risk assessment of perchlorate only characterizes a small portion of this public health issue and
does not explain the factors that cause the TIU to vary ± 55% in the Greer study population.
Furthermore, managing only the exposure to perchlorate and not managing the other NIS
stressors will have little effect of ensuring the %TIU is above the NOAEL for pregnant women.
This cumulative risk assessment estimates the %TIU(noael) = 49% for pregnant women. Only
by managing the exposure to all four NIS stressors can the %TIU be held above the %TIU(noael)
for pregnant women.

173

-------
10-P-0101

Using a cumulative risk assessment approach to characterize this public health issue, the
large variation in the %TIU observed in the Greer study can be easily explained. The NIS
stressor that can account for such a large variation in the %TIU observed in the Greer study
population is the dominant NIS stressor - iodide. If all the variation in %TIU was from the
variation in iodide nutrition alone, the Greer study population would be expected to have a UIC
in the range of 65 |ig/L and 225 |ig/L. This agrees with iodide variation observed in the NHANE
III survey, which identified the 10th to the 90th percentile UIC in the U.S. population to be 45
|ig/L to 386 |ig/L.

A 1.0% reduction in %TIU could be offset by a 1.0% increase in the UIC in pregnant
women. At a borderline iodide-deficient UIC of 100 |ig/L in pregnant women, a 1.0% increase
in %TIU is caused by the additional intake of about 1.1 jug of iodide. The American Thyroid
Association (ATA) recommends that pregnant and lactating women take a daily prenatal
vitamin/mineral supplement containing 150 |ig of iodide (ATA 2006). Taking an iodide-
containing prenatal vitamin/mineral supplement would easily offset a 1% reduction in %TIU.
Furthermore, since iodized salt in the United States contains about 100 ppm potassium iodide or
77 |ig iodide/gram of salt (Dunn 1998), the increased consumption of 14.3 mg of iodized salt
would provide the 1.1 |ig of iodide required to offset the 1% reduction in %TIU. Since a pinch
of table salt weighs about 460 mg, 14.3 mg of salt is equivalent to about l/32nd of a pinch of
salt. Since a smidgen of table salt weighs about 230 mg, 14.3 mg of salt is equivalent to about
1/16th a smidgen of salt.

174

-------
10-P-0101

10. Approaches to Address this Public Health Issue

10.1 %TIU Exposure Levels of Concern

Any effective remedy must ensure that pregnant and nursing women have a sufficient
TIU to prevent subtle mental deficits in their children. The cumulative risk assessment
determined that a %TIU(jnsufficient RfD) of 74% in pregnant and nursing women is the exposure
level at which the risk begins for this public health issue. As the %TIU decreases below 74%,
the level of risk increases. The risk levels of concern are characterized as follows:

• If the combination of exposures to the four NIS stressors induces a %TIU less than 74%
(i.e., the %>TIU(insufficient RfD) level), the possibility for mental deficits exists. At a typical
total NIS inhibition load, a %TIU corresponds to a UIC of 151 |ig/L in pregnant women.

• If the combination of exposures to the four NIS stressors induces a %TIU less than 49%
(i.e., the %THJ(Noael) level), the fetus and nursing infant are under thyroid stress (and the
thyroid-sensitive subpopulations are at greater risk) and the potential for mental deficits
increases. At a typical total NIS inhibition load, a %TIU corresponds to a UIC of 100
|ig/L in pregnant women.

• If the combination of exposures to the four NIS stressors induces a %TIU less than 24.5%
(i.e., the %THJ(Loael) level), mental deficits occur in a percentage of the offspring
population. The lower the %TIU, the frequency and severity of the mental deficits
increases. At a typical total NIS inhibition load, a %TIU corresponds to a UIC of 50
|ig/L in pregnant women.

10.2 Remedy Approach

The public health issue of potential subtle mental deficits in children is initiated by an
insufficient amount of iodide being taken up by the thyroid in pregnant women, fetuses, and
nursing infants. An excess exposure to any one (or more) of the four NIS stressors (i.e., lack of
iodide, thiocyanate, nitrate, and perchlorate) can result in a TIU less than 74%TIU in pregnant
women. Unfortunately, the current federal and State remedy being considered or implemented to
address this public health issue is the exposure to only one of the four NIS stressors. The federal
and State governments are considering or have implemented a maximum contaminant limit
(MCL) for perchlorate, which will limit the perchlorate exposure from drinking water.

Using a cumulative risk assessment approach, everyone in the U.S. population has a
continual exposure to all four NIS stressors. An excess exposure to any one (or more) of the four
NIS stressors can result in a TIU less than 74 %TIU in pregnant women. To address the risk, the
exposure levels to all four NIS stressor have to be considered and potentially managed. To
manage the risk, the proposed goal should be to ensure that the 90th percentile of pregnant
women have a %TIU greater than %TIU(RfD) of 74%. Currently, the exposure to NIS stressors
results in about half of all pregnant women having a %TIU less then the %TIU(RfD).

Furthermore, the proposed goal should be to ensure that >99th percentile of pregnant women
have a %TIU greater than the %TIU(noael) of 49%. Currently, the exposure to NIS stressors

175

-------
10-P-0101

results in about 29% of all U.S. pregnant women having a %TIU less then the %TIU(Noael)
value. Most troubling, the current exposure to NIS stressors results in up to 7% of all U.S.
pregnant women having a %TIU less then the %TIU(loael) of 24.5%.

10.2.1 Exposure Management of Each NIS Stressor

An adequate amount of iodide uptake in pregnant women, fetuses, and nursing infants is
addressed by preventing an excess exposure to any one (or more) of the four NIS stressors. The
cumulative risk assessment identified that the relative influence of each of the NIS stressors on
this public health issue can be described as follows:

Lack of iodide > SCN" » N03" » C104"

Where: " > " means "greater than"

" » " means "much greater than"

Therefore, the management of the lack of iodide stressor has the largest effect on the
%TIUs value observed in pregnant women, followed by the management of thiocyanate
exposure. By contrast, the management of nitrate and perchlorate will have the least effect on
the %TIUs observed in pregnant women. However, the exposure management to all four NIS
stressors is required to ensure pregnant women have a healthy %TIU level, which is greater than
74%.

10.2.1.1 Lack of Iodide Stressor

The cumulative risk assessment has identified that the lack of iodide NIS stressor is the
dominant stressor in this public health issue. Unfortunately, the iodide exposure assessment in
the sensitive group indicates that 6.9% of pregnant women are moderately iodide deficient and
an additional 22% are mildly iodide deficient. The iodide exposure assessment indicates that up
to 29% of the U.S. population of pregnant women is at a higher risk for mental deficits in their
children. Obviously, the amount of iodide in the diet of a significant portion of the U.S. pregnant
women population is dangerously low. Therefore, correcting poor iodide nutrition among
pregnant and lactating women is needed to ensure the healthy brain development of the fetus and
nursing infant.

This is where using a cumulative risk assessment approach to this public health issue is
critical. A cumulative risk assessment allows stressors that are outside EPA's legislative
mandates, authority, or expertise to be identified and addressed in cooperation with other federal
agencies (EPA 2003, xviii), health organizations, and technical disciplines. EPA authority
extends to the regulation of environmental contaminates. However, according to the Tonacchera
Model, the dominant stressor in this public health issue is the iodide nutrition level (i.e., the lack
of iodide in the diet) in pregnant and lactating women. Overseeing and correcting the iodide
nutrition level is clearly not in EPA's legislative mandate; however, addressing the iodide
deficiency in U.S. pregnant women is essential to effectively addressing this public health issue.

176

-------
10-P-0101

The NAS Committee was fundamentally aware of the crucial role of iodide nutrition
during pregnancy. The NAS Committee wrote in the summary of their report that "[The NAS
Committee] . . . recommends that consideration be given to adding iodide to all prenatal
vitamins" (NAS 2005, p 18). The NAS Committee also wrote that same statement as the last
line in its report (NAS 2005, p 196). The issue of whether prenatal vitamins should contain 150
|ig of iodide falls into the legislative authority the National Institutes of Health, the FDA (see
further discussion below), and the charter of the Institute of Medicine (IOM). The NAS
Committee's recommendation for adding iodide to all prenatal vitamins should or could be
addressed through the Federal Government's Perchlorate Interagency WorkGroup. However,
this NAS recommendation has not been acted upon within that environmental arena.

10.2.1.1.1 Iodide Supplementation During Pregnancy and Lactation

The NAS Committee, ATA, and leading thyroid experts have recommended iodide
supplementation during pregnancy and lactation. The NAS Committee recommends that
consideration be given to adding iodide to all prenatal vitamins (NAS 2005, p 18, 196). ATA
recommends that pregnant and lactating women take a daily prenatal vitamin/mineral
supplement containing 150 |ig of iodide (ATA 2006). Zimmerman and Delange have
encouraged both the European Thyroid Association and the European Office of ICCIDD to
recommend that supplement manufacturers include 150 |ig/day iodide in prenatal supplements
(Zimmermann 2004). Mild to moderate iodide deficiency during pregnancy adversely affects
thyroid function of the mother and newborn and mental development of the offspring; these
adverse effects can be prevented or minimized by iodide supplementation (Zimmermann 2004).

Unfortunately, not all prenatal vitamin supplements contain the recommended 150 |ig
iodide, and the use of iodide-containing prenatal vitamins is unacceptably low. The IOM
currently does not recommend including iodide in prenatal vitamins (Sullivan 2007). Therefore,
many prenatal vitamins do not contain iodide (Sullivan 2007). Only 1 of the 19 vitamins (i.e.,
5%) listed for prenatal use in the 2006 Physician's Desk Reference contained 150 jug of iodide,
while the other 18 vitamins listed contained no iodide (ATA 2006). Pearce identifies an abstract
that reports 44 of 69 (i.e., 64%) prenatal vitamins marketed in the United States contain some
level of iodide (Pearce 2007a). In the Boston study of 57 lactating women, only 41 (72%) took a
multivitamin, but only 3 lactating women (7%) took multivitamins containing iodide (Pearce
2007). The Boston study concludes that "47% of women sampled may have been providing
breast milk with insufficient iodine to meet infants' requirements" (Pearce 2007). Furthermore,
younger women of lower education and of certain ethnic groups tend not to use supplements on a
regular basis (Sullivan 2007).

The 234 pregnant women taking no iodide supplement in the NHANES III survey had a
median UIC of 141 |ig/L (ATA 2006). By contrast, the 100 pregnant women taking a daily 150
|ig iodide supplement in the NHANES III survey had a median urinary concentration of 169
|ig/L (ATA 2006). The fraction of women excreting excessive amounts of iodide (i.e., > 500
|ig/L) was not appreciably higher than those not taking a daily iodide supplement. ATA
concludes that taking a 150 |ig iodide supplement per day during pregnancy "appears to be safe"
(ATA 2006). Benefits of correcting iodide deficiency outweigh the risks (Delange 2005, p 283).

177

-------
10-P-0101

The risk from iodide intake is U-shaped, meaning that there is a risk from too little or too
much iodide. Our analysis identified that an iodide supplement of about 150 |ig of iodide per
day will reduce the risk from a low TIU during pregnancy and nursing. This is consistent with
Glinoer's recommendation for the use of a prenatal vitamin pill containing 100-200 |ig of iodide
per day (Glinoer 2005, p 1092). However, the responsibility of assessing the need for and
implementing prenatal vitamin use belongs to the medical community.

10.2.1.1.2 Participation of the Medical Community

A cumulative risk assessment allows stressors that are outside EPA's legislative
mandates, authority, or expertise to still be addressed by involving the expertise of other federal
agencies (EPA 2003, page xviii), health organizations, and technical disciplines. The medical
community has the authority to determine whether iodide-containing prenatal vitamins should be
recommended to address this public health issue. Our analysis identifies that taking iodide-
containing prenatal vitamin will reduce the risk from a low TIU during pregnancy and nursing.
Furthermore, our analysis identifies that iodide-containing prenatal vitamins should be taken at
least 3 months before pregnancy, during pregnancy, and during nursing.

In regard to the concern that pregnant women with a UIC above 150 |ig/L taking a
prenatal vitamin containing iodide might develop health problems from excess iodide, the 234
pregnant women taking no iodide supplement in the NHANES III survey had a median UIC of
141 |ig/L (ATA 2006). By contrast, the 100 pregnant women taking a daily 150 |ig iodide
supplement in the NHANES III survey had a median UIC of 169 |ig/L (ATA 2006). The
fraction of women excreting excessive amounts of iodide (i.e., > 500 |ig/L) was not appreciably
higher than those not taking a daily iodide supplement. ATA concludes that taking a 150 |ig
iodide supplement per day during pregnancy "appears to be safe" (ATA 2006). In general, the
benefits of correcting iodide deficiency outweigh the risks (Delange 2005, p 283).

An approach to avoiding this potential for excess iodide exposure during pregnancy is to
simply measure the UIC during prenatal care. If the UIC is below 150 |ig/L, the use of an
iodide-containing prenatal vitamin is warranted. If the UIC is above 150 |ig/L, the use of an
iodide supplement may not be necessary, and the potential risk of inducing excess iodide intake
is eliminated. On a related concern, if the pregnant woman's UIC is below 50 |ig/L (i.e.,
indicating moderate iodide deficiency), the medical community may need to consider more
aggressive intervention, such as iodide supplementation and levothyroxine (Glinoer 1995), to
avoid or minimize the severity of subtle mental deficits in her child.

The most appropriate preventative and therapeutic approach to avoiding excessive stress
on the pregnant mother's thyroid from the inadequate uptake of iodide is to systematically
increase the iodide supply as early as possible during gestation and continuing through lactation
if the mother intends to breast feed (Glinoer 2005, p 1092). This can be easily achieved by the
use of a multivitamin pill containing 100-200 |ig of iodide per day (Glinoer 2005, p 1092).
However, if an individual has been iodide deficient (i.e., mild or moderate) for a long time before
pregnancy, a lag period of about one trimester (i.e., 3 months) is inevitable before the thyroid
function normalizes following the onset of iodide supplementation (Glinoer 2005). This is the

178

-------
10-P-0101

reason for recommending taking iodide containing prenatal vitamins 3 months prior to
pregnancy, if possible.

The effectiveness of iodide supplementation during pregnancy can be observed in the six
randomized, controlled trials of iodide supplementation involving a total of 450 pregnant women
with mild to moderate iodide deficiency (Zimmermann 2004, Table 2). In all six trials, iodide
supplementation resulted in a significant increase in maternal UIC. The six trials found no
increase in maternal thyroid autoimmunity or in the prevalence or severity of postpartum thyroid
dysfunction. Iodide supplementation is efficacious for both the mother and newborn. Iodide
supplementation is associated with significantly reduced maternal thyroid volume and generally
lowers maternal TSH levels. Iodide supplementation prevents or minimizes the enlargement of
newborn thyroid volume and increased Tg levels. However, there are no clinical studies on the
effect iodide supplementation has on long-term outcomes, such as a child's mental development.

Results of Treatment of Mild Iodide Deficiency during Pregnancy with Iodide
Supplementation

The benefits of iodide supplementation can be observed in a 1995 prospective double-
blind, randomized trail study of 180 euthyroid women with excess thyroid stimulation who were
divided into three equal groups (Glinoer 1995). Excess thyroid stimulation was determined by
meeting the following criteria: serum Tg > 20 |ig/L and (fT4 index < 1.23 or T3/T4 ratio > 25
xlO"3). The screening of 2,000 pregnant women resulted in 180 pregnant women (i.e., 9% of the
screened population) meeting the excess thyroid stimulation criteria. Only pregnant women at
less than 16 weeks of gestation were enrolled in the study. Treatment began from the day of
enrollment to delivery. (Note: iodide supplementation in this study did not start 3 months before
pregnancy or at conception). Sixty women received a daily placebo (no treatment); another 60
women received a daily 100 |ig iodide supplement, and 60 women received a daily dose of 131
|ig iodide supplement and a daily dose of 100 |ig levothyroxine.

The results from the 1995 prospective double-blind, randomized trail study show the
benefits of supplementing pregnant women with iodide:

• Mean serum Tg in newborns of mothers with iodide supplementation was significantly
lower versus the controls (65 ± 6 |ig/L vs 113 ± 9 |ig/L (p = 0.0001), respectively).

• The mean UIC in newborns of mothers with iodide supplementation was significantly
higher versus controls (77 ± 8 |ig/L vs 43 ± 4 |ig/L (p = 0.0001), respectively). Even
with the improvement of 100 |ig iodide/day supplementation, the mean neonate UIC of
77 |ig/L falls still short of the WHO-recommended mean neonate UIC of 150 |ig/L
(Delange 2005). Furthermore, optimal neonate UIC is reported be as high as 180-225
|ig/L (Delange 2004). Since the study used only a 100 |ig dose of iodide/day
supplementation, the NAS Committee's and ATA's recommendation for prenatal
vitamins to contain 150 |ig iodide is supported by these data.

• The mean thyroid volume in newborns of mothers with iodide supplementation was
significantly less versus controls (0.76 ± 0.05|ig/L vs 1.05 ± 0.05 |ig/L (p = 0.0001),

179

-------
10-P-0101

respectively). In other words, the mean thyroid volume in newborns without maternal
iodide supplementation was 40% larger than newborns with maternal iodide
supplementation. Furthermore, glandular hyperplasia was already present at birth in 10%
of the neonates without maternal iodide supplementation (Glinoer 2005, p 1092). By
contrast, no instance of neonatal goiter is observed in neonates born to mothers with
iodide supplementation (Glinoer 2005, p 1092).

• Mothers with iodide supplementation have significantly higher iodide levels in their
breast milk versus controls (61 ± 10 |ig/L vs 29 ± 2 |ig/L (p = 0.001), respectively).
Breast milk less than 50 |ig/L is considered by WHO/UNICEF/ICCIDD to indicate
iodide deficiency in the population (Delange 1994 p,109). Clearly, iodide
supplementation dramatically improves the iodide levels in breast milk, which directly
improves the iodide nutrition of nursing infants in which critical brain development is
occurring.

These observations indicate that iodide supplementation during pregnancy reduces the
stress on the fetal thyroid so that the risk for developing subtle mental deficits is minimized.

10.2.1.2 Thiocyanate Stressor

The cumulative risk assessment has identified that the thiocyanate NIS stressor at typical
exposure concentrations is a medium-impact stressor in this public health issue. Common,
healthy vegetables in the diet are a significant and natural source of thiocyanates (Tonacchera
2004). Foods particularly rich in thiocyanate include cabbage, cauliflower, broccoli, Brussels
sprouts, corn (maize), mustard seed, and milk. For example, a nonsmoking individual consumed
100 grams of raw Brussels sprouts each day for a week to document a rise in thiocyanate serum
levels from 31 to 80 |imol/L. Furthermore, dietary thiocyanate exposure exhibits a seasonal
change of about 10 (amol/L presumably due to seasonal changes in the composition of the diet
(Foss 1986). Therefore, the human body is continuously and unavoidable exposed to
thiocyanate.

Thiocyanate may have a regulatory function in normal physiology (Middlesworth 1986).
Thiocyanate is continually synthesized in the normal rat. In the absence of all dietary intake of
thiocyanate in the rat, the thiocyanate serum concentration actually increased from 500 |ig/dl to
800 |ig/dl during the fasting (Middlesworth 1986, Figure 1). These data indicate that serum
thiocyanate in rats is maintained in a normal range. A "kidney threshold" is reported to exist in
humans at a thiocyanate serum threshold of 200-300 |imol/L (Tonnacchera 2004). These data
imply that humans also synthesize and maintain thiocyanate serum concentrations in a normal
range. Therefore, any remedy to significantly lower thiocyanate exposures - to decrease serum
thiocyanate concentration, which will increase the %TIU in adult and in particular pregnant
women - is not practical and will not be effective, due to the maintenance of thiocyanate serum
concentrations by the body.

The human nonsmoker thiocyanate exposure is typically in the range of 10-70 |imol/L in
the serum, and most occurs naturally through the consumption of healthy vegetables. Except for
the potential for occupational exposures to cyanide (e.g., electroplaters), the amount of

180

-------
10-P-0101

uncontrolled environmental cyanide exposure in the U.S. population is negligible (i.e., cyanide is
metabolized by the body into thiocyanate). By contrast, the major source of excess thiocyanate
exposure in humans occurs through smoking. The act of smoking increases the mean
thiocyanate serum concentration by a factor of 2.5x (i.e., from 40 to 100 |imol/L). The
thiocyanate exposure in smokers is typically in the range of 80-120 |imol/L. The increased risk
of fetal brain damage from smoking while pregnant is yet another reason to encourage pregnant
women not to smoke. Encouraging individuals not to smoke is clearly an activity outside EPA's
legislative mandate and authority.

In reviewing the potential for cyanide exposure in the U.S. population, the U.S. drinking
water limit (i.e., MCL) for cyanide is 0.2 mg/L (De Groef 2006). The consumption of 2 liters
water at the cyanide MCL per day would result in an exposure of 0.4 mg cyanide/day. This
amount of cyanide would be metabolized into 894 |ig thiocyanate/day (i.e., assuming 100%
conversion). Using the OPEC equation to convert exposure to 894 |ig thiocyanate/day into
perchlorate equivalent amount results in 51 |ig OPEC (i.e., the perchlorate RfD = 49 |ig/day).
Therefore, consumption of 2 liters of drinking water contaminated at the MCL for cyanide
induces the same amount of NIS inhibition as perchlorate exposure at the RfD.

10.2.1.3 Nitrate Stressor

The cumulative risk assessment has identified that the nitrate NIS stressor at typical
exposure concentrations is a weak impact stressor in this public health issue. Human exposure to
nitrate is both exogenous and endogenous. Exogenous human nitrate exposure is from diet
through the consumption of primarily vegetables. Nitrate is common in food (Tonacchera, 2004)
and occurs in green leafy vegetables. Nitrate is also added to processed meats as a preservative.
Nitrate commonly contaminates both surface water and groundwater sources of drinking water.
Furthermore, the endogenous human nitrate exposure is from the body's production of nitric
oxide, which is subsequently converted into nitrate by various types of cells in the body (NAS
1995, page 37). Therefore, humans are continuously and unavoidably exposed to nitrate.

Most nitrogenous materials in natural waters tend to be converted to nitrate, so all sources
of combined nitrogen, particularly organic nitrogen and ammonia, should be considered as
potential nitrate sources (EPA 2008). Primary sources of organic nitrates include human sewage
and livestock manure, especially from feedlots. The primary inorganic nitrates which may
contaminate drinking water are potassium nitrate and ammonium nitrate, both of which are
widely used as fertilizers (EPA 2008). According to the Toxics Release Inventory, releases to
water and land totaled over 112 million pounds from 1991 through 1993. The largest releases of
inorganic nitrates occurred in Georgia and California at 12.0 and 21.8 million pounds (EPA
2008). EPA's 1999 study on the occurrence of contaminants in Public Water Systems (PWSs)
regulated under the Safe Drinking Water Act identified that of the 8,988 groundwater PWSs
tested for nitrate, 85.9% had at least one analytical result above the minimum reporting limit,
11.4% had at least one analytical result above '/2 MCL, and 2.3% had at least one analytical
result above the MCL (EPA 1999, Appendix A).

In studies from Denmark and the United Kingdom, the average nitrate intake is
estimated to be about 40-50 mg/day for adults (Tonacchera 2004). This value is on the low end

181

-------
10-P-0101

of the nitrate intake estimates reported by the National Academy of Sciences (see Section 3.4
above; NAS 1995, p 35-43) and represents a bias low estimate of the average daily nitrate intake.
In the Western world, the typical nitrate serum concentration ranges from 10-140 |imol/L with
the mean nitrate serum concentration being between 30-50 |imol/L (Tonacchera 2004). The
Tonacchera Model indicates that at typical nitrate exposure level of 40 |imol/L, nitrate
contributes only about 11% of the body's typical total NIS inhibition load of 1.501 |imol/L.
Potentially reducing the nitrate exposure will only marginally improve the %TIU values in
humans and in particular pregnant women.

The U.S. drinking water MCL for nitrate is 10 mg of nitrate nitrogen per liter4 (EPA
1991b), which is equivalent to 44 mg of nitrate per liter (NAS 1989, p 5, 11, 45). The
conversion factor between these two analytical units is 1 mg of nitrate nitrogen per liter, equal to
4.429 mg of nitrate per liter. Therefore, the consumption of 2 liters water at the EPA nitrate
MCL per day would result in an exposure of 88 mg nitrate/day (i.e., this is about twice the
average daily intake of nitrate). Using the OPEC equation to convert exposure to 88 mg
nitrate/day into PEC results in a 587 |ig/day OPEC (i.e., the perchlorate RfD = 49 |ig/day).
Therefore, consumption of 2 liters of drinking water contaminated at the nitrate MCL induces 12
times the amount of NIS inhibition as perchlorate exposure at the RfD (De Groef 2006).

In 1999, EPA identified that 2.3% of the groundwater PWSs tested for nitrate had at least
one analytical result above MCL (EPA 1999, appendix A). Furthermore, in 1999, EPA
identified that 11.4% of groundwater PWSs tested for nitrate had at least one analytical result
above V2 MCL (EPA 1999, appendix A). The consumption of 2 liters of drinking water
contaminated at the V2 nitrate MCL induces 6 times the amount of NIS inhibition as perchlorate
exposure at the RfD. In regard to private groundwater wells, 9% of domestic groundwater wells
(irrespective of depth to groundwater) sampled by the U.S. Geological Survey's National Water-
Quality Assessment Program during 1993-2000 had nitrate levels above the EPA nitrate MCL of
44 mg/L for drinking water (Nolan 2002). This nitrate contamination of private groundwater
wells is a potential health issue for the estimated 42 million Americans that drink unregulated
groundwater from their own private wells. By comparison, the consumption of drinking water
contaminated with 3.7 mg/L of nitrate (i.e., 1/12th the nitrate MCL) has the equivalent NIS
inhibition as the consumption of drinking water contaminated at 24.5 ppb perchlorate.
Considering the nitrate contamination of both PWSs and private wells, about 10% of U.S.
population is consuming nitrate-contaminated water at levels that induce at least 6 times more
NIS inhibition than perchlorate exposure at the RfD. Therefore, lowering the public's nitrate
exposure provides a more meaningful opportunity to lower the public's NIS inhibition load (i.e.,
a more meaningful opportunity to lower public risk) than to lower the public's perchlorate
exposure below the perchlorate RfD.

4 The current EPA MCL for nitrate is drinking water is expressed as 10 mg nitrate nitrogen per liter (i.e., which is
the mass of nitrogen present in the nitrate form per liter), which is equivalent to 44 mg of nitrate per liter (NAS
1989, 5,11, 45). The conversion factor between these two analytical units is 1 mg of nitrate nitrogen per liter =
4.429 mg of nitrate per liter (NAS 1989, 10).

182

-------
10-P-0101

10.2.1.4 Perchlorate Stressor

The cumulative risk assessment has identified that the thiocyanate NIS stressor at the
RfD exposure concentrations is a very weak impact stressor in this public health issue. The
FDA's perchlorate TDS identified that perchlorate is present at very low levels in food, primarily
through the consumption of dairy and vegetables (Murray 2008, p 5). Furthermore, both
surface and ground sources of drinking water can be contaminated with perchlorate. Therefore,
the human body is continuously and unavoidably exposed to low levels of perchlorate.

The FDA TDS estimates a perchlorate intake of 5.4 to 6.8 |ig/day from food in 25- to 30-
year-old women (Murray 2008, table 5). The FDA TDS estimate of perchlorate intake agrees
with the CDC biomonitoring data that estimates the median perchlorate exposure in the U.S.
population is 4.6 |ig/day (Blount 2006). The Tonacchera Model indicates that a median
perchlorate exposure level of 4.6 |ig/day, observed in the CDC biomonitoring data, contributes
only about 0.09% of an adult body's typical total NIS inhibition load of 1.501 |imol/L.
Furthermore, the Tonacchera Model indicates that a perchlorate exposure at the RfD contributes
only 0.9% (less than 1%) of an adult body's typical total NIS inhibition load of 1.501 |imol/L.
Finally, the Tonacchera Model also estimated a 0.3% increase in %TIU in adults and 1.0%
increase in the %TIU of pregnant women if the actual perchlorate drinking water exposure is
reduced from 24.5 ppb to 6.1 ppb. Therefore, potentially reducing the maximum perchlorate
exposure in drinking water from 24.5 ppb to 6.0 ppb or lower will not significantly improve the
%TIU values in adults and, in particular, pregnant women.

The remedy of potentially decreasing the perchlorate DWEL from 24.5 ppb to 6.1 ppb
will help only a very small portion of the at-risk population. Conceptually, a theorectial 1.0%
increase in %TIU applied to the entire at-risk poplution would be expected to increase at most
2% of the pregnant woman population above of critical %TIU level of 49% (i.e., %TIU(noael) ),
which would leave about 27% of the pregnant woman population below the critical %TIU level.
However, the potential remedy of lowering the perchlorate DWEL from 24.5 ppb to 6.0 ppb will
increase the %TIU in only a small subset of this population. The EPA's Unregulated
Contaminant Monitoring Rule (UCMR 1) data indicate only 2.13% of the PWSs are
contaminated with perchlorate greater than 7 |ig/L, which effects a population size of
approximately 2.2 million people (EPA 2007, table 5). Therefore, about 0.7% of the U.S.
population (i.e., 2.2 million ^ 300 million x 100%) is potentially exposed to drinking water
containing > 7 ug/L and potentially able to benefit from the lowering of perchlorate exposure in
drinking water. Assuming pregnant women are evenly distributed throughout the U.S.
population, the act of lowering the perchlorate DWEL from 24.5 ppb to 6.0 ppb would be
expected to increase at most 0.014% of the annual pregnant woman population (i.e., 2% x 0.007)
above of critical %TIU level of 49%. This potential remedy leaves 28.99% of the estimated 29%
of U.S. pregnant woman population with a %TIU < 49% still at risk for subtle mental deficits in
their children. In other words, the act of lowering perchlorate drinking water exposure below
24.5 ppp provides a neglible decrease in the size of the at-risk population. By contrast,
implementing prenatal vitamins containing iodide has the potential to pull all 29% of the iodide-
deficient pregnant women in the United States successfully above the critical %TIU(noael) value,
and at a relatively low implementation cost.

183

-------
10-P-0101

A 1.0% reduction in %TIU could be offset by a 1.0% increase in the UIC in pregnant
women. At a borderline iodide-deficient UIC of 100 |ig/L in pregnant women, a 1.0% increase
in %TIU is caused by the additional intake of about 1.1 jug of iodide. ATA recommends that
pregnant and lactating women take a daily prenatal vitamin/mineral supplement containing 150
|ig of iodide (ATA 2006). Taking an iodide-containing prenatal vitamin/mineral supplement
would easily offset a 1% reduction in %TIU. Furthermore, iodized salt in the United States
contains about 100 ppm potassium iodide or 77 |ig iodide/gram of salt (Dunn 1998). The
increased consumption of 14.3 mg of iodized salt would provide the 1.1 |ig of iodide required to
offset the 1% reduction in %TIU. Since a pinch of table salt weights about 460 mg, 14.3 mg of
salt is equivalent to about l/32nd of a pinch of salt. Since a smidgen of table salt weights about
230 mg, 14.3 mg of salt is equivalent to about 1/16th a smidgen of salt.

In reviewing the exposure to NIS inhibitors in drinking water, about 11.4% of
groundwater PWSs tested for nitrate had at least one analytical result above V2 MCL, which
corresponds to an amount of NIS inhibition 6 times greater than the amount of NIS inhibition
from perchlorate exposure at the DWEL (i.e., at the RfD). Furthermore, about 2.3% of
groundwater PWSs tested for nitrate had at least one analytical result above MCL, which
corresponds to an amount of NIS inhibition 12 times greater than the amount of NIS inhibition
from perchlorate exposure at the RfD. Furthermore, these data indicate that a significant portion
of the U.S. population has a much higher risk from NIS inhibition from nitrate from drinking
water than the risk from NIS inhibition from perchlorate exposure from drinking water at the
RfD.

From a risk management standpoint, the Federal Government and the States should
weigh the effectiveness of potential remedies to this public health issue and the cost of
implementing each remedy against the health benefits to be derived by each remedy. The
following are three potential remedies that can potentially address this public health issue; each
has varying degrees of effectiveness:

• Add iodide to prenatal vitamin supplements and implement their use before and during
pregnancy and lactation.

• Reassess the nitrate drinking water MCL and consider lowering the MCL to decrease the
public's NIS inhibition exposure.

• Lower the perchlorate drinking water exposure level from 24.5 ppb to 6 ppb or lower.

184

-------
10-P-0101

11. Conclusion

The OIG Analysis of the scientific literature identified that the risk from perchlorate
exposure is only part of the larger public health issue that is defined by the subtle mental deficits
occurring in children born to mothers with low maternal TIU during pregnancy and nursing. The
TIU results from the combined biological effect of four NIS stressors acting on the thyroid:
thiocyanate, nitrate, perchlorate, and lack of iodide. Diet constantly exposes everyone to each of
the four NIS stressors, and an individual's TIU level is the result of the combined effect of all
four NIS stressors, not just perchlorate exposure. The OIG Analysis concludes that a single
chemical risk assessment of perchlorate is not sufficient to assess and characterize the combined
human health risk from all four NIS stressors. However, both EPA's draft perchlorate RfD from
the Argus rat study and the NAS Committee's recommended perchlorate RfD from the Greer
study used a single chemical risk assessment approach. By contrast, our cumulative risk
assessment better characterizes the nature and sources of risk affecting this public health issue.

All four NIS stressors meet EPA's risk assessment guidance requirements for conducting
a cumulative risk assessment using the dose-addition method. In the OIG Analysis, we
conducted a cumulative risk assessment and determined that the risk from each of the four NIS
stressors is not equal. The OIG Analysis also independently confirmed that EPA's perchlorate
RfD is conservative and protects human health, but limiting perchlorate exposure does not
effectively address this public health issue. Potentially lowering the perchlorate drinking water
limit from 24.5 ppb to 6 ppb does not provide a meaningful opportunity to lower the public's
risk. By contrast, addressing moderate and mild iodide deficiency that occurs in about 29% of
the U.S. pregnant and nursing population appears to be a more effective approach of increasing
TIU to healthy levels during pregnancy and nursing, thereby reducing the frequency and severity
of permanent mental deficits in children.

The NAS Committee recommended that"... consideration be given to adding iodide to
all prenatal vitamins" (NAS 2005, p 18). Our independent cumulative risk assessment came to a
similar conclusion. The most effective and efficient approach for reducing the health risk of
permanent mental deficits in children from low maternal TIU during pregnancy and nursing is to
require that iodide be added to all prenatal vitamins and to encourage women to take them before
and during pregnancy and while nursing.

185

-------
10-P-0101

Appendix A

"Whole Mixture" Approach to the
Cumulative Risk Assessment of Perchlorate

An alternative approach to conducting a cumulative risk assessment is the "whole
mixture" approach. Risk assessors can accomplish a whole mixture approach by developing a
statistical model on human epidemiological data. An example of this whole mixture approach
that has received a lot of attention is the Blount analysis (Blount 2006, Table 6) and the
subsequent Steinmaus analysis of the NHANES 2001-2002 epidemiological study. The Blount
analysis and the Steinmaus analysis represent statistical models fitted to the NHANES 2001-
2002 epidemiological data. The Blount analysis predicted a decrease of 1.64 |ig/dL in tT4 serum
from a 13 |ig/day perchlorate exposure in women > 12 yrs of age with a UIC <100 |ig/L due to
the NIS inhibition of perchlorate. These women are moderate to mildly iodide deficient. The
subsequent Steinmaus analysis (Steinmaus 2007) on the same NHANES 2001-2002
epidemiological data concluded that increasing thiocyanate exposure potentiates the rate of
decrease in tT4 with increasing perchlorate exposure.

Using EPA criteria, our review indicates that the whole mixture approach to risk
assessment of perchlorate (i.e., Blount analysis and the Steinmaus analysis) is not sufficiently
developed and corroborated to be either the basis for developing a perchlorate RfD or the basis
for establishing a potential perchlorate drinking water limit for the following reasons:

1. Section 3.1.3 of EPA's Draft Guidance on the Development, Evaluation, and Application
of Regulatory Environmental Models (November 2003) establishes the need to
corroborate all environmental models before risk assessors use the models as the basis for
rulemaking or regulation (EPA 2003a). The Blount and Steinmaus analyses represent a
statistical model of the toxicity of perchlorate; their findings must be corroborated with
other epidemiological data and other human exposure studies of perchlorate before they
should be used as the basis for rulemaking. No human epidemiological study has
corroborated the results of the Blount and Steinmaus analyses. To the contrary, the
available epidemiological data of perchlorate-exposed pregnant women with UIC <100
|ig/L contradict the findings of the Blount and Steinmaus analyses. These
epidemiological studies included the following:

a. A cohort of 789 pregnant women (i.e., 396 from Cardiff, Wales; 311 from Turin,

Italy; and 82 from Dublin, Ireland) with perchlorate exposure levels shows no
association between urinary perchlorate levels and free T4 levels in the pregnant
women as a group or in the subgroup of pregnant women with UIC <100 |ig/L
(Pearce 2007b).

b. A cohort of 230 pregnant women (i.e., 128 from Los Angeles, California; 102 from
Cordoba, Argentina) with low perchlorate exposure levels predicted by the Blount
analysis to induce a reduction of tT4 shows no association between urinary

A-1

-------
10-P-0101

perchlorate levels and free T4 levels in the pregnant women as a group or in the
subgroup of pregnant women with UIC <100 |ig/L (Pearce 2008).

c. From the 2005 Tellez cohort study (Tellez 2005) in three Chilean cities, a reanalysis
of the data identified 16 of the pregnant women had UIC <100 |ig/L (Gibbs 2008).
The reanalysis shows no association between urinary perchlorate levels and free T4
levels in this subset of pregnant women with UIC <100 |ig/L.

2. Pregnant women with a tT4 decrease of 3.2 |ig/dL from a normal tT4 level have
hypothyroidism. Researchers have observed that the children of these women have lower
cognitive, attention, and motor performance scores (Haddow 1999, tables 2 and 4).
However, we have identified adverse effects in children of pregnant women under less
thyroid stress than hypothyroidism (i.e., hypothyroxinemia and low TIU). Therefore, the
Blount analysis, which predicted a decrease of 1.64 |ig/dL in tT4 from a 13 |ig/day
perchlorate exposure in women with a UIC <100 |ig/L, is anticipated by our technical
review by ICF International to be sufficient to induce adverse effect in the children.

Thus, from the Blount analysis, 13 jug of perchlorate exposure per day could be
considered the LOAEL. However, the NAS Committee derived its recommended
perchlorate RfD from a NOEL of 0.007 |ig/kg-day. Assuming a 70-kg person and a
LOAEL of no less than 10 times greater than the NOEL, the estimated LOAEL
equivalent from the Greer study would be at least roughly 4,900 |ig/day. The Blount
analysis suggests that the perchlorate toxicity is significantly more toxic (i.e., more than
300 times more toxic) than what the NAS Committee and all previous perchlorate studies
have indicated. Typical human exposures to the other common NIS inhibitors -
thiocyanate and nitrate - induce substantially more NIS inhibition than 13 |ig/day of
perchlorate, but are not observed to induce thyroid hormone changes in the Blount
analysis. Therefore, the increased toxicity of perchlorate reported in the Blount analysis
would not be accounted for by the known mechanism of perchlorate toxicity: the
inhibition of NIS. Furthermore, the Blount analysis does not propose a new mechanism
of perchlorate toxicity that accounts for their reported increased toxicity of perchlorate.
Until researchers can explain the increased toxicity of perchlorate by a verified biological
mechanism, regulators should not use the Blount analysis as a basis for developing a
perchlorate RfD, nor should it be used as the basis for establishing a drinking water limit.

3. The NHANES 2001-2002 epidemiological urinary data are from a single, untimed spot
urinalysis and can vary due to dilution or concentration of the urine (i.e., due to an
individual's hydration status). A common technique to eliminate this variation in the spot
urinalysis is to normalize the iodide urine concentration against creatinine, resulting in
the analysis expressing iodide excretion as |ig of iodide/g of creatinine (i.e., |ig/g
creatinine) instead of |ig of iodide/L of urine. Lamm has successfully repeated Blount's
statistical analysis of the NHANES 2001-2002 epidemiological data only when his
analysis measures UIC as |ig/L (Lamm 2007). However, when Lamm repeats the
statistical analysis using only women of childbearing age (ages 15-44) and UIC measured
as |ig/g creatinine, the statistical significance of the relationship between decreasing tT4
with increasing perchlorate exposure in women with low UIC is lost. The statistical
significance of this relationship should be present regardless of whether the analysis

A-2

-------
10-P-0101

expresses UIC in |ig/L or |ig/g of creatinine. This inconsistency suggests that the
relationship between tT4 and perchlorate reported in the Blount analysis might be an
artifact in the NHANES dataset.

4. Section 3.2 of EPA's Supplementary Guidance for Conducting Health Risk Assessment of
Chemical Mixtures establishes a major requirement for the use of a whole mixture risk
assessment approach - the composition of the mixture must be stable over time; the
relative proportions of chemicals in the mixture are roughly constant so that the mixture
can be treated as though it were a single chemical (EPA 2000). Examples of EPA whole
mixture risk assessments include Araclor 1016, Araclor 1254, and coke oven emissions.
However, human exposure to each of the four NIS stressors varies independently based
on the dietary intake preferences of the individual. Therefore, the mixture of the four
NIS stressors is not a fixed, stable composition. Thus, a whole mixture approach is not
well suited to assessing the toxicity for this public health issue since exposure to each of
the four NIS stressors varies.

5. The OIG Analysis indicates that the toxicity of perchlorate is not fixed and cannot be
described by a single RfD; rather, the toxicity of perchlorate changes in response to the
exposure level of the other three NIS stressors, especially the lack of iodide stressor.

From clinical observations of populations with endemic cretinism and from the OIG
Analysis, at higher iodide nutrition levels, the body can tolerate a higher total goitrogen
load. The converse is also true: at lower iodide nutrition levels, the body's tolerance of
total NIS inhibition is much less. The Blount analysis also observes this trend in the
toxicity of perchlorate. The Blount analysis reports that 13 |ig/day of perchlorate
exposure induces toxicity (i.e., a decrease in tT4) in women with UICs < 100 |ig/L, but
the same 13 |ig/day of perchlorate exposure is nontoxic in women with UICs >100 |ig/L.
From both clinical observations (i.e., in Africa and India) and our cumulative risk
assessment, reducing NIS inhibition exposure cannot correct a low TIU caused by iodide
deficiency. Only the addition of iodide to the diet can correct a low TIU caused by iodide
deficiency.

6. Section 3.1.3.2 of EPA's Draft Guidance on the Development, Evaluation, and
Application of Regulatory Environmental Models defines robustness as the capacity of a
model to perform equally well across the full range of environmental conditions (EPA
2003a). This guidance further elaborates that if the data used to calibrate the model are
identical or statistically similar to the dataset used to corroborate a model, the study did
not provide an independent measure of the model's performance. Therefore, the
Steinmaus analysis does not corroborate the Blount analysis, because both studies use the
same NHANES 2001-2002 dataset. Furthermore, CDC plans to use the next NHANES
dataset to continue to examine this interface of environmental perchlorate exposure,
iodine levels, and tobacco smoke exposure (thiocyanate). However, repeating the
analysis in the next NHANES dataset would not represent an independent evaluation of
the robustness or the performance of the model because it is still a statistically similar
dataset.

A-3

-------
10-P-0101

7. The December 13, 2006, ATA Public Health Statement commented on the Blount
analysis. ATA identified that the NHANES 2001-2002 measured serum thyroxine as tT4
rather than as free T4. Free T4is the most frequently used clinical measurement and the
physiologically available form of thyroxine.

8. The scientific literature reports that several dietary NIS stressors biologically affect the
uptake of iodide, not to mention other chemical exposures affecting other subsequent
biological steps in the production of T4 that could contribute to the decrease in serum tT4
observed in the Blount analysis. The Blount analysis reports an R2 value of 0.240. R2 is
a statistical value called the "proportion of variation explained," and is the proportion of
scatter in a dataset that the statistical model explains. The R2 value of 0.240 reported in
the Blount analysis shows that perchlorate accounts for only 3% of the variation seen in
the serum tT4 (Charnley 2008). If only 3% of the tT4 change predicted by the Blount
analysis was from perchlorate, the toxicity level of perchlorate reported in the Blount
analysis would be more consistent with the other perchlorate studies and with the NAS-
recommended perchlorate RfD.

A-4

-------
10-P-0101

Appendix B

References

Ahidjo A, Tahir A, Tukur MA. 2006. Ultrasound determination of thyroid gland volume among
adult Nigerians. The Internet Journal of Radiology 2 (4).

http://www.ispub.com/ostia/index.php7xmlFilePathHournals/iira/vol4n2/thyroid.xml (accessed
February 13, 2008).

Alexander WD, Wolff J. 1966. Thyroidal iodide transport VIII, relation between transport
goitrogenic and antigoitrogenic properties of certain anions. Endocrinology 78:581-90.

Amitai Y, Winston G, Sack J, Lewis M, Blount BC, Valentin-Blasini L, Fisher N, Isreali A,
Leventhal A. 2007. Gestational exposure to high perchlorate concentrations in drinking water
and neonatal hhyroxine levels. Thyroid 17(9):843-50.

Anderson S, Pedersen KM, Pedersen IB, Laurberg P. 2001. Variations in urinary iodine
excretion and thyroid function. A 1-year study in healthy men. Eur J Endocrinol 144(5):461-65.

Aquaron R. 2000. Iodine content of non-iodized salts and iodized salts obtained from the retail
markets worldwide. 8th World Salt Symposium 2:935-40.

Argus. 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. Argus 1416-003. Argus
Research Laboratories, Inc., Horsham, PA.

American Thyroid Association (ATA). 2006. Iodine supplementation for pregnancy and
lactation - United States and Canada: Recommendations of the American Thyroid Association.
First draft initiated by Dr. John T. Dunn before his untimely death in 2004. Draft was completed
by members of the Public Health Committee of ATA. Thyroid 16(10):949-51.

Banerjee KK, Dishayee A, Marimuthu P. 1997. Evaluation of cyanide exposure and its effect
on thyroid function of workers in a cable industry. Journal Occupational Environmental
Medicine 39(3):258-60.

Baneijee KK, Marimuthu P, Bhattacharyya P, Chatterjee M. 1997a. Effect of thiocyanate
ingestion through milk on thyroid hormone homeostasis in women. British Journal of Nutrition
78:679-81.

Barbaresi W, Katusic S, Colligan R, Weaver A, Pankratz V, Mrazek D, Jacobsen S. 2004. How
common is attention-deficit/hyperactivity disorder? Towards resolution of the controversy:
Results from a population-based study. Acta Paediatr Suppl 93(445):55-59.

B-1

-------
10-P-0101

Barnes DG, Dourson M. 1988. Reference dose (RfD): Description and use in health risk
assessments. Regulatory Toxicology and Pharmacology 8:471-86.

Bianco AC, Larsen PR. 2005. Chapter 7: Intracellular pathways of iodothyronine metabolism.
In The Thyroid: A Fundamental Clinical Text. 9th ed. Philadelphia: Lippincott Williams &
Wilkins (ISBN 0-7817-5047-4).

Bleichrodt N, Born MP. 1994. Chapter 19: A metaanalysis of research on iodine and its
relationship to cognitive development. 279-285. In The Damaged Brain of Iodine Deficiency:
Cognitive, Behavioral, Neuromotor, Educative Aspects. JB Stanbury, Ed. New York:

Cognizant Communication Corp.

Bleichrodt N, Del Rey FE, Morreale de Escobar G, Garcia I, Rubio C. 1989. Iodine deficiency,
implications for mental and psychomotor development in children. 269-287. In Iodine and the
Brain. Delong GR, Robblins J, Condliffe PG, Eds. New York: Plenum Press.

Blount BC, Valentin-Blasini L, Osterloh JD, Mauldin JP, Pirkle JL. 2006a. Perchlorate
exposure of the U.S. population, 2001-2002. Journal of Exposure Science and Environmental
Epidemiology, 1-8.

Blount BC, Pirkle JL, Osterloh JD, Valentin-Blasini L, Caldwell L. 2006b. Urinary perchlorate
and thyroid hormone levels in adolescent and adult men and women living in the United States.
Environmental Health Perspectives 114(12): 1865-71.

Borak JB. 2002. Comment on: [California's] draft public health goal for perchlorate in
drinking water (peer review draft March 2002). Prepared on behalf of Lockheed Martin
Corporation. Jonathan Borak, MD, DABT, Associate Clinical Professor of Medicine and
Epidemiology, Yale School of Medicine, New Haven, CT.

Braverman LE, XueMei H, Pino S, Cross M, Magnani B, Lamm SH, Kruse MB, Engel A,
Crump KS, Gibbs JP. 2005. The effect of perchlorate, thiocyanate, and nitrate on thyroid
function in workers exposed to perchlorate long-term. Journal of Clinical Endocrinology &
Metabolism 90(2):700-706.

Bruce GM, Peterson MK, and Pleus RC. 2004. Comparative contribution of perchlorate and
anti-thyroid agents in American diets to iodide uptake inhibition. Paper presented at the 32nd
JANNAF Propellant Development & Characterization and 21st Safety & Environmental
Protection Joint Meeting. Seattle, WA, July 29, 2004.

Bruce G. 2005. Letter regarding perchlorate/nitrate/thiocyanate comparison written to David
Huber, U.S. EPA/Office Ground Water Drinking Water/Risk Management Division, July 13,
2005.

Calaciura F, Motta RM, Miscio G, Fichera G, Leonardi D, Carta A, Trischitta V, Tassi V, Sava
L, Vigneri R. 2002. Subclinical hypothyroidism in early childhood: A frequent outcome of

B-2

-------
10-P-0101

transient neonatal hyperthyrotropinemia. Journal of Clinical Endocrinology & Metabolism
87(7): 3209-14.

Callahan MA, Sexton K. 2007. If cumulative risk assessment is the answer, what is the
question? Environmental Health Perspectives, 115(5):799-806.

CarrascoN. 2005. Chapter 4: Thyroid hormone synthesis. In The Thyroid: A Fundamental
Clinical Text. 9th ed. Philadelphia: Lippincott Williams & Wilkins. ISBN 0-7817-5047-4.

Charnley G. 2008. Perchlorate: Overview of risks and regulation. Food and Chemical
Toxicology. (Epub ahead of print - March 10, 2008). doi:10.1016/j.fct.2008.03.006

Clewell RA, Merrill EA, Gearheart JM, Robinson PJ, Sterner TR, Mattie DR, Clewell III HJ.
2007. Perchlorate and radioiodide kinetics across life stages in the human: Using PBPK models
to predict dosimetry and thyroid inhibition and sensitive subpopulations based on developmental
stages. Journal of Toxicology and Environmental Health, Part A, 70:408-28.

Clewell RA, Merrill EA, Yu KO, Mahle DA, Sterner TR, Fisher JW, Gearhart JM. 2003.
Predicting neonatal perchlorate dose and inhibition of iodide uptake in the rat during lactation
using physiological-based pharmacokinetic modeling. Toxicology Sciences 74:416-36.

Crump C, Michaud P, Tellez R, Reyes C, Gonzalez G, Montgomery EL, Crump KS, Lobo G,
Becerra C, Gibbs JP. 2000. Does perchlorate in drinking water affect thyroid function in
newborns or school-age children? J Occup Environ Med, 42(6):603-12.

CDCP/NCHS. 2007. Centers for Disease Control and Prevention/National Center for Health
Statistics (CDCP/NCHS). http://www.cdc.gov/nchs/products/pubs/pubd/hestats/iodine.htm
(accessed July 25, 2007).

Dasgupta KP, Dyke JV, Kirk AB, Jackson WA. 2006. Perchlorate in the United States analysis
of relative source contributions to the food chain. Environ. Sci. Technol. 40:6608-14.

DeFur PL, Evans GW, Hubal EAC, Kyle AD, Morello-Frosch RA, Williams DR. 2007.
Vulnerability as a function of individual and group resources in cumulative risk assessment.
Environmental Health Perspectives 115(5): 817-24.

De Groef B, Decallonne BR, Van der Geyten S, Darras VM, Bouillon R. 2006. Perchlorate
versus other environmental sodium/iodide symporter inhibitors: Potential thyroid-related health
effects. Eur J Endocrinol 155(1): 17-25.

Delange FM, Dunn JT. 2005. Chapter 1 IE: Iodine deficiency. In The Thyroid: A Fundamental
Clinical Text. 9th ed. Philadelphia: Lippincott Williams & Wilkins. ISBN 0-7817-5047-4.

Delange FM. 2005a. Chapter 49: Endemic cretinism. In The Thyroid: A Fundamental Clinical
Text. 9th ed. Philadelphia: Lippincott Williams & Wilkins. ISBN 0-7817-5047-4.

B-3

-------
10-P-0101

Delange F. 2004. Optimal iodine nutrition during pregnancy, lactation and the neonatal period.
Int J Endocrinol Metab 2:1-12.

Delange F. 2001. Iodine deficiency as a cause of brain damage. Postgraduate Medical Journal
77:217-20.

Delange F. 1998. Screening for congential hypothyroidism used as an indicator of the degree of
iodine deficiency and of its control. Thyroid 8(12): 1185-92.

Delange F. 1994. The disorders induced by iodine deficiency. Thyroid 4(1): 107-28.

De La Vieja A, Dohan O, Levy O, Carrasco N. 2000. Molecular analysis of the sodium/iodide
symporter: Impact on thyroid and extrathyroid pathophysiology. Physiol Rev 80:1083-1105.

Demers LM, Spencer CA. Eds. 2003. Laboratory support for the diagnosis and monitoring of
thyroid disease. Laboratory Medicine Practice Guidelines prepared by the National Academy of
Clinical Biochemistry. Franklin, TN: Durik Adverstising, Inc. Available from the American
Association of Clinical Chemistry at www.AACC.org. Monograph originally published in
Thyroid, 13(1): 1-126.

Dewey KG, Lonnerdal B. 1983. Milk and nutrient intake of breast-fed infants from 1-6 months:
Relation to growth and fatness. J Pediatr Gastroenterol Nutr 2(3):497-506.

Dohan O, Portulano C, Basquin C, Reyna-Neyra A, Amzel LM, Carrasco N. 2007. The Na+/I-
symporter (NIS) mediates electroneutral active transport of the environmental pollutant
perchlorate. Proceedings of the National Academy Science 104(51):20250-55.

Dorato MA, Engelhardt JA. 2005. The no-observed-adverse-effect-level in drug safety
evaluations: Use, issues, and definition(s). Regulatory Toxicology and Pharmacology 42:265-
74.

Dorea JG. 2004. Maternal thiocyanate and thyroid status during breast-feeding. Journal of
American College of Nutrition 23:97-101.

Dome JLCM, Renwick AG. 2005. Forum: The refinement of uncertainty/safety factors in risk
assessment by the incorporation of data on toxicokinetic variability in humans. Toxicological
Sciences 86(l):20-26.

Dunn JT. 1998. Editorial: What's happening to our iodine? Journal of Clinical Endocrinology
and Metabolism 83(10):3398-3400.

Dusdieker LB, Stumbo PJ, Kross BC, Dungy CI. 1996. Does increased nitrate ingestion elevate
nitrate levels in human milk? Arch Pediatr Adolesc Med 150:311-14.

EPA 2010. ORD Assistant Administrator Paul Anastas memorandum titled: ORD: The Path
Forward (Dated March 4, 2010). Washington, DC: U.S. Environmental Protection Agency,

B-4

-------
10-P-0101

Office of Research and Development.

http://v26265ncav001.aa.ad.epa.gOv/opencms/export/ord@work/organization/labscentersoffices/
ioaa/The Path Forward.html

EPA. 2009. The U.S. Environmental Protection Agency's Strategic Plan for Evaluating the
Toxicity of Chemicals. EPA/100/K-09/001. Washington, DC: U.S. Environmental Protection
Agency, Office of the Science Advisor, Science Policy Council.

EPA. 2009a. EPA memorandum from OSWER Assitant Administrator to Regional
Administrators. Subject: Revised Assessment Guidance for Perchlorate, January 8, 2009.
http://www.epa.gov/swerffrr/documents/perchlorate memo 01-08-09.pdf.

EPA. 2009b. A Citizen's Guide to Radon: The Guide to Protecting Yourself and Your Family
From Radon. EPA 402/K-09/001. Washington, DC: U.S. Environmental Protection Agency,
Indoor Air Quality, January 2009. http://www.epa.gov/radon.

EPA. 2008. Consumer Factsheet on Nitrates/Nitrites. U.S. Environmental Protection Agency,
Office of Ground Water and Drinking Water.

http://www.epa.gov/OGWDW/contaminants/dw contamfs/nitrates.html (accessed March 4,
2008).

EPA. 2008a. Interim Drinking Water Health Advisory for Per chlorate. EPA 822-R-08-025.
Washington, DC: U.S. Environmental Protection Agency, Office of Water, Health and
Ecological Criteria Division, December 2008.

http://www.epa.gov/safewater/contaminants/unregulated/perchlorate.html.

EPA. 2007. Drinking Water: Regulatory Determinations Regarding Contaminants on the
Second Drinking Water Contaminant Candidate List— Preliminary Determinations. Published
in the Federal Register on May 1, 2007 (72 FR 24015).

EPA. 2007a. Concepts, Methods, and Data Sources for Cumulative Health Risk Assessment of
Multiple Chemicals, Exposures, and Effects: A Resource Document. EPA/600/R-06/013F.
Washington, DC: U.S. Environmental Protection Agency, Office of Research and Development,
National Center for Environmental Assessment, August 2007.

EPA. 2006. Peer Review Handbook. 3rd Ed. EPA/100/B-06/002. Washington, DC: U.S.
Environmental Protection Agency, Science Policy Council.

EPA. 2004. Estimated Per Capita Water Ingestion and Body Weight in the United States - An
Update. EPA-822-R-00-001. Washington, DC: U.S. Environmental Protection Agency, Office
of Water/Office of Science and Technology, October 2004.

EPA. 2003. Framework for Cumulative Risk Assessment. EPA/600/P-02/001F. Washington,
DC: U.S. Environmental Protection Agency, Office of Research and Development, National
Center for Environmental Assessment.

B-5

-------
10-P-0101

EPA. 2003 a. Draft Guidance on the Development, Evaluation, and Application of Regulatory
Environmental Models. Prepared by the Council for Regulatory Environmental Modeling.
Washington, DC: U.S. Environmental Protection Agency, Office of Science Policy, Office of
Research and Development, November 2003.

EPA. 2002. A Review of the Reference Dose and Reference Concentration Processes. Risk
Assessment Forum. EPA/630/P-02/002F. Washington, DC: U.S. Environmental Protection
Agency, December 2002.

EPA. 2002a. Perchlorate Environmental Contamination: Toxicological Review and Risk
Characterization (External Review Draft). NCEA-1-0503. Washington, DC: U.S.
Environmental Protection Agency, Office of Research and Development, January 16, 2002.

EPA. 2002b. Guidance on Cumulative Risk Assessment of Pesticide Chemicals That Have a
Common Mechanism of Toxicity. Washington, DC: U.S. Environmental Protection Agency,
Office of Pesticide Programs.

http://www.epa.gov/oppfeadl/trac/science/cumulative guidance.pdf.

EPA. 2000. Supplementary Guidance for Conducting Health Risk Assessment of Chemical
Mixtures. Risk Assessment Forum. EPA/630/R-00/002. Washington, DC: U.S. Environmental
Protection Agency.

EPA. 1999. A Review of Contaminant Occurrence in Public Water Systems. EPA 816-R-99-
006. Washington, DC: U.S. Environmental Protection Agency, Office of Water, November
1999.

EPA. 1997a. Chapter 7: Body Weight Studies, Table 7-1. In Exposure Factors Handbook,
Volume 1 - General Factors. U.S. Environmental Protection Agency, Office of Research and
Development/National Center for Environmental Assessment.
http://www.epa.gov/ncea/efh/pdfs/efh-chapter07.pdf (accessed August 5, 2007).

EPA. 1997b. EPA Administrator's memorandum to the Agency implementing "Cumulative
Risk Assessment Guidance - Phase I Planning and ScopingJuly 3, 1997.

EPA. 1997c. Guidance on Cumulative Risk Assessment. Part 1. Planning and Scoping.
Washington, DC: U.S. Environmental Protection Agency, Science Policy Council, July 3, 1997.

EPA. 1993. Reference Dose (RfD): Description and Use in Health Risk Assessments-
Background Document 1A, March 15, 1993. U.S. Environmental Protection Agency, Office of
Research and Development, National Center for Environmental Assessment.
http://www.epa.gov/iris/rfd.htm (accessed December 10, 2009).

EPA. 1992. Safeguarding the Future: Credible Science, Credible Decisions. A report of the
Expert Panel on the Role of Science at EPA. EPA/600/9-91/050. Washington, DC: U.S.
Environmental Protection Agency.

B-6

-------
10-P-0101

EPA. 1991. Guidelines for Developmental Toxicity Risk Assessment. Risk Assessment Forum.
EPA/600/FR-91/001. Washington, DC: U.S. Environmental Protection Agency. Published in
the Federal Register on December 5, 1991 (56 FR 63798).

EPA. 1991b. National Primary Drinking Water Regulations; Final Rule (56 FR 3526-3597).

Also listed in 40 CFR § 141.15.

EPA. 1986. Guidelines for Health Risk Assessment of Chemical Mixtures. Federal Register
51(185):34014-25. Also published by EPA as EPA/630/R-98/002.

EPA SAB. 2004. Science and Research Budgets for the U.S. Environmental Protection Agency
for Fiscal Year 2007 - An Advisory Report by the Science Advisory Board. EPA-SAB-ADV-06-
003. Washington, DC: U.S. Environmental Protection Agency, Office of the Administrator,
Science Advisory Board, March 30, 2006. http://www.epa.gov/science 1 /pdf/sab-adv-06-003.pdf

Farland WH, Jarabek AM. 2003. Perchlorate risk characterization: US EPA technical
perspectives. Presentation made to the NAS Committee to Assess the Health Implications of
Perchlorate Ingestion. Washington, DC, October 27, 2003.

FDA. 2007a. Preliminary estimation of perchlorate dietary exposure based on FDA 2004/2005
exploratory data, http://www.cfsan.fda.gov/~dms/clo4ee.html (accessed June 6, 2007).

FDA. 2007b. Overview of Dietary Supplements. http://www.cfsan.fda.gov/~dms/ds-
oview.html#regulate (accessed December 11, 2007).

Fenzi GF, Giusti LF, Aghini-Lombardi F, BartalenaL, Marcocci C, Santini F, Bargagna S,
Brizzolara D, Ferretti G, Falciglia G, Monteleone M, Marcheschi M, Pinchera A. 1990.
Neuropsychological assessment in schoolchildren from an area of moderate iodine deficiency. J.
Endocrinol. Invest. 13:427-31.

Fisher PW, L'Abbe M. 1980. Iodine in iodized table salt and sea salt. Can. Inst. Food Sci.
Technolo. J. 13(2): 103-4.

Fox MA, Tran NL, Groopman JD, Burke TA. 2004. Toxicological resources for cumulative
risk: An example with hazardous air pollutants. Regulatory Toxicology and Pharmacology 40:
305-11.

GAO. 2007. Government Auditing Standards (July 2007 Revision). GAO-07-713G.
Washington, DC: U.S. Government Accountability Office.

Gibbs G, van Landingham C. 2008. Urinary perchlorate excretion does not predict thyroid
function among pregnant women. Thyroid (Letter to the editor - in peer review; manuscript ID:
THY-2007-0377.R1).

B-7

-------
10-P-0101

Gibbs JP. 2006. A comparative toxicological assessment of perchlorate and thiocyanate based
on competitive inhibition of iodide uptake as common mode of action. Human and Ecological
Risk Assessment 12(1): 157-73.

Ginsberg G, Rice D. 2005. The NAS perchlorate review: Questions remain about the
perchlorate RfD. Environ Health Perspectives 113(9): 1117-19.

GlinoerD. 2005. Chapter 80: Thyroid disease during pregnancy. In The Thyroid: A
Fundamental Clinical Text. 9th ed. Philadelphia: Lippincott Williams & Wilkins. ISBN 0-
7817-5047-4.

Glinoer D, De Nayer P, Delange F, Lemone M, Toppet V, Spehl M, Grun J, Kinthaert J, Lejeune
B. 1995. A randomized trial for the treatment of mild iodine deficiency during pregnancy:
Maternal and neonatal effects. Journal of Clinical Endocrinology and Metabolism 80(l):258-69.

Glinoer D, Lemone M, Bourdoux P, De Nayer P, Delange F, Kinthaert J, Lejeune B. 1992.
Partial reversibility during late postpartum of thyroid abnormalities associated with pregnancy.
Journal of Clinical Endocrinology & Metabolism 74(2):453-57.

Glinoer D, Delange F, Laboureur I, De Nayer P, Lejeune B, Kinthaert J, Bourdoux P. 1992.
Maternal and neonatal thyroid function at birth in an area of marginally low iodine intake.

Journal of Clinical Endocrinology & Metabolism 75(3): 800-805.

Greer MA, Goodman G, Pleus RC, Greer SE. 2002. Health effects assessment for
environmental perchlorate contamination: The dose response for inhibition of thyroidal
radioiodine uptake in humans. Environ. Health Perspect. 110(9):927-37.

Greer MA, Stoot AK, Milne KA. 1966. Effect of thiocyanate, perchlorate and other anions on
thyroidal iodine metabolism. Endocrinology 79:237-47.

Haddow JE, Palomaki GE, Allan WC, Williams JR, Knight GJ, Gagnon JG, O'Heir CE, Mitchell
ML, Hermos RJ, Waisbren SE, Faix JD, Klein RZ. 1999. Maternal thyroid deficiency during
pregnancy and subsequent neuropsychological development of the child. The New England
Journal of Medicine 341(8):549-55.

Hansen PS, Brix TH, Bennedbaek FN, Bonnema SJ, Kyvik KO, Hegedus L. 2004. Genetic and
environmental causes of individual differences in thyroid size: A study of healthy danish twins.
The Journal of Clinical Endocrinology & Metabolism 89(5): 2071-77.

Hasuike Y, Nakanishi T, Moriguchi R, Yoshinaga O, Nanami M, Hama Y, Naka M, Miyagawa
K, Izumi M, Takamistsu Y. 2004. Accumulation of cyanide and thiocyanate in haemodialysis
patients. Nephrology Dialysis Transplant 19(6): 1474-79.

Hauth JC, Hauth J, Drawbaugh RB, Gilstrap LC, Pierson WP. 1984. Passive smoking and
thiocyanate concentrations in pregnant women and newborns. Obstet Gynecol 63(4):519-22.

B-8

-------
10-P-0101

Hollowell JG, Staehling NW, Flanders WD, et al. 2002. Serum TSH, T4, and thyroid antibodies
in the United States population (1988-1994): National Health and Nutrition Examination Survey
(NHANES III). Journal Clinical Endocinology & Metabolism 87:489-99.

Hollowell JG, Staehling NW, Hannon WH, Flanders DW, Gunter EW, Maberly GF, Braverman
LE, Pino S, Miller DT, Garbe PL, DeLozier DM, Jackson RJ. 1998. Iodine nutrition in the
United States. Trends and public health implications: Iodine excretion data from National
Health and Nutrition Examination Surveys I and III (1971-1974 and 1988-1994). J Clin
Endocrinol Metab 83(10):3401-8.

House 2007. Perchlorate: Health and Environmental Impacts of Unregulated Exposure.

Hearing before the Subcommittee on Environmental and Hazardous Materials of the Committee
on Energy and Commerce. House of Representative, 110th Congress, First Session. April 25,
2007. Serial No. 110-35. 38-495 PDF. Washington, DC: U.S. Government Printing Office.

Howdeshell KL. 2002. A model of the development of the brain as a construct of the thyroid
system. Environmental Health Perspectives 110(3):337-48.

IRIS. 2007. Glossary of IRIS Terms. EPA's Integrated Risk Information System (IRIS).
http://www.epa.gov/iriswebp/iris/gloss8.htm (accessed December 27, 2007).

Ivanac G, Rozman B, Skreb F, Brkljacic B, Pavic L. 2004. Ultrasonographic measurement of
the thyroid volume. Coll. Antropol. 28(1):287-91.

Kirk AB, Dyke JV, Martin CF, Dasgupta PK. 2007. Temporal patterns in perchlorate,
thiocyanate, iodide excretion in human milk. Environmental Health Perspectives 115(2): 182-86.

Kirk AB, Martinelango PK, Tian K, Dutta A, Smith EE, Dasgupta PK. 2005. Perchlorate and
iodide in dairy and breast milk. Environmental Science and Technology 39(7):2011-17.

Kooistra L, Crawford S, van Baar AL, Brouwers EP, Pop VJ. 2006. Neonatal effects of
maternal hypothyroxinemia during early pregnancy. Pediatrics 117(1): 161-67.

Ladenson PW. 2005. Chapter 66: Diagnosis of hypothyroidism. In The Thyroid: A
Fundamental Clinical Text. 9th ed. Philadelphia: Lippincott Williams & Wilkins. ISBN 0-
7817-5047-4.

Lamm, S.H., Hollowell, J.G., Engel, A., Chen, R. 2007. Perchlorate, thyroxine, and low urine
iodine association not seen with low creatinine-adjusted urine iodine among women of
childbearing age. Thyroid 17 (si), S-51.

Langer P. 1989. Normal thyroid size versus goiter - postmortem thyroid weight and
ultrasonographic volumetry versus physical examination. Endocrinology Exp 23(2):67-76.

B-9

-------
10-P-0101

Lombardi FA, Pinchera A, Antonangeli L, Rago T, Chiovato L, Bargagna S, Bertucelli B,
Ferretti G, Sbrana B, Marcheschi, Vitti P. 1995. Mild iodine deficiency during fetal/neonatal
life and neuropsychological impairment in Tuscany. J. Endocrinol. Invest. 18:57-62.

Laurberg P, Nohr SB, Pedersen KM, F|iglsang E. 2004. Iodine nutrition in breast-fed infants is
impaired by maternal smoking. The Journal of Clinical Endocrinology & Metabolism
89(1): 181-87.

Laurberg P, Andersen S, Knudsen N, Ovesen L, Nohr SB, Bulow Pedersen I. 2002. Thiocyanate
in food and iodine in milk: From domestic animal feeding to improved understanding of
cretinism. Thyroid 12(10):897-902.

Li R, Ogden C, Ballew C, Gillespie C, Grummer-Strawn L. 2002. Prevalence of exclusive
breastfeeding among U.S. infants: The Third National Health and Nutrition Examination Survey
(Phase II, 1991-1994). Research and Practice 92(7): 1107-10.

Marieb E. 1998. Chapter 18: Blood. In Human Anatomy & Physiology, 4th ed. Menlo Park,
CA: Benjamin/Cummings Science Publishing. ISBN: 0-8053-4360-1.

Martin JA, Hamilton BE, Sutton PD, Ventura SJ, Menacker F, Munson ML. 2003. Births:

Final data for 2002. National Vital Statistics Reports 52(10): 1-114. Published by CDC.

Massachusetts Department of Environmental Protection (MassDEP). 2006. Final Perchlorate
Regulations, Response to Comment # 21 from the New England Water Works Association, June
19, 2006.

McLanahan ED, Anderson ME, Campbell Jr. JL, Fisher JW. 2009. Competitive inhibition of
thyroid uptake of dietary iodide by perchlorate does not describe perturbations in rat serum total
T4 and TSH. Environmental Health Perspective 117(5):731-38.

McMurray WC. 1982. A Synopsis of Human Biochemistry: With Medical Applications.
Philadelphia: Harper & Row Publishers, Inc., ISBN 0-06-141642-8.

Merrill EA, Clewell RA, Gearhart JM, Robinson PJ, Sterner TR, Yu KO, Mattie DR, Fisher JW.
2003. PBPK predictions of perchlorate distribution and its effect on thyroid uptake of
radioiodide in the male rat. Toxicological Sciences 73:256-69.

Middlesworth, LV. 1986. Potential metabolic significance of blood thiocyanate.

Endocrinologia Experimentals 20:17-22.

Morreale de Escobar G, Obregon MJ, Del Rey E. 2004. Role of thyroid hormone during early
brain development. European Journal of Endocrinology 151:U25-U37. ISSN 0804-4643.

Murray CW, Egan SK, Kim H, Bern N, Bolger PM. 2008. U.S. Food and Drug
Administration's total diet study: Dietary intake of perchlorate and iodine. Journal of Exposure
Science and Environmental Epidemiology 2008:1-10.

B-10

-------
10-P-0101

Nafstad P, Magnus P, Stray-Pedersen B. 1995. Opposing placental gradients for thiocyanate
and cotinine at birth. Early Hum Dev 42(l):73-79.

National Academies Press (NAP). 2000. Appendix G: Biochemical indicators for iron, vitamin
A, and iodine from the third National Health and Nutrition Examination Survey. In Dietary
Reference Intake for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine, Iron,
Manganese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc, 690-691. Washington, DC:
National Academies Press. ISBN-13: 978-0-309-07290-8.

National Academy of Science (NAS). 2008. Science and Decisions: Advancing Risk
Assessment. NAS Committee on Improving Risk Analysis Approaches Used by the U.S. EPA.
Washington, DC: National Academies Press. ISBN-13: 978-0-309-12046-3.

NAS. 2008a. Phthalates and Cumulative Risk Assessment: The Task Ahead NAS Committee
on Improving Risk Analysis Approaches Used by the U.S. EPA. Washington, DC, National
Academies Press. ISBN-13: 978-0-309-12841-4.

NAS. 2007. Toxicity Testing in the Twenty-First Century: A Vision and a Strategy. Committee
on Toxicity Testing and Assessment of Environmental Agents. Washington, DC, National
Academies Press. ISBN-10: 0-309-10988-4.

NAS. 2005. Health Implications of Perchlorate Ingestion. Washington, DC: National
Academies Press. ISBN 0-309-09568-9.

NAS. 2001. Food Safety Policy, Science, and Risk Assessment: Strengthening the Connection:
Workshop Proceedings. Food Forum, Food and Nutrition Board. Washington, DC: National
Academies Press. ISBN 0-309-56512-X.

NAS. 1995. Nitrate and Nitrite in Drinking Water. National Research Council, Commission on
Life Sciences, Subcommittee on Nitrate and Nitrite in Drinking Water. Washington, DC:
National Academy Press. ISBN-10: 0-309-08370-2.

NAS. 1994. Science and Judgment in Risk Assessment. National Research Council,

Commission on Life Sciences, Committee on Risk Assessment of Hazardous Air Pollutants.
Washington, DC: National Academy Press.

NASDA 2009. NASDA October 8, 2009 letter to EPA Office of Water's August 19, 2009
request for supplemental perchlorate comments (Docket ID No. EPA-HQ-OW-2009-0297). The
National Association of State Departments of Agriculture (NASDA), 1156 15th Street NW, Suite
1020, Washington, D.C. 2005. http://www.nasda.org/cms/7196/20728/20741/24460.aspx

Nohr SB, Laurberg P, Borlum KG, Pedersen KM, Johannesen PL, Damm P, Fuglsang E,
Johansen A. 1994. Iodine status in neonates in Denmark: Regional variations and dependency
on material iodine supplementation. Acta Paediatr 83(6):578-82.

B-11

-------
10-P-0101

Nolan BT, Hitt KJ, Ruddy BC. 2002. Probability of nitrate contamination of recently recharged
groundwaters in the conterminous United States. Environmental Science and Technology,
36(10): 2138-45.

Obregon MJ, Escobar del Rey F, Morreale de Escobar G. 2005. The effects of iodine deficiency
on thyroid hormone deiodination. Thyroid 15(8):917-29.

Pearce EN, Spencer CA, Mestman J, Lee R, Bergoglio LM, Mereshian P, He X, Leung AM,
Braverman LE. 2008. "Thyroid Function is Not Affected by Environmental Perchlorate
Exposure in First Trimester Pregnant Women from California and Argentina." Presentation at
Annual Meeting of the Endocrine Society, San Francisco, CA, June 15-18, 2008.

Pearce EN, Leung AM, Blount BC, Bazrafshan HR, He X, Pino S, Valentin-Blasini L,
Braverman LE. 2007. Breast milk iodine and perchlorate concentrations in lactating Boston-
area women. Journal of Clinical Endocrinology & Metabolism 92(5): 1673-77.

Pearce EN. 2007a. National trends in iodine nutrition: Is everyone getting enough? Thyroid
17(9):823-27.

Pearce EN, Lazarus JH, Smyth PPA, He X, DaH'Amico D, Parkes AB, Burns R, Smith DF,
Maina A, Leung AM, Braverman LE. 2007b. Thyroid function is not affected by environmental
perchlorate exposure in first trimester pregnant women. Thyroid 17 (si), S-133 [abstract only],

Pearce EN, Bazrafshan HR, He X, Pino S, Braverman LE. 2004. Dietary iodine in pregnant
women from the Boston, Massachusetts area. Thyroid 14(4): 327-28.

Pollay M, Stevens A, Davis Jr. C. 1966. Determination of plasma-thiocyanate binding and the
Donnan ratio under simulated physiological conditions. Analytical Biochemistry 17:192-200.

Ryan PB, Burke TA, Hubal EAC, Cura JJ, McKone TE. 2007. Using biomarkers to inform
cumulative risk assessment. Environmental Health Perspectives, 115(5): 833-40.

Santiago-Fernandez P, Torres-Barahona R, Mulela-Marinez JA, Rojo-Martinez G, Garcia-
Fuentes G, Garriga J, Leon AG, Soriguer F. 2004. Intelligence quotient and iodine intake: A
cross-sectional study in children. J. Clin. Endocrinol. Metab. 89:3851-57.

Science Advisory Board (SAB). 2008. SAB Advisory on EPA's Draft Third Drinking Water
Contaminant Candidate List (CCL 3)(Draft 3, 2008) (EPA-SAB-07-00_). Committee Response
to Charge Question 4. SAB Drinking Water Committee on the Third Drinking Water
Contaminant Candidate List (CCL3). http J/www, epa. gov/ sab (accessed December 18, 2008).

SAB. 2007. SAB's February 28, 2007 memo to EPA Administrator Stephen L. Johnson
regarding Consu/tation on Enhancing Risk Assessment Practices and Updating EPA 's Exposure
Guide/ines. EPA-SAB-07-003.

B-12

-------
10-P-0101

Selva KA, Harper A, Downs A, Blasco PA, LaFranchi SH. 2005. Neurodevelopmental
outcomes in congenital hypothyroidis: Comparison of initial T4 dose and time to reach target T4
and TSH. Journal of Pediatrics 147:775-80.

Sexton K, Hattis D. 2007. Assessing cumulative health risks from exposure to environmental
mixtures - three fundamental questions. Environmental Health Perspectives, 115(5): 825-32.

Sol din OP, Tractenberg RE, Pezzullo JC. 2005. Do thyroxine and thyroid-stimulating hormone
levels reflect urinary iodine concentrations? Ther Drug Monit 27:178-85.

Steinmaus C, Miller MD, Howd R. 2007. Impact of smoking and thiocyanate on perchlorate
and thyroid hormone associations in the 2001-2002 National Health and Nutrition Examination
Survey. Environmental Health Perspectives 115(9): 1333-38.

Stoa KF. 1957. Studies on thiocyanate in serum with some supplementary investigations in
saliva, urine, and cerbrospinal fluid. 17. Bergen, Norway: University Bergen Medical
Yearbook.

Sullivan KM. 2007. Iodine supplementation for pregnancy and lactation: United States and
Canada: Recommendations of the American Thyroid Association. Thyroid 17(5):483-84.

Sullivan KM, May W, Nordenberg D, Houston R, Maberly GF. 1997. Use of thyroid
stimulating hormone testing in newborns to identify iodine deficiency. Journal of Nutrition
127:55-58.

Tajtakova M, Semanova Z, Tomkova Z, Szokeova E, Majoros J, Radikova Z, Sebokova
E, Klimes I, Langer P. 2006. Increased thyroid volume and frequency of thyroid
disorders signs in school children from nitrate polluted area. Chemosphere
62:559-64.

Tellez RT, Chacon PM, Abarca CR, Blount BC, Van Landingham CB, Crump KS, Gibbs JP.
2005. Long-term environmental exposure to perchlorate through drinking water and thyroid
function during pregnancy and the neonatal period. Thyroid 15(9):963-87.

Ting D, Howd RA, Fan AM, Alexeeff GV. 2006. Development of a health-protective drinking
water level for perchlorate. Environmental Health Perspectives 114(6):881-86.

Tonacchera M, Pinchera A, Dimida A, Ferrarini E, Agretti P, Vitti P, Santini F, Crump K, Gibbs
J. 2004. Relative Potencies and additivity of perchlorate, thiocyanate, nitrate, and iodide on the
inhibition of radioactive iodide uptake by the human sodium iodide symporter. Thyroid
14:1012-19.

UtigerRD. 1999. Editorial: Maternal hypothyroidism and fetal development. New England
Journal Medicine 341(8):601-2.

B-13

-------
10-P-0101

Vanderpas J, Bourdoux P, Lagasse R, Rivera M, Dramaix M, Lody D, Nelson G, Delange F,
Ermans AM, Thilly CH. 1984. Endemic infantile hypothyroidism in severe endemic goiter area
in Central Africa. Clinical Endocrinology 20(3):327-40.

Vanderpump MPJ. 2005. Chapter 19: The Epidemiology of Thyroid Diseases. In The
Thyroid: A Fundamental Clinical Text. 9th ed. Philadelphia: Lippincott Williams & Wilkins.
ISBN 0-7817-5047-4.

VetterJ. 2000. Plant cyanogenic glycosides. Toxicon 38:11-36.

Vermiglio F, Lo Presti VP, Moleti M, Sidoti M, Tortorella G, Scaffidi G, Castagna MG, Mattina
F, Violi MA, Crisa A, Artemisia A, Trimarchi F. 2004. Attention deficit and hyperactivity
disorders in the offspring of mothers exposed to mild-moderate iodine deficiency: A possible
novel iodine deficiency disorder in developed countries. Journal of Clinical Endocrinology &
Metabolism 89(12):6054-60.

Vermiglio F, Sidoti M, Finocchiaro MD, Battiato S, Presti VPL, Benvenga S, Trimarchi. 1990.
Defective neuromotor and cognitive ability in iodine-deficient schoolchildren of an endemic
goiter region in Sicily. Journal of Clinical Endocrinology and Metabolism 70(2):379-84.

Vitti P, Lombardi FA, Antonangeli L, Rago T, Chiovato L, Pinchera A, Marcheschi M,

Bargagna S, Bertuccelli B, Ferretti G, Sbrana B. 1992. Mild iodine deficiency in fetal/neonatal
life and neuropsychological performances. Acta Med Austriaca 19:57-59.

Wenzel KW, Lente JR. 1984. Similar effects of thionamide drugs and perchlorate on thyroid-
stimulating immunoglobulins in Graves' disease: Evidence against and immunosuppressive
action of thionamide drugs. Journal of Clinical Endocrinology Metabolism 58(l):62-69.

Wolff J. 1998. Perchlorate and the thyroid gland. Pharmacological Reviews 50(1):89-105.

World Health Organization (WHO). 2004. Hydrogen cyanide and cyanides: Human health
aspects. 61. Concise International Chemical Assessment Document.

Wyngaarden JB, Stanbury JB, Rapp B. 1953. The effect of iodide, perchlorate, thiocyanate, and
nitrate administration upon the iodide concentrating mechanism of the rat thyroid.

Endocrinology 52:568-74.

YenPM. 2005. Chapter 8: Genomic and nongenomic actions of thyroid hormones. In The
Thyroid: A Fundamental Clinical Text. 9th ed. Philadelphia: Lippincott Williams & Wilkins.
ISBN 0-7817-5047-4.

Zimmermann M, Delange F. 2004. Iodine supplementation of pregnant women in Europe: A
review and recommendations. European Journal of Clinical Nutrition 58:979-84.

B-14

-------
10-P-0101

Appendix C

List of Acronyms and Abbreviations

ADHD attention-deficit/hyperactivity disorder

ATA American Thyroid Association

BMC benchmark concentration

BMD benchmark dose

BMDL benchmark dose lower confidence limit

BV baseline visit

CDC Centers for Disease Control and Prevention

CG cyanogenic glycosides

CH congenital hypothyroidism

CHO Chinese hamster ovary

CI confidence interval

CIO4" perchlorate

CNF Chilean nitrate fertilizer

CNS central nervous system

d Cohen's J-value

DWEL drinking water equivalent level

EAR estimated average requirement

EPA U.S. Environmental Protection Agency

FDA Food and Dr|ig Administration

fT3 free triiodothyronine

fT4 free thyroxine

GAO Government Accountability Office

HA health advisory

HRL health reference level

HNA high-nitrate area

I" iodide

I2 iodine

ICCIDD International Council for the Control of Iodine Deficiency Disorders

IDA iodide-deficient area

IOM Institute of Medicine

IQ intelligence quotient

IRIS Integrated Risk Information System

ISA iodide-sufficient area

LNA low-nitrate area

LOAEL lowest-observed-adverse-effect-level

MCL maximum contaminant limit

mg milligram

mg/day milligram per day

mg/kg-day milligram per kilogram per day

mg/L milligram per liter

mU/1 milli-international-units per liter

NAP National Academies Press

C-1

-------
10-P-0101

NAS

National Academy of Sciences

NCEH

National Center for Environmental Health

NHANES

National Health and Nutrition Examination Survey

NIS

sodium iodide symporter

nmol/L

nanomole per liter

N03"

nitrate

NOAEL

no-observed-adverse-effect-level

NOEL

no-observed-effect-level

NPDWR

National Primary Drinking Water Regulation

NRC

National Research Council

OCEPA

Office of Communications, Education, and Public Affairs

OIG

Office of Inspector General

OPEC

oral perchlorate equivalent concentration

probability

PBPK

physiologically based pharmacokinetic

PEC

perchlorate equivalent concentration

pg/dL

picogram per deciliter

PII

plasma inorganic iodide

POD

point of departure

ppb

part per billion

ppm

part per million

RAIU

radioactive iodide uptake

RDA

recommended dietary allowance

RfD

reference dose

RPF

relative potency factor

RSC

relative cource contribution

SAB

Science Advisory Board

SCN"

thiocyanate

standard deviation

standard error of the mean

SPEC

serum perchlorate equivalent concentration

triiodothyronine

thyroxine

TDS

Total Dietary Study

thyroglobulin

Till

total iodide uptake

%TIU

percent total iodide uptake

TSH

thyroid-stimulating hormone (a.k.a. thyrotropin)

tT3

total triiodothyronine

tT4

total thyroxine

UCMR

Unregulated Contaminant Monitoring Rule

uncertainty factor

UIC

urinary iodide concentration

UNICEF

United Nations International Children's Emergency Fund

l-ig

microgram

|ig/day

mircrogram per day

C-2

-------
10-P-0101

|ig/dL microgram per deciliter

|ig/kg-day microgram per kilogram per day

|ig/L microgram per liter

|imol/L micromole per liter

|iU/ml micro-international-units per milliliter

USD A U. S. Department of Agriculture

WHO World Health Organization

WISC Wechsler Intelligence Scale for Children

C-3

-------
10-P-0101

Appendix D

Scientific Comments Received on the
OIG Scientific Analysis of Perchiorate
(External Review Draft)

For this appendix, go to the following:

www.epa.qov/oiq/reports/2010/20100419-10-P-0101 appD.pdf

-------
10-P-0101

Appendix E

OIG Response to Comments on
OIG Scientific Analysis of Perchiorate
(External Review Draft)

For this appendix, go to the following:

www.epa.qov/oiq/reports/2010/20100419-10-P-0101 appE.pdf

-------
10-P-0101

Appendix F

Distribution

Office of the Administrator

Assistant Administrator for Research and Development
Assistant Administrator for Water

Director, Office of Children's Health Protection and Environmental Education
Agency Follow-up Official (the CFO)