science in ACTION
BUILDING A SCIENTIFIC FOUNDATION FOR SOUND ENVIRONMENTAL DECISIONS
Prioritizing Endocrine-Disruptor Screening Using ToxPi
Visually translating the integration of ToxCast data
March 1. 2010
Impact Statement
This research paper presents ToxPi (Toxicological
Priority Index), a new weight-of-evidence framework for
profiling and prioritizing chemicals. It numerically
integrates various knowledge sources about chemical
specific properties (biological and chemical). This
numeric summation is displayed graphically in order to
help further understand the chemical. ToxCast is a multi-
year, multi-million dollar effort that uses advanced
science tools to help efficiently ($20K per chemical)
understand biological processes impacted by chemicals
that may lead to adverse health effects. ToxCast
currently includes 467 in vitro, high-throughput screening
assays which have assessed over 300 environmental
chemicals. This study applied ToxPi to ToxCast's 309
chemicals, which includes 52 of the chemicals currently
undergoing screening in Tier 1 of the EDSP. In the
future, ToxPi may be used to prioritize chemicals for
screening in the EDSP Tier 1 battery.
Understanding which chemicals can cause endocrine
disruption (ability of a chemical to interact with the body's
hormonal systems and cause adverse health outcomes
such as birth defects and cancer) is a high US EPA
priority. EPA has developed a group of tests (the EDSP
Tier I battery) capable of describing endocrine disruptor
potential and is in the process of acquiring data on
chemicals using these tests. EDSP assays are relatively
expensive (~$0,5M per chemical) and tools are needed
to help select which chemicals should be first in line for
further testing. This study applied ToxPi to multiple
potential endocrine disrupting measures and then ranked
chemicals for additional screening. The knowledge
sources used in this study included in vitro assays,
chemical properties and pathways. For this study, 90
ToxCast assays relevant to endocrine activity were
selected, and these assay results were then used to
ToxPi=f(In vitro assays +
XME/ADME
[36]
+Pathways)
Other NR
-ogP
'redieted
Caco-2
pes
Ingenuity
Pathways
KEGG
Pathways [12]
Figure 1 EDSP ToxPi Framework
Metabolizing Enzyme/ Absorption-Distribution-
Metabolism-Excretion (XME/ADME, 36 assays), and
Other Nuclear Receptor (NR, 38 assays).The three blue
slices represent 27 pathway effects indicated from the
assay results, and the orange slices represent the two
chemical properties used.
Figure 2 shows individual ToxPi scores for two
chemicals, Bisphenol A (BPA) and Tebuthiuron. BPA is
commonly used in plastics and Tebuthiuron is a
herbicide. The slice representing ER for BPA extends
farther from the circle's center when compared to the
Tebuthiuron indicating that BPA is more potent across
ER assays when compared to Tebuthiuron. BPA also
ranks above Tebuthiuron in all other ToxPi slices.
Fiqure2 ToxPi Profiles for Two Reference Chemicals
Study Description
ToxPi is shown visually as a circle with component slices
making up the circle. Figure 1 visually represents ToxPi
for endocrine activity and the circle's slices are color-
coded with shades of three different colors to show the
types of in vitro assays, chemical properties and
pathways used in this study. The 90 selected ToxCast
assays are represented by five green slices: Androgen
Receptor (AR, 5 assays), Estrogen Receptor (ER, 6
assays), Thyroid Receptor (TR, 5 assays), Xenobiotic
Figure 3 shows ToxPi results for all 309 ToxCast
chemicals, including EDSP chemicals that are part of
ToxCast. The blue boxes represent ToxCast chemicals
that are not on the EDSP tier 1 list and the red boxes are
1

-------
EDSP chemicals. ToxPi scores are ori the horizontal axis
(score from lowest to highest) and the individual
chemicals run along the vertical axis. EDSP chemicals
are represented throughout the ToxPi distribution. EDSP
chemical distribution is not surprising since these were
selected solely upon exposure considerations. In
comparison, ToxPi takes into account biological
information and provides a method for efficiently
prioritizing chemicals for further testing.
Figure 3 EDSP Chemicals of Interest Distribution
EDSP Chemical
i ToxCast Chemical (not on EDSP list)
309 Chemicals Sorted By ToxPi
• highest

Until data from EDSP screening is available, a good
ToxPi validity evaluation for endocrine disruption is to
select "reference" chemicals. The "reference" chemicals
have substantial research on their toxicological activities.
Therefore, there is an expectation about where these
chemicals should fall along the ToxPi distribution. Figure
4 shows ToxPi for reference chemicals along the ToxPi
distribution for all ToxCast chemicals. For assessing
ToxPi, reference chemical scores should provide
rankings that do not put known hazards near the bottom
of the priority list. As Figure 4 shows, potent reference
chemicals such as BPA, methoxychlor and its metabolite
HPTE are at the top of the ToxPi distribution as
expected.
Figure 4 Guidepost (spike-in) Chemicals
In the future, ToxPi predictions will be compared to the
EDSP Tier 1 screening results to test the validity of the
approach.
Conclusions
Combining multiple ToxCast data sources into an overall,
weight-of-evidence ToxPi resulted in a robust
prioritization method. Benefits of ToxPi include its:
•	Efficiency: ToxPi's advantage is that knowledge
sources used to prioritize chemicals comes from
more than one source which increases the likelihood
of detecting true endocrine active chemicals. This
improves efficiency by ensuring optimal use of costly
animal and human resources.
•	Flexibility: Because ToxPi is intended for relative
ranking, particular implementations of this framework
can be continually updated with new chemicals and
future data. Other knowledge sources such as in
vivo endpoints and exposure can be applied to this
framework to assess other types of possible
disruptions and chemicals
•	Collaborative Approach: ToxPi is amenable to
incorporating existing and emerging knowledge
sources and data from diverse sources, thereby
facilitating analysis across EPA resources and
offices.
•	Potential: Additional development of ToxPi based
upon forthcoming ToxCast results will be focused on
ranking chemicals for their potential to cause other
types of health effects such as reproductive toxicity,
birth defects, and cancer.
Background
The Food Quality and Protection Act (FQPA) and the
Safe Drinking Water Act (SDWA) requires EPA to
develop a screening program to assess the potential for
chemicals to interact with the endocrine system. As a
result, EPA's EDSP is responsible for screening
pesticide chemicals and environmental contaminants for
potential impacts to the endocrine systems of humans
and wildlife. The prioritization of chemicals for testing is a
goal shared by both EDSP and ORD.
References
Reif et al. (2010) "Endocrine Profiling and Prioritization of
Environmental Chemicals Using ToxCast Data"
Environmental Science and Technology
http://epa.gov/ncct/toxcast/
http: //www, e pa. g ov/en d o /
Originating Organization for Fact Sheet
EPA
Primary Author of Fact Sheet
Monica Linnenbrink
Secondary Authors of Fact Sheet
Robert Kavlock, David Dix and David Reif
First/Lead Author of Paper
David Reif
2

-------