"Strawman" Priority Setting Process for Endocrine Disruptor Screening and Testing

“Strawman” Priority Setting Process for Endocrine Disruptor
Screening and Testing
January 20, 1999
Background
The Endocrine Disruptor Screening and Testing Program has five major components:
1.
Sorting, in which chemicals are classified according to the availability of
information on each chemical's endocrine-disrupting potential.
2.
Priority setting, in which EPA will determine the priority order for entry into Tier
1 screening (T1S).
3.
Tier 1 screening, a battery of in vitro and in vivo assays designed to identify those
chemicals that are not likely to interact with the estrogen, androgen, or thyroid hormone
systems (EAT).
4.
Tier 2 testing, a battery of assays designed to determine whether a chemical may
have an effect in humans, fish or wildlife similar to that of naturally occurring hormones
and to identify, characterize, and quantify those effects for EAT effects.
5.
Hazard assessment, a weight-of-evidence evaluation of Tier 1 and Tier 2 results.
It is expected that the sorting will result in a relatively small number of chemicals
proceeding directly to Tier 2 testing or hazard assessment and that the vast majority of chemicals
will be placed in priority setting for Tier 1 screening.
EPA proposes that all chemicals would initially be subject to priority setting for T1S
except for chemicals with sufficient data to go directly to tier 2 testing or hazard assessment or
for the polymers and some exempted chemicals as described in the Endocrine Disruptor
Screening Program Proposed Statement of Policy published in the federal register on December
28, 1998.
Prioritizing Chemicals for Endocrine Disruptor Tier 1 Screening
The Endocrine Disruptor Screening and Testing Advisory Committee (EDSTAC) stated
that the priority setting system for T1S should be transparent, reflect the guiding principles
below for weighting data, and rely heavily, but not exclusively, on empirical data to set
priorities.
The guiding principles are that more weight should be given to biological data relevant to
large populations, disproportionately exposed populations, or susceptible populations. More
weight should be given to chemicals likely to be internalized by an organism and more weight
should be given to releases that are likely to lead to exposure. The EDSTAC also stated their

---

belief that production volume should not be used to rank industrial chemicals relative to
pesticides.
The EDSTAC felt that empirical data showing chemicals found in organisms should be
weighted heavily, that empirical evidence of effects should be weighted more heavily than
models, and that existing empirical data could be used to improve the predictive capability of
models.
The EDSTAC recommended a “compartment-based priority setting strategy” that builds
upon distinct exposure- and effects-related information categories and criteria as well as a
category of specially targeted priorities. The reason this approach was recommended over
approaches that strive to develop a single rank-ordered priority list that integrates all exposure-
and effects-related categories and criteria was the EDSTAC felt that the available data was very
different from one chemical to another. Using a compartment-based approach, chemicals with
similar data are compared against each other. The compartment based approach also helps to
avoid being totally driven by the “looking under the lamp post” phenomenon.
The EDSTAC listed the following categories and subcategories of information that
should be considered in developing the compartment-based approach.
A. Exposure-Related Information
1. Biological sampling data
2. Environmental, occupational, consumer product, and food-related data
(sampling and/or use data)
3. Environmental releases
4. Production volume
5. Fate and transport data and models
B. Effects-Related Information
1. Toxicological laboratory studies and databases
2. Epidemiologic and field studies and databases
3. Predictive biological activity or effects models (e.g. SAR, QSAR)
4. Results of high throughput pre-screening
C. Integrated Effects and Exposure Information
D. Specially Targeted Priorities
1. Mixtures
2. Naturally Occurring Non-Steroidal Estrogens
3. Nominations

---

The EDSTAC did not reach agreement on the definition or weighting of specific compartments.
Following the basic framework and guiding principles laid out in the EDSTAC report, EPA has
developed an initial “strawman” proposal for a compartment-based system.
Proposed Approach
In the proposed approach EPA uses the four categories proposed by the EDSTAC (i.e.
specially targeted priorities, exposure-related information, effects-related information, and an
integrated effects and exposure information category). Compartments will be defined for each of
these categories. It is assumed that a database will be developed to contain the rank-ordered lists
of chemicals within each compartment as well as the weights and algorithm to select chemicals
for entry into tier 1 screening. It is also assumed that selection of chemicals for T1S will occur in
phases.
EPA will need to determine how many chemicals to select in total for the first phase of
the screening and testing process (at this time that determination has not yet been made). The
EDSTAC felt that the total number of chemicals selected could have ramifications for the
weights and/or for the number of compartments. For example they stated in their report that if
the total number selected is relatively small, this could lead to having fewer compartments since
relatively few chemicals would be selected from each compartment.
In developing the strawman proposal, EPA adopted the following working definition of a
compartment:
All chemicals within a compartment share the feature(s) that define the compartment (e.g.
chemicals with TRI release data). The defining feature(s) of the compartment should, if
possible, be suitable for sorting chemicals within the compartment into a rank-ordered list.
The table below summarizes the “strawman” proposal. The table has the four categories
and multiple compartments under each category (except for the integrated effects and exposure
category). In order to select chemicals for screening, the following steps are required:
1. Define the compartments
2. Rank order chemicals within each compartment
3. Assign percentage weights to each compartment with all of the weights summing to
100%
4. Based upon the number of chemicals to be screened in phase I, use the weights and
rank-ordered lists of chemicals to select chemicals from each compartment.
5. As needed, review the chemicals selected from each compartment to ensure that
obviously incorrect selections are not made (e.g. a chemical ranked high for persistence
by the proposed modeling approach that in reality undergoes rapid hydrolysis).
6. Select and propose the list of phase I chemicals for entry
into T1S screening. (Note: The proposal of this list does not constitute the making of
“findings” for rule making purposes.)

---

As stated above, in the proposed approach there are four categories of chemicals and
multiple compartments within each category.
Specially Targeted Priorities
The first category, specially targeted priorities, is different from the others in that some
very specific recommendations were made by EDSTAC about chemicals to be screened or about
specific chemicals or mixtures which should be screened. There are three compartments in this
category.
Exposure-Related Information
The second category is the exposure-related information category. There are twelve
compartments in the exposure category. The compartments in this category contain information
relevant to predicting the likelihood of exposure to humans or to wildlife. The compartment
containing chemicals found in biological monitoring data may predict exposure with a relatively
high degree of reliability. Exposure is also likely to chemicals in food and drinking water,
chemicals in consumer/cosmetic products, and chemicals for which occupational exposure
standards have been developed. Exposure is also likely but perhaps less so for chemicals found
in environmental monitoring data. Exposure related information such as persistence in the
environment, bioaccumulation potential, release to the environment, and production volume give
an indirect indication of the likelihood of exposure.
Possible approaches to rank ordering lists of chemicals within compartments include
looking at the frequency of occurrence of chemicals in biological and environmental monitoring
studies, frequency of occurrence within drinking water and food monitoring studies, and
frequency of occurrence within consumer and cosmetic products. Alternatively, it may be
possible to use frequency of occurrence in combination with some sort of normalized
concentration or intake estimates to rank-order chemicals within some of these compartments.
This approach would allow the use of additional information (i.e. concentration data) in
establishing the rank-ordered lists. A compartment consisting of chemicals with occupational
exposure standards may be rank-ordered by the numerical value of those exposure standards.
Chemicals in compartments based on persistence in the environment may be rank-ordered by
half-life which may be predicted based upon measured or predicted physical-chemical properties
and fate along with the use of multimedia models. Likewise, bioaccumulation potential may be
estimated based upon estimated or measured physical-chemical properties and models.
Chemicals in compartments based on release to the environment and production volume may be
rank-ordered respectively by mass released and mass produced.
Effects-Related Information
The third category is the effects-related information category. There are seven
compartments in the effects category. The compartments in this category contain information
relevant to predicting the likelihood that the chemical can cause effects due to endocrine
disruption. When they are available, epidemiological and clinical data on chemicals showing
endocrine target organ effects in humans are relatively strong indicators of potential endocrine

---

disruptor effects in people. Studies using laboratory animals on reproductive or developmental
toxicity, carcinogenic effects in endocrine target tissues, and subchronic studies on effects in
endocrine target tissues may also be good indicators of a chemical’s potential to be an endocrine
disruptor in humans. In vitro studies such as the HTPS assays, and QSARs may also provide
useful information for screening chemicals for their endocrine disruption potential in humans.
For fish and aquatic invertebrates as well as mammalian and avian wildlife, studies showing
reproductive, growth and behavior effects provide useful information for screening a chemical’s
endocrine disruptor effects in these species. When data is not available, QSARs may in some
cases be useful here too.
Possible approaches to rank-ordering chemicals within the effects compartments include
ranking on a numerical scale where the data allow it (e.g. NOAEL and LOAEL data, HTPS data,
QSARs). In some cases all the data may allow is a positive or negative indication, in which rank
ordering would not be possible.
Integrated Effects and Exposure Information
In this “strawman”, we have not proposed any approaches for combining exposure and
effects into compartments in the integrated exposure and effects category. We are interested in
ideas on how we might do this. We are also interested in ideas about how we might deal with
chemicals that appear in multiple compartments (e.g. include a multi-hit compartment or, use
multiple hits as part of the rank ordering for those compartments that have a lot of ties?)
Hypothetical Example
As a hypothetical example of a compartment-based priority setting approach, suppose a
priority setting system consisted of three compartments, each of which contained chemicals in a
rank-ordered list. Furthermore, assume that the compartment #1 contained 100 chemicals,
compartment #2 contained 1000 chemicals, and compartment #3 contained 10,000 chemicals.
Assume that each compartment is given a equal weight of 33.3%. If a total of 99 chemicals were
to be selected, this means that 33 chemicals will be selected from each of the three
compartments. So, for compartment #1, the 33 highest ranked chemicals out of the 100
chemicals in compartment#1 will be selected. For compartment #2, the 33 highest ranked
chemicals out of the 1000 in compartment#2 will be selected. And finally for compartment #3,
the 33 highest ranked chemicals out of the 10,000 chemicals in compartment#3 will be selected.

---

Proposed Compartments for EDSP Priority Setting
Specially Targeted
Exposure &
Priorities
Exposure
Effects
Effects
Nominations
Human Biological
Epidemiological and clinical
Monitoring Data
data on endocrine target
organ effects.
EDSTAC
Ecological Biological
Reproductive/developmental
Recommended
Monitoring Data
toxicity - NOAELs/LOAELs
Mixtures
from studies in laboratory
animals.
Chemicals in food and
Carcinogenicity -
EDSTAC
drinking water
positive/negative results in
Recommended
endocrine target tissues
NONES
Chemicals in consumer
Subchronic toxicity -
and cosmetic products
NOAELs/LOAELs for
endocrine targets
Occupational exposure
High Throughput Pre- Screen
chemicals
test results (degree of
receptor binding)
Monitoring data
Ecotoxicity - field and
Surface and ground
laboratory studies
water
Monitoring data Indoor
Quantitative Structure-
and outdoor air
Activity Relationships
(QSARs) for estrogen
receptor binding
Monitoring data
Sediments/soil
Persistence
Bioaccumulation
potential
Environmental releases
Production and import
volumes
Below are definitions of the specially targeted priority compartments and a series of one
page descriptions of the exposure and effects compartments. The one page descriptions include a

---

definition of the compartment, a listing of relevant data bases, an estimate of the number of
chemicals in each compartment (where it was possible to estimate at this stage of the project)
and possible options for rank-ordering chemicals within each compartment.
Definitions of the Specially Targeted Priority Compartments
Nominations: Chemical substances or mixtures nominated by citizens who are disproportionately
exposed because of the group or community to which they belong, or because an ecosystem is
disproportionately exposed.
EDSTAC Recommended Mixtures: The EDSTAC recommended that representative samples of
six types of mixtures be screened and tested. The six types of mixtures are contaminants in
human breast milk, phytoestrogens in soy-based infant formulas, mixtures of chemical
substances most commonly found at hazardous waste sites, pesticide fertilizer mixtures found in
surface water and groundwater, drinking water disinfection byproducts, and gasoline.
EDSTAC Recommended Naturally Occurring Non-Steroidal Estrogens (NONEs): The EDSTAC
recommended that representative NONEs be screened and tested. These representative NONEs
are:
1) Isoflavones: genistein, daidzein, miroestrol, bichanin A, formononetin, equol;
2) Flavanones: kaemferol, naringenin;
3) Coumestans: coumesterol;
4) Dihydrochalcones: phoretin;
5) Triterpenes: betulafolienetriol (ginseng);
6) Lignans: enterolactone; and
7) Beta-resorcyclic lactones: zearalenone, zearalenol, zearanol
Definitions of Exposure Compartments
Name of compartment
Human Biological Monitoring Data
Definition of compartment
The chemicals in this compartment have all been found in humans in monitoring studies
Relevant databases and tools
The Third National Health and Nutrition Examination Survey (NHANES III) conducted by the
Center for Health Statistics/Centers for Disease Control and Prevention.
The National Human Adipose Tissue Survey conducted by the National Human Monitoring
Program of the Office of Pollution Prevention and Toxics/ United States Environmental
Protection Agency.

---

The National Human Exposure Assessment Survey (NHEXAS), conducted by EPA’s Office or
Research and Development monitored chemicals in blood and urine (this data is currently being
evaluated).
Approximate number of chemicals in the compartment
NHANES III measured 39 “priority toxicants” which included VOCs, pesticides and metabolites
in blood.
NHATS targeted 111 semivolatile organic chemicals (including some pesticides) in the 1986
chemical analysis of human adipose tissue.
NHEXAS had 56 target analytes that included pesticides, VOCs and metals.
Potential options for ranking within compartment
Rank chemicals by frequency of occurrence in human biological samples.
Alternatively, rank chemicals by body burden assuming a “standard human body” and using
measured concentration data from adipose tissue and blood.

---

Name of Compartment
Ecological Biological Monitoring Data
Definition of compartment
Ecological biological monitoring involves the sampling of a wide range of species and sample
matrices for exposure to environmental contaminants
Relevant databases and tools
The new Storage and Retrieval System(STORET) combines information from the STORET
Water Quality System(WQS) with that of the Biological Information System(BIOS) and the
Ocean Data Evaluation System(ODES).
Environmental Contaminant Data Management System which contains compilations of
analytical data of 100,000 samples of invertebrates, fish and wildlife.
Great Lakes Fish Monitoring Program contains a compilation of data from a 30-year
biomonitoring effort conducted by the Great Lakes Science Center of the U.S. Geological
Survey.
TSCATS 2.0 database which tracks the existence (but not the values) of environmental
monitoring data from TSCA submissions. This database, in the process of development, contains
data that used to be available in the OPPT CECATS database.
Approximate number of chemicals in the compartment
TSCATS 2.0 has about 8000 chemicals.
STORET houses approximately 25 million sample records and about 150 million analysis
records. The number of chemicals in STORET is unknown at this time. The utility of STORET
to this prioritization effort may be limited due to uncertainty about the quality of the data and the
difficulty of extracting it in a suitable form.
The Environmental Contaminant Data Management System contains about 625 compounds.
Great Lakes Fish Monitoring Program contains over 550 compounds. Total number of
compounds for this compartment, estimated assuming the numbers supplied above are close to
actual numbers, could be anywhere from 700- 1100 on up.
Potential Options for Ranking Within Compartment
Rank by frequency of appearance in biological tissues.

---

Name of compartment
Chemicals in food and drinking water
Definition of compartment
Direct and indirect food additives, contaminants found in food and drinking water
Relevant databases and tools
FDA lists of direct and indirect food additives; food inspection data from FDA and USDA;
drinking water monitoring data from EPA (Safe Drinking Water Information System (SDWIS)).
SDWIS has data on frequency of occurrence, concentrations found, and size of the population
served by the tested water systems.
Approximate number of chemicals in the compartment
FDA has approved over 3000 direct food additives and over 2500 indirect food additives. EPA
has drinking water monitoring data on about 100 chemicals. Total number of chemicals in
compartment (taking into account overlaps) is probably about 5000-6000.
Potential options for ranking within compartment
1. Rank by frequency of use/occurrence in food or water.
2. Rank by concentrations found in food or water.
3. Rank by number of persons exposed.

---