Talking Points Summarizing PSG Comments on EPA’s Consideration of
Perchlorate for the CCL 2 Regulatory Determinations

August 16,2007

EPA has been able to compile a wealth of data on health effects,
exposure (both biomonitoring and food), and occurrence that should now
allow the Agency to move forward with a preliminary determination on
perchlorate.

EPA’s RfD is Constructed to Protect the Most Sensitive Subpopulations 

EPA’s reference dose (RfD) 0.0007 mg/kg per day provides a safe level
that is very protective of all populations, including those identified
by the National Academy of Sciences (NAS) as the most sensitive.

First, the NAS panel departed from EPA’s traditional methodology and
relied on a more health-protective approach in its recommended RfD.

EPA’s traditional methodology bases the RfD upon an adverse effect
using either the No Observed Adverse Effect Level (NOAEL) or the Lowest
Observed Adverse Effect Level (LOAEL).

The NAS panel based its recommended RfD on a nonadverse event, the No
Observed Effect Level (NOEL).  EPA’s IRIS database also notes this in
stating, “Because this NOEL is for a nonadverse effect, this is
considered to be a more conservative and health-protective approach than
traditional hazard assessments.”  (EPA Integrated Risk Information
System (IRIS) database, p. 6).

Second, the NAS panel applied a full intraspecies factor of 10 to the
NOEL to protect even the most sensitive populations.

The NAS panel concluded that fetuses and preterm newborns were the most
sensitive populations, but also recognized that infants and developing
children were also sensitive populations that would also be afforded
protection by the recommended RfD.

Children and Neonates Protected by EPA’s RfD

Some have argued that other sensitive populations such as nursing
infants or young children are of greater concern and were not considered
by the NAS panel.  FDA’s preliminary food data and exposure assessment
as well as the NHANES data predict and show that children receive a
higher dose of perchlorate than adults.  

A recent study has reaffirmed the NAS panel’s conclusion that fetus is
the most sensitive population of concern and are more sensitive than
other groups including children.

Clewell et al., (2007) used PBPK modeling allowing the authors to
predict iodide and perchlorate kinetics in the human starting from fetal
life through adulthood.

PBPK modeling has been found to be useful in reducing uncertainty and in
filling scientific gaps in human health data on a substance.

The NAS perchlorate panel agreed with EPA in finding PBPK modeling
useful in determining human equivalent exposures.  The above model also
adopts the parallelogram approach found to be acceptable for use both by
EPA and the NAS panel. 

EPA research personnel have co-authored previous versions of this PBPK
study related to perchlorate.

Using perinatal rat models, published physiological parameters, and
chemical-specific parameters based upon a parallelogram approach, the
model was able to predict both iodide kinetics and serum perchlorate
concentrations.

The study finds that, at a given dose, the pregnant and lactating woman,
fetus, and nursing infant would have higher perchlorate serum
concentrations and greater thyroid iodide uptake inhibition (IUI) than
the nonpregnant adult or older child.  

Given the same dose, children will have serum perchlorate levels eight
to ten times lower than fetuses and neonates.  They will also have ½
the serum dose of adults.

Under the model, the fetus had higher perchlorate doses than the mother.
 Neonates were shown to have higher doses than the mother for low
external doses; however, the opposite was true for higher external
doses, which the authors believe is the result of the mammary NIS.

The study also found that while the fetus was predicted to receive the
greatest dose, the predicted extent of IUI at environmentally relevant
doses was not significant (1 percent at a dose 40 percent greater than
EPA’s RfD). 

The study looks at predicted serum perchlorate levels across life stages
and finds that at various external perchlorate doses, fetal dose is
higher than the corresponding doses for the neonate, child, and
nonpregnant adult.  Increased perchlorate levels during this timeframe
were found to cause higher predicted inhibition.

The study estimates that a maternal dose of 0.01 mg/kg per day would be
required to achieve 10% fetal inhibition -- a level 30 times higher than
EPA’s RfD.

Therefore, even if the child had a larger dose than the fetus, the child
would only have 3 percent IUI.  This level is essentially
indistinguishable from the 4 percent IUI estimated for adults.  

Put another way, if a child is exposed to the RfD dose in drinking water
and the 90th percentile dose from food consumption, the estimated IUI is
0.3 percent.  

EPA’s RfD Protective of Women with Low Iodine Levels

The publication of a study associating higher perchlorate concentrations
with lower levels of thyroid hormones in women has raised concerns that
the NAS and EPA’s RfD is not sufficiently protective of certain
subpopulations.   While further validation should be done to cure the
study’s known limitations, the findings support EPA’s RfD as being
protective of the population.  

The non-profit group, Toxicology Excellence for Risk Assessment (TERA),
has some noteworthy commentary on the conclusions in the Blount et al.,
(2006) (or “Blount 2006b”) thyroid association study.

Using the data and regression curves from Blount et al. (2006), TERA
showed that doses up to 1,000 times higher than the U.S. EPA perchlorate
reference dose (i.e., 0.45 - 0.68 mg/kg-day) will be required in even
iodine deficient women to raise serum TSH values to the upper limit of
the normal range.  

,700 μg/L is required, and to result in 1 percent of the population at
that abnormal level or higher, a urinary perchlorate level of 31,420
μg/L is required. Assuming that all urinary perchlorate arrives from
oral exposure, a urinary excretion rate of 1 L/day (a lower end of daily
adult urine excretion amount of 1-2 L/day), and a 70 kg body weight,
these urinary perchlorate levels are roughly equivalent to perchlorate
doses of 0.68 mg/kg-day and 0.45 mg/kg-day, respectively. These doses
are consistent with other human studies that demonstrate that doses of
0.5 mg/kg-day have no effect on thyroid hormone levels of normal,
healthy adults. These doses are also well above EPA’s current RfD of
0.0007 mg/kg-day. 

Using these assumptions, the average urinary perchlorate level in the
population, 2.84 μg/L, is roughly equivalent to an oral perchlorate
dose of 0.00004 mg/kg-day and 100 times lower than the dose in humans
that is known to have no effect on inhibition of iodide uptake (Greer et
al. 2002; NRC 2005). Therefore, while the associations between urinary
perchlorate and serum TSH and T4 observed in the paper are consistent
with perchlorate’s mode of action, they are not consistent with the
body of literature evaluating the effect of perchlorate dose on thyroid
changes.

TERA observes:

 In addition to the statistically significant relationship between serum
TSH and urinary perchlorate in women, regardless of dietary iodine
status, a statistically significant relationship also exists with body
mass index and age in iodine sufficient women, and with betablocker use
and premenarche in iodine deficient women. In addition to the
statistically significant relationship between serum T4 and urinary
perchlorate in iodine deficient women, and between serum T4 and urinary
nitrate in iodine sufficient women, a statistically significant
relationship also exists with menopause and total calorie intake in
iodine deficient women. 

 of 1.64 μg/dL in serum T4 concentrations associated with an increase
of urinary perchlorate from the 5th percentile to 95th percentile (0.19
to 13 μg/L) is a potentially sizable change, because the study is
cross-sectional, the positive relationship between urinary perchlorate
and serum TSH and T4 only suggests a significant association. The
statistically significant associations for the other independent
variables are also “real”; thus, the study suggests that thyroid
function is complex, not that perchlorate caused the observed increase
in TSH or decrease in T4. 

Since these other factors also play a role, a constellation of important
factors could alter the dose-response curve and for this reason the
quantitative use of the Blount et al. and NHANES data should proceed
very cautiously. Additional work is needed to determine whether some
unknown factor associated with perchlorate exposure, but not the factors
examined in this study, might be the cause of the observed changes in
TSH and T4.  

Further, the study’s impact is limited since it conflicts with the
extensive scientific record comprised of numerous, peer-reviewed studies
that show perchlorate has no measurable effect at low levels.  There is
no plausible biological mechanism that has been advanced that can be
tested in a scientific manner. 

In addition to those above, other recent publications continue to
validate EPA’s RfD as health protective to which the Agency should
give due consideration.

Tonacchera et al., 2004

Braverman et al., 2005

Crump and Gibbs 2005

Tellez et al., 2005

Gibbs 2006

Ting et al., 2006

Pearce et al., 2007.

Several limitations of Blount 2006b have been noted, including those
noted by the American Thyroid Association and NAS presenter Dr. Jonathan
Borak:

First, the findings of Blount 2006b are not suitable to show cause.

Second, the analysis is only a cross-sectional association study,
whereas the studies relied on by the NAS panel were based on
partially-controlled human perchlorate exposure, a more authoritative
form of scientific inquiry.

Third, the study measures total T4, not free T4.

Fourth, due to other missing data, perchlorate could be a surrogate for
an unknown variable.  While Blount et al. show that urinary perchlorate
levels predict serum T4 and TSH in women with low urinary iodine, they
also find that up to eight other independent variables also predict
serum T4 and TSH in women with low urinary iodine. Blount et al. also do
not address the adversity or biological significance of the predicted
serum hormone changes.

Fifth, the effects of all goitrogens on iodide uptake inhibition (IUI)
should be similar in direction and additive in magnitude; however, the
effects of the substances on IUI were found to be inconsistent in
Blount.

With this Level of Health Protection, EPA Can Rely on the Enormous
Scientific Data on Perchlorate for the Determination

Occurrence Data

Recent biomonitoring studies confirm that perchlorate is ubiquitous in
the US population at doses well below EPA’s reference dose (RfD). 
Drinking water exposure accounts for only a fraction of this measured
level.

Occurrence data reveals that perchlorate levels have significantly
declined over the last several years.

The California Department Health Services 2006 Monitoring Update shows a
significant decline in the number of detections. 

Monthly sampling data from the Metropolitan Water District of Southern
California (MWD) shows average annual perchlorate levels have declined
steadily from greater than 6 µg/L in 2000 to nondetects in 2007.

The occurrence data analysis in EPA’s notice involves the inclusion of
those water systems where only one detection was experienced during the
entire UCMR sampling period.  

Potentially results in a population estimate with less accuracy for use
in determining number of people impacted.

 

Goes beyond the recommendations of the NDWAC:

In its recommendations to the Agency, the National Drinking Water
Advisory Council (NDWAC) stated that EPA should consider: (1) the actual
and estimated national percent of [Public Water Systems] PWSs reporting
detections above half the health reference level: (2) the actual and
estimated national percent of PWSs with detections above the health
reference level; and (3) the geographic distribution of the contaminant.

Should EPA continue to use this approach, for purposes of determining
population impacts, the Agency should consider counting only that
proportion of the entire water system’s population served by the entry
point(s) experiencing the detect.  Thus, in situations where there’s
one entry point showing detect(s), only a percentage of the entire water
system’s population should be used equivalent to the percentage of the
number of entry points showing detects to the overall number of entry
points.  (i.e., if a water system has 5 entry points, and only one has
a detect, then only 20% of the population served by the water system
should be used by EPA – 1 entry point of 5 entry points = 20% or 20%
of the entire water system’s population).

EPA Options/Considerations for Moving Forward with a Regulatory
Determination

EPA’s proposal to use biomonitoring data to evaluate total perchlorate
exposure is a credible approach that should be adopted.

The National Health and Nutrition Examination Survey (NHANES)
biomonitoring data provides a better estimate of total perchlorate
exposure in the US population compared to extrapolating from food data.

Therefore, EPA should use the 2001-2002 NHANES perchlorate data to
determine directly whether regulation of perchlorate in drinking water
presents a meaningful opportunity for health risk reduction.

Its application yields a finding that there is no meaningful opportunity
for risk reduction by regulating perchlorate exposure in drinking water.


The comparative effect on IUI of perchlorate exposure in drinking water
to other dietary goitrogens is a key issue in determining whether there
is meaningful opportunity for risk reduction.

Nitrates and thiocyanates are goitrogens found in our diet and
contribute significantly more to IUI than does perchlorate.

Scientific support

Findings of Tonacchera, Braverman, and Gibbs

Environ report.

Other options:

Use of urinary biomonitoring total exposure value to estimate a relative
source contribution (RSC) is unnecessarily conservative and
mathematically inconsistent with past practice.  In addition, this
approach would not yield meaningful risk reduction.

Use of urinary biomonitoring data from exclusive bottled water drinkers
to estimate RSC would yield a value of essentially one.  Bottled water
contains essentially no perchlorate.  For tap water drinkers, food is
the source of at least 95% of their average perchlorate exposure.  Thus,
reducing the five percent derived from drinking water exposure through
an MCL would not make meaningful change to total exposure, especially
since total exposure is already well below the conservative RfD.

Use of food data to determine total exposure

Biomonitoring data is a better indicator of total exposure than
estimates calculated from data from food sampling and total dietary
surveys.

Enormous variety of food combinations creates large variations in
potential perchlorate exposure.  Assumption of unusual consumption
patterns could represent only a small fraction of the US population.

FDA’s preliminary exposure assessment would yield an RSC of
essentially one based upon a 90th percentile food dose.

Evaluation of data on perchlorate’s health effects, national
occurrence and exposure using the SDWA’s “meaningful risk
reduction” criteria reveals that an MCL for perchlorate is
unnecessary.

The HRL for perchlorate should be equal to the RfD’s drinking water
equivalent level (DWEL) based upon an RSC of essentially one as
discussed above.

-

)

1

5

t

u

±

²

À

Å

 

(

)

2

hZf

²

)

&

&

2

G

I

k

n



„

¡

¢

£

¤

hZf

)

¤

&

&

&

&

&

&

&

&

&

&

hç

&

&

&

&

&

&

&

&

&

&

&

&

&

hç

&

&

&

&

&

&

&

&

&

&

&

&

&

&

&

&

&

&

&

&

h=F

In comparing perchlorate to a selection of other substances from CCL 1
and CCL 2 where EPA has made a negative determination reveals that
perchlorate represents an even lower opportunity for meaningful risk
reduction based upon the occurrence and exposure data.

 Rebecca A. Clewell et al., 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
Stage, J. Toxicol. Environ. Health, 2007, at 408.

 See, E.A. Merrill et al., PBPK Model for Radioactive Iodide and
Perchlorate Kinetics and Perchlorate-Induced Inhibition of Iodide Uptake
in Humans, Toxicol. Sci., 2005, at 25, (EPA NCEA’s Annie Jarabek is a
study co-author).

 “Thoughts on the CDC Perchlorate Paper (Blount et al. 2006)”
(October 18, 2006)

 PAGE   

 PAGE   1 

