Site Specific Preliminary Remediation Goals Cleanup Goals For Tittabawassee River Floodplain Soil August 2014


SITE-SPECIFIC
PRELIMINARY REMEDIATION GOALS
(CLEANUP GOALS)

FOR

TITTABAWASSEE RIVER FLOODPLAIN SOIL

TITTABAWASSEE RIVER, SAGINAW RIVER & BAY
SUPERFUND SITE IN MICHIGAN



<2

AUGUST 2014

-------
TABLE OF CONTENTS

1.0 INTRODUCTION	1

2.0	LAND USE AND EXPOSURE SCENARIOS	3

2.1	Land Uses in the Floodplain	3

2.2	Exposure Scenarios	4

2.3	Exposure Receptors, Activities, and Pathways	4

2.4	Exposure Factors and Parameters	5

2.4.1	Exposure Frequency	5

2.4.2	Partition of Soil Exposure: Outdoor Soil and Indoor Dust	7

2.4.3	Dust Concentration	8

2.4.4	Oral Bioavailability/ Ingestion Absorption Efficiency	8

2.4.5	Dermal Absorption Efficiency (ABSd)	12

2.4.6	Skin Surface Area Parameter	13

2.4.7	Body Weight Parameter	14

3.0	DIOXIN TOXICITY	15

3.1	Non-Cancer Reference Dose	15

3.2	Cancer Slope Factor	16

3.3	EPA Policy for Application of Dioxin Toxicity Factors for Development of PRGs	16

4.0	DERIVATION OF PRG VALUES	16

4.1	Maintained Residential Areas PRG	17

4.2	Other Land Use Areas PRG	17

5.0	UNCERTAINTIES AND SENSITIVITY ANALYSIS	17

5.1	Exposure Frequency	17

5.2	Dust Concentration	19

5.3	Oral Bioavailability	19

5.4	Partition of Dermal Exposure: Outdoor Soil and Indoor Dust	20

6.0 REFERENCES	21

-------
TABLES, FIGURES, AND APPENDICES

TABLES

TABLE 1: Input Parameters Used for Computing Non-cancer PRGs for Exposure of	P 25
Residents to Dioxin in Soil

TABLE 2: Input Parameters Used for Computing Non-cancer PRGs for Exposure of Adult	P 27
Worker to Dioxin in Soil

TABLE 3: Apportioning Outdoor Exposure Days for Residents	P7

TABLE 4: Congener Distribution of Test Soil and Floodplain Soil	P 28

TABLE 5. Relative Bioavailability for Tittabawassee River Floodplain Test Soil and Animal	P 29
Feed Intake

TABLE 6: Skin Surface Areas	P 14

TABLE 7: Body Weights	P 15

TABLE 8: HQ Sensitivity Analysis Varying the Apportionment of Outdoor Exposure Days	P 19

TABLE 9: HQ Sensitivity Analysis Varying the Oral RBA	P 20

FIGURES

FIGURE 1. Current Land Use in the Floodplain	P 30

FIGURE 2. Desired Future Land Use in the Floodplain	P 30

FIGURE 3: Areas Where Residents May Be Exposed to Soil	P31

APPENDICES

APPENDIX A Review of Exposed Skin Surface Area Parameter for All Receptor Groups	P 32

APPENDIX B Review of Body Weight Parameter for All Receptor Groups	P 36

APPENDIX C Calculation of the Soil PRG for Residential Maintained Land	P 38

ii

-------
ACRONYMS

4-PeCDF	2,3,4,7,8-pentachlorodibenzofuran

ABSd	Dermal Absorption Coefficient

ARARs	Applicable or Relevant and Appropriate laws and regulations

CSF	cancer slope factor

Dioxin	can refer to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), dioxins and furans, or include

other dioxin-like chemicals (DLCs)

D/Fs	dioxins and furans

DLC	dioxin-like chemicals

Dow	The Dow Chemical Company

EFH	EPA's Exposure Factors Handbook

EPA	United States Environmental Protection Agency

EROD	7-ethoxyresorufin-O-deethylase

Floodplain	8-year floodplain of the Tittabawassee River

HQ	Hazard Quotient

IRIS	Integrated Risk Information System

kg	kilogram

LADD	Lifetime Average Daily Dose

LOAEL	lowest-observed-adverse-effect level

MDEQ	Michigan Department of Environmental Quality

NREPA	Natural Resources and Environmental Protection Act (Michigan)

NHANES	National Health and Nutrition Examination Survey

NTCRA	non-time critical removal action

OHEA	Office of Health and Environmental Assessment (EPA)

PBPK	physiologically-based pharmacokinetic

PCBs	polychlorinated biphenyls

pg	picogram

ppt	parts per trillion

PRGs	Preliminary Remediation Goals

RBA	Relative Bioavailability

RfD	Reference Dose

RME	reasonable maximum exposure

TCDD	2,3,7,8-tetrachlorodibenzo-p-dioxin

TCDF	2,3,7,8-tetrachlorodibenzofuran

TEQ	toxic equivalence

TSH	Thyroid Stimulating Hormone

UCL	upper confidence limit

UMDES	University of Michigan Dioxin Exposure Study

WHO	World Health Organization

-------
1.0 INTRODUCTION

The objective of this document is to present the technical basis for the site-specific Preliminary
Remediation Goals (PRGs) for human direct contact exposure (i.e., ingestion and dermal contact) to
Dioxin contamination in soil in the 8-year floodplain of the Tittabawassee River (Floodplain) at the
Tittabawassee River, Saginaw River & Bay site in Michigan. These site-specific PRGs were developed by
the United States Environmental Protection Agency (EPA), in consultation with the Michigan
Department of Environmental Quality (MDEQ).

PRGs are concentration goals for chemicals for specific medium (e.g., soil) and land use combinations at
Superfund sites. There are two general sources of chemical-specific PRGs: (1) concentrations based on
Applicable or Relevant and Appropriate laws and regulations (ARARs) and (2) concentrations based on
risk calculations. The PRGs developed in this document are reference soil concentrations derived from
site-specific risk-based calculations to provide protective human health risk levels for potential direct
contact exposure to soils within the Floodplain. Regulations for corrective action in Part 111 of the
Natural Resources and Environmental Protection Act, 1994 PA 451, as amended (NREPA), MCL
324.20101 etseq and the associated environmental protection standards under Part 201 of NREPA are
considered ARARs for the Floodplain. Those regulations allow for the use of either generic soil cleanup
numbers or the development of protective site-specific values. MDEQ has made a preliminary
determination that the PRGs proposed herein are site-specific values that meet their Part 111 and 201
ARARs for direct human contact to soil. PRGs are not necessarily final soil concentrations which can be
achieved for every location along the Floodplain. However, they are expected to serve as reference soil
concentrations to identify locations where response actions would be undertaken including: soil
removal/ disposal; soil covers/ barriers; and land use management/ institutional controls.

The Floodplain is part of the larger Tittabawassee River, Saginaw River & Bay site. At this time, EPA,
working with MDEQ, is addressing the Floodplain as a non-time critical removal action (NTCRA). The
Dow Chemical Company (Dow) is conducting site investigations and developing documents, under a
2010 Settlement Agreement, with the Agencies' oversight. The expected response options were briefly
described in the Tittabawassee River Floodplain Soil Alternatives Array (Dow 2013) and are described in
more detail in the Tittabawassee River Floodplain Response Proposal (Dow 2014). The Floodplain
NTCRA is part of a larger site-wide management plan. The management approach for the site includes
developing a set of prioritized actions intended to quickly reduce exposure to, and/or transport of,
impacted media. A residual risk assessment will be completed to assess the effectiveness of the
response actions and to determine whether there is a need for further actions in the Floodplain. At this
time, EPA anticipates that the residual risk assessment for the Floodplain will be conducted after some
upstream cleanup is done, but that it will occur before all Floodplain cleanup is complete. In accordance
with Superfund law and regulations, public comment will be taken on the Floodplain Response Proposal
(which is an engineering evaluation/cost analysis) and the PRGs before the Action Memorandum is
signed by EPA, in consultation with MDEQ.

The term "Dioxin" can refer to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), dioxins and furans, or include
other dioxin-like chemicals (DLCs). Four terms are used in this document:

• Dioxin is used as a general or umbrella term;

• D/F(s) refers to dioxins and furans only;

• DLC refers to dioxin-like chemicals which include D/Fs and other dioxin-like chemicals such as
coplanar polychlorinated biphenyls (PCBs);

• TCDD refers specifically to 2,3,7,8-tetrachlorodibenzo-p-dioxin.

-------
D/Fs and DLCs are found as mixtures in environmental samples. Many of these chemicals act through a
common mechanism(s) with both demonstrated and assumed additive toxicity. As a result, a total toxic
equivalence (TEQ) approach is used in accordance with EPA guidance and Michigan regulations (EPA
2010a). EPA and MDEQ will compare the D/F TEQ (i.e., measured Floodplain soil concentration) to the
proposed PRGs. Current understanding of the D/F TEQ is based on sampling primarily conducted
between 2006 and 2008 that measured D/F, and additional sampling will be conducted, as needed.
Approximately 10,000 D/F samples were assessed from about 2,000 locations in the Floodplain. A
subset of the samples was analyzed for dioxin-like coplanar PCBs and they were not detected in
Floodplain soil. Consequently, PCBs could be disregarded as contaminants for developing PRGs based
on Dioxin TEQ.

Because of the variety of land use activities and potential human receptor groups along the Floodplain,
multiple potential PRGs values were evaluated as reference levels along the Floodplain. PRGs are
derived from the application and combination of the following factors:

A) Identification of the land uses along the Floodplain (e.g., maintained residential property;
unmaintained property; agricultural land).

B) Identification of human exposure groups and applicable exposure scenarios (e.g., child soil
exposure at a residential property; adult soil exposure at occupational locations).

C) Definition and selection of the exposure factors which are needed to estimate the level of soil
contact and contaminant intake for the specific exposure scenarios (e.g., exposure frequency,
exposure duration, ingestion and dermal intake rates; bioavailability of contaminant from soil).
Numerical values for exposure factors can be selected based on site-specific information and
studies as well as non-specific "default" exposure assumptions obtained from EPA and/or MDEQ
guidance documents or published literature reference data.

D) Information on the toxicity of dioxin. EPA and MDEQ have performed an extensive review of the
health effects of dioxin in humans and in experimental animal models which support the
understanding of human toxicity. The toxicity review resulted in the development of numerical
toxicity factors which can be used to estimate health risks from the cancer and non-cancer
effects of dioxin exposure. Dioxin can cause both human cancer and non-cancer effects.
Consequently, both a cancer slope factor (CSF) and a non-cancer Reference Dose (RfD) are
needed to characterize the health effects of dioxin and to derive PRGs for cancer and non-
cancer effects.

The selected exposure factors and toxicity factors described above are combined into standardized
algorithms (equations) which calculate soil PRG values which the Agencies consider to be protective
target goals for human health from soil direct contact. In addition, based on risk guidance and practice
by EPA and MDEQ, two PRG values should be evaluated for a given exposure scenario:

A) PRGs for the non-cancer endpoint are derived based on achieving a target Hazard Quotient (HQ)
of 1 (one). If a measured soil concentration at given location is below the non-cancer PRG, that
location is considered unlikely to pose a significant human health risk for the applicable
exposure scenario.

-------
B) PRGs will also be assessed for the cancer risk endpoint. Under EPA risk assessment practice,
cancer risk PRGs are derived for a 1E-6 (1 in 1 million) to 1E-4 (1 in 10,000) excess individual
lifetime cancer risk range (as cited in the EPA National Contingency Plan; EPA 1991a). EPA's
preference is to establish initial PRGs based on a cancer risk of 1E-6. However, final PRG levels
may differ as long as they reflect a cancer risk within the target risk range and a non-cancer PRG
reflecting a HQ of 1 (EPA, 1997).

2.0 LAND USE AND EXPOSURE SCENARIOS

2.1 Land Uses in the Floodplain

There are about 4,500 acres in the Floodplain spread along both sides of 21 miles of the lower
Tittabawassee River. The Floodplain includes land adjacent to the river that experiences flooding when
the river water levels rise above the banks of the river. River water levels typically rise during heavy
rainfall events or spring snow melt periods. Based on site investigation results (i.e., the spatial change in
D/F TEQ levels), the portion of the Tittabawassee River floodplain that is generally flooded at least once
every 8 years (the "8-year floodplain") will be the focus of response alternatives developed for the
Floodplain (ATS 2009). The 8-year floodplain boundary was delineated by Dow using topography and
aerial photographs taken during the March 2004 flood event. The 8-year floodplain boundary is not a
"bright line" and the actual boundary will be refined as needed during design, based on the actual D/F
TEQ levels at a property. Additionally, riverbank areas are included as part of the Floodplain for the
purposes of this evaluation.

Land along the Floodplain of the Tittabawassee River is associated with multiple current and expected
future land uses. The diversity and distribution of current land use in the Floodplain is illustrated on
Figure 1. Additionally, EPA evaluated how the public would like to see the Floodplain used in the future,
and those results are shown on Figure 2. As indicated by Figure 1, a significant portion (approximately
54%) of the Floodplain is currently associated with undeveloped natural landscape and/or heavily
forested land with low human use activity. Approximately 18% of land is in active agricultural use for
crop production and 16% of landscape is covered by the Shiawassee National Wildlife Refuge. The
remaining land uses (approximately 12%) can be categorized into more frequent human use activity
areas including currently maintained residential land (5%), public parks/recreational areas (3%), and
commercial/retail property (4%). Federal, State and local regulations include some restrictions on
future development in the Floodplain. As shown on Figure 2, in the future the public wants to maintain
or increase natural areas, parks and recreational areas, and the Refuge (EPA 2013).

For the purpose of developing PRG values for the soil contact exposure pathway, consideration was
given to the various current and potential future land uses in the Floodplain, the expected level of
human activity in each land use type, and the expected distribution of sensitive receptor populations
across the land uses. Based on that review, the Floodplain land uses were categorized into two general
types for the purpose of evaluating PRGs:

A) Maintained Residential Areas - Current residential properties may have portions in the
Floodplain that are maintained for frequent residential activity including open spaces for
gardening, playing, or recreational activity. This is the land use type for which the sensitive
young child receptor (i.e., age 1 to 6 years) is expected to experience the highest frequency and
opportunity for direct soil contact exposure (EPA 2002). It should be noted that it is typical for

-------
the houses and house perimeters to be out of the Floodplain and to have soil D/F levels much
lower than D/F levels within the Floodplain (see Figure 3).

B) Other Land Use Areas - all other Floodplain land use types associated with possible direct soil
contact exposure for receptors. Based on the expected types of activities and the expected
frequency of human activity, these land use types were subdivided as follows for consideration
in developing PRGs:

1) Residential Unmaintained Land - Current residential properties with portions clearly not
maintained for frequent residential activity. Residential unmaintained Land is expected to
be characterized by woodlots, brush, wetlands and other areas not subject to regular
mowing or other maintenance.

2) Other Unmaintained Land - Current Floodplain areas that are generally low use and
unmaintained, typically wooded or brushy. These areas constitute the bulk of the floodplain
acreage. These areas will have less frequent human exposure and the most frequent
potential receptors for soil contact exposure would be recreating and/or trespassing older
children or teenagers.

3) Agricultural Land - Lands currently or historically used for crop production. The potential
exposed receptor is an adult farm worker who has direct contact with soil during activities
such as plowing, seeding, and harvesting.

4) Shiawassee National Wildlife Refuge - Land dedicated to conservation management. The
receptor with the highest potential for exposure is an adult worker on the Refuge.

5) Park Land - Lands available for open public access and recreation that could have other
attractive features such as ball fields, play areas and/or trails. Children and adult recreators
would have exposure, but the receptors with the most frequent opportunity for direct soil
contact are expected to be adult site workers.

6) Commercial Land - Typically, the Floodplain portion of many commercial properties would
be considered Other Unmaintained Land. These areas include a subset of lands available for
public access and recreation (e.g., a golf course) where children, teenagers, and adults could
be exposed. However, the receptor with the most frequent opportunity for soil contact is
expected to be an adult site worker.

2.2 Exposure Scenarios

According to EPA guidance, the quantitative estimation of health risk is made by developing Exposure

Scenarios which are defined by a combination of the following:

• Identification of a human receptor for the chemical contaminant;

• Identification of the receptor location and activity which leads to contaminant exposure;

• Definition of the exposure pathway(s) which result in chemical contaminant contact followed by
intake or absorption;

• Selection of quantitative exposure factors and parameters which are used to calculate an
estimate of the dose of the chemical contaminant over the appropriate time period;

• Application of the chemical contaminant Toxicity Factor which is combined with the calculated
dose to derive the risk level.

2.3 Exposure Receptors, Activities, and Pathways

The following situations require evaluation based on the land uses described earlier:

A) Maintained Residential Areas - Direct soil contact exposure leads to contaminant intake and
absorption through the pathways of incidental soil ingestion and dermal exposure to outdoor

-------
soils and indoor dust. For evaluating the non-cancer risk endpoint, this is the land use type for
which the sensitive young child receptor (i.e., age up to 6 years) is expected to experience the
highest frequency and opportunity for direct soil contact exposure. For evaluating the cancer
risk endpoint, the sensitive receptor is a resident assumed to live at a residential location for a
total exposure duration of 30 years, with potential exposure time age-averaged over child and
adult time periods up to age 30 (EPA 2002).

B) Residential Unmaintained Land - Direct soil contact exposure leads to contaminant intake and
absorption through the pathways of incidental soil ingestion and dermal exposure to outdoor
soils and indoor dust. For evaluating the non-cancer risk endpoint, this is the land use type for
which an older child (age 7-11 years) and teenager (age 12-20 years) are expected to experience
a higher frequency and opportunity for direct soil contact exposure compared to a young child.

C) Park Land and Other Unmaintained Land - Direct soil contact exposure leads to contaminant
intake and absorption through the pathways of incidental soil ingestion and dermal exposure to
outdoor soils. For evaluating the non-cancer risk endpoint of a recreator, an older child (age 7-
11 years) and teenager (age 12-20 years) are expected to experience the highest frequency and
opportunity for direct soil contact exposure.

D) Agricultural Land, Refuge Land, Park Workers, and Commercial Land - Direct soil contact
exposure leads to contaminant intake and absorption through the pathways of incidental soil
ingestion and dermal exposure to outdoor soils. The receptor requiring evaluation is an adult
worker (age 21 and older) for whom the non-cancer and cancer risk endpoint applies. For
evaluating the cancer risk endpoint, the adult worker is assumed to have Floodplain soil
exposure for a duration of 25 years.

2.4 Exposure Factors and Parameters

As explained earlier (Section 2.2), quantitative exposure factors and parameters are needed to calculate
an estimate of the dose of the chemical contaminant over the appropriate time period. Two types of
exposure factors are employed for risk evaluation: generic or default values and site-specific values.
Default values are used for factors that can be estimated from population statistics (e.g., body weight;
dermal surface area) or human activity studies (e.g., soil ingestion rate; residential exposure duration).
Factors which were assigned default values are shown in Tables 1 and 2.

Site-specific values are derived from local/regional information or studies which are valid to apply to the
specific circumstances at the site, to reduce uncertainty or variability in the risk evaluation. The site-
specific exposure factors and parameters are also summarized in Tables 1 and 2, and discussed in detail
below. Based on evaluation of the Floodplain exposure scenarios and available information, the
following exposure factors were selected to be defined using site-specific information or studies.

2.4.1 Exposure Frequency
Soil and dust exposure frequency are both evaluated on a site-specific basis. Outdoor days include both
soil and dust exposure. Indoor days include only dust exposure. These exposure frequencies apply to
both the ingestion and dermal components.

2.4.1.1 Local Climate - Determining Outdoor vs. Indoor Exposure Frequency
Local climate data is used to assign outdoor soil exposure frequency to only those days without snow
cover (e.g., <1 inch) and/or without frozen soil (e.g., soil temperature >32°F). These endpoints are used

-------
to determine the reasonable maximum exposure (RME) for the following reasons: 1) the studies that
serve as the basis of the soil ingestion rates were conducted primarily in the summer or fall, but did not
exclude days of inclement weather or days without outdoor play time; and 2) days with precipitation
events may also represent days with outdoor activities depending on timing and amount of
precipitation, and type of outdoor activity. Therefore, the conditions of snow cover and/or frozen soils
are used to define the number of days with only indoor dust exposure.

The available local climate data (NOAA 2010 and MSU 2010) for the 2005-2009 period indicate the
number of days with either snow cover or frozen soils (<32°F) to be 90 days (88.8-91.8 days as a range to
account for days with missing data), resulting in 275 days (365 minus 90 days) when the soil is not frozen
or there is less than an inch of snow cover. Consequently, the site-specific value selected for outdoor
exposure days for the PRG derivations is 260 days per year (275 minus 15 days assumed to be spent
away from home). This site-specific exposure frequency value is actually greater than the MDEQ
Statewide default value of 245 days per year. The site-specific value for indoor only exposure days for
the PRGs derivation is the 90 days per year with either snow cover or frozen soils.

2.4.1.2 Residential Exposure Scenarios - Apportioning Outdoor Exposure Days
As discussed above, the site-specific value for outdoor exposure days is 260 days per year. EPA and
MDEQ applied this exposure frequency in two ways while calculating the PRGs for residential exposure
scenarios (see Section 4): 1) First, PRGs were calculated for the sensitive young child receptor assuming
that all 260 days of outdoor exposure could take place on residential soil with some amount of
contamination. 2) Second, once that PRG was calculated, the 260 days of outdoor exposure were
apportioned to account for different expected land uses (and concentrations) within a residential
property.

The site-specific residential exposure scenario considers the variation in soil concentration data and
exposure potential for most of the Floodplain residential properties. Houses are generally not in the
Floodplain. The typical residential property on the Tittabawassee River floodplain has three different
types of areas where direct contact soil exposure may occur (See Figure 3):

•	An area around the house perimeter outside the Floodplain that has low soil concentrations that
are less than 50 parts per trillion (ppt). (Zone A-l on Figure 3)

•	Maintained Residential Areas within the Floodplain that upon completion of the cleanup will
meet the proposed Maintained Residential PRG. (Zone B)

•	Residential unmaintained land within the 8-year floodplain with varying Dioxin concentrations.
Upon completion of the cleanup, these areas will meet the proposed Other Land Use Areas PRG.
(Zone C)

Exposure frequency is used to represent proportional amount of exposure time spent in the different
areas described above. The amount of time spent in each of these areas was considered for different
age groups for the residential receptors as follows and as summarized in Table 3 below:

•	Young child receptor (1-6 years) - the RME is considered to spend most of the time around the
house and in the maintained area within the 8-year floodplain. Less time is spent in the
unmaintained or other use areas, and then only when accompanied by older sibling or adult.
The expectation is that a young child would not be allowed to play unsupervised in
unmaintained areas adjacent to the river.

•	Older child receptor (6-12 years) and teenage receptor (12-21 years) - the RME is considered to
have more independence to play/spend time in unmaintained and other land use areas in the

6

-------
Floodplain, including weekends during the school year and five out of seven days per week
when school is not in session, with the remainder of the days split between the house perimeter
outside the Floodplain and the maintained areas within the Floodplain.

• Adult receptor (>21 years) - the RME is considered to spend weekend days in the unmaintained
area, with the remainder of the days split between the house perimeter outside the Floodplain
and the maintained areas within the Floodplain.

Receptor Age
Group

Total
Outdoor
Exposure
Frequency
(days/year)

Exposure
Frequency for
Residential Area
Outside Floodplain
(days/year)

Exposure Frequency
for Maintained
Residential Area
Inside Floodplain
(days/year)

Exposure Frequency for

Residential
Unmaintained or Other
Land Use Area Inside
Floodplain (days/year)

Young child
(1-6 years)

260

121 (47%)

121 (47%)

18 (7%)

Older child
(7-11 years)

260

76.5 (29%)

76.5 (29%)

107 (41%)

Teenager
(12-21 years)

260

76.5 (29%)

76.5 (29%)

107 (41%)

Adult
(>21 years)

260

93 (36%)

93 (36%)

74 (28%)

Table 3: Apportioning Outdoor Exposure Days for Residents

For most residential properties in the Tittabawassee River floodplain these RME scenarios are
adequately protective. A very few residential properties are almost completely inside the Floodplain;
therefore soil around the house perimeter may have elevated TEQ. However, most of these residences
have already been cleaned up by complete excavation and backfill with clean soil at background levels.
In addition, there may be a very few non-residential property uses that do not fit the exposures
considered for the Other Land Use PRG. For these properties, a property specific evaluation may be
necessary to determine an exposure scenario that takes into consideration greater exposure to
contamination in the Floodplain. Alternatively the proposed PRGs for the Maintained Residential and
Other Land Use areas may be appropriate in such cases.

2.4.1.3 Exposure Frequency for Non-Residential Scenarios
For non-residential property, an adult worker scenario assumes a person who attends a workplace
located where all of the potential soil exposure is within the impacted Floodplain. The worker was
assessed for 186 days of outdoor soil exposure (five days per week for the 260 days based on the
climate data; Section 2.4.1.1) and 245 days of indoor dust exposure (i.e., 245 total work days split into
186 days with both outdoor soil/indoor dust exposure and 59 days with only indoor dust exposure).

2.4.2 Partition of Soil Exposure: Outdoor Soil and Indoor Dust
Estimates of total daily incidental soil ingestion exposure are typically modeled or assumed to be
composed of soil from two sources: outdoor soil and indoor dust. A subset of annual days will only
have indoor dust exposure. Consequently, an estimate for indoor daily dust exposure (ingestion and
dermal) is needed. Dust is composed partially of outdoor soil transported into the home by various
processes (e.g., tracking on shoes or clothes; pet traffic; open doors and windows). No verified sources
of data to derive a site-specific value or site-specific partition ratio of outdoor soil to indoor dust could
be found. Consequently, a partition ratio of forty five-fifty five (45:55) between soil and dust exposure
was selected based on the recommendation found in the EPA Exposure Factors Handbook (EFH) (EPA,

7

-------
2011). This ratio is included in the calculations for soil exposure on outdoor days, with all of the
exposure for indoor days coming from indoor dust only.

The EFH recommended partition ratio of forty five-fifty five (45:55) between outdoor soil and dust
exposure relates to soil ingestion. Because appropriate data is not currently available, the EFH does not
recommend a partition ratio for soil and dust exposure via dermal contact. However, based on
professional judgment, the Agencies believe that some split between soil and dust exposure for outdoor
days is also appropriate for dermal contact. For the purposes of the Floodplain PRG calculations, the
ingestion partition ratio of forty five-fifty five was also used for soil and dust exposure via dermal
contact on outdoor exposure days (i.e., 260 days/year). This is discussed further in Section 5.0,
Uncertainties and Sensitivity Analysis.

2.4.3 Dust Concentration

After determining that indoor dust should be modeled as a separate source of contaminant exposure,
EPA and MDEQ sought an approach for how a site-specific Dioxin TEQ concentration in indoor dust could
best be determined. The University of Michigan Dioxin Exposure Study (UMDES) collected site-specific
data from residential properties in the Floodplain for both indoor dust and outdoor soil (UMDES 2008).
This data has been considered for determining appropriate indoor dust concentrations for exposure
related to the soil contamination. Options for using the available data include: 1) a dust-to-soil ratio;
and 2) use of the dust concentration data directly (i.e., 95% upper confidence limit or UCL of the mean
concentration).

A dust-to-soil ratio has been considered for other sites to address indoor dust concentrations, so this
approach was explored first using the UMDES data. The UMDES developed a linear regression model
that resulted in a dust-to-soil ratio of 0.2. However, the Agencies do not believe that this linear
regression model should be used to develop the site-specific PRG because the model is not transparent
and may be confounded with collinear parameters and sampling weights, and consequently, may not be
acceptable for determining the appropriate dust/soil concentration ratio. Additionally, paired soil/dust
data is not available from UMDES, only summary statistics. Since most of the properties along the
Tittabawassee River floodplain have large variations in soil concentrations, using a dust-to-soil ratio
from summary statistics is challenging. Therefore, using a dust-to-soil ratio to estimate dust
concentrations does not appear to be the preferred approach for this site.

Another approach would be to use the dust concentration data directly, instead of a dust-to-soil ratio.
The UMDES analyzed dust from 207 Floodplain residences. The Agencies have opted to use a fixed
Dioxin TEQ dust concentration based on this dataset (mean = 35 ppt, median = 15 ppt, 95% UCL of the
mean = 50 ppt). The Agencies typically use a 95% UCL of the mean to represent the concentration data
for an individual exposure unit. Consequently, a value of 50 ppt is used as the dust Dioxin TEQ
concentration for the derivation of PRGs.

2.4.4 Oral Bioavailability / Ingestion Absorption Efficiency

Oral bioavailability is the proportion of an ingested chemical that is absorbed from the gastrointestinal
tract into the bloodstream and tissues. Bioavailability of D/F from contaminated soil could be influenced
by several factors including the source of the D/F, the soil type, and weathering. If there is evidence that
the bioavailability of the D/F in soil compared to that of the test medium in the critical study on which
the RfD and/or CSF are based is less than 100%, then an adjustment to the bioavailability is appropriate
(EPA 2010b). The ratio of D/F bioavailability in site soil compared to the D/F bioavailability of a non-soil

-------
reference medium is referred to as the Relative Bioavailability (RBA) and becomes the basis for a site-
specific adjustment factor.

EPA issued a report which reviewed the available information from the published literature on the
bioavailability of dioxins in soil (EPA 2010b). This review included the Dow rat and swine pilot and
follow-up studies (Budinsky et al, 2008). The primary objectives of the EPA literature review and data
analysis were to: 1) Identify and summarize the best available studies that could be used for estimating
RBA in soils that contain multiple D/F congeners; 2) Determine if data from the best studies are
adequate to conclude that RBA for D/Fs in soil is less than 100%; and 3) Determine if data from the
studies are adequate to recommend a quantitative nationwide default RBA value (e.g., central tendency;
high-end) for application to site-specific risk assessments. The Report identified three well conducted
studies in which quantitative RBA estimates were made for soils containing multiple congeners and for
soils tested in more than one species (Budinsky et al. 2008; Finley et al. 2009; Wittsiepe et al. 2007).

The analysis of the studies supported the following conclusions: 1) The RBA of D/F mixtures in soils can
be expected to be less than 100% based on comparison to a lipid or organic solvent used as the
reference material (e.g., corn oil); 2) Available estimates of soil dioxin RBA are not adequate for
recommending a nationwide default RBA value to use in risk assessments as an alternative to 100% or
actual site-specific values; 3) RBA varies with the level of congener chlorination in a manner that
suggests species differences for the RBA of chlorinated congeners; and 4) The available data and
protocols are not adequate to determine a preferred animal model for predicting soil RBA in humans.

One of EPA's conclusions is that current information is not sufficient to determine a preferred animal
model or bioassay protocol for predicting soil RBA in humans. Part of that determination is based on
conflicting results observed for RBA with increasing levels of dioxin chlorination. Rodents appear to
have lower soil RBA with increasing chlorination. Based on the available swine studies, including the
Dow studies, swine appear to have higher soil RBA with increasing dioxin chlorination (Budinsky et al.
2008). The EPA Report also made a case that RBA studies in swine should be considered as valid for
making estimates of RBA in soil. The Report states: "While it is not an objective of this report to evaluate
a preferred animal model, there are several potential strengths with using swine for estimating RBA of
dioxins in soil. As demonstrated for lead bioavailability, similarities between the physiology and
anatomy of juvenile swine and human gastrointestinal tracts make swine a suitable model for predicting
RBA in humans (USEPA 2007). However, it is important to note that juvenile swine are appropriate for
estimating lead bioavailability because the primary concern is exposure to young children, as compared
to PCDD/Fs where all life stages are of interest. Swine and rats also differ in the distribution of absorbed
PCDD/Fs. Similar to humans, swine accumulate higher levels in adipose tissue relative to the liver,
whereas, the distribution in rats tends to show the opposite trend (Budinsky et al. 2008; Thoma et al.
1989, 1990)." (EPA 2010b; page 28).

Consequently, EPA and MDEQ performed a more detailed review of bioavailability studies conducted in
rats and swine by Dow on a soil sample from the Tittabawassee River floodplain.

First, the Agencies reviewed the data on D/F distribution in floodplain soil samples and concluded that
the selected test soil evaluated for bioavailability by Dow was adequately representative of the
floodplain soils of concern. Dow provided evidence showing that a high proportion of the D/F TEQ
measured in floodplain soils is strongly associated with particulate anthropogenic black carbon that was
produced and released downstream into the Tittabawassee River during the chloralkali production
process (Chai et al., 2011). In addition, Chai et al. (2007) measured specific surface areas of bulk

-------
floodplain soils and their sub-fractions and their associated D/F TEQs. No correlation was observed
between D/F TEQ distribution and the higher specific surface areas present in the finer sub-fractions of
the soil. The Floodplain soil data support the conclusion that, at this site, the anthropogenic black
carbon and the D/F TEQ adsorbed to black carbon control bioavailability; and natural organic matter
(which can vary by soil type) is much less important (ATS, 2007; Chai et al, 2007).

Second, the Agencies reviewed the Dow studies on RBA of D/F congeners in soil using two animal
models, Sprague-Dawley rats and juvenile swine. The studies were designed to measure the
bioavailability of the five D/F congeners which contribute the highest portion of TEQ (> 90%) for the
Floodplain soils (Dow, 2005). The following is a summary of the pilot studies conducted in rats and
swine, and a follow-up study conducted in rats with the Floodplain soil (Dow 2006).

A) The pilot study measured liver and adipose tissue D/F congener levels in soil-fed animals and
control animals. The objectives were to evaluate the study designs including the number of
animals per dose group and to confirm the analytical methods necessary to detect the D/Fs
retained in the liver and adipose tissue of both animals. Soil was administered for 30 days as a
soil/feed mixture for rats and as soil wrapped in a dough ball for swine. The control animals
ingested matched doses of the same five congeners in a corn oil vehicle. (The dosing for control
rats included a corn oil reference. The control swine were dosed with corn oil vehicle in a
gelatin capsule which was wrapped in a dough ball.) The results estimated the RBA for
individual congeners based on the comparison between the fraction of the dose retained by the
soil-fed group and the vehicle-dose group. Then a TEQ-weighted estimate of RBA for each
species was obtained by weighting the individual congener RBA estimates by their respective
contribution to the TEQ concentration of the floodplain soil sample.

B) The pilot study included measurement of 7-ethoxyresorufin-O-deethylase (EROD) which is linked
to liver cytochrome P450 enzyme activity. The purpose was to evaluate whether a different
level of enzyme induction was occurring between the soil-fed animals and the control animals.
The control group rats showed higher EROD activity compared to soil fed rats. The higher EROD
induction in the controls led to speculation that the observed RBA estimates may be elevated
because control rats experienced increased metabolic activity which could result in faster
elimination of the D/F congeners in the control rats. (No significant differences in hepatic EROD
activity were observed among the swine treatment groups.)

C) A follow-up study in rats with the Floodplain soil and several dose-matched corn oil controls was
designed to evaluate whether the increased EROD activity was influencing the RBA results (Dow,
2006). The follow-up study demonstrated that when controls with similar dose and EROD
activity were used to estimate RBA, the results were not different from those of the pilot study
for four of the five congeners. The only congener that appeared to be significantly elevated in
the pilot study was 2,3,7,8-tetrachlorodibenzofuran (TCDF). Because TCDF is known to have a
much shorter half-life than the other D/F congeners, it is more likely to have increased
elimination. Since TCDF is an important congener for the Tittabawassee River floodplain soil, it
was concluded that the rat pilot study data should not be used. Only the follow-up study data
for the rat should be used. The follow-up study had two control doses (50% and 80% of the soil
fed dose) that appeared to match the EROD activity measured in the soil fed rats. Consequently,
it was determined that an average of the RBA calculated using each of these two controls would
be appropriate to use for the rat data.

-------
D) One of the important conclusions of the follow-up study was that there appeared to be a true
difference in RBA between rats and swine. It is not clear whether the difference could be due to
species absorption differences and/or soil dosing differences (i.e., soil mixed with normal feed
for rats vs. soil within dough balls given as a bolus intake for swine).

Consequently, at the present time, it is not clear which animal model would better represent the human
oral bioavailability of dioxin in Floodplain soil. Therefore, EPA and MDEQ concluded that it would be
appropriate to use an average of the two species RBA results for the Floodplain soil.

Since the bioavailability from soil must be evaluated relative to the test medium for the critical toxicity
study, the appropriate values to use for the Tittabawassee River floodplain are:

A) For the EPA non-cancer RfD - use the oil gavage control in rats and the dough-ball control in the
swine.

B) For the EPA and MDEQ CSFs - the test medium was rodent feed for the rat study which serves as
the basis for both the current EPA and MDEQ CSFs. Therefore, a soil bioavailability relative to
rat feed is appropriate for dioxin cancer risk assessment with the current CSFs. (For floodplain
soils, the 2,3,4,7,8-pentachlorodibenzofuran (4-PeCDF) congener contributes more to the TEQ
(32-38%) than any of the other congeners and appears on average to be representative of the
TEQ oral bioavailability in the rat. Therefore, using the relative bioavailability of 4-PeCDF
between the feed reference and the oil gavage reference to adjust the TEQ- weighted relative
bioavailability for use with the CSF is appropriate.)

The final consideration for the bioavailability values is how to address different congeners that are
contributing to the TEQ. A TEQ-weighted average from the five congeners evaluated in the
bioavailability study is used to represent the total TEQ bioavailability. That approach is valid for the
following reasons: a) the five congeners in the test soil contributed 92% of the TEQ for the soil that was
used in the bioavailability study; and b) the five congeners tested for bioavailability represent 87-89% of
the average TEQ for the large collection of Tittabawassee River floodplain soil samples. The distribution
of the five congeners in the test soil is provided in Table 4 and compared to the average congener
distribution of soil samples collected along the Tittabawassee River floodplain.

As a result of the above considerations, the oral bioavailability values were derived as follows. For the
RBA to use with the 2012 EPA RfD:

• Compute the average of the reported TEQ-weighted rat RBA values from the follow-up study
using the 50% (0.5x) and 80% (0.8x) oil gavage controls.

• Compute the average of the reported TEQ-weighted swine RBA values at both half the detection
level and the full detection level, and

• Use the average of the rat and swine averages.

For the RBA to use with the CSFs:

• Compute the average of the reported TEQ-weighted rat RBA values from the follow-up study
using the 50% (0.5x) and 80% (0.8x) oil gavage controls adjusted by dividing by the feed relative
to oil gavage control for 4-PeCDF.

• Compute the average of the reported TEQ-weighted swine relative to dough ball bioavailability
values at both half the detection level and the full detection level, and

• Use the average of the rat and swine averages.

-------
These values are displayed in Table 5. The values recommended for the Tittabawassee River floodplain
as best available information for oral bioavailability are:

•	0.43 (43%) relative to oil bioavailability for use with the 2012 EPA RfD; and

•	0.51 (51%) relative to feed bioavailability for use with either the EPA or MDEQ CSF.

2.4.5 Dermal Absorption Efficiency (ABSd)

Dermal Absorption (also known as "Percutaneous Absorption") refers to the amount of chemical that
can enter into the body's circulatory system after application to the skin. Dermal absorption is the
translocation of a substance across the skin to the point where it is introduced into the circulation via
the capillaries which perfuse the dermis. The process entails sequential diffusion of the chemical
through two differentially selective barriers commonly referred to as the "stratum corneum" and the
"epidermis." The stratum corneum is the outer layer of cornified non-viable cells which provides the
primary barrier to chemical translocation. Chemical substances must diffuse through the lipid-rich
intercellular matrix of the stratum corneum in order to reach the thicker viable epidermis layer. The
stratum corneum is the rate-limiting diffusion barrier for hydrophilic (water soluble and ionic)
substances; and the epidermis is the rate-limiting diffusion barrier for the lipophilic (fat soluble/water
insoluble) substances.

Dermal absorption of organic compounds such as TCDD, PCBs, and some pesticides can be extensive
when they are applied to skin in neat solutions. Absorption has also been measured, but at a
considerably lower rate, for some of these chemicals when applied to skin in a soil matrix (e.g., TCDD,
PCBs, benzo[a]pyrene). EPA evaluated the original experimental studies on TCDD dermal absorption to
the skin of rats in vivo and to rat and human excised skin preparations in vitro (EPA 1992). The available
studies indicated that a higher level of total TCDD absorption over a specified time period was observed
for low organic carbon soils (0.45%) spiked with radiolabeled TCDD compared to high organic carbon
soils (11%). Based on the results of the absorption studies on rats in vivo and rat and human excised
skin preparations in vitro, the human dermal absorption rate was estimated from the following
relationship:

Human in vivo ABS = (Human in vitro ABS) x (Rat in vivo ABS)

(Rat in vitro ABS)

The approach above assumes that the ratio of in vivo to in vitro measured absorption fractions for a
specific contaminant will be the same in humans as in animal species. The validity of this approach
depends on similarities in skin structure and pharmacokinetic processes between animals and humans.
Recognizing unavoidable differences between mammalian species, the above relationship is still
considered to be the best available approach for making the human dermal absorption estimate based
on experimental studies under controlled conditions.

Based on EPA's evaluation of the best available published studies, the following range was
recommended for the dermal absorption efficiency estimate for TCDD in soil: 0.1 % to 3%. The high end
of the range was recommended for soils with low organic carbon content, and the low end of the range
for soils with high organic carbon content (EPA 1992).

Since the original evaluation, EPA performed and published an additional study which is suitable for
updating the original recommendation for TCDD dermal absorption (EPA 2008a). The purpose of the
new study was to conduct a more complete set of experiments using fully characterized soil and

12

-------
including intravenous administration, rat in vitro skin testing, and human skin in vitro testing. The
experiments were designed to allow comparison of the following experimental conditions: In vivo and in
vitro experiments; TCDD applied in neat form and in soil; absorption across intact in vitro rat and human
skin samples; and absorption from soil with low and high organic carbon contents. Ultimately, the study
discusses dermal absorption efficiencies that should be considered when evaluating human exposure to
TCDD contaminated soils.

In the updated EPA study, eight groups of dermal absorption measurements were conducted (two rat in
vivo; six rat in vitro or human in vitro samples). TCDD was applied in neat solution (high dose at 250
Hg/cm2 and low dose at 10 ng/cm2) or sorbed on a low total organic soil (0.5%) or high total organic soil
(11%) at 1 ppm (10 ng TCDD/10 mg soil/cm2). Risk assessments generally assume that dermal
absorption from a single soil exposure occurs for up to 24 hours (i.e., soil remains on skin up to 24 hours;
EPA 2004). After a 24 hour exposure time, the percent of TCDD dose absorbed compared to the starting
TCDD dose on low organic soil was 7.9% (rat in vivo), 3.8% (rat in vitro) and 0.5% (human in vitro). The
percent of dose absorbed from TCDD on high organic soil was 0.1% (human in vitro). Human skin was
observed to be three to four times less permeable to TCDD than rat skin across a range of doses and
exposure times. Using the algorithm mentioned previously, the human skin in vivo absorption efficiency
was estimated to be 1.0% for the low organic soil. After adjustments to account for differences
between in vitro and in vivo results and adjusting for application to monolayer loads, the 24-hour TCDD
absorption value recommended for human skin was 1.9% from low organic soil and 0.24% from high
organic soil.

Consequently, for the purpose of deriving PRG values a dermal ABSd value of 0.02 (i.e., 2.0% rounded up
from 1.9%) was used for the following reasons:

1) The 2% value is recommended for soils with low organic carbon content. The average total
organic carbon content for Tittabawassee River Floodplain soils is < 1% which corresponds well
with the low organic carbon content soil from the EPA 2008 study.

2) The 2% value is considered to be the best available dermal absorption estimate for human skin
for application to a risk assessment employing the 2012 EPA non-cancer RfD. The EPA non-
cancer RfD value is based on epidemiology studies on TCDD exposure where the measured
human serum TCDD level resulted from the likely combination of oral and dermal absorption
pathways.

2.4.6 Skin Surface Area Parameter
In order to account for dioxin contaminant intake through dermal exposure, values need to be selected
for the parameter known as skin surface area. The parameter is needed for estimating contaminant
absorption through the dermal pathway and is dependent on the type of receptor (e.g., child, adult) and
the exposure scenario under consideration (e.g., residential, worker).

EPA has reviewed the published studies and other available information on skin surface area in order to
provide recommendations for incorporating this parameter into Superfund risk assessments. Two
factors need to be evaluated in order to derive a value for the skin surface area parameter: 1) Total
body surface area; and 2) Fraction of total body surface area attributed to specific exposed body parts.
These two factors are combined to estimate the surface area available for exposure.

The recommendations for total body surface area are presented in the EFH, Chapter 7: Dermal
Exposure Factors (EPA 2011). The EFH reviews the latest National Health and Nutrition Examination

-------
Survey (NHANES) empirical data on body surface area and presents the metadata by age groups. The
recommendations for fractions of total body surface area attributed to body parts are presented in the
EPA Risk Assessment Guidance for Superfund: Supplemental Guidance for Dermal Risk Assessment Part E
Final (EPA 2004).

The above data sources were consulted in order to derive estimates of exposed skin surface area
classified by the receptor group and exposure scenario. In addition, EPA, in consultation with MDEQ,
determined which body parts would be expected to present exposed skin surfaces under the climate
conditions and activity practices expected in the Floodplain areas. The combinations of receptor
groups/exposure scenarios and the skin surface estimates are shown in the table below:

Receptor/Exposure Scenario

Total Body
Surface Area*
(sq cm)

Body Parts Available for
Exposure

Exposed Skin
Surface Area*
(sq cm)

Young Child/Residential/Recreational

6840

face, neck, hands, forearms,
lower legs

2052

Older Child/Residential/Recreational

10800

face, neck, hands, forearms,
lower legs, feet

3920

Teenager/ Residential/Recreational

17150

face, neck, hands, forearms,
lower legs, feet

6260

Adult/Residential/Recreational

19780

face, neck, hands, forearms,
lower legs

5618

Adult/Worker

19780

face, neck, hands, forearms

3026

Table 6: Skin Surface Areas

*Details of the derivation are presented in Appendix A

2.4.7 Body Weight Parameter
In order to make estimates of dioxin contaminant intake, values need to be selected for the body weight
of each receptor group. Body weight values are used in the calculation of average daily dose for the
non-cancer endpoint and lifetime average daily doses for the cancer risk endpoint.

EPA has reviewed the published studies and other available information on body weight in order to
provide recommendations for incorporating this parameter into Superfund risk assessments. The
primary EPA data review and recommendation documents to inform the body weight parameter
include: 1) Child-Specific Exposure Factors Handbook (2008b), Chapter 8: Body Weight; and 2) Exposure
Factors Handbook (EPA 2011), Chapter 8: Body Weight Studies. Reference 1 describes the EPA review
of the NHANES data (1999-2006) on body weight and presents the recommended metadata for a
number of age groups for children up to age 21. This reference recommends treating the adult as a
person 21 years of age or older. Reference 2 describes the EPA review of the NHANES data (1999-2006)
on body weight for adults and presents the recommended metadata for a number of adult age groups.

The above data sources were consulted in order to derive estimates of body weight classified by the
receptor group and the corresponding age range needed for calculating contaminant dose estimates.
The combinations of receptor groups and body weights are shown in the table below:

14

-------
Receptor Group

Age Range

Body Weight (kilograms) *

Young Child

1-6

16.2

Older Child

7-11

31.8

Teenager

12-21

63.4

Adult

21-70

81.8

Table 7: Body Weights

*Details of the derivation are presented in Appendix B

For the adult body weight shown above, the high end of adult age is truncated at 70 years because the
calculation of Lifetime Average Daily Dose (LADD) is defined to correspond to a lifetime age of 70 years.
The LADD is a parameter needed to calculate the cancer risk component of a PRG value (EPA 1989;
1991b).

3.0	DIOXIN TOXICITY

When information is available for multiple adverse effects of a hazardous substance, an evaluation of
both cancer and non-cancer adverse health effects is necessary to determine the most sensitive effect
for developing PRGs (EPA 1991b).

3.1	Non-Cancer Reference Dose

Relatively new information regarding prenatal and postnatal health effects attributed to dioxin exposure
and changes in risk assessment practices have resulted in the necessity to more closely consider the
potential for non-cancer adverse effects in developing the current PRGs. Based on this information, EPA
developed an oral RfD that was finalized in February 2012 and posted to the Integrated Risk Information
System (EPA, 2012a). The associated toxicity assessment published at the same time is the final version
of the non-cancer portion of the EPA Reanalysis of Key Issues Related to Dioxin Toxicity and Response to
National Academy of Sciences (NAS) Comments, Volume 1 (EPA, 2012b). The 2012 EPA RfD is the best
available information for assessment of non-cancer endpoints. In addition, because the adverse effects
captured by the RfD are related to early-life exposures, the appropriate sensitive receptor is a child.
Therefore, a young child receptor is used to develop the non-cancer direct contact PRGs.

EPA derived the 2012 RfD of 7.0E-10 mg/kg-day based on two human epidemiology studies
demonstrating altered thyroid function (Baccarelli et al, 2008) and impaired adult male reproductive
function (Mocarelli et al, 2008) associated with prenatal and postnatal exposure toTCDD, respectively.
The Baccarelli study evaluated serum Thyroid Stimulating Hormone (TSH) levels in neonates born to
mothers who were exposed to TCDD 17-29 years prior to pregnancy because of a 1976 chemical plant
explosion in Seveso, Italy. The adverse effect was identified as an increase in TSH levels above the
World Health Organization standard of 5 pi-units TSH per mL of serum, which indicated dysregulation of
thyroid hormone metabolism. The Mocarelli study reported decreased adult sperm concentrations and
decreased motile sperm counts in men who were 1-9 years old in 1976 at the time of initial exposure to
TCDD from the Seveso accident.

The 2012 RfD uses intake rates derived using the Emond et al. (2005) human physiologically-based
pharmacokinetic (PBPK) model from serum concentrations reported in the studies. For the Baccarelli et
al. study, EPA used the study's regression model to estimate a maternal plasma TCDD concentration at
the neonatal TSH level of concern, and the Emond human PBPK model under the gestational scenario to
determine the maternal intake rate lowest-observed-adverse-effect level (LOAEL) of 2.4 xlO"8 mg/kg-

15

-------
day. For the Mocarelli et al. study, since it was not clear whether the effects were related to the peak
exposure or to the average exposure, EPA used the average of the estimated intake rates for both to
derive an intake rate LOAEL of 2.0 x 10"8 mg/kg-day. EPA decided that the two studies could be regarded
as co-critical studies with a LOAEL converging around the value of 2.0 x 10"8 mg/kg-day. To derive the
RfD value, the LOAEL was adjusted downward by a 30x uncertainty factor: 10 for the LOAEL and 3 for
the possibility of within human variability.

3.2 Cancer Slope Factor

EPA is concerned with addressing potential cancer risk for dioxin exposure. However, EPA has not yet
determined a final CSF for evaluating cancer risk. EPA is continuing with development of a new final CSF
as part of the ongoing EPA Dioxin Reassessment/Reanalysis (EPA 2009). That effort will take some
additional time, and no projected completion date is available. In the absence of a final dioxin CSF, EPA
policy calls for conducting reviews of the available EPA and non-EPA sources of scientific information to
determine if an appropriate interim CSF value can be recommended for use in Superfund risk
assessments. Priority will be given to those sources of information that are publicly available, have a
transparent analysis of original data, and which have been subjected to peer review (EPA 2003). On that
basis, EPA identified two candidates for use as valid interim CSF values:

A) EPA's Office of Health and Environmental Assessment (EPA 1985) developed an oral cancer
slope factor of 1.56E-04 (pg/kg-day)"1. This was based on the combined incidence of lung,
palate, and nasal carcinomas, and liver hyperplastic nodules or carcinomas in female rats in the
study by Kociba et al. (1978).

B) California EPA (CalEPA 1986 and2002) developed an oral cancer slope factor of 1.3E-04 (pg/kg-
day)"1. This was based on the occurrence of hepatocellular adenomas and carcinomas in male
mice in a study by the National Toxicology Program (NTP 1982).

EPA cited the value from OHEA as the preferred value to use as an interim CSF because it is derived from
the evaluation of all tumors types confirmed in the test animals (EPA 2009).

3.3 EPA Policy for Application of Dioxin Toxicity Factors for Development of PRGs

For the goal of applying the best current science as the basis for its cleanup actions, EPA announced that
the Agency will use the final RfD for TCDD to address cleanup projects under Superfund. Application of
the new RfD will apply to the development of site-specific PRGs. The following statement is found at
http://www.epa.gov/superfund/health/contaminants/dioxin/dioxinsoil.html

"Dioxin-contaminated sites cleaned up based on the new non-cancer RfD are not expected to
need additional cleanup when a new EPA cancer toxicity value for dioxin is published in EPA's
Integrated Risk Information System (IRIS). This is because we anticipate that dioxin cleanup
levels based on the new non-cancer RfD will be within the cancer risk range currently used by
EPA's Superfund and RCRA cleanup programs."

4.0 DERIVATION OF PRG VALUES

PRGs were calculated for a variety of direct contact Floodplain soil exposure scenarios. The calculations
followed standard EPA and MDEQ algorithms and used a combination of both standard default and the
site-specific input parameters discussed herein. Potential PRGs were calculated to assess both non-
cancer risks to meet a HQ = 1 and cancer risks to meet EPA's risk range. Additionally, under MDEQ risk

-------
assessment regulations (Part 201 of NREPA), cancer risks are assessed for 1E-5 (1 in 100,000) using a CSF
derived from EPA's Great Lakes Water Quality Initiative for TCDD (EPA 1995).

Based on these calculations, EPA and MDEQ are proposing two site-specific PRGs for the Floodplain soil:
1) Maintained Residential Areas; and 2) Other Land Use Areas. The PRGs are based on the most
sensitive receptor and exposure scenario within each land use. Thus, the PRGs will be protective for all
other direct contact receptors/scenarios.

4.1 Maintained Residential Areas PRG

For Maintained Residential Areas of the Floodplain, EPA and MDEQ are recommending a PRG of 250 ppt
TEQ.

The most sensitive receptor and endpoint for the Maintained Residential Areas of the Floodplain is a
young child for non-cancer effects. The proposed PRG is based on calculated values to achieve a HQ of
1, assuming that all of the child's soil exposure is to the maintained portion of the property in the
Floodplain. The calculations are shown in Appendix C. The proposed cleanup level for Maintained
Residential Areas is 250 ppt TEQ based on the site-specific assumptions described in this memorandum,
adjusted to account for exposures to other areas of the residential property and other uncertainties.
PRGs were calculated for older residents and for cancer endpoints, and the calculated values were less
stringent than 250 ppt TEQ.

4.2 Other Land Use Areas PRG

For Other Land Use Areas of the Floodplain, EPA and MDEQ are recommending a PRG of 2,000 ppt TEQ.
This PRG will apply to direct contact exposure in: Residential Unmaintained Land; Other Unmaintained
Land; Agricultural Land; Shiawassee National Wildlife Refuge; Park Land; and Commercial Land.

The most sensitive receptor and endpoint for the Other Land Use areas is also a young child for non-
cancer effects. The calculated value to achieve a HQ of 1 is 2,000 ppt. This value is based on the site-
specific assumptions described in this memorandum, including the apportionment of time in various
areas discussed in Section 2.4.1.2. PRGs (cancer and non-cancer) were calculated for older residents
and adult workers, and the calculated values were less stringent than 2,000 ppt. PRGs were also
quantitatively calculated for older children, teen or adult recreators, and those values are less stringent
than the residents of the same age group.

5.0 UNCERTAINTIES AND SENSITIVITY ANALYSIS

Both EPA and MDEQ have published documentation regarding the uncertainties associated with toxicity
factors, the default exposure factors and parameters, and the algorithms used to calculate PRGs. All of
those uncertainties apply to the site-specific PRGs calculated herein. This discussion will focus on
uncertainties and some sensitivity analyses around the three most significant site-specific exposure
factors and parameters: Exposure Frequency; Dust Concentration; and Oral Bioavailability. Additionally,
although the factor is not as significant, there is also a discussion of the uncertainties associated with
partitioning dermal exposure between outdoor soil and indoor dust on outdoor exposure days.

5.1 Exposure Frequency

Recent local climate data supports the site-specific value for outdoor exposure days of 260 days per year
(Section 2.4.1.1). EPA and MDEQ believe that this is the best available information to use at this time.
However, Executive Order 13653 of November 1, 2013, among other things, directs Federal Agencies to
integrate consideration of climate change in managing lands and waters (FedCenter 2013). The Order

-------
calls for "adaptive learning, in which experiences serve as opportunities to inform and adjust future
actions." The Superfund program is consistent with the concept of adaptive learning. As discussed
above, this Floodplain response action is a NTCRA. In the future, a remedial decision(s) will be made for
the Floodplain. Climate change will be considered, as needed, as a component of the remedial decision-
making. Superfund also requires a Five-Year Review, during which the continued protectiveness of
remedies is evaluated. EPA anticipates that if there is significant climate change in the Floodplain that
calls into question the site-specific exposure frequency, it can be evaluated in the Five-Year Review.

Of more immediate concern is the apportionment of outdoor exposure days between the house
perimeter, maintained area in the Floodplain, and unmaintained area in the Floodplain (Section 2.4.1.2).
The Floodplain is characterized by residential properties which are large and complex with opportunities
for multiple use patterns and occupancy rates. There are currently no default exposure frequency
factors which can be assigned to a multi-use property. Therefore, professional judgment needs to be
applied, considering site-specific circumstances.

For a number of reasons, EPA and MDEQ believe that the partitioning used to calculate the Other Land
Use PRG is a conservative approach. First, there is another zone where residents may contact soil - the
non-house perimeter (e.g., unmaintained areas) outside of but adjacent to the Floodplain, which have
soil TEQ concentrations that are significantly below the 250 ppt TEQ PRG (Zone A-2 on Figure 3).
However, to provide a conservative evaluation, this area was excluded from the apportionment of
outdoor exposure days. Second, there are periods when the Floodplain is inaccessible due to flooding.
The Agencies did not try to reduce the frequency of outdoor day exposure because of "flood days."
Rather, EPA and MDEQ believe that this adds another degree of conservatism to the calculations. Third,
in the PRG calculations time is split evenly between the maintained residential areas in and out of the
Floodplain. An argument could be made that in many cases more time is spent around the house
perimeter, outside of the Floodplain. During implementation of interim response action exposure
controls at Floodplain properties, residents were interviewed about their use of the Floodplain.
Generally, the reported frequency of use is less than (and sometimes much less than) the exposure
frequency included in the PRG calculations. Although this information is informative, the interviews
were not conducted as a formal survey, so the results are considered to be an anecdotal line of
evidence.

As discussed in Section 4, potential non-cancer health effects were the driver behind the PRGs. Unlike
carcinogenic compounds where EPA has established an acceptable risk range, for non-carcinogenic
chemicals the HQ does not reflect a range (i.e., HQ = 1). However, it is reasonable to consider the HQ
within the framework of uncertainties related to the RfD. In the discussion of uncertainty included in
EPA's definition of the RfD, EPA defines the RfD as:

"...an estimate (with uncertainty spanning perhaps an order of magnitude) of daily exposure to
the human population (including sensitive subgroups) that is likely to be without an appreciable
risk of deleterious effects during a lifetime..." (EPA 2012c)

Because there is this range of uncertainty around the RfD, and other uncertainties, the Agencies have
conducted a quantitative sensitivity analysis around the apportionment of outdoor exposure days
between the house perimeter, maintained area in the Floodplain, and unmaintained area in the
Floodplain (Zones A-l, B, and C on Figure 3). Standard algorithms that calculate the HQ were applied,
the PRG values presented in Section 4 were held constant, and the exposure frequency in different
property zones was varied. The results are shown in Table 8. As discussed in Section 2.4.1.2 above, the

-------
young child is considered to spend most of the time around the house and in the maintained area within
the 8-year floodplain. Although this sensitivity analysis calculates HQs for up to 10% of the young child's
time spent in the unmaintained portions of the Floodplain, EPA and MDEQ do not believe that this is
likely at this particular site - the results are simply to present risk management information. From a risk
management perspective, EPA and MDEQ believe that this sensitivity analysis supports the
reasonableness of the site-specific exposure frequencies used to calculate the PRGs.

Young Child - Unmaintained Exposure
Frequency (% of soil exposure time)
Zone C

Proportion of Exposure Frequency to
Maintained Areas In vs. Out of
Floodplain (Zone B:Zone A-l)

Calculated HQ