vvEPA
United States
Environmental Protection
Agency
The Chemical Landscape of New Approach Methodologies
for Exposure
Kristin
Center for	Computational
Office of Research	and Dev
Environmental Protection Agency
APCRA Public Webinar
March

-------
SEPA
United States
Environmental Protection
Agency
Disclaimer
The views expressed in this presentation are those of the authors and do not necessarily reflect the views or
policies of the U.S. EPA
2 of 24
Office of Research and Development

-------
Exposure in the APCRA Initiative
United States
Environmental Protection
Agency
Identify available New Approach Methodologies (NAMs) for exposure-relevant
domains
Examine the landscape of exposure data (both traditional and NAMs) for an
inventory of chemicals relevant to APCRA partners
¦	Identify key information or activities that would enable or enhance fit-for-purpose
exposure estimates, predictions, or assessments and provide recommendations
¦	Provide exposure metrics to support the APCRA inventory and hazard-focused
case study activities
3 of 24
Office of Research and Development

-------
SEPA
United States
Environmental Protection
Agency
US EPA Office of Research and
Development Center for
Computational Toxicology and
Exposure
Kathie Dionisio
Annette Guiseppi-Elie
Kristin Isaacs
Katherine Phillips
Jon Sobus
Elin Ulrich
John Wambaugh
Barbara Wetmore
4 of 24
Office of Research and Development
Contributors
Health Canada
European Chemicals

Agency
Angelika Zidek
Andreas Ahrens

-------
SEPA
United States
Environmental Protection
Agency
5 of 24
Office of Research and Development
Risk is Multifaceted
Regulatory bodies are tasked with evaluating risks
associated with 1000s of chemicals in commerce
¦ For example, as of 2019 there were ~40,000
chemicals on EPA's TSCA Inventory
Evaluating chemicals for risk to humans or the
environment requires information on hazard and
exposure potential
Exposure potential quantifies the degree of contact
between a chemical and a receptor
, ¦ Toxicokinetic information is required to bridge hazard
\ and exposure (what real-world exposure is required to
produce an internal concentration consistent with a
potential hazard?)

-------
4>EPA
United States
Environmental Protection
Agency
EPA's ExpoCast Project
6 of 24
Office of Research and Development
Risk is Multifaceted
mg/kg BW/day
Potential
Hazard from in
vitro with
Reverse
Toxicokinetics
Potential
Exposure
Lower Medium Higher
Risk Risk Risk

-------
4>EPA
United States
Environmental Protection
Agency
Forecasting Exposure is a Systems Problem
Forward
Models

n
CHEMICAL


USE


and



I ^
Consumer
Products and
Durable Goods
Chemical Manufacturing
Evaluation
EXPOSURE
ENVIRONMENTAL
SURVEILLANCE
and
BIOMONITORING
TOXICOKINETICS
Critical Exposure-Relevant Domains
Chemical use and release. Provides critical
information for identifying chemical sources,
exposure pathways, and relevant predictive
models for a given chemical.
Exposure
Media
Outdoor Air, Soil, Surface
Ground Water
ECOLOGICAL
v
Human
Biomarkers
of Exposure
RECEPTORS n^\^vv
Sampling
Ecological
Flora and Fauna


Biomarkers
of Exposure
Media occurrence, environmental surveillance,
andbiomonitoring. Provides exposure data for
evaluating predictive models.
Exposure estimates. Predictions of chemical
intake in mg/kg/day that can be compared with
hazard information to inform risk.
Toxicokinetics. Provides real-world exposure
context to in vitro high-throughput screening
data and biological receptor monitoring
information.
7 of 24
Office of Research and Development

-------
4>EPA
United States
Environmental Protection
Agency
Eight Classes of NAMs for Exposure
Chemical descriptors that provide information on chemicals
in an exposure context (e.g., how chemicals are used)
Machine-learning approaches that use these descriptors to
fill gaps in existing data
High-throughput exposure models for various pathways
High-throughput measurements to fill gaps in monitoring
data
High-throughput approaches for measuring and predicting
chemical toxicokinetics
New evaluation frameworks for integrating models and
monitoring to provide consensus exposure predictions
All these pieces together provide the tools for high-
th roughput chemical prioritization
JjyL
ELSEVIER
Current Opinion in Toxicology
Volume 15, June 2019, Pages 76-92
New approach methodologies for exposure
science
John F. Wambaugh 1 A E3 Jane C. Bare 2, Courtney C. Carignan Kathie L. Dionisio 4 Robin E.
Dodson 5, OlivierJollfet 6, Xiaoyu Liu David E. Meyer 2, Seth R. Newton 4, Katherine A. Phillips 4,
Paul S. Price 4 Caroline L. Ring s, Hyeong-Moo Shin 9, Jon R. So'bus 4, Tamara Tal 10, El in M. UI rich
4, Daniel A. Vallero 4, Barbara A. Wetmore 4 Kristin K. Isaacs 4
0 Show more
https://doi.Org/10.1016/j.cotox.2019.07.001
Get rights and content
8 of 24
Office of Research and Development

-------
g pp/\
Characterizing the Chemical Landscape for
United States
Environmental Protection
Agency
Exposure NAMs
¦	"APCRA inventory" - case study chemical list
•	6621 chemical substances compiled by APCRA partners for potential use in
retrospective or prospective case studies
•	Compiled from regulatory lists from EPA, Health Canada, ECHA, EFSA, NICNAS
¦	Investigated the coverage of this inventory
¦	"Traditional" exposure data
•	Regulatory reporting
•	Targeted monitoring data
•	Regulatory exposure assessments
•	In-vivotoxicokinetic information
¦	Exposure NAMs across all four domains
Office of Research and Development

-------
4>EPA
United States
Environmental Protection
Agency
Traditional and NAM Exposure Datasets
NAM dataset
Voluntary
Regulatory or agency data
reporting of chemical use
vvEPA
New quantitative and qualitative
chemical use descriptors from EPA's
Chemicals and Products Database
(CPDat, Dionisio et al., 2018)
Machine learning models for
chemical function
(Phillips et al. 2017)
Chemical	Environments
r	Use	Surveillance
and	and
Release	Biomonitoring
Toxicokinetics Exposure J
Estimates /
IVIVE

10 of 24
Office of Research and Development

-------
4>EPA
United States
Environmental Protection
Agency
Traditional and NAM Exposure Datasets
NAM dataset
Voluntary
Regulatory or agency data
reporting of chemical use
vvEPAi
1USGS
science for a changing world
New quantitative and qualitative
chemical use descriptors from EPA's
Chemicals and Products Database
(CPDat, Dionisio et al., 2018)
Machine learning models for
chemical function
(Phillips et al. 2017)
Traditional (targeted) monitoring
data for various environmental
media from publicly available
monitoring databases
Non-Targeted analysis studies for various
environmental media from EPA and the EU
(Newton et al. 2018, Rager et al. 2016, Sjerps et al.
2016, Phillips et al. 2018)
Machine learning models
for media occurrence
Office of Research and Development
Toxicokinetics
X IVIVE
Exposure
Estimates

-------
4>EPA
United States
Environmental Protection
Agency
Traditional and NAM Exposure Datasets
NAM dataset
1USGS
science for a changing world
Traditional (targeted) monitoring
data for various environmental
media from publicly available
monitoring databases
Non-Targeted analysis studies for various
environmental media from EPA and the EU
(Newton et al. 2018, Rager et al. 2016, Sjerps et al,
2016, Phillips et al. 2018)
Machine learning models
for media occurrence
Cumulative
Estimated Daily
Intakes
Chemicals Management
Plan Environmental and
Consumer Assessments
Publicly Available Traditional
Assessments from Regulatory
Bodies
ena-

•to"™™*"*-'*
Consensus Modeling of Median Chemical Intake for the U.S.
Population Based on Predictions of Exposure Pathways
Caroline L Ring/'8'00 Jon A. Amot,"1" Deborah H. Bennett,1' L Peter P. Egeghy,' Peter Fantkc,®® 1
Lei Huang,* Kristin K. Isaacs, Olivier Jolliet,*'- Katherine A. Phillips,''" Paul S. Price,
Hyeong-Moo Shin,'1 - John N. Wcstgate,1 R. Woodrow Sctzer,' and John F. Wambaugh* '®

ere yj
IS A Danmarks
l| ij m l#l Tekniske
» ~ Universitet
High-Throughput Models for Various
Pathways and Consensus
Predictions from a Collaborative
Modeling Study (Ring et al., 2019)
12 of 24
Office of Research and Development

-------
4>EPA
United States
Environmental Protection
Agency
Traditional and NAM Exposure Datasets
NAM dataset
Voluntary
Regulatory or agency data
reporting of chemical use
vvEPAi
1USGS
science for a changing world
New quantitative and qualitative
chemical use descriptors from EPA's
Chemicals and Products Database
(CPDat, Dionisio et al., 2018)
Machine learning models for
chemical function
(Phillips et al. 2017)
In-silico machine learning models
for protein binding
and clearance (Sipes et al, 2017,
Ingle et al. 2018)
In-vitro protein binding
and clearance (Wetmore et al. 2015, Pearce et
al. 2017, Wambaugh et al 2019a.)
Traditional (targeted) monitoring
data for various environmental
media from publicly available
monitoring databases
Non-Targeted analysis studies for various
environmental media from EPA and the EU
(Newton et al. 2018, Rager et al. 2016, Sjerps et al.
2016, Phillips et al. 2018)
Machine learning models
for media occurrence
In-vivo toxicokinetic parameters collected from
the literature (Sayre et al., 2019)
Cumulative
Estimated Daily
Intakes
Chemicals Management
Plan Environmental and
Consumer Assessments
Publicly Available Traditional
Assessments from Regulatory
Bodies
High-Throughput Models for Various
Pathways and Consensus
Predictions from a Collaborative
Modeling Study (Ring et al., 2019)
/////
* -* ///
13 of 24
Office of Research and Development

-------
oEPA
United States
Environmental Protection
Agency
Traditional Release Reporting
Information
Traditional Use Reporting
Information
Machine-Learning QSUR Models for
Function
Chemical Use and Release
Chemical Use
Descriptors Developed Using
Informatics Approaches
APCRA Inventory
6621 Inventory Chemicals
iC NAM
Office of Research and Development
The number of chemicals for which release data are available is still
limited

-------
oEPA
Chemical Use and Release
United States
Environmental Protection
Agency
Traditional Release Reporting
Information
Traditional Use Reporting
Information
Machine-Learning QSUR Models for
Function
Chemical Use
Descriptors Developed Using
Informatics Approaches
APCRA Inventory
ML Models for function
allow for extrapolation to
data poor chemicals
iC NAM
Office of Research and Development
6621 Inventory Chemicals
The number of chemicals for which release data are available is still
limited

-------
4>EPA
United States
Environmental Protection
Agency
Media Occurrence, Environmental Surveillance,
and Biomonitoring
Traditional Targeted
Monitoring Data
Non-Targeted Studies in _J
Several Media jr=J
Positive Prediction of
Occurrence in Different.
Media from Machine
Learning Models
J|i bi^ii ¦ ¦
i In	¦¦. mi
I III	I II ? V !¦
I'.lJ J
r~jt'i
mi ii i ill ii mil
il	¦ mi
APCRA Inventory
llll I II
mini
1 ¦¦
Uir.'i
il i ¦ i ¦
n jj
11 ii
¦ i i
11 ^
11
in in
i1 r i

. .ii
vr . .
6621 Inventory Chemicals
16 of 24
¦ Traditional monitoring very limited
Office of Research and Development

-------
4>EPA
United States
Environmental Protection
Agency
Media Occurrence, Environmental Surveillance,
and Biomonitoring
Traditional Targeted
Monitoring Data
Non-Targeted Studies in _J
Several Media *¦—
Positive Prediction of
Occurrence in Different.
Media from Machine
Learning Models
111 ^1^11 ¦
y
r ii ii
•j'ldffr'1 ,M
i—tt'i
nli n i iii ii inn i
APCRA Inventory
A limited number of non-
targeted studies in media have
provided data for many
additional chemicals
6621 Inventory Chemicals
17 of 24
¦ Traditional monitoring very limited
Office of Research and Development

-------
oEPA	Exposure Predictions
United States
Environmental Protection
Agency	r— 		
Traditional
Assessments
HT Exposure Models
for Pathways (Ring et al. 2019)
Consensus
Predictions -t
(Ring et a I. 2019)
Positive Prediction for Various
Exposure Pathways
(Ring et al, 2019)
APCRA Inventory
•k NAM
Office of Research and Development
18 of 24
6621 Inventory Chemicals
High-throughput exposure models covering different exposure pathway classes
have generated exposure estimates for large numbers of chemicals compared to
traditional assessments.

-------
4>EPA
Exposure Predictions
United States
Environmental Protection
Agency
Traditional
Assessments
HT Exposure Models
for Pathways (Ring et al. 2019)
Consensus
Predictions -t
(Ring et a I. 2019)
Positive Prediction for Various
Exposure Pathways
(Ring et al, 2019)
APCRA Inventory
•k NAM
Office of Research and Development
6621 Inventory Chemicals
High-throughput exposure models covering different exposure pathway classes
have generated exposure estimates for large numbers of chemicals compared to
traditional assessments.

-------
SEPA	Toxicokinetics
United States
Environmental Protection
Agency
In Vivo TK Data
High-Throughput In Vitro TK Data
In Silico (QSAR) TK Parameters
Tox21
ToxCost
APCRA Inventory
6621 Inventory Chemicals
•kNAM
^^9 Office of Research and Development
20 of 24
High throughput in vitro measurement of toxicokinetics has expanded the
quantity and domain of chemicals with data, allowing for the development or
refinement of in silico models

-------
oEPA
Toxicokinetics
United States
Environmental Protection
Agency
In Vivo TK Data
High-Throughput In Vitro TK Data
In Silico (QSAR) TK Parameters
Tox21
ToxCost
APCRA Inventory
In silico approaches have
expanded the availability of
HTTK parameters to nearly all
chemicals tested for in vitro
bioactivity (96% of Tox21 and
89% ofToxCast) allowing for in
vitro to in vivo extrapolation of
6621 Inventory Chemicals
High throughput in vitro measurement of toxicokinetics has expanded the
quantity and domain of chemicals with data, allowing for the development or
ic NAM	refinement of in silico models
21 of 24
Office of Research and Development

-------
SEPA
United States
Environmental Protection
Agency
Summary
In all exposure-relevant domains, high-throughput NAMs have substantially increased the number of
chemicals for which data are available and improved coverage of chemical inventories.
Methods for estimating chemical releases (quantitative estimates of emission into different
environmental compartments) are needed; predictions for releases can reduce uncertainty in HT
exposure models that currently rely on production volume as surrogates for emission rates.
Methods should be developed for addressing mixtures or UVCBs. Approaches are needed for
identifying representative compositions or structures for multicomponent substances, and for making
use of this information in in silico modeling (i.e., QSAR) frameworks.
Measurement NAMs (i.e., non-targeted approaches) have the potential to substantially increase the
scope of evaluation datasets for predictive exposure models.
Continuing to develop and refine NAMs for exposure and toxicokinetic domains will improve the
quality of and expand the scope of risk-based metrics available for chemical prioritization.
22 of 24
Office of Research and Development

-------
g pp/\
Ongoing Exposure NAM Evaluation Activities
United States
Environmental Protection
Agency
Will aid in assessing fit-for-use of exposure NAMs in various regulatory contexts
(classification and labelling, prioritization, first-tier versus full assessments)
Comparison of Quantitative Use Relationship (QSUR) models for chemical function with
industry reported data
•	EPA's Chemical Data Reporting for Industrial Uses (Public)
•	ECHA Plastics Additives Initiative (PLASI)
•	Health Canada Chemicals Management Plan Information Gathering
Comparison of traditional exposure assessments (Health Canada Chemicals
Management Plan) to high-throughput model predictions
•	Consumer Assessments
•	Environmental media (i.e., ambient/far-field)
23 of 24
Office of Research and Development

-------
SEPA
United States
Environmental Protection
Agency
References
1.Dionisio KL, Phillips K, Price PS, Grulke CM, Williams
A, Biryol D, Hong T, Isaacs KK. The Chemical and
Products Database, a resource for exposure-relevant
data on chemicals in consumer products. Sci Data.
2018 Jul 10;5:180125.
2.Ingle BL, Veber BC, Nichols JW, Tornero-Velez R.
Informing the Human Plasma Protein Binding of
Environmental Chemicals by Machine Learning in the
Pharmaceutical Space: Applicability Domain and
Limits of Predictability. J Chem Inf Model. 2016 Nov
28;56(11):2243-2252.
3.Newton	SR, McMahen RL, Sobus JR, Mansouri K,
Williams AJ, McEachran AD, Strynar MJ. Suspect
screening and non-targeted analysis of drinking water
using point-of-use filters. Environ Pollut. 2018
Mar;234:297-306. doi: 10.1016/j.envpol.2017.11.033.
4.Pearce	RG, Setzer RW, Strope CL, Wambaugh JF,
Sipes NS. httk: R Package for High-Throughput
Toxicokinetics. J StatSoftw. 2017 Jul 17;79(4):1-26.
5.Phillips	KA, Wambaugh JF, Grulke CM, Dionisio KL,
Isaacs KK. High-throughput screening of chemicals as
functional substitutes using structure-based
classification models. Green Chem. 2017;19(4):1063-
1074.
6.Phillips	KA, Yau A, Favela KA, Isaacs KK, McEachran
A, Grulke C, Richard AM, Williams AJ, Sobus JR,
Thomas RS, Wambaugh JF. Suspect Screening
Analysis of Chemicals in Consumer Products. Environ
Sci Technol. 2018 Mar 6;52(5):3125-3135.
7.Rager	JE, Strynar MJ, Liang S, McMahen RL, Richard
AM, Grulke CM, Wambaugh JF, Isaacs KK, Judson R,
Williams AJ, Sobus JR. Linking high resolution mass
spectrometry data with exposure and toxicity forecasts
to advance high-throughput environmental monitoring.
Environ Int. 2016 Mar;88:269-280.
8.Ring	CL, Arnot JA, Bennett DH, Egeghy PP, Fantke P,
Huang L, Isaacs KK, Jolliet O, Phillips KA, Price PS,
Shin HM, Westgate JN, Setzer RW, Wambaugh JF.
Consensus Modeling of Median Chemical Intake for
the U.S. Population Based on Predictions of Exposure
Pathways. Environ Sci Technol. 2019 Jan
15;53(2):719-732.
9.Sayre R, Wambaugh J, and Grulke C. Database of
Pharmacokinetic Time-Series Data and Parameters for
Evaluating the Safety of Environmental Chemicals.
Presented at American Chemical Society Spring
Meeting, Orlando, FL, March 31 - April 04, 2019.
10.Sjerps RMA, Vughs D, van Leerdam JA, Ter Laak
TL, van Wezel AP. Data-driven prioritization of
chemicals for various water types using suspect
screening LC-HRMS. Water Res. 2016 Apr 15;93:254-
264
11.Sipes NS, Wambaugh JF, Pearce R, Auerbach SS,
Wetmore BA, Hsieh JH, Shapiro AJ, Svoboda D,
DeVito MJ, Ferguson SS. An Intuitive Approach for
Predicting Potential Human Health Risk with the Tox21
10k Library. Environ Sci Technol. 2017 Sep
19;51 (18): 10786-10796.
12.Wambaugh	JF, Wetmore BA, Ring CL, Nicolas CI,
Pearce RG, Honda GS, Dinallo R, Angus D, Gilbert J,
Sierra T, Badrinarayanan A, Snodgrass B, Brockman
A, Strock C, Setzer RW, Thomas RS. Assessing
Toxicokinetic Uncertainty and Variability in Risk
Prioritization. Toxicol Sci. 2019a Dec 1;172(2):235-
251.
13.Wambaugh	JF, Bare JC, Carignan CC, Dionisio KL,
Dodson RE, Jolliet O, Liu X, Meyer D, Newton S,
Phillips KA, Price PS, Ring CL, Shin H, Sobus JR, Tal
T, Ulrich E, Vallero D, Wetmore BA, Isaacs KK. New
approach methodologies for exposure science, Current
Opinion in Toxicology, Volume 15, 2019b, Pages 76-
92.
14.Wetmore	BA, Wambaugh JF, Allen B, Ferguson SS,
Sochaski MA, Setzer RW, Houck KA, Strope CL,
Cantwell K, Judson RS, LeCluyse E, Clewell HJ,
Thomas RS, Andersen ME. Incorporating High-
Throughput Exposure Predictions With Dosimetry-
Adjusted In Vitro Bioactivity to Inform Chemical
Toxicity Testing. Toxicol Sci. 2015 Nov; 148(1): 121-36.
24 of 24
Office of Research and Development

-------