Collaborative Research Program to

Modernize the Process and Bring

Innovative Science to Evaluate
New Chemicals Under TSCA
Katie Paul Friedman, PhD
Paul-friedman.katie@epa.gov
Presented at the U.S. EPA Collaborative Research Program to Support New Chemical Review Public
Meeting

April 20, 2022

 Mapping the proposed research areas to the
process
2 Machine 1) Update and refine chemical categories
readable
data 2) Develop and expand databases containing TSCA chemical information
3) Develop and refine QSAR and predictive models for physical-chemical
Curation properties, environmental fate/transport, hazard, exposure, and
toxicokinetics
Lit 4) Explore ways to integrate and apply NAMs in New Chemical
mining 1 Chemical Assessments
categories
5) Develop a TSCA New Chemicals Decision Support Tool to modernize
and analogs
the process

2 5
3
Many types of Collate
Accessible QSARs Display Decide Report
Data Info

4 In vitro
assays and
models
 Research Challenge Approach Expected Outcome​(s)
Area​
1 Update and Currently 56 Systematically define chemical This will increase the
Refine TSCA categories and analogues for read- efficiency of new chemical
Chemical categories, across using structural (and other) reviews and promote the
Categories ​ last updated boundaries; physical-chemical use of the best available
2010 properties; structural alerts for data to protect human
hazard, fate, exposure, and/or health and the
functional uses; existing hazard environment.
data; and/or, in vitro mechanistic
and toxicokinetic data from NAMs

3
 Building on existing guidance and tools, we will seek systematic
and quantitative approaches to categories and analogs

• Existing guidance provides foundations for both
categories and QSARs, e.g.:
• Guidance on Grouping of Chemicals, second
edition Series on Testing and Assessment No.
194, 2014
National PFAS Testing Strategy:
Identification of Candidate Per- and
Poly-fluoroalkyl Substances (PFAS)
for Testing (October 2021)

• Chemical categories can be defined to both maximize

read-across and capture structural diversity
• Use of structure descriptors, bioactivity data,
and/or apical toxicity outcomes to define
similarity of local chemical neighborhoods
• Examination of data-poorness and data-richness
by structure

https://www.epa.gov/system/files/documents/2021-10/pfas-natl-test-strategy.pdf
4
 Public tools for systematic read-across have been evolving and
may be useful
• Work by Patlewicz, Shah, and colleagues for objective read-across (now GenRA v3)
• Quantitative evaluation of similarity and confidence in predictions
• CompTox Chemicals Dashboard and Python package (genra-py)
• Interactive workflow to:
o search for target or draw it;
www.comptox.epa.gov/dashboard

o define fingerprints for similarity and number of analogs;

o Hybrid descriptors now available;
o Examine what data exist for source analogs;
o Inspect the consistency, concordance, and range of effects for analogs
o Understand confidence in the prediction(s)

5
 Research Challenge Approach Expected Outcome​(s)
Area​
2 Develop and Existing TSCA Extract and curate available The TSCA CBI information will
Expand information is not TSCA CBI study information be combined with publicly
Databases computationally available sources to expand
Containing accessible or Continue extraction and the amount of information
TSCA Chemical easily searchable curation of physical-chemical available, enhancing chemical
Information​ property, environmental fate, reviews and enabling efficient
hazard, and exposure sharing of chemical
information (non-CBI) in ORD information across EPA.
databases Safeguards for CBI will be
maintained as appropriate in
Map information in ORD this process.
databases to standardized
reporting templates and store
in an International Uniform
Chemical Information
Database (IUCLID)
6
 Curating existing data into computationally accessible resources
is ongoing in ORD and prepares them for operational use

Physical-chemical properties, environmental fate,

multi-media monitoring, functional uses

ChemProp ChemExpo
MMDB
DB

Human health and ecological hazard

Thomas et al. 2019 10.1093/toxsci/kfz058

ToxVal ToxRef ECOTOX

Curating data into computationally

accessible formats supports efforts to
establish confidence in NAMs, Toxicokinetics
characterize uncertainty and variability, Mapping existing databases to IUCLID
and develop software and tools to inform formats enables international
chemical safety. CvT HTTK collaboration and data sharing.

7
 Research Challenge Approach Expected Outcome​(s)
Area​
3 Develop and Currently used models Develop and update Updated models that
Refine QSAR are not always publicly QSAR and predictive reflect the best available
and Predictive accessible, easy to models using science, increase
Models for update with additional existing data and transparency, and a
Physical- chemicals, or the best curated data from process for updating
Chemical performing for all Research Area #2 these models as science
Properties, chemistries allows.
Environmental Evaluate models to
Fate/Transport, determine the best
Hazard, suite for use by
Exposure, and OPPT for regulatory
Toxicokinetics purposes

8
 Existing use of QSAR and structure alerts can be enhanced by
ongoing work in ORD to publish QSARs for real-time prediction
Existing QSAR and structure
profiling strategies, such as:
• EPI Suite
• ECOSAR
• OncoLogic
• OECD QSAR Toolbox
structure-based profilers

Existing OECD guidance, such

as: https://comptox.epa.gov/
Guidance document on the dashboard/predictions
validation of quantitative
structure-activity relationships • Work by Martin, et al (TEST models and software, User guide, 2020:
models (2007)
https://www.epa.gov/sites/default/files/2016-05/documents/600r16058.pdf);
• Moving now to machine learning methods and their consensus
predictions
• Make models and their performance reports publicly available,
including evaluation of applicability domain
9
 Research Challenge Approach Expected
Area​ Outcome​(s)
4 Explore Ways Reduction in the use Develop and A suite of NAMs that
to Integrate of vertebrate animals could be used by
evaluate a suite of
and Apply in accordance with in vitro NAMs for external stakeholders
NAMs in New TSCA Section 4(h)​ informing new for testing and data
Chemical chemical submissions under
Assessments Many PMN evaluations TSCA as well as
submissions are data informing and
poor Use mechanistic expanding new
and toxicokinetic in chemical categories
Amended TSCA vitro NAMs to
requires affirmative inform and refine
determination chemical
regarding categories in
unreasonable risk Research Area #1
10
 NAMs may address key data gaps, future predictive
models, and/or chemical categories: Tier 1
Thomas et al. 2019 10.1093/toxsci/kfz058 Nyffeler et al. 2020 10.1016/j.taap.2019.114876

Cell Painting
Creating fingerprints
A Tier 1 NAM, high-throughput phenotypic profiling (aka, Cell Painting), measures a large number of cellular morphological features that
can be used to inform point-of-departure, molecular initiating events, and bioactivity fingerprints.
11
 NAMs may address key data gaps, future predictive
models, and/or chemical categories: Tier 1-2
Thomas et al. 2019 10.1093/toxsci/kfz058 Houck et al. 2021 10.1016/j.tox.2021.152789 Ietswaart et al. 2020 10.1016/j.ebiom.2020.102837;

We can learn from use

of targeted assays for a
safety pharmacology
approach in use by
cosmetics and
pharmaceuticals
companies

Tier 1-2 screening technologies can fill gaps related to molecular initiating events, such as nuclear receptor targets, genotoxicity, and/or
oxidative stress.

12
 NAMs may address key data gaps, future predictive
models, and/or chemical categories: Tier 2-3
Thomas et al. 2019 10.1093/toxsci/kfz058 Faber et al. 2020 10.1093/toxsci/kfaa085 Gamble, Deisenroth, et al. in review

Tier 2-3 screening technologies can fill gaps related to different routes of administration (e.g., inhalation) and/or evaluation of
specific toxicities such as developmental toxicity.
13
 Research Challenge Approach Expected Outcome​(s)
Area​
5 Develop a Searching, Build proof of A decision support tool that will
TSCA New collating, and concept software efficiently integrate all the data
Chemicals integrating data workflow that streams (e.g., chemistry, fate,
Decision for new chemical integrates all data exposures, hazards) into a final
Support assessments is streams in a new risk assessment and
Tool​ ​to inefficient and chemical risk transparently document the
Modernize costly decision context
decisions and assumptions
the Process
made. This will facilitate the new
chemicals program tracking
decisions over time and
evaluating consistency within
and across chemistries.

14
 Examples of next generation risk assessment
workflows that incorporate NAMs
Beal et al. 2020, “Implementing in vitro bioactivity
Baltazar et al. 2020, “A Next Generation Risk Assessment Case Study for Coumarin in
data to modernize priority setting of chemical
Cosmetic Products.” 10.1093/toxsci/kfaa048
inventories.” 10.14573/altex.2106171

• OPPT New Chemicals Division has a specific process and decision context
• Customized, modular decision support tool will be needed
• Blend existing data and models with NAMs
• Software that allows user to interact with data, make selections, and
record these selections
 Components of a proof-of-concept decision support tool may include
modules to display, select, and download key information for
assessment
Start with structure, SMILES, CAS RN, DSSTox ID Hazard Data Landscape
Physicochemical QSAR hazard
Structure, properties predictions
Identify
substructure, and
Metabolites and key
similarity

and analogs
degradants data (or

Category
In vivo hazard data
gaps)
and
Reports and
In vitro data decision
PODs
tracking

Use and Exposure Landscape Integrated Toxicity Views (e.g., developmental and
reproductive toxicity; carcinogenicity)
Exposure and Fate
Data and Predictions Identify key
exposure Structure alerts Applicable in vitro
Functional Use Data values data
QSARs for specific
and Predictions In vivo data
toxicity types
Dive deeper where possible on specific effects
16
 Beginning the work
• Research will be initiated in FY23.
• Research during the 2023-2026 timeline will provide a foundation for
continued improvement to meet the needs of new chemicals evaluation.
• The research is expected to complement the EPA NAMs Work Plan, in
which we seek to establish scientific confidence and demonstrate
application of NAMs as well as develop NAMs that fill critical information
gaps.
• Partnerships:
• OPPT, ORD, and NIH (DNTP/NIEHS, NICEATM, and NCATS)
• Other regulatory partners, such as ECHA

17
 EPA Core Planning Team
OCSPP ORD OGC
Jeff Dawson Annette Guiseppi-Elie Sonja Rodman
Anna Lowit Russell (Rusty) Thomas Don Sadowsky
Madison Le Katie Paul Friedman Scott Albright
Cathy Fehrenbacher Samantha Jones Victoria Clarke
Louis (Gino) Scarano Beth Owens Chris Kaczmorek
Keith Salazar Alison Harrill Devi Chandrasekaran
Tala Henry
Meghan Tierney
Stan Barone
Kellie Fay
Martin Phillips
Keith Avery

18
 EPA ORD scientists
Kristin Isaacs Shaun McCullough
Caroline Ring Mark Higuchi
Katherine Phillips Dan Villeneuve
John Wambaugh Chad Deisenroth
Antony Williams Joshua Harrill
Grace Patlewicz Richard Judson
Todd Martin Sean Watford
Dale Hoff Jill Franzosa
Brett Blackwell Tim Buckley
Richard Kolanczyk Sid Hunter
Risa Sayre Peter Egeghy
Norm Adkins Michael DeVito
Madison Feshuk Jonathan Wall
19