US EPA CSS-HERA Board of Scientific Counselors Chemical Safety Subcommittee Meeting February 2-5, 2021 CSS Session 1B Slides


Note: this document may contain some
elements that are not fully accessible to
users with disabilities. If you need
assistance accessing any
information in this document, please
contact ORD_Webmaster@epa.gov.

US EPA CSS-HERA
Board of
Scientific
Counselors
Chemical Safety
Subcommittee
Meeting

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

SW(

$

<

3J

\

SB



ro

z

111

p*

PR0^°

The work presented within represents US EPA Office of Research and Development research
activities. Material includes both peer reviewed, published results and work-in-progress
research. Please do not cite or quote slides.

-------
Table of Contents

High-Throughput Exposure Models and the Systematic Empirical Evaluation of Models (SEEM)
Framework (John Wambaugh)	3

High-Throughput Toxicokinetic Models and In Vitro-ln Vivo Extrapolation (IVIVE) (Barbara
Wetmore)	25

NAMsfor Exposure: Non-Targeted Analysis (Jon Sobus)	49

The work presented within represents US EPA Office of Research and Development research activities. Material
includes both peer reviewed, published results and work-in-progress research. Please do not cite or quote slides.

-------
Xs,EPA

United States

Environmental Protection
Agency

ExpoCast

exposure forecasting

High Throughput Exposure Models

and the Systematic Empirical Evaluation of Models (SEEM) Framework

John Wamb

US EPA CSS-HERA Board of Scientific Counselors
Chemical Safety Subcommittee Meeting

The views expressed in this presentation are those of the author(s) and do not necessarily
reflect the views or policies of the US EPA.

February 2-5, 2021

-------
C PDA

United States	Stakeholder Need

Environmental Protection
Agency

as stated in research plan
Chemical exposure scenarios and pathways:

Chemical evaluations require information to estimate
exposure via a variety of high-priority pathways, including
scenario-specific data and models particular to consumer
products and materials in the indoor environment, as well
as occupational, ambient and ecological pathways.

2 of 20

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
5 PPA

Exposure Pathways

Environmental Protection | /

Agency

Other Industry

orkers

USE and RELEASE

Household
Products and
Durable Goods

I al -I

Chemical Manufacturing and Processing

Environmental
Release

Direct Use Residen ial Use C cupationa

(for example, surface cleaner) {for exami e, flooring) Use

MEDIA
EXPOSURE

(MEDIA + RECEPTOR)

'ndoor Air, Dust, Surfac.
Near-Field Near-Field

Direct Indirect

RECEPTOR Population

ener

Populatio

Waste

Occupational Dietary

Ecological

Food Drin'
-------
4>EPA

United States
Environmental Protection
Agency

Properties of High- II hroughput Exposure Models

1) Capable of handling many chemicals with minimal
descriptive information

2) Cover one or more relevant exposure routes

3) Allow for integration with models for other pathways

4) Scientifically plausible

ELSEVIER

Current Opinion in Toxicology

Available online 31 July 2019
In Press, Journal Pre-proof (?)

' Toxicotogy

New Approach Methodologies for Exposure
Science

John F. Wambaugh 1A Jane C. Bare 2, Courtney C. Carignan 3, Kathie L. Dionisio 4 Robin E.
Dodson 5' 6, Olivier Jolliet1, Xiaoyu Liu 8, David E. Meyer2, Seth R. Newton 4, Katherine A. Phillips 4,
Paul S. Price 4, Caroline L. Ring 9, Hyeong-Moo Shin 10,Jon R. Sobus 4, Tamara Tal u, Elin M. Ulrich
4, Daniel A. Vallero 4, Barbara A. Wetmore 4, Kristin K. Isaacs 4

5) Allow for the assessment of interindividual and ntraindividual variation in exposure

6) Amenable to integration within statistical frameworks that quantify uncertainty

7) No more complicated than necessary

4 of 20

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
4>EPA

United States
Environmental Protection
Agency

Existing HT Models for Key Pathways

Consumer (Near-Field) Pathways

Ambient (Far-Field) Pathways

Dietary Pathways

SHEDS-HT (Isaacs et a I2014)

RAIDAR-ICE (Li et a\., 2018)

RAIDAR-IC1L

Risk Assessment,
IDentification And Ranking
Indoor & Consumer Exposure

FINE (Shin et at., 2015)

Indoor Air (Mj

4±-

Carpet (Mt) Vinyl Floors (M„)

UseTox (Rosenbaum et a I., 2008)

Fate
factor

Ecotox
Effect
factor

a* t;

Damage to
aquatic ecosystems

Intake
- fraction iF
iF=XF»FF

Human
Effect
factor
EPhuman

RAIDAR (Arnot et at, 2006,
2008)

UseTox (Rosenbaum et al. (2008)

Fate
factor -

Ecotox
Effect
factor

SHEDS-HT (Biryol et a I., 2017)

-10-6-6-4
Log inferred median exposures (mg/kg-BW/day)

5 of 20

Office of Research and Development

Figure from Kristin Isaacs

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
&EPA

Consensus Exposure Predictions with

United States
Environmental Protection
Agency

Different exposure models
incorporate knowledge,
assumptions, and data

(MacLeod et al., 2010)

We incorporate multiple models
(including SHEDS-HT, USEtox,
RAIDAR) into consensus
predictions for 1000s of
chemicals within the Systematic
Empirical Evaluation of Models

fSEEM) (wambaugh et al., 2013, 2014, Ring
eta!., 2019)

Evaluation is like a sensitivity
analysis: What models are
working? What data are most
needed?

the SEEM Framework

Space of
Chemicals

Subset of
Chemicals with
Biomonitoring
Data

v.-_

Apply calibration and estimated
uncertainty to other chemicals

Expos
Infere
(Reve
Dosimc

¦M

m
DC

0
-si
flj

-u

Estimate
Uncertainty

Calibrate
models

Different
Chemicals

Model 1

01©

Available Exposure Predictors

Evaluate Model Performance
and Refine Models

6 of 20

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
&EPA

United States
Environmental Protection
Agency

Ensemble Predictions

We can use ensemble methods to make more stable models and characterize
uncertainty

Ensemble learning techniques in the machine learning
paradigm can be used to integrate predictions from
multiple tools. Pradeep (2016)

Office of Research and Development

Hurricane Path Prediction is an
Example of Integrating Multiple Models

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

'Ensemble methods are learning algorithms that construct a set of classifiers and
then classify new data points by taking a (weighted) vote of their predictions."
Dietterich (2000)

Ensemble systems have proven themselves to be very
effective and extremely versatile in a broad spectrum
of problem domains and real-world applications
(Polikar, 2012)

-------
&EPA

United States
Environmental Protection
Agency

Arnot j

UNIVERSITY OF

MICHIGAN

UC DAVIS

UNIVERSITY OF CALIFORNIA

UNIVERSITY OF

TEXAS

i ARLINGTON

Danmarks
m m Tekniske
» ~ Univetsitet

£
<

I \W ?

\T0-°

PRO"^

Ring et al. (2018)

SEEM3 Collaboration

Jon Arnot, Deborah H. Bennett, Peter P, Egeghy, Peter Fantke, Lei Huang, Kristin K. Isaacs, Olivier Jolliet, Hyeong-
Moo Shin, Katherine A. Phillips, Caroline Ring, R. Wood row Setzer, John F. Wambaugh, Johnny Westgate

Predictor

Reference(s)

Chemicals
Predicted

Pathway(s)

EPA Inventory Update Reporting and Chemical Data
Reporting (CDR) (2015)

US EPA (2018)

7856

All

Stockholm Convention of Banned Persistent Organic
Pollutants (2017)

Lallas (2001)

248

Far-Field Industrial and
Pesticide

EPA Pesticide Reregistration Eligibility Documents
(REDs) Exposure Assessments (Through 2015)

Wetmore et al. (2012, 2015)

239

Far-Field Pesticide

United Nations Environment Program and Society for
Environmental Toxicology and Chemistry toxicity model
(USEtox) Industrial Scenario (2.0)

Rosenbaum et al. (2008)

8167

Far-Field Industrial

USEtox Pesticide Scenario (2.0)

Fantke et al. (2011, 2012, 2016)

940

Far-Field Pesticide

Risk Assessment IDentification And Ranking (RAIDAR)
Far-Field (2.02)

Arnot et al. (2008)

8167

Far-Field Pesticide

EPA Stochastic Human Exposure Dose Simulator High
Throughput (SHEDS-HT) Near-Field Direct (2017)

Isaacs (2017)

7511

Far-Field Industrial and
Pesticide

SHEDS-HT Near-field Indirect (2017)

Isaacs (2017)

1119

Residential

Fugacity-based INdoor Exposure (FINE) (2017)

Bennett et al. (2004), Shin et al. (2012)

645

Residential

RAIDAR-ICE Near-Field (0.803)

Arnot et al., (2014), Zhang et al. (2014)

1221

Residential

USEtox Residential Scenario (2.0)

Jolliet et al. (2015), Huang et al.
(2016,2017)

615

Residential

USEtox Dietary Scenario (2.0)

Jolliet et al. (2015), Huang et al. (2016),
Ernstoff et al. (2017)

8167

Dietary

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
*>EPA

United States
Environmental Protection
Agency

SEEM3 Considers Pathway of Exposure

We organize models by the
exposure pathways they cover

We calibrate predictors based
on ability to explain median
NHANES exposure rates

General Population
Median
Chemical Exposure
(mg/kg BW/day)

Pathway

Residential

Ring et al. (2018)

Office of Research and Development

Dietary

Far-Field
Pesticides

Far-Field
Industrial

Unknown

Chemical-
Specific
Pathway
Relevancy

Yes/No ~

Yes/No

Predictors

9 of 20

Average Unexplained Residential
SHEDS-HT Direct Residential
SHEDS-HT Indirect Residential
USETox
FINE

RAIDAR-ICE
Production Volume

Average Unexplained Dietary
SHEDS-HT Dietary
Production Volume
USETox

Average Unexplained Pesticide

Pesticide REDs

USETox

Stockholm Convention
Production Volume

Average Unexplained Industrial

USEtox

RAIDAR

Stockholm Convention
Production Volume

Average Unexplained Overall

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
&EPA

United States
Environmental f
Agency

1041

jQ
C5J

OS
¦CT5

H5
CD

1 -

10'

fB
T3
;L>

O
f5

3
O.

£ 10

-.3.

Ring et al. (2018)

Consensus Modeling of Median Chemical Intake

Intake Rate

> 0.1 mg/kg BW/day

1976 chemicals

¦ 0

1 D

Chemical Rank
Office of Research and Development

Path way (s)

Dietary

~ Dietary, Industrial
O Dietary, Pesticide

Dietary, Pesticide, Industrial
Dietary, Residential
¦ Dietary, Residential, Industrial

• Dietary, Residential Pesticide
Dietary, Residential Pesticide, Industrial!

~ Industrial

~ Pesticide
Pesticide, Industrial
Residential
Residential, Industrial!

Residential, Pesticide
Residential, Pesticide, industrial
Unknown

Of 687,359 chemicals
evaluated, 30% have
less than a 50%
probability for exposure
via any of the four
pathways and are
considered outside the
"domain of
applicability"

10fl

5
cm

3'
m

1 fl-

OI
T3
CD

TO
3
CL

-s.

Intake Rate

< 0.1 mg/kg BW/day

685,383 chemicals

< 1 ng/kg BW/day
681,574 chemicals

103 2x105 4x10s 6x1De
Chemical Rank

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
&EPA ExpoCast SEEM Models: Required Building

United States I I O

Blocks for the Output

Environmental Protection
Agency

Machine-learning models for filling gaps from
Supporting Models structure when no data are available

Exposure Factor Datasets

Composition and use/release data

Office of Research and Development „ . . , US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

Slide from Kristin Isaacs

-------
&EPA

United States
Environmental Protection
Agency

ExpoCast SEEM Models: Required Building

Blocks for the Output

Individual HT Pathway Models

Modell Model 2 Model 3 Model 4 Model 5

for example, SHEDS-HT, HT ChemSteer,
external models

Supporting Models
Exposure Factor Datasets

Office of Research and Development

Slide from Kristin Isaacs

Machine-learning models for filling gaps from
structure when no data are available

Composition and use/release data

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
&EPA

United States
Environmental Protection
Agency

ExpoCast SEEM Models: Required Building

Blocks for the Output

Monitoring Data for Evaluating and Calibrating

the Predictors

Including NHANES biomonitoring and
USGS water datosets

Individual HT Pathway Models

Modell Model 2 Model 3 Model 4 Model 5

for example, SHEDS-HT, HT ChemSteer,
external models

Supporting Models
Exposure Factor Datasets

Office of Research and Development

Slide from Kristin Isaacs

Machine-learning models for filling gaps from
structure when no data are available

Composition and use/release data

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
&EPA

United States
Environmental Protection
Agency

ExpoCast SEEM Models: Required Building

Blocks for the Output

Consensus SEEM Predictions
for Receptor

Monitoring Data for Evaluating and Calibrating

the Predictors

Including NHANES biomonitoring and
USGS water datosets

Individual HT Pathway Models

Modell Model 2 Model 3 Model 4 Model 5

for example, SHEDS-HT, HT ChemSteer,
external models

Supporting Models
Exposure Factor Datasets

Office of Research and Development

Slide from Kristin Isaacs

Machine-learning models for filling gaps from
structure when no data are available

Composition and use/release data

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
&EPA

United States
Environmental Protection
Agency

ExpoCast SEEM Models: Required Building

Blocks for the Output

Consensus SEEM Predictions*
for Receptor-

Monitoring Data for Evaluating and Calibrating

the Predictors

*New Approach
Methodologies for Exposure:

Application to Real Decision Contexts

Including NHANES biomonitoring and
USGS water datasets

Individual HI Pathway Models*

Modell Model 2 Model 3 Model 4 Model 5

for example, SHEDS-HT, HT ChemSteer,
external models

Supporting Models*
Exposure Factor Datasets

Office of Research and Development

Slide from Kristin Isaacs

Machine-learning models for filling gaps from
structure when no data are available

Composition and use/release data

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
&EPA

United States
Environmental Protection
Agency

Formatting Occupational Exposure Models for HI Use

We have developed consensus models for consumer
and some ambient pathways, but ecological and
occupational consensus models are ongoing

Many predictors for these pathways exist, but they are
not typically oriented for high throughput capacity, for
example EPA's ChemSTEER (Chemical Screening Tool
for Exposures and Environmental Releases)

Command Line Occupational Exposure Tool (CLOET) a
command line tool that allows use of ChemSTEER v3.0
in a high throughput manner

Multiple scenarios for each model have been run and
tested against ChemSTEER GUI to test for model
fidelity.

EPA-OPPT 1-Hand Dermal
Contact with Liquid

EPA-OPPT 2-Hand Dermal
Contact with Liquid

EPA-OPPT 2-Hand Dermal
Immersion with Liquid

EPA-OPPT 2-Hand Dermal
Contact with Solids

EPA-OPPT 2-Hand Dermal
Contact with Container Surfaces

User-defined
Dermal

EPA-OPPT Small
Volume Solids Handling

OSHA PEL-limiting Model
for Substance-specific Particulates

OSHA Total
PNOR PEL-limiting

OSHA Respirable
PNOR PEL-limiting

EPA-OPPT Automobile
OEM Spray Coating

EPA-OPPT Automobile
Refinish Spray Coating

EPA-OPPT UV
Roll Coating

Dermal Models

Exposure Scenario
¦¦ high
¦¦ low

10"2 10"1

Average Daily Dose (mg/kg-BW/day)

Inhalation Models

icr5 io-4 i
-------
&EPA

United States
Environmental Protection
Agency

" wo-Stage Occupational Exposure Model

OSHA's chemical exposure health data set for air samples was used to build a two-stage
model that predicts 1) if a chemical is likely to be detected in air and 2) what the likely
concentration would be

OPERA physicochemical property distributions across NAICS sector and subsectors are
included as input distributions to the models in addition to the OSHA data

OPERA Property
Predictions

Detect /
detect Model

Bayesian Hierarchical Regression allows
us to organize our predictions (either
detect/non-detect or concentration) by
NAICS Sector and/or Subsector

Air Concentration
Model

Minucci et al,

in preparation

Non-detects

Office of Research and Development

logio[Conc. (mg/m3)]

Other Services
Manufacturing
Construction

Agriculture, Forestry, Fishing and Hunting
Real Estate and Rental and Leasing
Retail Trade
Educational Services
Health Care and Social Assistance
Wholesale Trade
Professional, Scientific, and Technical Services
Transportation and Warehousing
Arts, Entertainment, and Recreation
Information

Admin., Support, Waste Manage, and Remediation Services
Mining, Quarrying, and Oil and Gas Extraction
Accomodation and Food Services
Public Administration
Utilities

Finance and Insurance

0.2 0.4 0.6
Probability of detection

0.8 -0.5 0.0 05 L0 1.5 2.0
Air concentration (log mg/m3)

Slide from Katherine Phillips

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
&EPA

United States
Environmental Protection
Agency

EcoSEEM Metamodel for Surface
Water Chemical Concentrations

l_>

to
CD

.CD

.2 ~o

tj CD

1— o

Q_ >

Chemical-
specific

information

(/)

Sayre et al,
in preparation

-a
o

1/1

cz
ro

ro
CD

-i—>

Release (loading) and
fate predictors

(P)

n, n.

In v.=m0+

m ikIn '

o
ro

CD
<_>
c
o
<_>

"ro
c
o

4—>
TO
C

-t—'
CJ

* .*

- •

EcoSEEM

~ -*¦

metamodel

(m)

Office of Research and Development

L0 ¦ Lu Lu "Ljk^\Ljiyki

j=1 k~ 1

Slide from Risa Sayre

v: *¦-
v

* V '.v

V»- ' '

i/>

cz
o
'¦*—>
TO

CD
C_)

o
<_>

QJ
-»—>
TO

CD
<_>
TO

USGS/EPA water
monitoring data

(y)

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
*>EPA

United States
Environmental Protection
Agency

EcoSEEM Evaluating Predictive Ability of

HT Surface Water Models

The strength of the correlation
between each combination of
release and fate model
predictions and the observed
water concentrations allows
model calibration

The most informative pair for
bulk concentrations was USEtox
freshwater model using loadings
from NPV

Sayre et al,
in preparation

USEtox urban air*NPV-

Ubttox rural air NrV"

UoLtox natural soil Nrv-

• - U S Etox fre s h wate r*TRI -
ro

ts U S Etox fre s hwate r* S H E D S-dtd -

-------
*>EPA

United States
Environmental Protection
Agency

Outlook

SEEM metamodels have been developed for consumer and some
ambient pathways (Ring et al., 2018) and ecological and
occupational consensus models are in development

Estimates of exposure, with
appropriately estimated
uncertainty, allow quantitative
prioritization of potential
chemical risk (Wetmore et al.,

2015; Ring et al., 2017)

EPA's
ExpoCast
Project

Chemical Risk

Dose-Response

(Toxicokinetics
/Toxicodynamics)

Exposure

mg/kg BW/day

Potential
Hazard from
in vitro HTS
and HTTK

Exposure
Forecasts
from SEEM
Consensus
Meta-Models

Lower
Risk

Medium Higher
Risk Risk

20 of 20

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
smi

ExpoCast Project
(Exposure Forecasting)

Center for Computational Toxicology and Exposure

Linda Adams
Lucas Albrecht
Matthew Boyce
Miyuki Breen
Alex Chao
Daniel Dawso
Mike Devito
Alex East
Lindsay Eddy

mat

LaU-

Colin Guider
Mike Hughes
Victoria Hull *
Kristin Isaacs
Richard Judson
Jen Korol-Bexell
Anna Kreutz Vs
Charles Lowe
Seth Newton

Christopher Eklund Alii Phillips
Peter Egeghy Katherine Phillips

Marina Evans
Alex Fisher * ^
Rocky Goldsmith
Louis Groff
Chris Grulke

Paul Price
Tom Purucker
Ann Richard
Caroline Ring

Risa Sayre
Mark Sfeir
Marci Smeltz
Jon Sobus
Zach Stanfield
Mike Tornero-Velez
Rusty Thomas
Elin Ulrich
Dan Vallero
Barbara Wetmore
John Wambaugh
Antony Williams

Hongwan Li
Xiaoyu Liu
Zachary Robbins
Mark Strynar

Collaborators

Arnot Research and Consulting
Jon Arnot
Johnny Westgate
Integrated Laboratory Systems
Xiaoqing Chang
Shannon Bell

National Toxicology Program

Steve Ferguson

Karnel Mansouri

Ramboll

Harvey Clewell

Silent Spring Institute

Robin Dodson

Simulations Plus

Michael Lawless

Southwest Research Institute

Alice Yau

Kristin Favela

Summit Toxicology

Lesa Ay I ward

Technical University of Denmark

Peter Fantke

Unilever

Beate Nicol

Cecilie Rendal

Ian Sorrell

United States Air Force
Heather Pangburn
Matt Linakis

University of California, Davis
Deborah Bennett
University of Michigan
Olivier Jolliet

University of Texas, Arlington
Hyeong-Moo Shin
University of Nevada

Ui • '

University of North Carolina, Chapel Hill
Julia Rager
Marc Serre

-------
*>EPA

United States
Environmental Protection
Agency

Arnot, J.A, et al. 2006. Screening level risk assessment model for
chemical fate and effects in the environment. Environ. Sci.

Technol. 40: 2316-2323

Arnot, J.A.; Mackay D. 2008. Policies for chemical hazard and risk
priority setting: Can persistence, bioaccumulation, toxicity and
quantity information be combined? Environ. Sci. Technol. 42:
4648-4654. DOI: 10.1021/es800106g

Barber MC, et al (2017). "Developing and applying metamodels of
high resolution process-based simulations for high throughput
exposure assessment of organic chemicals in riverine
ecosystems." Science of the Total Environment, 605, 471-481.

Bennett, D. H.; Furtaw, E. J., Fugacity-based indoor residential
pesticide fate model. Environmental Science & Technology 2004,
38, (7), 2142-2152.

Biryol D, et al.. High-throughput dietary exposure predictions for
chemical migrants from food contact substances for use in
chemical prioritization. Environment international. 2017 Nov
l;108:185-94.

Dietterich, Thomas G. "Ensemble methods in machine
learning." International workshop on multiple classifier systems.
Springer, Berlin, Heidelberg, 2000. Ernstoff, A. S, et al., High-
throughput migration modelling for estimating exposure to
chemicals in food packaging in screening and prioritization tools.
Food and Chemical Toxicology 2017,109, 428-438.

Fantke P, Jolliet O. Life cycle human health impacts of 875
pesticides. The International Journal of Life Cycle Assessment.
2016 May l;21(5):722-33.

Fantke P, et al.. Dynamic toxicity modelling based on the USEtox
matrix framework. InSETAC Europe 25th Annual Meeting 2015 (pp.
33-34). SETAC Europe.

Huang, L, et al., A review of models for near-field exposure
pathways of chemicals in consumer products. Science of The Total
Environment 2017, 574, 1182-1208.

Office of Research and Development

References

Huang, L.; Jolliet, O., A parsimonious model for the release of volatile
organic compounds (VOCs) encapsulated in products. Atmospheric
Environment 2016, 127, 223-235.

Isaacs KK, et al.. SHEDS-HT: an integrated probabilistic exposure
model for prioritizing exposures to chemicals with near-field and
dietary sources. Environmental science & technology. 2014 Nov
4;48(21):12750-9.

Jolliet, O , et al., P., Defining Product Intake Fraction to Quantify and
Compare Exposure to Consumer Products. Environmental Science
& Technology 2015, 49, (15), 8924-8931.

Lallas PL. The Stockholm Convention on persistent organic
pollutants. The American Journal of International Law. 2001 Jul
l;95(3):692-708.

Li L, et al.. A model for risk-based screening and prioritization of
human exposure to chemicals from near-field sources.
Environmental science & technology. 2018 Nov 8;52(24):14235-44.

MacLeod, Matthew, et al. "The state of multimedia mass-balance
modeling in environmental science and decision-making." (2010):
8360-8364.Minucci et al, in preparation

Minucci, Jeff et al. "High Throughput Model for Occupational
Exposure"

Polikar, Robi. "Ensemble learning." Ensemble machine learning.
Springer, Boston, MA, 2012. 1-34.

Pradeep, Prachi, et al. "An ensemble model of QSAR tools for
regulatory risk assessment." Journal of cheminformatics 8.1
(2016): 1-9.

Rosenbaum RK, et al. USEtox—the UNEP-SETAC toxicity model:
recommended characterisation factors for human toxicity and
freshwater ecotoxicity in life cycle impact assessment. The
International Journal of Life Cycle Assessment. 2008 Nov
1;13(7):532.

Ring, Caroline L., et al. "Identifying populations sensitive to
environmental chemicals by simulating toxicokinetic
variability." Environment international 106 (2017): 105-118.

EPA ORD Publications in Bold

Ring CL, et al. Consensus modeling of median chemical intake for
the US population based on predictions of exposure pathways.
Environmental science & technology. 2018 Dec 5;53(2):719-32.

Say re et a I, "Consensus Model for Predicting Chemical Surface
Water Concentrations"

Shin, H.-M., et al., Intake Fraction for the Indoor Environment: A
Tool for Prioritizing Indoor Chemical Sources. Environmental
Science & Technology 2012, 46, (18), 10063-10072.

Shin, Hyeong-Moo, et al. "Risk-based high-throughput chemical
screening and prioritization using exposure models and in vitro
bioactivity assays." Environmental science & technology 49.11
(2015): 6760-6771

Wambaugh, John F., et al. "High-throughput models for exposure-
based chemical prioritization in the ExpoCast
project." Environmental science & technology 47.15 (2013): 8479-
848.

Wambaugh, John F., et al. "High Throughput Heuristics for
Prioritizing Human Exposure to Environmental
Chemicals." Environmental science & technology (2014).

Wambaugh JF, et al. New approach methodologies for exposure
science. Current Opinion in Toxicology. 2019 Jun l;15:76-92.

Wetmore, B. A., et al. Incorporating High-Throughput Exposure
Predictions With Dosimetry-Adjusted In Vitro Bioactivity to
Inform Chemical Toxicity Testing. Toxicol. Sci. 2015,148 (1),
121-36.

Zhang, X.; Arnot, J. A.; Wania, F., Model for screening-level
assessment of near-field human exposure to neutral organic
chemicals released indoors. Environmental Science & Technology
2014, 48, (20), 12312-12319.

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
Xs,EPA

United States

Environmental Protection
Agency

ExpoCast

exposure forecasting

High-Throughput Toxicokinetic Models
and In Vitro-ln Vivo Extrapolation (IVIVE)

Barbara A.

US EPA CSS-HERA Board of Scientific Counselors
Chemical Safety Subcommittee Meeting

February

The views expressed in this presentation are those of the author
and do not necessarily reflect the views or policies of the US EPA

-------
*>EPA

United States
Environmental Protection
Agency

NAMs for Exposure
Toxicokinetics

Acceptance and use of in vitro data
by uncertainties associated with exposure characterization and metabolism

Many in vitro systems:

• lack consideration of biotransformation capabilities

• Overestimation of hazard for chemicals rapidly cleared in vivo

• Underestimation of hazard for chemicals bioactivated in vivo

• lack consideration of exposure route

• lack consideration of susceptible populations / life stages

• In vitro potency estimates are often not adjusted for chemical availability
in the in vitro system (ie, in vitro disposition)

Recent Agency Case Study Finding: i

TK data availability rate limiting factor in TSCA screening for
chemical prioritization

Toxico
kinetics

Office of Research and Development

*"A Proof-of-Concept Case Study Integrating Publicly Available
Information to Screen Candidates for Chemical Prioritization under TSCA';

-------
&EPA

United States
Environmental Protection
Agency

a
a

Hepatocytes

Basolateral

Caco-2 cells

Transporter assays

rCYP3A4

7 CYP3A4 C
' '

ClrCTS3A4

In Vitro-ln Vivo Extrapolation (IYIYE)
I. In WtroToxicokinetic Assays

Plasma Protein
Binding (fu)

Apparent
permeability (Papp)

Renal clearance
Renal reuptake

Isozyme-specific
clearance
(hepatic, renal, intestinal)

Office of Research and Development

IVIVE originally used arid vetted in pharma applications
HT-IVIVE approach uses

- hepatic clearance

- plasma protein binding

- conservative assumptions
Predictions consistently protective of human health

Internal Concentration
Predictions Given a Set
Administered Dose

Ongoing efforts will:

- Incorporate additional TK inputs for better predictivity

- Assess impact of transporter involvement

- Evaluate extent of population variability

- Employ experimental measures to develop predictive

tools

Rotroff et al., Tox Sci., 2010
Wetmore et a!., Tox Sci., 2012
Wetmore et al., Tox Sci., 2014

Wetmore et al., Tox Sci., 2015 Wambaugh et al., 2019
Wambaugh et al., Tox Sci., 2015 Smelte et al., in preparation
Honda et al., 2019 Kreutz et al., in preparation

-------
*>EPA

United States
Environmental Protection
Agency

InVitro-ln Vivo Extrapolation
II. Physiologically-basedToxicokinetic Modeling

Oa I

11—

¦Tr* T bmm

fine U.J

1*.. Tl—.

O ¦

Rnl L-i h-di

Uh Dbd

ii .

—i

0k™

Evolving Capabilities

• Augmentation of PBTK models based on need

• Expanding to incorporate additional TK data
(intestinal, renal compartments)

• Incorporating additional exposure routes

• Incorporating additional pathways (gestational)

• Incorporating demographic info to expand
population-based info (variability)

"httk": Open-source modeling package

Modeling Platform incorporates:

- chemical-specific inputs (TK data, physico-chemical)

- physiologic inputs (blood flow rates, tissue size)

into Simulations set up for:

- populations of interest

- exposures of interest

- Capturing variability (within or across populations)
Based on variations in the physiologic inputs (Monte Carlo)

Office of Research and Development

Pearce et al, 2017, J Statistical Software

-------
&EPA

United States
Environmental Protection
Agency

NAMs for Prioritization

Integrating Hazard,TK, and Exposure

-3

+->
c

_0J

CO
>

"=3

0J
+->
03

CD
CtO

9. J?; .

Wiifl i 0

10-

Chemicals Monitored by CDC NHANES

High throughput in vitro
screening can be used to
estimate doses needed to
cause bioactivity

Exposure intake rates can
be inferred from
biomarkers

mg/kg

3W/day

Office of Research and Development

Wambaugh et oi, 2014
Wetmore et ol., 2015
Ring etal. (2017)
And others...

Potential
Hazard from
in vitro with
Reverse
Toxicokinetic
s

Potential
Exposure
Rate

Lower
Risk

Medium Higher
Risk Risk

-------
^EEst Toxicokinetics and IVIVE - Stakeholder Needs

Environmental Protection
Agency

OngoinR Development of Toxicokinetic and IVIVE Tools for use in NAMs

Primary goal: to provide a human exposure-dose context for bioactive
concentrations from NAMs for hazard testing

¦ TK Methods across TSCA landscape - including challenging chemistries, emerging contaminants

¦ Incorporating more exposure routes and pathways

¦ Tools to characterize exposures to sensitive populations and life stages

¦ Characterize in vitro disposition across TSCA landscape

¦ Tools to identify, quantitate and/or reduce sources of uncertainty

Secondary goal: to provide open-source data and models for evaluation and use by the
broader scientific community

¦ Concomitant incorporation of above tools and data in HTTK package

¦ Databases with in vitro, in vivo data for use in IVIVE evaluations, in silico tool development

6 of 24

Office of Research and Development

-------
*>EPA

United States
Environmental Protection
Agency

Rapid Exposure Modeling and Dosimetry

HTTK: Open-Source
Platform

Predictive Tools

Plasma protein binding

Hepatic clearance
Transporter Involvement
Isozyme Involvement

Databases

in vitro TK data
In vivo TK data (CvTdb)

TK Data Generation
in vitro:

More chemicals, chemistries
Species expansion (rat, human)
TK assay expansion (intestinal, renal)
in vivo:

Rat (cross-species extrapolation)
Office of Research and Development

Model Expansion

Multi-compartment; PBTK
Exposure routes
Gestational pathway
Incorporating new TK data streams

Refinement

IVIVE / IVIVC efforts
In Vitro Disposition
Best Practices

Population Variability

NHANES; physiology
Toxicokinetic variability

Uncertainty / Variability
Assessments

Bayesian approaches
Experimental uncertainty
Biologic variability

-------
&EPA

United States
Environmental Protection
Agency

- In VitroToxicokinetic Data Generation -

PFAS: Using NAMs to Fill Information Gaps

oEPA

CPA UW1I0O1 February 2019

EPA's Per- and Polyfluoroalkyl
Substances (PFAS) Action Plan

Goals:

• Generate data to support development and
refinement of categories and read-across
evaluation

• Incorporate substances of interest to Agency

• Characterize mechanistic and toxicokinetic
properties of the broader PFAS landscape

Data collection;
Pre-defined
structural
categories

Step 0:
Characterizing the
PFAS library

Maximizing Read-across

On Wkgrp-31 list;
Availability of in vivo data

On EPA-PFAS list;
Availability of in vivo data

Step 1: Select
substances from
categories of greatest
interest to the Agency

Step 2: Select
substances from
categories of interest
the Agency

+2 structural categories

, iii.im..
Availability of in vivo data

Availability of
in vivo data

^ r

Capturing Structural Diversity

EPA interest
in vivo data lacking

Characterizing the
PFAS Landscape

Step 3: Select
substances from
remaining categories
with in vivo data

Step 4: Select
substances from
categories of interest
to the Agency

+6 structural categories" J

*2 categories contained only 1
chemical, so were not included

+5 structural categories

Step 5: Select
substances from
remaining
categories

+10 structural categories

53 structural
categories

53 substances: 12 categories

9 substances:

6 categories

13 substances

10 categories

Research Area 1: What are the human health and ecological effects of
exposure to PFAS?

as high throughput and computational
approaches to explore different chemical
categories of PFAS... to inform hazard

characterization, and to promote prioritization

8 of 24

Office of Research and Development

-------
&EPA

United States
Environmental Protection
Agency

- In VitroToxicokinetic Data Generation -

Category-Based Analyses of Toxicokinetic Data

Category-Based Analysis of
Plasma Protein Binding Data

PFAS TK data: -150 PFAS

- Hepatic clearance

- Plasma protein binding

- Renal transporter activity
IVIVE, modeling, TK NAMs

Preliminary set: Plasma protein
binding data across 50+ PFAS

75% of PFAS:

FU<0.Q5

r 100

-75

-50

-25

O
C

3
c_

Si
<"

8'*

!,H

Hepatic Clearance Data

PFOA

PFHA

"i"

p = 0.007

PFBA

p = 0.0007

2H, 2H, 3H, 3H-PFOA

Perfluorooctanesulfonamide

I'*

Tim* (ran)

CLint = 0.9 mL/min/kg

<&* q>«

Heptafluorobutyramide

I'l

f 0.8-
| 0 6-

Txp_cats

Distribution of F,,

Office of Research and Development

* 0 4-

I °2'
I

PFAS - Amides
Octafluoroadipamide

i
*
O 1.0-

Tim* (mtn)

CLint = 33.5 mL/min/kg

Perfluorooctanesulfonamide

! 0.5-

ztr

| 0.6-

e> 0.2-
<

I 0.0.

-------
*>EPA

United States

Environmental

Agency

Protection

^ 10"

CD
V)

-6.

- Predictive Tool Development -

In vitro TK measurements are being employed in model development and evaluation.
Plasma protein binding (fu); hepatic clearance (Clint) underway; others to follow.
In silico predictions for fu (plasma protein binding)

A A

¦i-J-
¦£

u ¦ %

»A
•

O 10

¦'

QSAR ^

• ADMet

A Dawson .0
OPERA "g

ol 10

CD
>

¦ •

•

A —

-5

10 d
Predicted f,

10"

Office of Research and Development

-2.

-3.

*
*

This method
uses nearest
neighbors, and
many evaluation
chemicals are in
training set

ADMet

Dawson

QSAR

OPERA

Dawson et al. submitted
Pradeep etal., 2020
Tornero-Velez etal., underway
Sipes et al., 2017

-------
&EPA

United States
Environmental Protection
Agency

RESEARCH ARTICLE

- Model Expansion -
Gestational Pathway

Empirical models for anatomical and
physiological changes in a human mother and
fetus during pregnancy and gestation

Dustin F. Kapraun 1 *, John F. Wambaugh 2. R. Woodrow Setzer 2, Richard
S. Judson 2

1 National Centet for Environmental Assessment, US Environmental Protection Agency, Research Trtangle
Park. North Carolina. Un itsd States, of America, 2 National Center for Computational ToxicoJogy, US
Environmental Protection Agency. Research Triangle Park, North Carolina, United Slates of America

* kafwai in rli istln & ppa qnv

Table 1. Itemized comparison of selected publications that contain one or more formulae related Id human gestation and pregnane}'.

^Mamiseript^

Presents original data'

Presents original compiled data' set(s)

Presents original models' based on compiled data sets of Abdulfalil et al. f2al

Presents original models' based on compiled data sets of Abduljalil et al. liji

(+) Employs and thoroughly describes rigorous statistical methods tor parameter' estimation

(+) Employs and thoroughly describes rigorous statistical methods for model' selection

(+) Presents original models" for multiple maternal compartments

(+) Presents original models' for multiple fetal compartments

(+) Presents models that reflect a biologically accurate depiction of the fetal circulatory system*

(+) Presents explicit models' tor "rest of body" compartments that yield feasible (e.g.. non-negative)
values for all relevant time points

(+) Systematically compares original models' with previously published models"

(-1 Presents models that contain errors or inconsistencies identified in the current manuscript

[21]

Q5]

[25]

[28]

[221

[3JJ

li]

[11]

50-

40-

§30

-------
&EPA

New HU-PBTK Models: Development is Done with ^ Branches

United States

Environmental

Agency

Protection

Gas Inhalation
Exposure Route

EPA, USAFSAM

Standard httk 1,10.0 PBTK Model

f\juid>

Lung Tissue

Lung Blood

Gut Tissue

Gut Blood

•gw-

CL,

Liver Tissue

Liver Blood

% t

metabolism

Kidney Tissue

Tissue Blood

Rest-of-Body

Rest-of-Body Blood

richly

^^e^usecT

Non-Exposed Skin Tissue

Non-Exposed Skin Blood

Exposed Skin Blood

Exposed Skin Tissue
*

Media

Dermal Exposure Route

EPA, Unilever, INERIS

¦*!

Inhaled air

Lung Arterial Blood

Gut Tissue

Gut Blood

Liver Tissue

Liver Blood

Qgf

Kidney Tissue

Tissue Blood

Rest-of-Body

Lung Tissue

Lung Blood

-K>—~

Office of Research and Development

on a Git Repository

4 Qgur

¦^gut

Oliver

^^kidne^

Rest-of-Body Blood

^•richly

¦^¦lung

Human Gestational Model

EPA, FDA

Venous Blood

=3 O

^ "

Arterial Blood

Fetus

Aerosol Inhalation
Exposure Route
(with APEX model)

EPA, USAFSAM

| Inhaled Aerosol L

I *

Gut Lmrien ~|

Gut Tissue

Gut Blood

Liver Tissue

Liver Blood

metabolism

Qgf

Kidney Tissue

Tissue Blood

Rest-of-Body

Lung Tissue

Lung Blood

Venous Blood

Arterial Blood

Mother

Oliver

Rest-of-Body Blood

^richly

m Q|uns

P 03

-------
&EPA

United States
Environmental Protection
Agency

- Database Development -
CvTdb: An In Vivo TK Database

EPA has developed a public database of concentration
vs. time data across several species for building,
calibrating, and evaluating TK models

Effort ongoing, but to date includes:

¦ 198 analytes (EPA, National Toxicology Program,
literature)

¦ Routes: Intravenous, dermal, oral, sub-cutaneous,
and inhalation exposure

Standardized, open-source curve fitting software
invivoPKfit used to calibrate models to all data

https://github.com/USEPA/CompTox-ExpoCast-invivoPKfit

CvTdb Link: https://github.com/USEPA/CompTox-PK-CvTdb

Office of Research and Development

expired air

38 17

442 147

Other: 12 7

Sayre et al. (2020)

feces 4 1
urine 59 14

www.nature.com/scientificdata

SCIENTIFIC DATA!

•) Check lor i/paaws I

OPEN Database of pharmacokinetic time-
datadescriptor series data and parameters for 144
environmental chemicals

Risa R. SayreO1,2'30, John F. Wambaugh 1 & Christopher M. Grulke

-------
&EPA

United States
Environmental Protection
Agency

- HTTK Platform -
Open-Source Tools and Data for HTTK

https://CRAN.R-proiect.orq/packaqe=httk

. Chemical-specific in vitro data have been obtained from r
experiments. Both physiologically-based ("PBTK") and empirical (for example, one compartment) "TK" m<
parameterized with the data provided for thousands of chemicals, multiple exposure routes, and various spei
of systems of ordinary differential equations which are solved using compiled (C-based) code for speed. A J
included, which allows for simulating human biological variability (Ring et al., 2017 T. These functions and data provide a set
vivo extrapolation ("IVTVE") of 1]
dosimetry (also known as "RTK"

R package "httk"

downloads 1071/month

2.0.3

Depends: R(>2.10)

Imports: deSolve. msm. data .table, survey, mvtnorm. tninenorm. stats, graphics, utils, magrittr. £

Suggests: ggplot2. kni.tr. rmarkdown. R.rsp. GGallv. gplots, scales. EnvStats. MASS. RColorBrev

classhit. ks, stringr. reshape. reshaoe2. gdata, viridis. CensRegMod, gmodels. colors gat
dglyr, forcats. smatr. stools. gridExtra
Published: 2020-09-25

Author: JohnWambaugh [aut. ere], Robert Pearce [aut], Caroline Ring [ant], Greg

Sfeir [aut]. Matt Linakis [aut], Jimena Davis [ctb], James Siuka [ctb], Nisha Si
Wetmore [ctb], Woodrow Setzer , [ctb]

Maintainer: John Wambaugh

BugReports: https://github.com/J-rSEPA/CompTox-ExDoCast-httk

Open source, transparent, and peer-
reviewed tools and data for high
throughput toxicokinetics (httk)
Available publicly for free statistical
software R

Allows in vitro-in vivo extrapolation
(IVIVE) and physiologically-based
toxicokinetics (PBTK)

Human-specific data for 987 chemicals
Described in Pearce et al. (2017a)

-------
*>EPA

United States
Environmental Protection
Agency

- HTTK Platform -
Modules within R Package "httk"

Feature

Description

Reference

Chemical Specific In Vitro
Measurements

Metabolism and protein binding for ~1000
chemicals in human and ~200 in rat

Wetmore et al. (2012,
2013, 2015), plus
others

Chemical-Specific In Silico
Predictions

Metabolism and protein binding for ~8000
Tox21 chemicals

Sipes et al. (2017)

Generic toxicokinetic models

One compartment, three compartment,
physiologically-based oral, intravenous, and
inhalation (PBTK)

Pearce et al. (2017a),
Linakis et al. (2020)

Tissue partition coefficient
predictors

Modified Schmitt (2008) method

Pearce et al. (2017b)

Variability Simulator

Based on NHANES biometrics

Ring et al. (2017)

In Vitro Disposition

Armitage et al. (2014) model

Honda et al. (2019)

Uncertainty Propagation

Model parameters can be described by
distributions reflecting uncertainty

Wambaugh et al.
(2019)

15 of 24

Office of Research and Development

-------
&EPA

United States
Environmental Protection
Agency

- In Vitro Disposition -

A Tox21 Cross Partner Project (EPA, NTP, FDA)

An Experimental Evaluation of Mass Balance Models

describing in vitro partitioning and disposition

- Pilot study completed

- 20 chemical case study underway

- Chemical levels quantitated across 5 in vitro compartments

Preliminary Design and Data

Armitage et al. 2014 PMID 25014875

Diagram of in v

'/"fro compartments Head space

DMSO
(if present) ^

Sorption to
vessel wall

OOCX

, st Test medium

Serum constituents
^ (if present)

Dissolved

V OM J
^ Cells/tissue

^OOOOO

Table 2, Plate Matrix

Test Plate

Test Plate Barcode

Plating Condition

Exposure Duration (hr)

Measured Compartment

TC00000013

Medium - cells

Medium

Medium - cells

Plastic

TC00000014

Medium + cells

Medium

Medium + cells

Plastic + Cells

TC00000015

Medium + cells

Whole Well Crash

TC00000016

Medium - cells

Medium

Medium - cells

Plastic

TC00000017

Medium + cells

Medium

Medium + cells

Plastic + Cells

TC00000018

Medium + cells

Whole Well Crash

TC00000019

Medium - cells

Medium

Medium - cells

Plastic

TC00000020

Medium + cells

Medium

Medium + cells

Plastic + Cells

TC00000021

Medium + cells

Whole Well Crash

r is

: 10

Figure 1. Conceptual representation of an in vitro test system.

ROSI RIF OMEP APAP NP14B CARB THIA TPP FlUS ATRZ
¦ Media ¦ Whole Well

Office of Research and Development

-------
*>EPA

United States
Environmental Protection
Agency

Target Consumers
Population,

ToxCast

Providing the

Informing TSCA

General Population

ToxCast

HTTK
Oral
Route

Dermal
Route
Needed

Evaluation Data:
NHANES

Human

ExpoCast/SEEM

Many Exposure
Predictors

HTTK
Oral
Route

Aerosol
Route
Needed

V V!

Evaluation Data:
NHANES

Human

ExpoCast/SEEM

Many Exposure
Predictors

Pathways

Covered Consumer Ambient

Office of Research and Development

Pieces for Prioritization

-------
*>EPA

United States
Environmental Protection
Agency

Target Consumers
Population,

ToxCast

Providing the Pieces for Prioritization

Informing TSCA

Highly Exposed and
Sensitive Populations

General Population

ToxCast

Workers

ToxCast

Gestational

ToxCast

HTTK
Oral
Route

Dermal
Route
Needed

Evaluation Data:
NHANES

Human

ExpoCast/SEEM

Many Exposure
Predictors

HTTK
Oral
Route

Aerosol
Route
Needed

r v

Aerosol
Route
Needed

Gestational
Model
Needed

Evaluation Data:
NHANES

Human

ExpoCast/SEEM

Many Exposure
Predictors

Pathways

Covered Consumer Ambient

Office of Research and Development

Occupational

ExpoCast/SEEM

Evaluation Data:
NHANES

Demographic Human

ExpoCast / SEEM

Many Exposure
Predictors

Occupational

Multiple

-------
*>EPA

United States
Environmental Protection
Agency

Target Consumers
Population,

ToxCast

Providing the Pieces for Prioritization

Informing TSCA

Highly Exposed and
Sensitive Populations

General Population

ToxCast

Workers

ToxCast

HTTK
Oral
Route

Dermal
Route
Needed

HTTK
Oral
Route

Aerosol
Route
Needed

r v

Aerosol
Route
Needed

Evaluation Data:
NHANES

Human

ExpoCast/SEEM

Many Exposure
Predictors

Evaluation Data:
NHANES

Human

ExpoCast/SEEM

Many Exposure
Predictors

Pathways

Covered Consumer Ambient

Office of Research and Development

Occupational

ExpoCast/SEEM

Occupational

Gestational

ToxCast

Gestational
Model
Needed

Multiple

Informing EDSP

Ecological (Fish)

ToxCast +
SeqaPass/
LC50 Models

EPI
Suite
BCF*

HTTK
Fish
Needed

Evaluation Data:
USGS Surface Water

Ecological

ExpoCast/SEEM

Ambient

-------
&EPA

United States
Environmental Protection
Agency

TK and IVIVE Projects and Relationships

CSS Products Supporting Models/Data

Outputs

2.6.4: New Methods/Data

Generic Derma! Model

Address

2.6.5: Exposure Routes

Uncertainty

2.6.6: Life-stage and Sens. Pop.

2.6.7: QSAR Models

2.6.8: In Vitro Distribution

2.6.9: Uncertainty Experiments

2.6.10: Parent-Metabolite

2.6.11: HTTK Fish

2.6.12: HTTK-AOP Model

Office of Research and Development

Generic Aerosol Model

Generic Parent-Metabolite
Model

Generic Human
Gestational

Generic Aquatic
Species Model

TK/TD Model

Challenging
Chemistries

New Exposure
Routes

Sensitive Pop's
and Lifestages

Applications

IVIVE for Gen. Pop.

Risk Workflows
(OPPT,OLEM, MN)

New R Package
"httk" Release

Occupational Risk
IVIVE

Ecological Risk IVIVE

-------
4>EPA

United States
Environmental Protection
Agency

International Collaborations

-Accelerating the Pace of Chemical Risk Assessment (APCRA)

In Vitro Bioactivity, HTI K, and In Vivo toxic Doses

SOT Ssa£.

academic .oup com/toxsci

Utility of In Vitro Bioactivity as a Lower Bound Estimate
of In Vivo Adverse Effect Levels and in Risk-Based
Prioritization

Katie Paul Friedman a ,** Matthew Gagne,' Lit-Hsin Loo,' Panagiotis
Karamertzanis,5 Tatiana Netzeva,5 Tomasz Sobanski,5 Jill A- franzosa,1 Ann
M. Richard," Ryan R Lougee,'- Andrea Gissi,® Jia-YingJoey Lee,' Michelle
Angrish, Jean Lou Dome. Stiven Foster," Kathleen Raffaele,' Tina
Bahadori,1 Maureen R. Gwinn,* Jason Lambert,* Maurice Whelan,*" Mike
Rasenberg,8 Tara Barton-Madaren,' and Russell S. Thomas • *

•National Center lot Computational Toxicology. Office of Research and Development, US Environmental
Protection Afrncy, Research Triangle Part. NC 777\ I. 'Healthy Enveonmens and Consumer Safely Branch,
Health Canada. Government of Canada, Ottawa. Ontario, Canada, KLA0K9. InaowtBia In Food and
Chtm teal Safety Programme and kotnfonrutks Institute, Agetvy for Science. Technology and Research,
Singapore. 138671. Singapore. 'Computational Assessment ttak. European Chemicals Agency. European
Chemicals Agency Annan katu 11. PO Box «a>. n 00121 Hebinki. Uuelmia. Finland , ^National Health and
Enwronmental Effects Seacard! Laboratory, Office of Research and Development, US Environmental
Protection Apncy. Research Triangle Part NC 27711, Oak Ridge Institute for Sdence and £dnotion. US.
Derailment of Eneijy. Oak Ridge, TN 3783J, USA. 'National Center for Environmental Aasewment. Office of
Reseach and Development. US EiivSanmejital Protection Agency, Washington, DC 20001 and Research
Triangle P®k. NC 27711: "SoentjflcCommlttee and Emerging Risb Unit Department of Rarit Assessment and
SdentAc Ataistante. Via Carlo Magno IA.43126 Parm* Raly.'Office of land and Emergency Management.
US- Environmental Protection Aprncy, Waahingjon. OC, 200CH. and "European Ca mmission, Joint Research

Centre 0RQ. Via Enrico Fermi, Z7«, I

mMoacontyiiatain ar

International case study with EPA, ASTAR,
ECHA, Health Canada." and EFSA

ExpoCast PODnam

A f

^ "

(PODTradj^oria| POD

EFSA

r+,m -

PODhc)

••

-A. '

•- ^ /¦
• AA id* A "

r _ 4 u u • ¦

• *2
~' ^ •

•V "if" '

—

-5 -4 -3 -2-10 1
Iog10 mg/kg-bw/day

For ~89% of the
chemicals,
PODnam was
conservative.

(~ 100-fold on
average), but
less conservative
than a TTC

Chemicals where
PODnam was not
conservative
enriched in
OPs/carbamates

21 of 24

Office of Research and Development

Paul-Friedman et al. 2020

-------
vvEPA Additional Efforts and Outreach

United States
Environmental Protection
Agency

Additional Efforts

• In vitro TK data generation: Ongoing, internal (>400 TSCA, incl. 150 PFAS) and external (>215); as needed on
program office-initiated efforts (Office of Chemical Safety and Pollution Prevention, Office of Water)

• In vivo TK: rat in vivo studies for comparative assessments and IVIVE evaluation (Hughes et oi, underway)

• Dermal Route: permeability/partitioning models completed (Evans et oi), integration with HTTK begun

• Bioavailability: incorporation of Caco-2 data in IVIVE (Honda et oi, 2019; Honda et oi, in preparation)

• Transporters: TK renal transporter data generation for PFAS IVIVE modeling (Smeltz et oi, underway)

• Sensitive Populations/Variability: Isozyme-specific chemical evaluations to evaluate TK variability and supply in
silico predictive efforts (Kreutz etoi, underway,/; Correlated Monte Carlo approach to incorporate physiologic
variability (Ring et ai, 2017)

• Parent-Metabolite HTTK: NTA data for metabolism of ToxCast chemicals generated by contractor and being
analyzed (Boyce etol. underway)

Stakeholder Outreach and Collaborations

• CompTox Chemicals Dashboard: Contains ADME data for >1000 chemicals.

• 2020 SOT: "New Data and Tools for Understanding Chemical Distribution In Vitro" - Nynke Kramer and John Wambaugh

• FIFRA SAP "The use of new approach methodologies (NAMs) to derive extrapolation factors and evaluate developmental
neurotoxicity for human health risk assessment" - Incorporation of in vitro TK / HTTK

• Integration of high throughput hazard, exposure, and TK NAMs into proposed TSCA workflows (white paper, peer review)

• APCRA Collaborations - HTTK case study (underway) and NAM prospective case study (underway)

• Ongoing collaborations with Health Canada, US Geological Survey, and MN Department of Health
Office of Research and Development

-------
SEPA

United States
Environmental Protection
Agency

• ArmitageJ.M. etal. "Application of mass balance models and the
chemical activity concept to facilitate the use of in vitro toxicity data
for risk assessment." Environmental Science & Technology (2014) 48,
9770-9779.

• Bell, S.M., et al. "In vitro to in vivo extrapolation for high throughput
prioritization and decision making." Toxicology In Vitro (2018) 47:
213-227.

• Honda, G.S., et al. "Using the concordance of in vitro and in vivo data
to evaluate extrapolation assumptions." PloS One (2019) 14(5):
e0217564.

• Jamei, M., et al. "The Simcyp® population-based ADME simulator."
Expert Opinion on Drug Metabolism & Toxicology (2009) 5(2):211-
223.

• Kapraun, D., et al. "Empirical Models for anatomical and physiological
changes in a human mother and fetus during pregnancy and
gestation." PloS One (2019). 14(5):e0215906.

• Linakis, M. et al. "Development and Evaluation of a high throughput
inhalation model for organic chemicals." Journal of Exposure Science
& Environmental Epidemiology (2020) 30:866-877.

• Lukacova, V. etal. "Prediction of modified release pharmacokinetics
and pharmacodynamics from in vitro, immediate release, and
intravenous data." The AAPSjournal (2009) 11(2): 323-334.

• Mansouri, K., et al. "OPERA models for predicting physicochemical
properties and environmental fate endpoints." Journal of
Cheminformatics (2018). 10(1): 10.

• National Research Council. Risk Assessment in the Federal
Government: Managing the Process Working Papers. (1983). National
Academies Press.

• National Research Council. Issues in Potable Reuse: The viability of
augmenting drinking water supplies with reclaimed water. (1998).
National Academies Press.

References

• Paul-Friedman, K. et al. "Utility of In Vitro Bioactivity as a Lower
Bound Estimate of In Vivo Adverse Effect Levels and in Risk-Based
Prioritization." Toxicological Sciences (2020) 173(l):202-225.

• Pearce, R.G„ et al. "httk: R Package for High-Throughput
Toxicokinetics." Journal of Statistical Software, (2017a) 79(4):l-26.

• Pearce, R.G., et al. "Evaluation and calibration of high-throughput
predictions of chemical distribution to tissues." Journal of
Pharmacokinetics and Pharmacodynamics 44.6 (2017b): 549-565.

• Pradeep, P., et al. "Using chemical structure information to develop
predictive models for in vitro toxicokinetic parameters to inform high-
throughput risk assessment." Computational Toxicology (2020) Epub.

• Ring, C.L, et al. "Identifying populations sensitive to environmental
chemicals by simulating toxicokinetic variability." Environment
International 106 (2017): 105-118.

• Rotroff, D.M., et al. "Incorporating human dosimetry and exposure
into high-throughput in vitro toxicity screening." Toxicological
Sciences (2010) 117(2): 348-358.

• Sayre, R. et al., "Database of pharmacokinetic time-series data and
parameters for 144 environmental chemicals", Scientific Data (2020)
7(1): 1-10.

• Schmitt, W. "General approach for the calculation of tissue to plasma
partition coefficients." Toxicology in Vitro (2008) 22(2): 457-467.

• Shibata, Y., et al. Prediction of hepatic clearance and availability by
cryopreserved human hepatocytes: an application of serum
incubation method. Drug Metabolism and Disposition (2002), 30(8),
892-896

• Sipes, N.S., et al. "An intuitive approach for predicting potential
human health risk with theTox2110k library." Environmental Science
& Technology (2017) 51(18): 10786-10796.

• Sobels, F. H. 'The parallelogram; an indirect approach for the
assessment of genetic risks from chemical mutagens." In: Progress in
Mutation Research, Vol. 3. 1982. (K. C. Bora, G. R. Douglas, and E. R.
Nestman, Eds.), Elsevier, Amsterdam, pp. 233-327.

• Tan, Y.-M., et al. "Reverse dosimetry: interpreting trihalomethanes
biomonitoring data using physiologically based pharmacokinetic
modeling." Journal of Exposure Science & Environmental
Epidemiology (2007) 17(7):591-603.

• Thomas, Russell S., et al. "Incorporating Monte Carlo simulation into
physiologically based pharmacokinetic models using advanced
continuous simulation language (ACSL): a computational
method." Toxicological Sciences (1996) 31(l):19-28.

• Wambaugh, J.F., et al. "Evaluating In Vitro-ln Vivo Extrapolation of
Toxicokinetics." Toxicological Sciences (2018) 163(1): 152-169.

• Wambaugh, J.F., et al. "Assessing Toxicokinetic Uncertainty and
Variability in Risk Prioritization" Toxicological Sciences, 172(2), 235-
251.

• Wang, Y.-H. "Confidence assessment of the Simcyp time-based
approach and a static mathematical model in predicting clinical drug-
drug interactions for mechanism-based CYP3A inhibitors." Drug
Metabolism and Disposition (2010) 38(7):1094-1104.

• Wetmore, B.A., et al. "Integration of dosimetry, exposure and high-
throughput screening data in chemical toxicity

assessment." Toxicological Sciences (2012) 125(1):157-174.

• Wetmore, B.A., et al. "Relative impact of incorporating
pharmacokinetics on predicting in vivo hazard and mode of action
from high-throughput in vitro toxicity assays." Toxicological Sciences
(2013) 132(2) 327-346.

• Wetmore, B.A., etal. "Incorporating high-throughput exposure
predictions with dosimetry-adjusted in vitro bioactivity to inform
chemical toxicity testing." Toxicological Sciences (2015) 148(1): 121-
136.

23 of 24

Office of Research and Development

-------
smi

ExpoCast Project
(Exposure Forecasting)

Center for Computational Toxicology and Exposure

Linda Adams
Lucas Albrecht
Matthew Boyce
Miyuki Breen
Alex Chao

Colin Guider
Mike Hughes
Victoria Hull
Kristin Isaacs
Richard Judson

j/K

Daniel Dawson 1* Jen Korol-Bexell•*

Mike Devito Anna Kreutz;

Alex East Charles Lowe

Lindsay Eddyv Seth Newton

Christopher Eklund Alii Phillips

Peter Egeghy Katherine Phillips

Marina Evarts Tom Purucker

Alex Fisher * Ann Richard

Rocky Goldsmith Caroline Ring
Louis Groff ; "

Chris Grulke

Risa Sayre
Mark Sfeir
Marci Smeltz
Jon Sobus
Zach Stanfield
Mike Tornero-Velez
Rusty Thomas
Elin Ulrich
Dan Vallero
Barbara Wetmore
John Wambaugh
Antony Williams

.mn V

Hongwan Li
Xiaoyu Liu
Zachary Robbins
Mark Strynar

CPHEA
Paul Price

v£J£5

Collaborators

Arnot Research and Consulting
Jon Amot

Johnny Westgate

Integrated La bora
Kamel Mansouri
Xiaoqi

oratory Systems

iaoqing Chang
Shannon Bell

National Toxicology Program

Steve Ferguson

Nisha Sipes

Ramboll

Harvey Clewell

Silent Spring Institute

Robin Dodson

Simulations Plus

Michael Lawless

Southwest Research Institute

Alice Yau

Kristin Favela

Summit Toxicology

Lesa Ay I ward

Technical University of Denmark

Peter Fantke

Unilever

Beate Nicol

Cecilie Rendal

Ian Sorrell

United States Air Force
Heather Pangburn
Matt Linakis

University of California, Davis
Deborah Bennett
University of Michigan
Olivier Jolliet

University of Texas, Arlington
Hyeong-Moo Shin
University of Nevada

U~i ¦ • '

i Li

University of North Carolina, Chapel Hill
Julia Rager
Marc Serre

-------
v>EPA

United States
Environmental Protection
Agency

ExpoCast

exposure forecasting

NAMs for Exposure:
Non-Targeted Analysis

Jon Sobus,

US EPA CSS-HERA Board of Scientific Counselors
Chemical Safety Subcommittee Meeting

February 2-5, 2021

-------
o CDA

United States Stakeholder Needs

(OCSPP; EPA Regions)

Environmental Protection
Agency

Chemical safety evaluations require an improved
understanding of chemical exposure scenarios and pathways

High-priority exposure data needs -> consumer products,
indoor environments, occupational settings,
ambient environments, ecological pathways

2 of 22

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
Challenges

• High-quality exposure data are unavailable for many chemicals

• Measurement data traditionally generated using "targeted" methods

• Targeted analytical methods:

- Require a priori knowledge of chemicals of interest

- Produce data for few selected analytes (lOs-lOOs)

- Require standards for method development & compound quantitation

- Are blind to emerging contaminants

- Can't keep pace with the needs of 21st century chemical safety evaluations

&EPA

United States
Environmental Protection
Agency

3 of 22

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
Research Objective

Environmental Protection w

Agency

Rapid Exposure Modeling and Dosimetry Output 2.7:

Develop, evaluate, and apply non-targeted analysis (NTA)

methods, alongside targeted monitoring methods, to identify
critical sources and pathways of human and ecological exposures

Key Question:

Are NTA methods suitable to meet the needs of
21st century chemical safety evaluations?

4 of 22

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
5 EDA

United States General NTA Workflow Steps

Environmental Protection
Agency

1) Prioritize "molecular features"

2) Correctly assign formulas

3) Correctly assign structures

4) Predict chemical concentrations

5) Determine chemical sources

(4)
12 Mg/mL

TTTTTT~rT~T n 12 « ii is it i: i's A iiiiih a ii % £ oi ii ii 33 si x % i: k k « « « « k

Coats (';) vs. Acqtiisiticn Tune (nw)

Samples

High-
Resolution MS

: A_ Chromatoqram ResJb.

4 a a i &[%]\ a ^

-ESI ElCiSSO 2529) Scan Fras-S0€N V&AjasDataOS d

1 Sample
1 Ionization Mode

300 Extracted "Molecular Features"

5 of 22

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
c CDA

United States Ongoing Research Activities

Environmental Protection
Agency

• Evaluate NTA State-of-the-Science

- EPA's Non-Targeted Analysis Collaborative Trial (ENTACT)

• Develop and Disseminate Guidance Materials

- Benchmarking and Publications for NTA (BP4NTA)

• Build Tools to Ensure Transparency & Reproducibility

- NTA Study Reporting Tool (NTA SRT)

- EPA NTA Web Application (NTA WebApp)

• Address Priority Data Needs with Proof-of-Concept Applications

6 of 22

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
oEPA

United States
Environmental Protection
Agency

Evaluating N' 'A Science-of-the-Science

How variable are tools and results from lab to lab?
Are some methods/workflows better than others?
How does sample complexity affect performance?
What chemical space does a given method cover?
How sensitive are specific instruments/methods?

OQQi
OlCIBI
OQQI

ooo

m Om

II oo

ii o
ii o«

¦ I OB

t OlOOOOOl ikQar
• O IftQH 'O O

Oo
o

ooooo

ommt o

o - o

¦ooo

0 s**"!! I
OOO

OOOOO
ic.oai

EPA's Non-Targeted Analysis

7 of 22

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
*>EPA

United States
Environmental Protection
Agency

ENTACT Study Design (Part I)

30 global participants, 19 results submitted to date
10 synthetic mixtures of ToxCast substances (n=1269)

CO
o

<4—

CD
.Q

400
350
300
250
200
150
100
50
0

¦ 5 NT A method replicates

¦ Grade A - replicate 90 set
Grade A - unique to mix

¦ Grade A - all isobaric set (replicated)

¦ Grades B,C - lower purity mix

Mill

499 500 501

502

503 504

505

506 507 508

Mixture Number

Replication in
substance spikes
offers a unique
means to assess

NTA method
reproducibility!

Ulrich et al. 2019. doi: 10.1007/s00216-018-1435-6

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
oEPA

United States
Environmental Protection
Agency

1.20

1.00

^ 0.80

4—»

0.60

TJ
O

What % observed?
(of those spiked)

X-Axis

What % correct?
(of those observed)

Y-Axis

What % consistent?
(of those correct)

9 of 22

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
*>EPA

Take-Away Messages from ENTACT

(to date...)

United States
Environmental Protection
Agency

• Lack of transparency in methods/results reporting

• Method procedures change over short time increments

• Biased self-reporting highlight strengths, mask weaknesses

• Blinded ToxCast mixtures allow for NTA performance assessment

• Standard performance measures highly variable across labs/methods

• Standard performance assessment methods/benchmarks must be adopted

• Benchmarks require input/consensus from NTA community

• Community focus must be on QA/QC and guidance (and innovation)

10 of 22

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
^pPA Developing and Disseminating

Guidance Materials

Environmental Protection
Agency

BP4NTA Borne out of 2018 ENTACT workshop

~100 U.S. and international members
- Government, academia, and industry

5P4NTA

BENCHMARKING AND PUBLICATIONS
FOR NON-TARGETED ANALYSIS

Working Group Objectives:

- Short term define common NTA terms, concepts, and performance metrics

- Short term provide recommendations on research & reporting best practices

- Long term establish proficiency testing levels (ASTM/ISO)

Products (including 3 manuscripts):

- Website with key resources and links: https://nontargetedanalvsis.org/

- Guidance documents with definitions & supporting info

- NTA Study Reporting Tool" to standardize reporting (proposals & manuscripts)

11 of 22

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
*>EPA

BuildingTools to Ensure

United States
Environmental Protection
Agency

Environmental Protection

Transparency & Reproducibility

The "NTA Study Reporting Tool" (NTA SRT):

• Standardized framework for reviewing quality of NTA reporting

• Aids NTA study design and review (proposals & manuscripts)

• Follows chronology of typical NTA studies with detailed examples

• Scale-based scoring (numeric & colorimetric) for individual study attributes

• HTML interactive version via BP4NTA website (hyperlinks supporting docs.)

• Fillable PDF version available for download (via website)

• Comment box for periodic updates/revisions (via website)

• Working with journal editors for initial testing and deployment

12 of 22

Office of Research and Development

US EPA CSS-HERA BOSC Meeting - February 2-5, 2021

-------
*>EPA

United States
Environmental Protection
Agency

NTA Study Reporting Tool (draft version)

o
c
o

>>
"O
3

CO
.<

Study Sections &

Example Information

Numeric & Rationale/Notes

Colorimetric