Future Directions Workshop: Advancing the Next Scientific

Future Directions

Workshop: Advancing

the Next Scientific

Revolution in

Toxicology

April 28-29,2022

Thomas Hartung, Johns Hopkins University, University of Konstanz,

and Georgetown University

Ana Navas-Acien, Columbia University

Weihsueh A. Chiu, Texas A&M University

Prepared by:

Kate Klemic, Virginia Tech Applied Research Corporation

Matthew Peters, Virginia Tech Applied Research Corporation

Shanni Silberberg, Ofce of the Under Secretary of Defense

(Research & Engineering), Basic Research Ofce

Future Directions Workshop series

Workshop sponsored by the Basic Research Ofce, Ofce of

the Under Secretary of Defense for Research & Engineering

CLEARED

For Open Publication

Department of Defense

OFFICE OF PREPUBLICATION AND SECURITY REVIEW

Apr 26, 2023

Contents

Preface iii

Executive Summary 1

Introduction 4

Toxicology Research Challenges 6

Toxicology Research Advances and Opportunities 8

Toxicology Research Trajectory 15

Accelerating Progress 19

Conclusion 20

Glossary 21

Bibliography 22

Appendix I—Workshop Attendees 27

Appendix II—Workshop Agenda and Prospectus 34

iii

Innovation is the key

to the future, but basic

research is the key to

future innovation.

– Jerome Isaac Friedman,

Nobel Prize Recipient (1990)

Preface

Over the past century, science and technology has brought

remarkable new capabilities to all sectors of the economy,

from telecommunications, energy, and electronics to medicine,

transportation and defense. Technologies that were fantasy

decades ago, such as the internet and mobile devices, now

inform the way we live, work, and interact with our environment.

Key to this technological progress is the capacity of the global

basic research community to create new knowledge and to

develop new insights in science, technology, and engineering.

Understanding the trajectories of this fundamental research,

within the context of global challenges, empowers stakeholders

to identify and seize potential opportunities.

The Future Directions Workshop series, sponsored by the

Basic Research Directorate of the Ofce of the Under Secretary

of Defense for Research and Engineering, seeks to examine

emerging research and engineering areas that are most likely to

transform future technology capabilities. These workshops gather

distinguished academic researchers from around the globe

to engage in an interactive dialogue about the promises and

challenges of each emerging basic research area and how they

could impact future capabilities. Chaired by leaders in the eld,

these workshops encourage unfettered considerations of the

prospects of fundamental science areas from the most talented

minds in the research community.

Reports from the Future Direction Workshop series capture

these discussions and therefore play a vital role in the discussion

of basic research priorities. In each report, participants are

challenged to address the following important questions:

• How will the research impact science and technology

capabilities of the future?

• What is the trajectory of scientic achievement over the next

few decades?

• What are the most fundamental challenges to progress?

This report is the product of a workshop held April 28-29, 2022, at

the Basic Research Innovation Collaboration Center in Arlington,

VA on the future of toxicology research. It is intended as a

resource for the S&T community including the broader federal

funding community, federal laboratories, domestic industrial

base, and academia.

Executive Summary

1 hps://pubmed.ncbi.nlm.nih.gov/20574894/

In the nearly two decades since the human genome

was sequenced, the eld of toxicology has undergone a

transformation, taking advantage of the explosion in biomedical

knowledge and technologies to move from a largely empirical

science aimed at ensuring the absence of harmful effects to a

mechanistic endeavor aimed at elucidating disease etiology

based on an understanding of the biological responses to

chemicals (including biochemistry) and the impact on organ

systems. However, a substantial gap remains between the

promise of mechanistic toxicology and its actual impacts on

improving human health. Toxicology continues to work in a

largely reductionist paradigm of single endpoints, chemicals,

and biological targets, whereas it is known that biology and

pathobiology involve complex interactions across each of these,

with the additional recognition that social stressors also have

biological consequences. At the same time, the pace of scientic

and technical advances has resulted in a deluge of models and

data for understanding toxicological exposure, hazard, and

risk that is increasingly challenging to evaluate, integrate and

interpret. A critical need, therefore, exists to understand how to

leverage these new frontiers in toxicology to achieve the desired

long-term impact of improving human health. This fundamental

problem addresses the question of what exposures, now or

in the future, can contribute to disease and calls for a Human

Exposome Project.

The 2007 National Research Council report on Toxicity Testing

for the 21st Century—a Vision and a Strategy

(Tox-21c) was a

watershed moment for US toxicology, changing the discussion

from whether to change to when and how to change. With

knowledge in the life sciences doubling every seven years since

1980 and every 3.5 years since 2010, as well as publications

doubling every fteen years (Bornmann, 2021; Densen, 2011), we

now have about 16 times as much knowledge and twice as many

publications as in 2007.

The Future Directions in Toxicology Workshop convened on April

28-29, 2022, in Arlington, VA, to examine research challenges

and opportunities to usher toxicology into a new paradigm

as a predictive science. Hosted by the Basic Research Ofce

in the Ofce of the Under Secretary of Defense for Research

and Engineering, this workshop gathered 20 distinguished

researchers from across academia, industry, and government

to discuss how basic research can advance the science of

toxicology. The workshop aimed at the next generation of a

vision for toxicology, “Toxicity Testing for the 21st Century

2.0—Implementation” that extends the vision of the 2007 report

and adapts it to scientic and technological progress. This report

is the product of those discussions, summarizing current research

challenges, opportunities, and the trajectory of toxicological

science for the next twenty years.

The vision developed at the workshop foresees toxicology

developing into a Human Exposome Project that better

integrates the exposure side of disease, focusing on real-

world exposures affecting diverse populations over time. Thus,

changing the principal approach from a hazard-driven to an

exposure-driven paradigm. This new paradigm identies the

relevant human or ecological exposures and then bases the

risk assessment process on exposomics, forming an exposure/

mechanism hypothesis from the multi-omics imprint in biouids

and tissues, and biomonitoring, the large-scale sampling

and measurement of biospecimen. This new paradigm also

incorporates negligible exposures and is focused on ensuring

safety instead of predicting toxicity. Another critical aspect of this

paradigm is the inclusion of disruptive research technologies,

such as microphysiological systems, the bioengineering of

organ architecture and functionality to model (patho-)physiology,

and articial intelligence/machine learning to process the

complex data generated for informed decisions. Ultimately, we

need to integrate the evidence provided by these technologies,

especially through probabilistic risk assessment. An evidence-

based toxicology approach ensures condence and trust in the

process by which scientic evidence is assessed for the safety of

chemicals for human health and the environment.

The workshop was organized around three key areas that are

likely to transform toxicology: 1) employing an exposure-driven

approach, 2) utilizing technology-enabled techniques, and 3)

embracing broad-scale evidence integration. These three key

areas are expected to have a huge impact on the development

of three key long-term public health goals: 1) Precision Health,

2) Targeted Public Health Interventions and Environmental

Regulations and 3) Safer Drugs and Chemicals, through their

distinct perspectives and long-term goals.

Exposure-driven Toxicology

Exposure-driven assessments were not covered in the 2007

Tox-21c report and were only the subject of a parallel NRC

report, but the needs for integration into toxicology, for example

through exposomics, are increasingly evident. Exposure-

driven toxicology, focused on real-world exposures and gene-

environment interactions that affect diverse populations can

contribute to addressing the three aims identied during the

workshop: 1) precision health through the identication of

environmental exposures for improved health outcomes in

specic populations, 2) targeted public health interventions and

environmental regulations to address those environmentally-

driven health outcomes and 3) the identication of safer drugs

and chemicals. Precision health aims for individual, personalized

preventive interventions, and pharmaceutical and non

pharmaceutical therapies. Targeted public health Interventions

and environmental regulations must address population and

spatial-temporal variability in genome and epigenome, as well as

exposome. Safer drugs and chemicals shall be attained through

in vitro/in silico chemical screening, in vitro/in silico clinical trials

and identifying intrinsic and extrinsic susceptibilities.

Workshop participants envision a Tox-21c 2.0 that reects real-

world based exposure designs (in silico, cellular, organoids,

models, organisms, longitudinal epidemiological studies). It will

include population-scale measurements that are based on readily

available biobanks and ecobanks that inform on the distribution

of thousands of chemical and non-chemical stressors in relevant

populations (general population, relevant subgroups, disease

cohorts). Noteworthy, the exposomics approach potentially

can interrogate all types of stressors, not just chemicals, that

actually perturb biology and change biomarkers in body uids.

Study designs and computational approaches will be aligned to

provide interpretable and actionable results. Ethical issues, policy

implications, community engagement, and citizen participation

will keep pace with and inform the technology, rather than being

exclusively reactive to the technology. In the near-term, a critical

rst implementation step for exposure-driven toxicology and

precision health is to scale-up mass spectrometry technology for

high-quality inexpensive assessment of thousands of chemicals

that can be tagged to exogenous exposures including non-

chemical stressors. Libraries that tag key information for those

chemicals (metadata layering) will need to be expanded and

developed to facilitate interpretation, and to guide preventive

strategies, interventions, and policy recommendations. In the

mid-term, technologies will be required that link the exposome

with health outcomes, and leverage longitudinal studies and

biobanks retrospectively and prospectively, ensuring “FAIR”-ness

(Findability, Accessibility, Interoperability, and Reuse of digital

assets)

. In the long-term (20 years), we envision that exposome-

disease predictions and exposome-targeted prevention, and

treatment solutions will become part of the toxicology and

public health practice landscape, leveraging also other ~omics

technologies, genomic information, and clinical characteristics.

Technology-enabled Toxicology

The workshop participants discussed technological advances

over the last 10-15 years of great relevance to toxicology in

three key areas: cell and tissue biology, bioengineering, and

computational methods. Workshop participants noted that

while biological technologies, such as stem cell engineering,

have emerged as routine, commercial enterprises for

biomedical research, their potential in toxicology could be

further expanded through a) reliable and genetically-diverse

cell sourcing, b) improved protocols to differentiate patient-

derived stem cells into adult cell phenotypes across essential

tissues, c) integrative and non-invasive biomarkers, d) integration

of dynamic physiology and pathophysiology outcomes, e)

population heterogeneity and susceptibility through life-courses,

and f) biological surrogates for non-chemical stressors. On

the bioengineering side, workshop participants noted that

technological capabilities, such as microphysiological systems

(MPS), have shown many successes in the laboratory but need

to be further developed to 1) include a variety of models of

increasing architectural complexity (monolayer/suspension

cultures, organoids and multi-organoid systems) for different

stages of drug/chemical development, 2) better represent

healthy and diseased populations by a personalized multiverse

2 hps://www.go-fair.org/fair-principles/

of possible futures, 3) codify platform standardization, 4)

increase throughput, 5) demonstrate validation against in vivo

outcomes, 6) incorporate perfusion and biosensors with near

real-time outputs, and 7) develop automated fabrication. In

addition, the workshop participants noted that the emergence

of “big data” and “big compute” has revolutionized much of

biology, through the ability to analyze and interpret complex and

multi-dimensional information. Computational capabilities and

models are of utmost importance for toxicology, serving as the

key enabling technology. For instance, AI/Machine Learning has

emerged as a key technology to support data mining, predictive

modeling, hypothesis generation, and evidence interpretation

(e.g., explainable AI). Data acquisition and data-sharing following

the FAIR principles is key to unleashing these opportunities. The

emergence of these needs in toxicology necessitates widespread

use and understanding of these technologies combining

them with expert knowledge to yield augmented intelligence

workows. Moreover, given the quantity of information generated

and consumed by these new technologies, the workshop

participants agreed that there is a need for comparable,

compatible, integrable multi-omic databases, quantitative in vitro

to in vivo extrapolation, and the development of in silico “digital

twins” of in vitro and in vivo systems.

Evidence-integrated Toxicology

Workshop participants discussed the key challenge of integrating

data and methods (evidence streams) in test strategies,

systematic reviews, and risk assessments. They agree that

evidence-based toxicology and probabilistic risk assessments are

emerging solutions to this challenge. Evidence integration across

evidence streams (epidemiological, animal toxicology, in vitro, in

silico, non-chemical stressors, etc.) is expected to play a key role

in translating evidence into knowledge that can inform decision-

making. The group developed a vision to conduct complex

rapid/real-time evidence integration by combining advancements

made in data-sharing, and application of articial intelligence

(e.g., natural language processing), with the transparency

and rigor of systematic reviews. To implement this vision, the

workshop participants identied a need for collaborative, open

platform(s) to transparently collect, process, share, and interpret

data, information, and knowledge on chemical and non-chemical

stressors. Creating these platforms is foundational for rapid and

real-time evidence integration and will empower all steps of

protection of human health and the environment. Several needs

were identied to create this platform: 1) software development

to create dynamic and accessible interfaces, 2) denitive

standards and key data elements to facilitate analysis of meta-

data and automated annotation, and 3) consideration for quality

control.

In conclusion, the workshop advocates a paradigm shift

to “Toxicology 2.0” based on the evidence integration of

emerging disruptive technologies, especially exposomics,

microphysiological systems, and machine learning. To date,

exposure considerations typically follow the identication of a

hazard. Future Tox-21c 2.0 must be guided by the identication

of relevant exposures through exposomics. The adaptation to

technical progress, especially microphysiological systems and AI,

requires harmonization of reporting and quality assurance. The

key challenge lies in the integration of these different evidence

streams. evidence-based medicine can serve as a role model with

systematic reviews, dened data search strategies, inclusion and

exclusion criteria, risk-of-bias analysis, meta-analysis, and other

evidence synthesis approaches. While this is mostly applicable

to existing data and studies, a new challenge is the prospective

application for the composition of test strategies (Integrated

Test Strategies—ITS, Integrated Approached to Testing and

Assessment—IATA, and Dened Approaches—DA). A key role for

Probabilistic Risk Assessment was also identied. The participants

also emphasized the need to ensure validation of these new

approaches, as well as expand training, communication, and

outreach. Ultimately, it calls for expanding the approach to a

Human Exposome Project.

Introduction

In 1983, the US National Research Council (NRC) Committee

on the Institutional Means for Assessment of Risks to Public

Health published a foundational study titled “Risk Assessment

in the Federal Government: Managing the Process” (NRC,

1983), commonly referred to as the Redbook. As Lynn

Goldman put it, the Redbook “has created a framework for

incorporation of toxicology into environmental decision-making

that has withstood the test of time” (Goldman, 2003). It was

complemented by the 2009 report “Science and Decisions:

Advancing Risk Assessment,” aka the Silverbook (NASEM,

2009), which highlighted some challenges in the process.

Parallel work by another NRC committee resulted in the 2007

NRC report on “Toxicity Testing in the 21st Century” (NRC,

2007), or Tox 21c, which developed “a vision and a strategy” to

transform toxicological sciences. The gap analysis of the report

has not really changed, as toxicology is still “time-consuming

and resource-intensive, it has had difculty in meeting many

challenges encountered today, such as evaluating various life

stages, numerous health outcomes, and large numbers of

untested chemicals” and needs “to use fewer animals and

cause minimal suffering in the animals used”. The 2007 report

was complemented by the NRC reports “Exposure Science in

the Twenty-rst Century: A Vision and a Strategy” (NRC, 2012)

and NASEM (2017a) “Using 21st Century Science to Improve

Risk-Related Evaluations.” The Tox-21c and subsequent

reports have changed the debate about safety and risk

assessment of substances in the US and beyond, and led to a

remarkable number of initiatives and programs (Krewski 2020).

The resulting diversity in approaches combined with an ever-

accelerating availability of disruptive technologies calls for a

re-conceptualization of the future of toxicology. The way forward

is to dissolve the dichotomy of hazard and exposure sciences,

embrace the disruptive technological advances, and foster

evidence integration from these evidence streams.

In the nearly two decades since the human genome

was sequenced, the eld of toxicology has undergone a

transformation, taking advantage of the explosion in biomedical

knowledge and technologies to move from a largely empirical

science aimed at ensuring the absence of harmful effects to a

mechanistic endeavor aimed at elucidating disease etiology

and biological response pathways induced by exposures.

However, a substantial gap remains between the promise of

mechanistic toxicology and the actualization of the eld as a

predictive science. For instance, high-throughput in vitro and

in silico toxicity testing remains largely focused on prioritization

of individual chemicals for future investigation allowing to

focus limited resources on the one hand, but which may on the

other hand provide a false sense of safety for “de-prioritized”

chemicals. Specically, these efforts, as well as those aimed at

translating such data into hazard or risk have been hampered

by inadequate coverage of important biological targets given

the limitations of current in vitro methods to simulate in vivo

metabolism or predict effects in different tissues and across

different life stages (Ginsberg 2019), inadequate consideration

of population heterogeneity, and aiming still to provide

assurances of safety rather than quantication of effects across

the population. Furthermore, there has been little progress in

understanding the complex interactions among chemicals and

between chemicals and other intrinsic and extrinsic factors that

affect population health, such as genetics and non-chemical

stressors, including marginalization and other social determinants

of health.

In practice, toxicology largely remains a process based

on reductionist paradigm (Figure 1, left side), classifying

individual chemicals for individual hazards, and investigating

simplistically “linear” mechanistic pathways based on

individual biological targets. Although signicant research and

development investment has been made in improving the

throughput of toxicology through the advent of in vitro and in

silico technologies, the vast majority of these efforts to make

toxicity testing faster, cheaper, and perhaps more relevant are

still fundamentally “one at a time” approaches that feed into

“one at a time” risk assessments and ultimately “one at a time”

decisions. Thus, they ultimately only address a narrow slice of the

human-relevant experiences of toxicity, where 1) all exposures

are time-dependent mixtures of chemical and non-chemical

stressors, 2) every individual has unique susceptibilities and

baseline conditions, and 3) multi-factoral, multi-causal outcomes

are the norm (Figure 1).

This report advocates for a fundamental shift to a holistic

paradigm where toxicology embraces complexity rather than

sweeping it under the rug. Against this backdrop, the workshop

was organized around three main research areas (Figure 2) that

are key to enabling this paradigm shift.

First, whereas both traditional mammalian toxicity testing, and

high-throughput screening assays largely focus on one chemical/

mechanism/outcome at a time, this paradigm shift envisions

toxicology to be exposure-driven, addressing real-life exposure

scenarios in which multiple agents, including social determinants

of health, work together to affect multiple mechanistic pathways

and health outcomes. Additionally, this paradigm shift requires

replacing individual assays that are genetically/epigenetically/

exposomically homogeneous with multiplexed systems that

incorporate inter-cell/tissue interactions on a backdrop of

population variability. Thus, toxicology will become Technology-

enabled, leveraging technological advances from genetics to

bioengineering to enable the characterization of toxicity in

integrated in vitro/in silico platforms across the landscape of

genomics, epigenomics, life-stage, and non-chemical stressors.

Finally, with respect to risk, this paradigm shift requires moving

away from single study-based binary (safe/unsafe) decision-

making to integrating diverse data across multiple data streams

to reach a probabilistic assessment (Maertens, 2022) across

multiple outcomes across the population. Thus, especially with

the emergence of “big data” along with “big compute,” toxicity

will be evidence-integrated, combining multiple evidence

streams across diverse sources of structured and unstructured

information.

The rest of this report summarizes the discussion from the

workshop relating to research challenges, research opportunities,

and the ultimate trajectory to achieve the vision of a holistic,

predictive toxicology.

Figure 1 The proposed paradigm shift in Toxicology research.

Figure 2 The three workshop topics and the expected long-term impacts.

Enabling characterization of toxicity across

genetics, life stage, and non-chemical stressors

and pathobiology of intermediate states,

perturbations, and outcomes while increasing

accuracy, precision, relevance, and domains

of applicability

Frontiers of

Toxicology

Exposure-

driven

Technology-

enabled

Evidence-

integrated

Driven by real-life exposure scenarios and how

multiple agents work together to affect multiple

mechanistic pathways and health outcomes.

Long-Term Impacts

•Safer Chemicals and Drugs through

in vitro/in silico chemical screening,

in vitro/in silico clinical trials,

identification of intrinsic and

extrinsic susceptibilities

•Precision Health through individual,

personalized preventive

interventions, pharmaceutical and

non-pharmaceutical therapies

•Targeted Public Health Interventions

& Environmental Regulations

addressing population & spatial-temporal

variability in genome, epigenome, and

exposome, as well as their socio-economic

consequences

Integrating across diverse sources of structured

and unstructured information with enhanced

access, management, evaluation, and

communication.

Current Reductionist Paradigm

•One chemical at a time

•One endpoint at a time

•One biological target at a time

•Single genetic background tested

•Straight, linear mechanistic pathways

•Interactions simplified or ignored

•Safe vs. unsafe dichotomy

A New Holistic Paradigm

• Directly addressing interactions among

▪ Chemicals

▪ Non-chemical stressors

▪ Heterogeneous populations

▪ Social determinants of health

▪ Life stages, including developmental

origins of disease

•Mechanisms integrated into the complex

physiological networks

•Quantifying probabilistically the impacts on

incidence and severity of human disease

Human-Relevant

Experiences of

Toxicity

A Proposed Paradigm Shift:

Embracing the Multi-Factorial, Multi-Casual Nature of Toxicity

Toxicology Research Challenges

For many decades the discussion of changing toxicological

processes was driven by ethical issues of animal use and the

desire to develop and validate so-called “Alternative Methods.”

In the last two decades, it has become increasingly clear that

there are many more reasons to rethink the toolbox of risk

sciences (Hartung, 2017a), namely:

• Long duration and low throughput do not match testing or

public health needs (Hartung and Rovida, 2009; Meigs, 2018)

• Uncertainty in extrapolating results to humans (NASEM, 1983

[the “Red Book”], 1994, 2009 [Science and Decisions])

• Only single chemical/endpoint at a time; does not account

for multiple exposures and non-chemical exposures,

including social determinants (Jerez and Tsatsakis, 2016;

Bopp, 2019; NASEM, 2009 [Science and Decisions])

• Does not account for inter-individual variability (NAS, 2016)

• Does not incorporate associated socio-economic costs and

benets (Chiu, 2017; Meigs, 2018; NASEM, 2009 [Science

and Decisions])

The ongoing transition in terminology in the eld from

“Alternative Methods” to “New Approach Methods” reects

this broader motivation for change. Tox-21c embraced these

challenges and developed a framework of an essentially

mechanistic toxicology of perturbed pathways combined with

quantitative in vitro-to-in vivo extrapolation to human exposure

(Hartung, 2018). A roadmap of consequential steps was

suggested (Hartung, 2009a; Hartung, 2009b).

Workshop participants discussed these overarching challenges

to predictive toxicology and dened the key challenges for

each area as:

Exposure-driven Toxicology

Populations are exposed to multiple environmental agents,

including chemical agents through air, water, food, soil, and

non-chemical agents such as noise, light, and social stressors

(e.g., racism, socioeconomic deprivation, climate). Therefore,

toxicological research that embraces an exposure-driven

approach, characterizing real-life exposure scenarios, including

exposure mixtures and how these agents work together affecting

multiple mechanistic pathways and health outcomes, is needed. A

key opportunity is the expansion of exposomic approaches (Sille,

2020; Huang, 2018; Escher, 2020) to include this broader landscape

of exposures. The workshop participants highlighted three primary

challenges to achieving a more exposure-driven approach:

• Real-world exposures: understanding the interplay

of environmental and social stressors with genetic and

molecular variants

• Predictive intervention: understanding the contributions

of this research toward the identication and evaluation of

effective interventions

• Targeted populations: the inclusion of the affected

communities through participatory research efforts

There are several reasons why an exposure-driven approach

has not yet been embraced. First, many relevant exposures

are not yet fully characterized as we lack the tools and

technologies needed to characterize these exposures, as well

as to understand the health implications. In addition, there has

not yet been a successful engagement of the key stakeholders,

foremost the populations that are directly affected by these

exposures, that is needed for the success of preventive

interventions. However, there are currently substantial advances

coming in these areas and we can easily anticipate substantial

progress in the years to come.

Technology-enabled Toxicology

Predictive toxicology requires expanding the “toolbox” in

several directions. The workshop participants identied the key

challenges to developing the toolbox as:

• Broader model systems: As adverse outcomes involve

interactions of the environment (see above), genes, and

life stage, we need our “model systems” to cover “gene”

and “life stage” more broadly than currently possible using

traditional animal studies (e.g., typically inbred strains) or

even most current high-throughput testing assays (e.g.,

typically based on genetically homogeneous immortalized

cell lines). Example technologies include genetically

diverse population-based in vitro and in vivo resources, and

expansion of experimental designs to cover different stages

of development, as well as developmental origins of health

and disease.

• Access to the intermediate state: Additionally, our

approaches currently cluster at the beginning (e.g.,

high-throughput assays) and the end (e.g., in vivo apical

endpoints) of the pathophysiological process, neglecting

the modulating and stochastic factors that inuence

outcomes that lie between. Thus, approaches that provide

access to intermediate states, perturbations, and outcomes

are needed to better understand the progression to

disease. Example technologies include novel biomarkers,

microphysiological systems (MPS, encompassing organoid

and organ-on-chip technologies), and in silico models (e.g.,

systems toxicology/virtual experiments, AI/ML).

• Assessment tools: We lack the ability to characterize

the predictive accuracy, precision, and relevance of new

approaches or to understand their domains of applicability.

Evidence-integrated Toxicology

Toxicology is currently transitioning from a data-poor to a

data-rich science with the curation of legacy databases, “grey”

information on the internet, mining of scientic literature, sensor

technologies, -omics, robotized testing, high-content imaging,

and others. The workshop participants identied the key

challenges to evidence-integrated toxicology as:

• Information sources: There are no established methods

or consensus on how to handle new types of information

sources (which may be incomplete) or how to weigh

evidence strength, risk of bias, quality scoring, etc., or how

to integrate the evidence streams.

• Validation/Verication: In the case of probabilistic risk

assessment, sources of evidence are already integrated,

resulting in a more holistic probability of risk/hazard, so the

challenge is to determine how to validate real-life, t for

purpose, ground-truthing, qualication, and triangulation,

and communicate these probabilities.

• Data Science: We have not yet adopted best practices

for data curation and storage, data mining, analysis, and

visualization.

Toxicology Research Advances and Opportunities

The workshop participants anticipate exciting new research

advances on the path to achieving the vision of a holistic,

predictive toxicology that addresses real-life exposure scenarios,

leverages technological advances, and integrates multiple

evidence streams across diverse sources. This section presents

those advances and opportunities according to the three

workshop themes.

Exposure-driven Toxicology

The risks of developing chronic diseases are attributed to

both genetic and environmental factors, e.g., 40% of 560

diseases studied had a genetic component (Lakhani, 2019)

while 70 to 90% of disease risks are probably due to differences

in environments (Rappaport and Smith, 2010). The Human

Genome has been at the center of medical research for the last

forty years, but not many major diseases can be explained or

treated as a result.

Understanding exposure effects and genome x exposure

(GxE) interactions are thus central to the future of medicine.

The original concept of the exposome (Wild, 2016; Vermeulen,

2020), encompassing all exposures of an individual over time,

seems to be impractical and unfeasible as a goal (Figure 3).

The National Academies of Sciences report (NRC, 2010) has

even elaborated on this concept. The 180 million synthesized

chemicals, 350,000 of which are registered for marketing in the

19 most developed countries (Wang, 2020) and myriad natural

and breakdown products seem to make it impossible to measure

and study their effects on humans and the environment. Current

approaches in cells or animals can cost from several thousand to

a million dollars per substance and health effect (Meigs, 2018).

Worldwide toxicity testing covers only a few hundred substances

comprehensively and costs about $20 billion per year. In addition,

human and ecological exposure to substances does not occur

in isolation of single substances or in any constant exposure

scheme. To understand it all or at least a lot of it seems like an

impossible mission.

This has led to a hazard-driven approach to toxicology, i.e., an

established hazard is followed up with exposure considerations

to assess risk. The thresholds of toxicological concern (TTC)

(Hartung, 2017b) concept has been introduced to make

pragmatic use of this by establishing the fth percentile of

lowest-observed (LOEL) or no-observed effect levels (NOEL)

and adding a safety factor of 100. This essentially sets a limit

of possible toxicity at one-hundredth of the point of departure

below 95% of relevant chemicals. For instance, there is a

potential role for TTC to abrogate risk assessment where

exposure and/or bioavailability (internal TTC) (Hartung and

Leist, 2008; Partosch, 2015) are negligible (Wambaugh, 2015),

thus showing a path for how substances could be triaged

according to their negligible exposure. However, TTC-type

approaches are still a “one chemical at a time” paradigm, may

not account for exposures varying temporally or across the

population, and do not address potential interactions among

the thousands of substances to which people are constantly

exposed.

In recent years, the concept of the exposome has been

proposed to capture the diversity and range of environmental

exposures (e.g., inorganic and organic chemicals, dietary

constituents, psychosocial stressors, physical factors), as well

as their corresponding biological responses (Vermeulen,

2020). While measuring those exposures throughout the

lifespan is challenging, technology-enabled advances, such

as high-resolution mass spectrometry, network science, and

numerous other tools provide promise that great advances in

the characterization of the exposome are possible. Indeed,

the exposome approach has achieved traction in recent times

because of the availability of -omics technologies (Sille, 2020)

as discussed below. A NIEHS workshop (Dennis, 2017) saw the

following advantages of an exposome approach:

• Agnostic approaches are encouraged for detection of

emerging exposures of concern

• Techniques, and development of techniques promote

identication of unknown/emerging exposures of concern

• Links exogenous exposures to internal biochemical

perturbations

• Many features can be detected (> 10,000) for the cost of a

single traditional biomonitoring analysis.

• Includes biomolecular reaction products (e.g., protein

adducts, DNA adducts) for which traditional biomonitoring

measurements are often lacking or cumbersome

• Requires a small amount of biological specimen (~100 μL or

less) for full-suite analysis

Figure 3 The exposome concept. [Adapted from Vermeulen, 2020].

• Enables detection of “features” that are linked to exposure

or disease for further conrmation

• Encourages techniques to capture short-lived chemicals

• Aims to measure biologically meaningful lifetime exposures,

both exogenous and endogenous, of health relevance

A number of research studies have started to apply these targeted

and untargeted technologies to characterize those complex

exposures and how they impact health and disease, and which

relevant pathways are affected. For instance, birth cohort studies

are attempting to characterize those complex and cumulative

exposures during critical windows that are of increased importance

for long-term human health (Figure 4). Those exposures are not

limited to chemical exposures and consider non-chemical stressors

throughout the lifespan. The concept of cumulative exposures is

critical, as some communities are disproportionally exposed to a

cumulation of chemical and non-chemical exposures which over

time can result in adverse health outcomes.

Figure 4 The early life exposome. Examples of relevant exposures and

their exposure patterns during pregnancy and childhood, including

1) persistent organic pollutants (POPs), 2) mercury, lead, 3) arsenic, 4)

secondhand smoke, 5) air pollution, noise, 6) UV radiation, seasonal

exposure to chemicals, 7) non-persistent pollutants. Other exposures

such as psychosocial stressors could follow different exposure patterns.

[Adapted from Robinson, 2015].

These laboratory and exposure sciences advances also require,

in parallel, advancement in biostatistics and data science, to

maximize the information that can be obtained from those

high-dimensional data. For instance, elastic-net regularization

regression is becoming a popular machine learning tool that can

be used to identify the relevant predictors from these complex

sets of exposure data. For instance, these high-dimensional

models were of great relevance to identifying key factors

associated with endogenous intermediate pathways (e.g.,

inammation, protein damage, oxidative stress, and others) in

a pregnancy cohort from Massachusetts called the LIFECODES

cohort (Aung, 2021). These types of cohort studies with complex

exposure data, in diverse populations, prospective follow-up, and

high-quality health outcome data will continue to grow and will

become key tools to advance exposure-driven toxicology.

Technology-enabled Toxicology

The near exponential growth in biotechnology and

bioengineering over the last few decades has created numerous

technologies that could be applied or leveraged in toxicology.

Here we highlight three complementary technology areas that

have the potential to vastly increase the coverage, biological

relevance, and depth of data available for assessing the human

health effects of chemical exposures.

Induced Pluripotent Stem Cell Technologies

The discovery that somatic cells can be reprogrammed

to become pluripotent, recognized by the Nobel Prize in

Physiology or Medicine in 2012, has led to a vast array of

advances in biomedical science, from basic cell biology to

regenerative medicine. Thus, induced pluripotent stem cells

(iPSCs) are among the most substantial research advances of

the 21st century, on par with the sequencing of the human

genome, with thousands of publications per year utilizing this

technology (Figure 5).

Figure 5 Annual growth in publications for “induced pluripotent stem

cells” query in PUBMED, as of July 2022.

In principle, such technologies would enable one to generate

unlimited cells and tissues that retain the genetic information

of the original donor. Cardiomyocytes were one of the rst

functional cell types to be successfully differentiated from iPSCs

and have gone from research lab to commercial application

preclinical safety evaluation of xenobiotics in less than a decade

(Burnett, 2021). They have been found to be useful in identifying

cardiotoxicity hazards for both drugs and environmental

chemicals and are key components of a broad FDA-led initiative

2010 2015 2020

Year of Year

500

1000

1500

2000

2500

3000

Count

Annual Pubmed Hits for

"Induced Pluripotent Stem Cells"

(CiPA)

to address drug-induced arrhythmias, see Figure 6.

However, even for this relatively “mature” technology of

iPSC-derived cardiomyocytes, a number of limitations remain,

including their expressing a more fetal-like phenotype and

challenges in routine and reproducible differentiation from

individual patients. These challenges are even more pronounced

for other cell types, as discussed below.

Nonetheless, the potential for iPSCs to revolutionize biomedical

science overall, and toxicology in particular, is well recognized,

especially when coupled with the rapid development of

advanced in vitro and microphysiological technology.

In vitro and Microphysiological Systems (MPS) Technologies

As discussed in “Toxicology Research Challenges”, there is an

increasing recognition that meeting the needs of toxicology will

require expanding beyond the use of traditional preclinical in

vivo rodent models. These technologies have been termed “New

Approach Methods” (Environmental Protection Agency, European

Chemicals Agency) or "Alternative Methods" (Food and Drug

Administration), and all have the aim of increasing the rigor and

predictivity of toxicity assessments while reducing the reliance

on vertebrate models. Much of the progress in the last 15 years

has been on high-throughput in vitro systems, exemplied by the

Tox21 Consortium

, which is a federal collaboration among U.S.

Environmental Protection Agency, National Toxicology Program,

National Center for Advancing Translational Sciences, and the

Food and Drug Administration focusing on “driving the evolution

of Toxicology in the 21st Century by developing methods to

rapidly and efciently evaluate the safety of commercial chemicals,

pesticides, food additives/contaminants, and medical products.”

This effort made use of commercially available assay platforms

across a wide range of targets, testing almost 10,000 compounds.

The screening data generated across a wide diversity of chemicals

and potential mechanisms of toxicity has resulted in hundreds of

publications, with many lessons learned as to the opportunities

and challenges in high-throughput screening data (Richard, 2021).

3 hps://cipaproject.org

4 hps://tox21.gov

5 hps://www.fda.gov/science-research/about-science-research-fda/advancing-alternave-methods-fda

6 hps://mpsworldsummit.com

The workshop participants agreed that more advanced in

vitro technologies, in particular microphysiological systems

(Marx, 2016; Marx, 2020; Roth, 2022), represent the next great

opportunity to advance toxicology (NASEM, 2021). An MPS

model has been dened as one that “uses microscale cell culture

platform for in vitro modeling of functional features of a specic

tissue or organ of human or animal origin by exposing cells

to a microenvironment that mimics the physiological aspects

important for their function or pathophysiological condition.”

These may include a wide variety of types of platforms, from

mono-cultures to co-cultures and organoids, and also include

so-called “organ-on-chip” models that include an engineered

physiological micro-environment with functional tissue units

aimed at modeling organ-level responses. These “chip” models

consist of four key components:

• microuidics to deliver target cells, culture uid, waste

discharge

• living cell tissues in either 2D or 3D, including

scaffolding, physical, or chemical signals to simulate the

microenvironment physiologically

• a system for delivering the drug or chemical, either through

the same as the microuidics delivering culture uid, or via a

separate channel (e.g., air-liquid interface)

• a sensing component that may be embedded (e.g.,

electrodes), visual (via transparent materials), or assayed

from efuent

The mushrooming of MPS models has been fueled by stem

cell technologies, 3D cultures (Alepee, 2014), microuidics

(Bhatia and Ingber, 2014), sensor technologies (Clarke, 2021),

bioprinting (Fetah, 2019) and others. Figure 7 shows different

ways of producing 3D cultures, which are key to creating organ

architecture and functionality as key features of MPS. Notably, the

MPS eld has most recently started to organize itself by annual

global meetings and an International MPS Society.

2006 2007 2011 2016 2017 2018 2019

•Generation of iPSCs

from mouse fibroblasts

•Use of patient-specific iPSC-

CMs for disease modeling

•Use of iPSC-CMs for drug

toxicity testing

•Generation of iPSCs from human fibroblasts

•Generation of IPSC-derived

CMs from human fibroblasts

•Use of iPSC-CMs from

patients to recapitulate

clinical toxicity of a drug

•Use of iPSC-CMs for

non-drug toxicity testing

•Use of a healthy population of

iPSC-CMs for drug toxicity testing

•Use of a healthy population of

iPSC-CMs for environmental

chemical toxicity testing

Figure 6 Developmental timeline of induced pluripotent stem cells derived cardiomyocytes (iPSC-CMs) for toxicity testing. [Source: Burnett, 2021]

Figure 7 Ways to generate 3D cultures.

MPS models have been developed for nearly every human organ,

and several have been linked together into multi-organ platforms

(Hargrove-Grimes, 2021) (Figure 8).

Figure 8 “Man-on-a-chip.” [Source: Materne, et al., 2013]

Moreover, applications have been reported in drug development,

disease modeling, personalized medicine, and assessment of

environmental toxicants. However, many translational challenges

remain that hinder the application of MPS in toxicology

(Andersen, 2014; Watson, 2017; Nitsche, 2022). Efforts continue

to improve external validation, reproducibility, and quality

control and to enable technology transfer. Noteworthy, Good

Cell and Tissue Culture Practice (GCCP 2.0, Pamies et al., 2022)

has expanded these standards to MPS. Overall, the throughput

remains low, and the cost remains high, hampering broader

application of these technologies, particularly as benchmarking

against simpler in vitro systems has not always revealed

sufcient improvements to warrant the additional time, cost, and

complexity. Nonetheless, emerging efforts to dene appropriate

“context of use” cases for MPS are promising through the

continued interactions among researchers, regulators, and the

private sector (Hargrove-Grimes, 2021; NAS, 2021).

Imaging and Other High-content Measurement Technologies

The high complexity of MPS and consequential lower throughput

make them an ideal match to high-content measurement

technologies, which provide through a comprehensive analysis

of the biological system maximum insight into the Adverse

Outcome Pathway (AOP) in play.

High-content imaging (HCI) combines automated microscopy

with image analysis approaches to simultaneously quantify

multiple phenotypic and/or functional parameters in biological

systems. The technology has become an important tool in the

elds of toxicological sciences and drug discovery because it

can be used for mode-of-action identication, determination

of hazard potency, and the discovery of toxicity targets and

biomarkers (van Vliet, 2014). In contrast to conventional

biochemical endpoints, HCI provides insight into the spatial

distribution and dynamics of responses in biological systems.

This allows the identication of signaling pathways underlying

cell defense, adaptation, toxicity, and death. Therefore, high

content imaging is considered a promising technology to

address the challenges for the Tox-21c approach. Currently, HCI

technologies are frequently applied in academia for mechanistic

toxicity studies and in pharmaceutical industry for the ranking

and selection of lead drug compounds or to identify/conrm

mechanisms underlying effects observed in vivo.

Several ~omics technologies such as genomics, transcriptomics,

proteomics, metabolomics, lipidomics etc. represent further high-

content technologies allowing deep phenotypic characterization

and mechanistic analysis. Hartung and McBride (2011) suggested

earlier the use of combined orthogonal ~omics technologies to

map pathways of toxicity (PoT) (Kleensang, 2014). Noteworthy,

the PoT concept is reminiscent of the AOP approach, which were

both proposed independently in 2011. However, there are some

fundamental differences (Hartung, 2017c): AOP are designed by

experts largely based on their understanding and review of the

literature; they are for this reason very much biased by current

knowledge/belief and typically not quantitative and difcult to

validate experimentally. AOP are narrative, low level of detail,

and largely a linear series of events. PoT, in contrast, are deduced

from experimental data, especially pathway analysis from

untargeted ~omics technologies. PoT are dened on molecular

level with high level of detail, integrating emerging information,

mainly describing network perturbation. They can be studied

further by interventions in the experimental system and often

allow quantitative description. This process is not free of biases

either and the most promising combination of different omics

technologies is still early in development.

Brain Slices

Re-Aggregating Cultures

Colonization of Scaffold

Layered Co-Culture

Hanging Drop

Transwell Culture

Bioprinting

Evidence-integrated Toxicology

Toxicology is at the intersection of application and basic science

serving as an integrator of health sciences and public health. While

this is a very powerful position, the dominance of the regulatory

perspective constrains stakeholders. Traditionally, regulation needs

predictions regarding single chemicals, but as a consequence,

risk assessors are stuck in a system where they are tackling one

chemical at a time. Because this is how toxicologists are trained

and how regulatory requirements are formulated, toxicology has

been shaped into a “one-chemical-at-a-time” science. The entire

ecosystem of regulators, the public, and private industry have

ended up focusing on understanding the impacts of each specic

chemical on health or environmental outcomes, even leading

to the creation of trade associations devoted solely to a single

chemical. Breaking out of this paradigm, at minimum, requires

that toxicologists share the data they collect so it can be assessed

and integrated with other data to create a more holistic view of

a chemical’s risk prole. If a risk assessor choses a new tool, its

integration requires broader discussion with regulators and often

regulations must ultimately be updated. Ideally, this discussion and

any information on the tool is public where others can comment

on it. Since there is no centralized effort to do so currently, data

sharing falls to the individual toxicologist and is not necessarily a

common practice. True change has to come from moving public

understanding and regulatory requirements with the eld as one;

which is very difcult to do at the same time.

Big Data

Eighty-four percent of all data in the world has been produced

in the last six years. The scientic literature on the interaction of

humans alone is enormous. For illustration: PubMed is estimated

to cover 25% of biomedical literature. This database includes

about one million new articles per year, of which ~100,000 describe

exposures and ~800,000 include some effects of a substance

on a biological system. Grey literature, such as the internet,

databases of legacy data, -omics technologies, robotized testing,

sensor technologies, image analysis etc. continuously add to

this knowledge base. A critical challenge is in sharing of these

data, which has been a notorious problem in toxicology. Often

information is only in the possession of companies and shared with

regulators in condence, if at all. Not only do we have to overcome

these hurdles, but we also need to establish data collection and/or

meta data standards. This refers in essence to the FAIR principles,

i.e., to make data available in a way that others can use them.

Toxicology is thus currently moving from a data-poor to a data-

rich science, though too many things are still siloed. Raw data

is often behind paywalls or regulatory walls, which can include

being shielded from the public with claims that it contains

condential business information. Consequently, we only see the

tip of the iceberg and data is often not accessible.

Adding to this, no ontologies or metadata allowing people to

make use of each other’s data are available. We generate a lot

of it every day, and generally do not know how to integrate it

unless it is highly curated. Data can be structured by chemical

identity. With more and more data available, the eld of

toxicology becomes dynamic and needs consistent support, e.g.,

to host a central database online. Such a tool needs agreement

regarding how to take data from across datasets. Data needs to

be shareable and usable for machine learning. Ideally, a real-

time assessment would be implemented based on monitoring (for

example integrating data via application programming interfaces

(API)), but such broader integration is hindered by various levels of

technology used by and available in practice. A possible steward,

semantics, standards, and denition of the level of information

needed for human prediction are required. Initially, the focus might

be on narrow chemical spaces with many studies/replicates.

A central problem of safety assessments is how to dene

something as safe. Typically, we have enough data to say

something is toxic, but when do we know enough to say it is

safe? The absence of evidence is no evidence of absence; i.e., a

lack of evident toxicity does not mean that it could not manifest

under different circumstances that are not adequately covered

in the test systems. This calls, on the one hand, for post-

marketing surveillance as done for drugs after market entry, or

more generally for alertness towards consumer feedback and

new scientic ndings.

Systematic Review Methods

A central problem of toxicology is evidence integration (enabling

integration of diverse, cross-disciplinary sources of information)

as more and more methodologies and results, some conicting

and others difcult to compare, are accumulating. This is a

challenge faced in more and more risk assessments, but also in

many systematic review methods that need to combine different

evidence streams (NASEM, 2011, 2017b, 2021; Woodruff and

Sutton, 2014; Samet, 2020; EPA, 2020; EFSA and EBTC, 2018).

Evidence integration is needed on very different levels of data,

studies and to other stressors, as well as across evidence streams.

The central opportunities are in quality assessment and AI,

especially natural language processing (NLP). There needs to

be a common platform, especially on the data side for dynamic

modeling (“dynamic data requires dynamic models”), sharing,

quality control, hardware and software, standards, metadata,

automated annotation, continuous adaptation to AI progress

(e.g. explainable AI), role model evidence-based medicine,

composition of test strategies, validation, and probabilistic risk

assessment. This collaborative open platform to transparently

collect, process, share, and interpret data, information and

knowledge on chemical and non-chemical stressors will enable

real-time and rapid evidence integration, empowering all

steps of protection of human health and the environment.

The combination of tests and other assessment methods

in integrated testing strategies (Hartung, 2013; Tollefsen,

2014; Rovida, 2015), a.k.a. IATA or DA by the Organisation for

Economic Co-operation and Development, needed to integrate

different types of evidence.

Role of Machine Learning and Articial Intelligence

We need evidence integration on the levels of data, information,

knowledge, and ultimately action. A system for integrating across

different levels of information that are each integrated within

their own space, requires broad integration of quality information

with the proper infrastructure to support this. The vision is to

create an infrastructure with harmonized agreement on the

levels of information and for what they may be best suited. For

this and its broad use, more toxicologists with computational

skills are needed. We also need common vocabularies across

different levels of information, data architectural standards for

release and utilization, real-time integration (e.g., through APIs),

and annotation at different levels. Such annotation requires

the connection of raw data to study metadata and the use of

language that a computer can digest by NLP through ontologies,

standardized “controlled” vocabularies, harmonized templates

such as IUCLID (https://iuclid6.echa.europa.eu/), and a library of

synonyms. A major question is what can be done to make sure

data is encoded/tagged to make it useful? High-quality training

sets for annotation to build knowledge graphs, causal networks,

etc. need to be developed. However, the use of annotation/

structured databases is an old way of looking at things. It is

almost impossible to get people to conform to data annotation

guidelines, so instead the eld will need to embrace methods to

manage unstructured data.

We might rather spend energy on building better NLP and deep

learning technologies to analyze unstructured data. The enormous

progress on NLP in recent years means that we are moving very

close to surpassing the Turing Test, if we have not already. The

Turing Test is a deceptively simple method of determining whether

a machine can demonstrate human intelligence. If a machine can

engage in a conversation with a human without being detected as

a machine, it has demonstrated human intelligence. Over the last

two years, enormously large models have been trained (Hoffmann,

2022). They use 140 to 530 billion parameters and 170 billion to 1.4

trillion training tokens. Some of these models claim to have been

trained on the entire Internet. They can respond in real-time to

questions with high accuracy, write articles indistinguishable from

those by human authors and even write code for computers. The

rst impact of the NLP breakthrough is that human knowledge

becomes machine-readable. Our vision is that this enables the

creation of similar models to virtually grasp the interaction of

organisms with chemical substances.

Toxicologists can read and extract information better, but

a computer can do this faster on many more sources. We

now need to train computers to be as good as humans in

interpreting data. AI is the best tool for evidence integration,

and evidence integration must become the standard for risk

and safety assessments. The big question is: how are people

going to use this information generated by AI? Here we need

to separate our vision from its implementation. Toxicological

research areas and associated S&T advances can overcome

hurdles to enable toxicology as a predictive science via evidence

integration. From the explosion in the use of machine Learning

and data science, the emerging use of NLP, knowledge graphs,

and next-generation-omics analytics we need to move to

explainable AI, embrace reinforcement learning and modern

database management. The platform to be established will

7 hps://www.ebtox.org

need an IT architecture, hardware and software, continuous

deployment/support, decision support tools, expert systems etc.

Evidence-based methodologies as furthered by evidence-based

toxicology

(e.g., systematic review principles, risk of bias, meta-

analysis, quality scoring, probabilistic approaches) can serve as

role models for objective and transparent handling of evidence.

Besides making sense of evidence pieces, such a platform can

also guide the composition and validation of Testing Strategies

(IATAs, DAs, AOP networks) and extraction of human relevant

reference datasets.

Probabilistic Approaches

Recognizing that as science delivers only probability rather

than absolutes, probabilistic tools lend themselves to all of

these (Maertens, 2022; Chiu and Paoli, 2020), will enable us to

move away from black/white, toxic/non-toxic dichotomies, as

well as better support life cycle and socioeconomic analyses

that require evaluation of incremental benets or risks rather

than “bright line” evaluations (NASEM, 2009; Chiu, 2017;

Fantke, 2018, 2021). Substantial progress on developing and

implementing probabilistic risk assessment approaches has

been made in the last 10 years (Chiu and Slob, 2015; Chiu,

2018), with the publication of guidance from the WHO/IPCS

(World Health Organization & International Programme on

Chemical Safety, 2018). Conceptually, this involves replacing

the xed values currently used for both the initial 'point

of departure” dose, as well as the “uncertainty factors”

with distributions that reect the state of the scientic

understanding, incorporating and combining uncertainties

quantitatively through statistical approaches (see Figure 9

for example applied to the Reference Dose). Several case

studies illustrating the broad application of probabilistic

approaches have been demonstrated (Blessinger, 2020; Chiu,

2018; Kvasnicka, 2019). Moreover, this conceptual approach

to deriving toxicity values probabilistically can be extended

to non-animal studies (Chiu and Paoli, 2020), as well as to

incorporating population variability through genetically diverse

models described above (Chiu and Rusyn, 2018; Rusyn, 2022).

In this way, probabilistic approaches provide a framework that

facilitates integration across different data types and sources.

Point of

Departure

Divide

by 10

Divide

by 10

Reference

Dose

Traditional Approach

POD

100

RfD =

Test System

(e.g., experimental animal, in vitro assay)

Inter-Species or

IVIVE Adjustment

Intra-Species

Variability

“Typical” Member of

Human Population

Common Conceptual Model

“Sensitive” Member of

Human Population

Dose or Concentration Distribution with

Effect Size M in Test System

Inter-Species or

IVIVE Distribution

Median to I

Percentile

Distribution

Dose Distribution with

Effect Size M in Median Human

Probabilistic Approach

Dose Distribution with

Effect Size M in I

Percentile Human

Reference Dose (RfD):

An estimate (with uncertainty spanning perhaps an order of

magnitude) of a daily oral exposure to the

human population

(Including sensitive subgroups

) that is likely to be without

an appreciable risk

of deleterious effects during a lifetime.

Probabilistic RfD (PrRfD):

A statistical lower confidence limit on the human dose

that at which a fraction I of the population shows an effect

of magnitude (or severity) M or greater (for the critical

effect considered).

Figure 9 Illustration of the transition from deterministic to probabilistic approaches when deriving reference doses from toxicity data. The

“Traditional Approach” refers to the practice attributed to Lehman and Fitzhugh (1954) to derive a "safe dose” by taking the dose level without

signicant effects in an animal study (a “point of departure”) and dividing by a “safety factor” of 100. The “Common Conceptual Model” is an

abstraction of this this procedure, whereby information from a test system (whether animal study or other type of data) is rst adjusted to the

“typical” in vivo human, and then adjusted to account for human variability in susceptibility, thereby deriving dose level that is protective of

“sensitive” members of the human population. The “Probabilistic Approach” further incorporates quantitative uncertainty and variability into this

conceptual model, using probability distributions at each step instead of single numbers, so that the result is a distribution (reecting incomplete

knowledge) for the dose that would cause on effect of magnitude “M” in the “I”th most sensitive percentile of the human population. [Adapted

from World Health Organization & International Programme on Chemical Safety., 2018]

Toxicology Research Trajectory

Recognizing the broad research advances and opportunities that

have arisen in the last 15 years since the 2007 Tox-21c report,

the workshop participants outlined their vision for the future

research trajectory needed to fulll the promise of transforming

toxicology into an exposure-driven, technology-enabled,

evidence-integrated eld that can better address population and

precision health while ensuring safe pharmaceuticals and a safer

environment. For each of the three research areas, participants

delineated a 5-, 10-, and 20-year plan for building capabilities

that would facilitate this transformation.

Exposure-driven Toxicology

Workshop participants identied several major areas of research

focus to advance exposure driven-toxicology in the coming

decades: 1) real-world-based exposure designs, 2) population-

scale measurements, 3) strategies to ask the right questions, and

4) consideration of ethical and policy implications.

Real-world-based Exposure Designs

Developments in this area are needed to allow for better in

silico, cellular, organoids, model organisms, as well as full

populations-based longitudinal studies. These developments will

allow studies to be conducted in a way that supports prediction

of environmental transport and fate (including chemical

transformations, inter-species comparisons, the application of

the understanding of exposure levels and exposure mixtures)

relevant to the population and its sub-groups. They will also

allow us to apply that knowledge to the experimental setting.

By using this real-world-based exposure design, the results

of different approaches to answering similar questions will be

easier to compare, and make it easier to utilize triangulation as a

key strategy for assessing the health effect and relevant toxicity

pathways of chemical and non-chemical exposures.

Population-scale Measurements

To understand the relevant exposures that lead to disease in

general and specic populations, additional efforts are needed to

develop biobanks (including biological specimens) and ecobanks

(including environmental samples) that inform on the distribution

of thousands of chemicals and non-chemical stressors in relevant

populations. Factors of interest include relevant exposure

scenarios, sociodemographic conditions, and relevant disease or

health status. Beyond human populations, the inclusion of animals

and the ecosystem for real-world exposure assessment is of

relevance to human environmental health, as well as environmental

health and toxicology, more broadly. Recent studies, for instance,

have shown that exposure assessment efforts in companion

animals, such as cats using non-invasive silicon tags, can contribute

to the assessment of ame retardants in homes, and their potential

role in feline hyperthyroidism (Poutasse, 2019).

Ask the Right Questions

One of the complexities in the current eld of omics technology

is how to prioritize the right questions in a way that leads

to the correct computational approach. For instance, the

question might be related to the total mixture, or to specic

components of a mixture. Thinking strategically and with the

right stakeholders (community, policymakers, interdisciplinary

scientists), will contribute to developing those right questions

in ways that are most useful for society and respecting

privacy concerns that many have regarding the unintended

consequences of data sharing.

Ethical and Policy Implications

An important amount of the workshop discussion focused

on aspects related to the ethical and policy implications of

toxicological research including the disproportionate burden

of exposures affecting disadvantaged communities. Groups

discussed the need for research to address those concerns by

incorporating elements of community engagement, citizen

support and environmental justice that must keep pace with

the technology.

Anticipated Capabilities

Regarding the key anticipated capabilities for exposure-driven

toxicology, the workshop participants anticipated the following

achievements as shown in Table 1 and described here:

At 5 years:

• Scale-up technology for high quality inexpensive assessment

of 1000-5000 chemicals that can be tagged to exogenous

exposures including non-chemical stressors. Technology is

currently slow and throughput is not high enough, which

makes exposomic approaches expensive.

• Develop libraries that tag key information for those chemicals

(meta data layering) to ensure their interpretation. There is

currently a lack of validation for many chemical signatures that

can be identied with untargeted technologies as to their

prediction of health effects.

• These technology problems can be solved through effort

and investment, similar to the genome project.

At 10 years:

• The availability of scalable technology for exposomics to

achieve high throughput, that is also cheap, sensitive, and

specic will allow us to apply this exposure-based approach

to longitudinal studies and biobanks.

• Studies that can be both retrospective and prospective

ensuring “FAIR” ness and linking exposome with health

outcomes.

• Retrospective studies will allow us to go back decades and

leverage biobanks. At the same time, we will be able to plan new

prospective studies to evaluate the exposures of the future.

20 years:

• Exposome-disease prediction will integrate detailed exposure-

based information with health outcome data in large scale

and numerous populations. We will achieve a great level of

precision in disease prediction that will be environment-based

and can also leverage gene-environment interactions.

• This knowledge will provide us with new forms of exposome

targeted prevention and treatment.

Table 1: Timeline for Key Exposure-driven Toxicology Developments

Key Capability Near-term (5-yr) goal Mid-term (10-yr) goal Long-term (20-yr) goal

Analytical

chemistry

Exposome assays (1000—

5000k/person)

High throughput exposome

assays (10,000/person)

Exposome disease prediction

Metabolomics,

toxicology

Reference exposome library

(meta-data layering)

Organ specic disease associated

Exposome targeted

treatment and prevention

Epidemiology,

clinical research

Disease associated metabolites

Retrospective and prospective

studies ensuring “FAIR” and linking

exposome with health outcomes

Exposome targeted treatment

Technology-enabled Toxicology

Workshop participants anticipate new technological capabilities

in two key research areas to fulll the promise of transforming

toxicology (see Table 2). These include biological capabilities

to provide a diversity of cells and tissues and bioengineering

capabilities to develop relevant and reproducible assays.

Moreover, in each, a set of supporting computational capabilities

will need to be developed.

Biological Capabilities

The critical path for biological capabilities lies in the

understanding of heterogeneity and susceptibility throughout

the life course at multiple scales from cells to the whole

organism. As genetics have turned out to be a much smaller

factor in outcomes than originally anticipated, there is a

need to better understand how non-genetic factors, such

as epigenetic differences and social and environmental

stressors, individually and collectively modulate development,

pathology, and pathophysiology. Because of the diversity in

the human population, an important resource for enabling this

understanding will be reliable and reproducible sources of cells

from multiple tissues representative of the population. Cell

sourcing is particularly important because it is likely that in vitro

microphysiological and other bioengineered systems (discussed

below) will play an essential role in untangling these complex

interactions. It is also recognized that in parallel, computational

capabilities, including multi-omic databases and advances in

interpretable AI, will need to be developed to move biological

capabilities forward.

Bioengineering Capabilities

With respect to bioengineering capabilities, the main

hurdles are: the lack of validated and standardized platforms

with automated fabrication, and the lack of availability of

individualized cell differentiation to enable personalized

toxicological evaluation. It is recognized that a range of t-

for-purpose models, ranging from simple suspensions and

monolayers to fully vascularized and innervated multi-organ

microphysiological systems, will be developed over time.

However, for any of them to be personalized, cell differentiation

protocols are needed that can enable creation of multiple

tissues from iPSCs from any individual. Coupled with the

biological capabilities understanding heterogeneity and

susceptibility, these bioengineering capabilities would enable

modeling of the diversity of the human population through

time. If the goals of automated fabrication and low cost are

also achieved, then a “multiverse”-type model platform is

envisioned, in which each person could have numerous “chip-

based twins” that could predict a range of possible future states

depending on different future exposures. As with biological

capabilities, a parallel set of computational capabilities will

be required, with the goal of creating “digital twins” to go

alongside the suite of “chip-based twins.”

Table 2: Timeline for Key Technology-enabled Toxicology Developments

Key Capability Near-term (5-yr) goal Mid-term (10-yr) goal Long-term (20-yr) goal

Biological

Representative and reliable

sources of human cells

National repository of human

cells representing key tissues

International repository of

human cells representing

all major tissues

Repository of human cell

types that is representative

of population diversity

Determinants of

heterogeneity and

susceptibility to toxicants

Role of genetic, epigenetic,

and social determinants

Understanding impact

of timing of exposure

and life stage

In vitro/in silico models for

heterogeneity and susceptibility

Understanding of

pathology, pathophysiology,

and development at

multiple scales

Faithful in vitro cell

differentiation coupled with

integrated multi-organ systems

Non-invasive, label-

free biomarkers

Integrated biomarkers

for modeling normal and

diseased organismoids.

Computational support for

biological capabilities

Multi-omic databases that are

comparable and computable

Integrated multi-omic databases coupled with interpretable AI

Evidence-integrated Toxicology

Evidence integration (animal studies, human studies, in vitro,

and all other types of studies) is needed at different levels of

integration, including: raw data, reports/scientic papers (meta-

analysis), and data on exposure and hazard. We integrate raw

data by interpreting it, and transforming it into information. We

integrate different pieces of information in our studies and reports

and create knowledge from the body of available studies and

papers. Ultimately, we act based on this knowledge, but where

do we integrate? Integration across evidence levels requires a

systematic review, based on structured data submissions. Curated

data sources and models fall between raw data and study reports.

The hope is that twenty years from now we will have a system

for integrating all these data and factors. This would enable

better risk management for public health with enormous societal

consequences. But we will also need to address the challenges of

communicating the results. Any evidence integration that does not

lead to simple classication is a potential communication problem.

However, chemicals cannot be simply put into black and white

bins (toxic vs. non-toxic) because they exist on a spectrum best

characterized by the probability of hazard.

To make Big Data common and encourage its use broadly

in toxicology in 10-20 years, we must incentivize people to

share information that today is still considered proprietary to

businesses. A lot of companies are already required to submit

structured information as regulatory data, but there needs to be

regulatory change to make the information accessible (possibly

with acceptable “blinding” of certain data). A key role of public-

private partnerships for data sharing, accessibility, modeling,

and cross-sector harmonization was identied. The application

of blockchain technology to encrypt pharma/health data and

federated model building should be explored. For the animal

part of toxicology, we have a moral obligation to make results

public. Experimenters received an exemption from society to

do something they normally would not (harm an animal), so

they owe society the data/outcome in return. The OECD QSAR

Toolbox is a pioneering tool that could serve as a model since

industry and regulators are less often using tools offered by

vendors, especially those tools which use commercial data that’s

not publicly available. In addition, there are differing levels

of transparency requirements (open access vs. open data vs.

open source). Cost could also be an issue, and promoting data

collection, processes, and trusted data brokers will be needed.

Training models with synthetic data sets as done more often with

patient/clinical data might be a promising avenue to explore.

The FAIR principles (Wilkinson et al., 2016) give important

guidance for the accessibility of structured and annotated data

to make them useful. Efcient utilization relies on real-time data

integration and updates as well as structured, annotated input

data to unlock the information contained.

The opportunities come from Big Data (characterized by the

3V's, or volume, variety, and velocity). It is expected that NLP for

data extraction will play a big role. Currently, most AI is a black

box. To make progress, we will need to move from this paradigm

to an explainable AI. Increasingly there is a need to combine

prospective and retrospective testing strategies. Policy does not

keep pace with the speed of technology. Strategies for enabling

evidence integration, such as better integration of various

~omics data from mechanistic toxicology, ~omics data layering

at the individual level is needed. We need to understand how to

incorporate less understood but equally important variables into

the equation, such as microbiome, circadian rhythm, and age. A

goal is to develop mechanistic models by changing from data-

driven to mechanism-driven. Challenges include:

• Finding incentives and reward structures in institutions to

encourage integration of big data approaches

• Dening relevant toxicants information to move through

three areas of measurement: time, space, and population.

This includes expansion to global populations.

• Small portable technologies to capture genetic and

environmental heterogeneity

Table 2: Timeline for Key Technology-enabled Toxicology Developments

Key Capability Near-term (5-yr) goal Mid-term (10-yr) goal Long-term (20-yr) goal

Bioengineering

Individual-specic cell

differentiation protocols

Reproducible cell differentiation

for all essential tissues for

at least one donor

Multi-donor cell differentiation

for all essential tissues

Individualized cell differentiation

for all essential tissues

Menu of t-for-purpose in

vitro model platforms.

Models representing

population diversity

Models representing

both health and

diseased populations

Models capable of

modeling “multiverse” of

potential future states

Validated, reproducible,

and affordable multi-organ

microphysiological “chips”

Platform standardization and

automated fabrication for a

dozen commercial platforms,

validated for individual organs

Commercial/off-the-

shelf single-organ chips;

validated multi-organ

chips; demonstrated

personalize chips

Validated personalized

“multiverse” chip

Computational support

to bioengineering

Quantitative in vitro to in

vivo extrapolation coupled

with in silico modeling

of model platforms

Digital/in silico

organ modeling

Personalized digital twins

• Designing a resource-efcient tiered strategy composed of

various methods for gathering human data for supervised

learning

• Identication of enabler technologies that make analyses

affordable and reliable

• Integration of human, animal, in vitro, etc. studies in

cohesive projects, allowing crosstalk and feedback

• Testing, data analysis, treatment and prevention informed by

predictive modeling (recursively applied to testing)

• Cloud-based computing platforms for continuous data and

model integration

• Common language so that information can move between

the three areas of exposure, technology, and evidence-

integration

• Robust informatic infrastructures: graph databases and novel

ML over structured and non- structured

• Need for Quality Assurance/Quality Control and validation

• Quality control of MPS reporting input into AI

Table 3 Timeline for Key Evidence-integrated Toxicology Developments

Key Capability Near-term (5-yr) goal Mid-term (10-yr) goal Long-term (20-yr) goal

CompTox literacy

Trainers and trainees

with CompTox skills

Establish curricula broadly

Highly skilled CompTox

workforce

NLP for decision

making

Annotation enabling resources

(e.g., synonyms, ontologies,

standardized vocabularies); NLP

to automatically retrieve/parse

study methods [unstructured data];

NLP to learn about relationships

in currently structured data

NLP to automatically

extract/parse study results;

accessible networks of NLP-

dened causal networks

NLP to automatically interpret/

combine study results; NLP

associations create datasets

that inform risk assessments

Data sharing

Global use of IUCLID and other

structured repositories and APIs

Platform of networked annotated

databases; community

engagement in exposure/health

data sharing and tracking

Real-time update and analysis

via networked platform

Explainable AI for

evidence integration

Explainable AI algorithms

and modeling pipelines

Interactive decision-support tool

that integrate evidence streams

Widespread implementation

of AI/ML in decision making;

probabilistic risk assessment

Accelerating Progress

8 hps://marianamazzucato.com/books/the-entrepreneurial-state

9 hps://web.ornl.gov/sci/techresources/Human_Genome/project/budget.shtml

10 hps://web.ornl.gov/sci/techresources/Human_Genome/project/economics.shtml

Realizing the Tox-21c 2.0 vision needs the Entrepreneurial State

(i.e., massive government investment) to create dedicated grants

for research, training, and implementation, in addition to efforts

to overcome institutional barriers. The evident role model is the

Human Genome Project, which largely transformed biomedical

science (Hood, 2013). Its costs have been estimated at up to

$3 billion.

A Battelle report from 2011 appraised the large and

widespread economic and functional impacts

: “Between 1988

and 2010 the human genome sequencing projects, associated

research and industry activity—directly and indirectly—generated

an economic (output) impact of $796 billion, personal income

exceeding $244 billion, and 3.8 million job-years of employment.

In the 2013 update, these numbers increased to economic

(output) impact of $965 billion, personal income exceeding

$293 billion, and 4.3 million job-years of employment.” The

transformation of toxicology envisioned here as a Human

Exposome Project (HEP) represents a similar opportunity,

promising to identify an even larger fraction of causes of disease

and opening up new opportunities for prevention and cure.

Governance: The necessarily multi-disciplinary and international

character of toxicology requires strategic steering. A cross-

agency alliance in the US could play a central role. New metrics

for success that measure impact/implementation must be

developed. Public-private partnerships are the most promising

avenue given their economic prospects. In its build-up,

dedicated preparatory programs, infrastructure, partnerships,

and engagement with the community have to be developed.

To start, a centralized effort is needed to facilitate data sharing

and to create the analysis platforms discussed in the evidence

integration section.

Education and Training: The challenge in educating and

training skill set development is a cross-cutting issue. The lack

of multidisciplinary skills was mentioned across all workshop

groups, as was the need to strike a balance between what you

can and cannot share (privacy, IP, crowdsourcing). A continued

dialogue through workshops, papers, strategic plans, policy

advice, and implementation is needed. The implementation

especially requires training in computational toxicology skills and

user comfort with computational tools. Most university curricula

are not up-to-date in this respect. The younger generation

is very receptive. A portfolio of CompTox training materials

would be helpful. It is also a communication challenge to both

the workforce and through outreach/engagement (two-way to

understand needs; multi-directional communication and training

on all levels is key) to the public and decision-makers.

Communication: The assembly of the components for

transforming toxicology and revamping risk sciences is a premier

communication challenge. Already, personal risk and public

health are difcult to communicate to the general public and

policymakers. Identifying drivers of disease must present itself

not as an anti-industry stance, but a societal need with business

opportunities. Communication with end-users to understand and

overcome institutional barriers is required. We need to articulate

in terms of the problems we are trying to solve.

Conclusion

This workshop is visionary, looking 10-20 years into the future.

Incredibly powerful novel methodologies exist to revamp

toxicology; the challenge is their implementation. We have to

enable toxicology to keep pace and benet from cross-sector

advances. There is a need for proof-of-concept examples

around information retrieval, evidence integration, quality

assessment, and decision support. A special opportunity lies in

crowdsourcing. Investment in technical infrastructure will facilitate

decision support tools (interpretable, actionable, probabilistic,

exible) that integrate multiple evidence streams.

The implementation of this vision is based on several key

expectations:

1. Biomonitoring and exposomics can evolve and be scaled to

make toxicology and environmental health more exposure-

driven

2. Relevant human model systems can be bioengineered to

study disease etiologies and interventions, especially of

exposure, as microphysiological systems

3. Computational approaches allow us to scale assessments of

chemicals and drugs

4. Evidence integration from these disruptive technologies can

guide risk assessment and management

Long-term Impacts

Ultimately advances in these areas would enable transformation

in toxicology with lasting impacts in three major ways:

• Safer Chemicals and Drugs. Much of toxicology today is

focused on ensuring the safety of xenobiotic exposures,

whether they be pharmaceuticals intentionally administered

or chemicals to which one is incidentally exposed through

the environment, consumer products, or occupation. Thus,

the most direct impacts of predictive toxicology would be

improvements in safer chemicals and drugs through higher

throughput and/or higher relevance in vitro or in silico

assays, particularly ones that are better at identifying intrinsic

and extrinsic susceptibilities.

• Precision Health. Similarly, there is substantial investment

already in precision medicine in the form of personalized

drug treatment. However, the research described here could

broaden this to the concept of precision health, which would

not only include personalized pharmacological treatment,

but also personalized preventive interventions and non-

pharmaceutical therapies. Moreover, while current efforts in

precision medicine focus on pharmacogenomics or poly-

pharmacy, advances in the frontiers of toxicology discussed

above could enable that individualization to extend to

the epigenome and exposome, as well as to interventions

connected with community health and well-being including

access to green-spaces, clean air and water.

• Targeted Public Health Interventions and Environmental

Regulations. Finally, the concepts of precision health

could be expanded to support public health with better

targeted public health interventions and environmental

regulations. This would better elucidate the toxicological

impacts of the genome, epigenome, and exposome,

all in combination. Thus, not only could there be

better assurance that interventions and regulations are

protective of the most vulnerable, but novel approaches

may also be revealed that enable better targeting of such

measures so that scarce resources can be allocated to

achieve the greatest overall benet.

Ultimately, the workshop participants envision a future for

toxicology as a Human Exposome Project in which collaborative,

technology-enabled open platforms transparently generate,

collect, process, share, and interpret data, information, and

knowledge of real-world chemical and non-chemical stressors to

enable real-time and rapid evidence integration, empowering all

steps of protection of human health and the environment.

Glossary

This is the section provides the denitions for terms used in the body of the report (see also Ferrario et al.,2014; Sille et al., 2020).

Adverse Outcome Pathway (AOP): An AOP is a sequence of events from the exposure of an individual or population to a chemical

substance through a nal adverse (toxic) effect at the individual level (for human health) or population level (for ecotoxicological

endpoints). The key events in an AOP should be denable and make sense from a physiological and biochemical perspective. AOPs

incorporate the toxicity pathway and mode of action for an adverse effect. AOPs may be related to other mechanisms and pathways as

well as to detoxication routes.

Biomarker: Indicator signaling an event or condition in a biological system or sample and giving a measure of exposure, effect, or

susceptibility.

Biomonitoring: The measurement of the body burden of toxic chemical compounds, elements, or their metabolites, in biological

substances

Evidence-based toxicology (EBT): EBT is a process for transparently, consistently, and objectively assessing available scientic

evidence in order to answer questions in toxicology. Particularly EBT: a) promotes the consistent use of transparent and systematic

processes to reach robust conclusions and sound judgments; b) displays a willingness to check the assumptions upon which current

toxicological practice is based to facilitate continuous improvement; c) recognizes the need to provide for the effective training and

development of professional toxicologists; d) acknowledges a requirement for new and improved tools for critical evaluation and

quantitative integration of scientic evidence; e) embraces all aspects of toxicological practice, and all types of evidence of which use

is made in hazard identication, risk assessment, and retrospective analyses of causation; f) ensures the generation and use of best

scientic evidence; g) includes all branches of toxicological science: human health assessment, environmental and ecotoxicology, and

clinical toxicology; h) has the potential to address concerns in the toxicological community about the limitations of current approaches

to assessing the state of the science; i) acknowledges and builds upon the achievements and contributions of Evidence Based

Medicine/Evidence Based Health Care.

Exposome: Concept describing the totality of exposure experienced by an individual during their life and the health impact of those

exposures (Wild, 2005), redened (Miller and Jones, 2014): The cumulative measure of environmental inuences and associated

biological responses throughout the lifespan, including exposures from the environment, diet, behavior, and endogenous processes.

Hazard: 1) A biological, chemical, or physical agent with the potential to cause an adverse health effect (European Commission, 2002).

2) The inherent characteristic of a material, condition, or activity that has the potential to cause adverse effects to people, property, or

the environment.

Metabolomics/Metabonomics: Evaluation of cells, tissues, or biological uids for changes in metabolite levels that follow exposure

to a given substance in order to determine the metabolic processes involved, to evaluate the disruption in intermediary metabolic

processes that results from exposure to that substance, or to determine the part of the genome that is responsible for the changes.

Note: Although “metabolomics” and “metabonomics” are frequently used as synonyms, there is a growing consensus that there is a

difference in that “metabolomics” places a greater emphasis on comprehensive metabolic proling, while “metabonomics” is used to

describe multiple (but not necessarily comprehensive) metabolic changes caused by a biological perturbation.

Risk assessment: A scientically based process consisting of four steps: hazard identication, hazard characterization, exposure

assessment, and risk characterization.

Threshold of toxicological concern (TTC): Human exposure threshold value for a group of chemicals below which there should be no

appreciable risk to human health.

Toxicokinetics: Generally, the overall process of the absorption (uptake) of potentially toxic substances by the body, the distribution

of the substances and their metabolites in tissues and organs, their metabolism (biotransformation), and the elimination of the

substances and their metabolites from the body. In validating a toxicological study, the collection of toxicokinetic data, either as

an integral component in the conduct of non-clinical toxicity studies or in specially designed supportive studies, in order to assess

systemic exposure.

Validation: The process by which the reliability and relevance of a particular approach, method, process, or assessment is established

for a dened purpose.

Bibliography

Alépée N., Bahinski T., Daneshian M., De Wever B., Fritsche E., Goldberg A., Hansmann J., Hartung T., Haycock J., Hogberg H.,

Hoelting L., Kelm J.M., Kadereit S., McVey E., Landsiedel R., Leist M., Lübberstedt M., Noor F., Pellevoisin C., Petersohn D.,

Pfannenbecker U., Reisinger K., Ramirez T., Rothen-Rutishauser B., Schäfer-Korting M., Zeilinger K. and Zurich M-G. (2014) “State-of-

the-art of 3D cultures (organs-on-a-chip) in safety testing and pathophysiology—a t4 report” ALTEX, 31:441-477.

Andersen M., Betts K., Dragan Y., Fitzpatrick S., Goodman J.L., Hartung T., Himmelfarb J., Ingber D.E., Jacobs A., Kavlock R., Kolaja K.,

Stevens J.L., Tagle D., Taylor D.L. and Throckmorton D. (2014) “Developing microphysiological systems for use as regulatory tools -

challenges and opportunities” ALTEX, 31:364-367.

Aung M.T., Yu Y., Ferguson K.K., Cantonwine D.E., Zeng L., McElrath T.F., Pennathur S., Mukherjee B., Meeker J.D. (2021) “Cross-

Sectional Estimation of Endogenous Biomarker Associations with Prenatal Phenols, Phthalates, Metals, and Polycyclic Aromatic

Hydrocarbons in Single-Pollutant and Mixtures Analysis Approaches” Environ Health Perspect, 129(3):37007. Epub 2021 Mar 24.

PMID: 33761273; PMCID: PMC7990518. doi: 10.1289/EHP7396

Bhatia, S.N. and Ingber, D.E. (2014) “Microuidic organs-on-chips. Nature biotechnology” 32, 760–772.

Blessinger T, Davis A, Chiu WA, Stanek J, Woodall GM, Gift J, Thayer KA, Bussard D. (2020) “Application of a unied probabilistic

framework to the dose-response assessment of acrolein.” Environ Int. 2020 Aug 5;143:105953. PMID: 32768806. doi: 10.1016/j.

envint.2020.105953.

Bopp S.K., Kienzler A., Richarz A., van der Linden S.C., Paini A., Parissis N. and Worth A.P. (2019) “Regulatory assessment and

risk management of chemical mixtures: challenges and ways forward” Critical Reviews in Toxicology, 49(2):174-189. DOI:

10.1080/10408444.2019.1579169

Bornmann, L., Haunschild, R. & Mutz, R. (2021) Growth rates of modern science: a latent piecewise growth curve approach to model

publication numbers from established and new literature databases. Humanit Soc Sci Commun 8, 224. https://doi.org/10.1057/

s41599-021-00903-w

Burnett S., Blanchette A., Chiu S.A. and Rusyn, I. (2021) “Human induced pluripotent stem cell (iPSC)-derived cardiomyocytes as an in

vitro model in toxicology: strengths and weaknesses for hazard identication and risk characterization” Expert Opinion on Drug

Metabolism and Toxicology, 17:8, 887-902, DOI: 10.1080/17425255.2021.1894122

Chiu W.A. (2017) “Chemical risk assessment and translation to socio-economic assessments” OECD Environment Working Papers,

March 14; (117). http://dx.doi.org/10.1787/a930054b-en

Chiu W.A., Axelrad D., Dalaijamts C., Dockins C., Shao K., Shapiro A.J., and Paoli G. (2018) “Beyond the RfD: Broad Application

of a Probabilistic Approach to Improve Chemical Dose-Response Assessments for Noncancer Effects.” EHP, DOI: https://doi.

org/10.1289/ehp3368

Chiu, W. A., Ginsberg, G.L., and Pullen, K. (2010) “The Exposome: A Powerful Approach for Evaluating Environmental Exposures and

Their Inuences on Human Disease”

Chiu W.A. and Paoli G.M. (2020) “Recent Advances in Probabilistic Dose-Response Assessment to Inform Risk-Based Decision Making.”

Risk Anal. PMID: 32966629 doi: 10.1111/risa.13595.

Chiu W.A. and Rusyn I. (2018) “Advancing chemical risk assessment decision-making with population variability data: challenges and

opportunities” Mamm Genome, doi: 10.1007/s00335-017-9731-6. [Epub ahead of print] PubMed PMID: 29299621

Chiu W.A. and Slob W. (2015): “A Unied Probabilistic Framework for Dose-Response Assessment of Human Health Effects” EHP, DOI:

http://dx.doi.org/10.1289/ehp.1409385

Clarke G.A., Hartse B.X., Asli A.E.N., Taghavimehr M., Hashemi N., Shirsavar M.A., Montazami R., Alimoradi N., Nasirian V., Ouedraogo

L.J., and Hashemi N.N. (2021) “Advancement of Sensor Integrated Organ-on-Chip Devices” Sensors 21, 1367

Densen P. (2011) "Challenges and opportunities facing medical education." Trans Am Clin Climatol Assoc. 122:48-58. PMID: 21686208;

PMCID: PMC3116346.

Dennis K.K., Marder E., Balshaw D.M., Cui Y., Lynes M., Patti G.J., Rappaport S.M., Shaughnessy D.T., Vrijheid M., and Barr D.B. (2017)

“Biomonitoring in the era of the exposome” Environ Health Perspect 125, 502-510. doi:10.1289/EHP474

Environmental Protection Agency. (2021) “A Review of U.S. EPA’s ORD Staff Handbook for Developing IRIS Assessments: 2020 Version”

Washington, DC: The National Academies Press, https://doi.org/10.17226/26289.

European Food Safety Authority and Evidence-Based Toxicology Collaboration. (2018) “Evidence integration in risk assessment:

The science of combining apples and oranges” EFSA Scientic Colloquium 23, Supporting Publication 16, EN-1396. doi:10.2903/

sp.efsa.2018.EN-1396

Escher B.I., Stapleton H.M. and Schymanski E.L. (2020) “Tracking complex mixtures of chemicals in our changing environment” Science,

80 (367), 388-392. 10.1126/science.aay6636

Fantke P., Aylward L., Bare J., Chiu W.A., Dodson R., Dwyer R., Ernstoff A., Howard B., Jantunen M., Jolliet O., Judson R., Kirchhübel N.,

Li D., Miller A., Paoli G., Price P., Rhomberg L., Shen B., Shin H-M., Teeguarden J., Vallero D., Wambaugh J., Wetmore B.A., Zaleski

R., and McKone T.E. (2018) “Advancements in Life Cycle Human Exposure and Toxicity Characterization” Environ Health Perspect,

126(12):125001. doi: 10.1289/EHP3871. PubMed PMID: 30540492.

Fantke P., Chiu W.A., Aylward L., Judson R., Huang L., Jang S., Gouin T., Rhomberg L., Aurisano N., McKone T. and Jolliet O. (2021)

“Exposure and Toxicity Characterization of Chemical Emissions and Chemicals in Products: Global Recommendations and

Implementation in USEtox” Int J Life Cycle Assess, 26(5):899-915. doi: 10.1007/s11367-021-01889-y. Epub 2021 Apr 5. PMID:

34140756; PMCID: PMC8208704.

Farhat N., Tsaioun K., Saunders-Hastings P., Morgan R.L., Ramoju S., Hartung T., Krewski D. (2022) “Systematic review in evidence-based

risk assessment” ALTEX, 39(3), 463–479. doi: 10.14573/altex.2004111

Ferrario D., Brustio R. and Hartung T. (2014) “Glossary of reference terms for alternative test methods and their validation” ALTEX,

31:319-335.

Fetah K., Tebon P., Goudie M.J., Eichenbaum J., Ren L., Barros N., Nasiri R., Ahadian S., Ashammakhi N., Dokmeci M.R., and

Khademhosseini A. (2019) “The emergence of 3D bioprinting in organ-on-chip systems” Prog Biomed Eng 1, 012001.

Ginsberg G.L., Fedinick K.P., Solomon G.M., Elliott K.C., Vandenberg J.J., Barone S., and Bucher J.R. (2019) “New Toxicology Tools and

the Emerging Paradigm Shift in Environmental Health Decision-Making” Environ Health Perspect, 127(12):125002. doi: 10.1289/

EHP4745. Epub 2019 Dec 13. PMID: 31834829; PMCID: PMC6957281.

Hartung T. (2009a) “Toxicology for the twenty-rst century” Nature, 460:208-212. doi:10.1038/460208a

Hartung T. (2009b) “Toxicology for the 21st century: Mapping the road ahead” Tox. Sci., 109:18-23.

Hartung T. (2017a) “Evolution of toxicological science: the need for change” International Journal of Risk Assessment and

Management, 20:21-45.

Hartung T. (2017b) “Thresholds of Toxicological Concern—setting a threshold for testing where there is little concern” ALTEX, 34:331-

351. doi: 10.14573/altex.1707011

Hartung, T. (2017c) “Utility of the adverse outcome pathway concept in drug development” Expert Opin Drug Metab Toxicol 13, 1-3. do

i:10.1080/17425255.2017.1246535

Hartung T. (2018) “Perspectives on in vitro to in vivo extrapolations” Journal of Applied In vitro Toxicology, 4:305–316. Doi: 10.1089/

aivt.2016.0026

Hartung T. and McBride M. (2011) “Food for thought… on mapping the human toxome” ALTEX, 28, 83-93. doi: 10.14573/

altex.2011.2.083

Hartung, T. and Leist, M. (2008) “Food for thought … on the evolution of toxicology and phasing out of animal testing” ALTEX 25, 91-

96. doi:10.14573/altex.2008.2.91

Hartung T. and Rovida C. (2009) “Chemical regulators have overreached” Nature, 460:1080-1081.

Hernández A.F. and Tsatsakis A.M. (2017) “Human exposure to chemical mixtures: Challenges for the integration of toxicology with

epidemiology data in risk assessment” Food and Chemical Toxicology, 103: 188-193. https://doi.org/10.1016/j.fct.2017.03.012

Hoffmann S. and Hartung T. (2005) “Diagnosis: Toxic!—Trying to apply approaches of clinical diagnostics and prevalence in toxicology

considerations” Toxicol. Sci, 85, 422-428. doi: 10.1093/toxsci/k099

Hoffmann S., de Vries R.B.M., Stephens M.L., Beck N.B., Dirven H.A.A.M., Fowle J.R., Goodman J.E., Hartung T., Kimber I., Lalu M.M.,

Thayer K., Whaley P., Wikoff D., and Tsaioun K. (2017) “A primer on systematic reviews in toxicology” Arch Toxicol 91, 2551-2575.

doi:10.1007/s00204-017-1980-3

Hood, L. and Rowen, L. (2013) “The Human Genome Project: big science transforms biology and medicine” Genome Med 5, 79.

https://doi.org/10.1186/gm483

Huang H., Wang A., Morello-Frosch R., Lam J., Sirota M., Padula A., and Woodruff T.J. (2018) “Cumulative Risk and Impact Modeling on

Environmental Chemical and Social Stressors” Curr Envir Health Rpt 5, 88–99. https://doi.org/10.1007/s40572-018-0180-5

Kleensang A., Maertens A., Rosenberg M., Fitzpatrick S., Lamb J., Auerbach S., Brennan R., Crofton K.M., Gordon B., Fornace A.J.

Jr., Gaido K., Gerhold D., Haw R., Henney A., Ma’ayan A., McBride M., Monti S., Ochs M.F., Pandey A., Sharan R., Stierum R.,

Tugendreich S., Willett C., Wittwehr C., Xia J., Patton G.W., Arvidson K., Bouhifd M., Hogberg H.T., Luechtefeld T., Smirnova L., Zhao

L., Adeleye Y., Kanehisa M., Carmichael P., Andersen E.M., Hartung T. (2014) “Pathways of Toxicity” ALTEX, 31:53-61. doi: 10.14573/

altex.1309261

Kvasnicka J., Stylianou K.S., Nguyen V.K., Huang L., Chiu W.A., Burton G.A., Semrau J., and Jolliet O. (2019) “Human Health Benets

from Fish Consumption vs. Risks from Inhalation Exposures Associated with Contaminated Sediment Remediation: Dredging of

the Hudson River” Environ Health Perspect, 127(12):127004. doi:10.1289/EHP5034. Epub 2019 Dec 13. PubMed PMID: 31834828;

PubMed Central PMCID: PMC6957280.

Krewski D., Andersen M., Tyshenko M.G., Krishnan K., Hartung T., Boekelheide K., Wambaugh J.F., Jones D., Whelan M., Thomas

R., Yauk C., Barton-Maclaren T. and Cote I. (2020) “Toxicity Testing in the 21st Century: Progress in the past decade and future

perspectives” Arch Toxicol, 94:1–58.

Krewski, D., Saunders-Hastings, P., and Baan, R., et al. (2022) “Workshop Report: Development of an evidence-based risk assessment

framework” ALTEX, in press.

Krewski, D., Saunders-Hastings, P., and Arzuga, X., et al. (in preparation) “Development of a framework for evidence synthesis:

Workshop report”

Lakhani, C. M., et al. (2019) “Repurposing large health insurance claims data to estimate genetic and environmental contributions in 560

phenotypes” Nat Genet 51, 327–334.

Lynn R. Goldman. (2003) “The Red Book: A Reassessment of Risk Assessment, Human and Ecological Risk Assessment: An International

Journal” 9:5, 1273-1281. https://doi.org/10.1080/10807030390248492

Materne E-M., Tonevitsky A.G., and Marx U. (2013) “Chip-based liver equivalents for toxicity testing—organotypicalness versus cost-

efcient high throughput” Lab Chip, 13, 3481 DOI: 10.1039/C3LC50240F

Maertens A., Golden E., Luechtefeld T.H., Hoffmann S., Tsaioun K. and Hartung T. (2022) “Probabilistic Risk Assessment—the Keystone

for the Future of Toxicology” ALTEX, 39:3-29. doi:10.1,4573/altex.2201081

Marx U., Akabane T., Andersson T.B., Baker E., Beilmann M., Beken S., Brendler-Schwaab S., Cirit M., David R., Dehne E-M., Durieux I.,

Ewart L., Fitzpatrick S.C., Frey O., Fuchs F., Grifth L.G., Hamilton G.A., Hartung T., Hoeng J., Hogberg H., Hughes D.J., Ingber D.E.,

Iskandar A., Kanamori T., Kojima H., Kuehnl J., Leist M., Li B., Loskill P., Mendrick D.L., Neumann T., Pallocca G., Rusyn I., Smirnova

L., Steger-Hartmann T., Tagle D.A., Tonevitsky A., Tsyb S., Trapecar M., van de Water B., van den Eijnden-van Raaij J., Vulto P.,

Watanabe K., Wolf A., Zhou X. and Roth A. (2020) “Biology-inspired microphysiological systems to advance medicines for patient

benet and animal welfare” ALTEX, 37:364-394. doi: 10.14573/altex.2001241

Marx U., Andersson T.B., Bahinski A., Beilmann M., Beken S., Cassee F.R., Cirit M., Daneshian M., Fitzpatrick S., Frey O., Gaertner C.,

Giese C., Grifth L., Hartung T., Heringa M.B., Hoeng J., de Jong W.H., Kojima H., Kuehnl J., Luch A., Maschmeyer I., Sakharov D.,

Sips A.J.A.M., Steger-Hartmann T., Tagle D.A., Tonevitsky A., Tralau T., Tsyb S., van de Stolpe A., Vandebriel R., Vulto P., Wang J.,

Wiest J., Rodenburg M. and Roth A. (2016) “Biology-inspired microphysiological system approaches to solve the prediction dilemma

of substance testing using animals” ALTEX, 33:272-321. doi: 10.14573/altex.1603161

Meigs L., Smirnova L., Rovida C., Leist M., and Hartung T. (2018) “Animal testing and its alternatives—the most important omics is

economics” ALTEX, 35:275-305. doi: 10.14573/altex.1807041

National Academies of Sciences, Engineering, and Medicine. (2009) “Science and Decisions: Advancing Risk Assessment” Washington,

DC: The National Academies Press. https://doi.org/10.17226/12209

National Academies of Sciences, Engineering, and Medicine. (2011) “Review of the Environmental Protection Agency’s Draft IRIS

Assessment of Formaldehyde” Washington, DC: The National Academies Press. https://doi.org/10.17226/13142

National Academies of Sciences, Engineering, and Medicine. (2016) “Interindividual Variability: New Ways to Study and Implications for

Decision Making: Workshop in Brief” Washington, DC: The National Academies Press. https://doi.org/10.17226/23413

National Academies of Sciences, Engineering, and Medicine. (2017b) “Application of Systematic Review Methods in an Overall Strategy

for Evaluating Low-Dose Toxicity from Endocrine Active Chemicals” Washington, DC: The National Academies Press, https://doi.

org/10.17226/24758.

National Academies of Sciences, Engineering, and Medicine. (2017a) “Using 21st Century Science to Improve Risk-Related Evaluations”

Washington, DC: The National Academies Press. https://doi.org/10.17226/24635

National Academies of Sciences, Engineering, and Medicine. (2021) “Microphysiological Systems: Bridging Human and Animal

Research: Proceedings of a Workshop—in Brief” Washington, DC: The National Academies Press. https://doi.org/10.17226/26124

National Research Council. (1983) “Risk Assessment in the Federal Government: Managing the Process” Washington, DC: The National

Academies Press. https://doi.org/10.17226/366

National Research Council. (2007) “Toxicity Testing in the 21st Century: A Vision and a Strategy” Washington, DC: The National

Academies Press. https://doi.org/10.17226/11970

National Research Council. (2012) “Exposure Science in the 21st Century: A Vision and a Strategy” Washington, DC: The National

Academies Press. https://doi.org/10.17226/13507

Nitsche K.S., Müller I., Malcomber S., Carmichael P.L. and Bouwmeester H. (2022) “Implementing organ-on-chip in a next-generation

risk assessment of chemicals: a review” Arch Toxicol 96, 711–741.

Pamies D., Leist M., Coecke S., Bowe G., Allen D., Gstraunthaler G., Bal-Price A., Pistollato F., DeVries R., Hogberg H.T., Hartung T. and

Stacey G. (2022) “Guidance Document on Good Cell and Tissue Culture Practice 2.0 (GCCP 2.0)” ALTEX, 39:30–70. doi: 10.14573/

altex.2111011

Partosch F., Mielke H., Stahlmann R., Kleuser B., Barlow S., and Gundert-Remy U. (2015). “Internal threshold of toxicological concern

values: Enabling route-to-route extrapolation: Arch Toxicol 89, 941-948. doi:10.1007/s00204- 014-1287-6

Poutasse C.M., Herbstman J.B., Peterson M.E., Gordon J., Soboroff P.H., Holmes D., Gonzalez D., Tidwell L.G., and Anderson K.A.

(2019) “Silicone Pet Tags Associate Tris (1,3-dichloro-2-isopropyl) Phosphate Exposures with Feline Hyperthyroidism” Environ Sci

Technol. 53(15):9203-9213. doi: 10.1021/acs.est.9b02226. Epub 2019 Jul 10. PMID: 31290326; PMCID: PMC7330886

Rappaport S.M. and Smith M.T. (2010) “Environment and Disease Risks” Science, 330: 460–461. doi:10.1126/science.1192603

Richard A.M., Huang R., Waidyanatha S., Shinn P., Collins B.J., Thillainadarajah I., Grulke C.M., Williams A.J., Lougee R.R., Judson

R.S., Houck K.A., Shobair M., Yang C., Rathman J.F., Yasgar A., Fitzpatrick S.C., Simeonov A., Thomas R.S., Crofton K.M., Paules

R.S., Bucher J.R., Austin C.P., Kavlock R.J., Tice R.R. (2021) “The Tox21 10K Compound Library: Collaborative Chemistry Advancing

Toxicology” Chem Res Toxicol. 34(2):189-216. doi: 10.1021/acs.chemrestox.0c00264. Epub 2020 Nov 3. PMID: 33140634; PMCID:

PMC7887805

Robinson, O., and Vrijheid, M. (2015) “The Pregnancy Exposome” Curr Envir Health Rpt 2, 204–213. https://doi.org/10.1007/s40572-015-

0043-2

Roth A., M.P.S.-W.S. Berlin, Marx U., Vilén L., Ewart L., Grifth L.G., Hartung T., Ingber D.E., Mendrick D.L., Steger-Hartmann T., and

Tagle D.A. (2021) “2019 Human microphysiological systems for drug development” Science, 373:1304-1306

Rusyn I., Chiu W.A., and Wright F.A. (2022) “Model systems and organisms for addressing inter- and intra-species variability in risk

assessment.” Regul Toxicol Pharmacol. 2022 May 27:105197. doi: 10.1016/j.yrtph.2022.105197. Epub ahead of print. PMID: 35636685.

Samet J.M., et al. (2020) “The IARC Monographs: Updated procedures for modern and transparent evidence synthesis in cancer hazard

identication” Journal of the National Cancer Institute, 112(1):30-37. 10.1093/jnci/djz169

Sillé F.C.M., Karakitsios S., Kleensang A., Koehler K., Maertens A., Miller G.W., Prasse C., Quiros-Alcala L., Ramachandran G., Rappaport

S.M., Rule A.M., Sarigiannis D., Smirnova L., and Hartung T. (2020) “The exposome—a new approach for risk assessment” ALTEX, 37:

3-23. doi: 10.14573/altex.2001051

van Vliet E., Daneshian M., Beilmann M., Davies A., Fava E., Fleck R., Julé Y., Kansy M., Kustermann S., Macko P., Mundy W., Roth A.,

Shah I., Uteng M., van de Water B., Hartung T. and Leist M. (2014) “Current approaches and future role of high content imaging in

safety sciences and drug discovery” ALTEX, 31:479-493.

Vermeulen R., Schymanski E.L., Barabási A.L., and Miller, G.W. (2020) “The exposome and health: Where chemistry meets biology”

Science, 367(6476):392-396. doi: 10.1126/science.aay3164. PMID: 31974245; PMCID: PMC7227413

Wambaugh J.F., Wetmore B.A., Pearce R., Strope C., Goldsmith R., Sluka J.P., Sedykh A., Tropsha A., Bosgra S., Shah I., Judson R.,

Thomas R.S., and Setzer R.W. (2015) “Toxicokinetic triage for environmental chemicals” Toxicol Sci 147, 55-67. doi:10.1093/toxsci/

kfv118

Wang, Z., Walker, G. W., Muir, D. C. G., and Nagatani-Yoshida, K. (2020) “Toward a global understanding of chemical pollution: a rst

comprehensive analysis of national and regional chemical inventories,” Environ Sci Technol 54, 2575–2584. http://doi.org/10.1021/

acs.est.9b06379

World Health Organization & International Programme on Chemical Safety. (2018) “Guidance document on evaluating and expressing

uncertainty in hazard characterization” 2nd ed. World Health Organization. https://apps.who.int/iris/handle/10665/259858. License:

CC BY-NC-SA 3.0 IGO

Wild, C.P. (2019) “The global cancer burden: necessity is the mother of prevention” Nature Reviews Cancer 19: 123-124. https://doi.

org/10.1038/s41568-019-0110-3

Wilkinson M.D., Dumontier M., Aalbersberg I.J., Appleton G., Axton M., Baak A., Blomberg N., Boiten J-W., da Silva Santos L.B.,

Bourne P.E., Bouwman J., Brookes A.J., Clark T., Crosas M., Dillo I., Dumon O., Edmunds S., Evelo C.T., Finkers R., Gonzalez-Beltran

A., Gray A.J.G., Growth P., Goble C., Grethe J.S., Heringa J., Hoen P.A.C., Hooft R., Kuhn T., Kok R, Kok J., Lusher S.J., Martone M.E.,

Mons A., Packer A.L., Persson B., Rocca-Serra P., Roos M., van Schaik R., Sansone A-A., Schultes E., Sengstag T., Slater T., Strawn

G., Swertz M.A., Thompson M., van der Lei J., van Mulligen E., Velterop J., Waagmeester A., Wittenburg P., Wolstencroft K., Zhao J.

and Mons B. (2016) “The FAIR Guiding Principles for scientic data management and stewardship” Sci Data 3, 160018. https://doi.

org/10.1038/sdata.2016.18

Woodruff T.J., Sutton P. (2014) “The Navigation Guide systematic review methodology: A rigorous and transparent method for

translating environmental health science into better health outcomes” Environmental Health Perspectives, 122(10):1007-1014.

Appendix I—Workshop Attendees

Workshop Co-chairs

Ana Navas-Acien Columbia University

Weihsueh Chiu Texas A&M University

Thomas Hartung Johns Hopkins University

Workshop Participants

Tony Atala WakeForest University

Dana Dolinoy University of Michigan

Lauren Heine ChemForward

Salman Khetani University of Illinois at Chicago

Marianthi Kioumortzoglu Columbia University

Nicole Kleinstreuer NIEHS NTP

Koren Mann McGill University

Uwe Marx TissUse

Patrick McMullen Scitovation

Gary Miller Columbia University

Katie Paul- Friedman US Environmental Protection Agency

Jennifer Sass NRDC

Kris Thayer US Environmental Protection Agency

Cavin Ward-Caviness US Environmental Protection Agency

Cheryl Walker Baylor University

Katrina Waters Pacic Northwest National Laboratory

Hao Zhu Rutgers University

Government Observers

Bindu Nair OUSD(R&E), Basic Research Ofce

Jean Luc Cambrier OUSD(R&E), Basic Research Ofce

Shanni Silberberg OUSD(R&E), Basic Research Ofce

Daniel Osburn OUSD(R&E), Basic Research Ofce

Betsy Melebrink OUSD(R&E), Basic Research Ofce

Anna Lowit Environmental Protection Agency

Mark Johnson US Army Public Health Center

Rabih Jabbour US Army Edgewood Chemical Biological Center

Natalie Vinas US Army Engineer Research and Development Center

Louis Scarano Environmental Protection Agency

Rachel Gooding Department of Homeland Security, Chemical Security Analysis Center

VT-ARC Team

Matthew Bigman Virginia Tech Applied Research Corporation

Jordan Brown Virginia Tech Applied Research Corporation

Christina Houfek Virginia Tech Applied Research Corporation

Kate Klemic Virginia Tech Applied Research Corporation

Lynne Ostrer Virginia Tech Applied Research Corporation

Workshop Participant Short Biography

Anthony Atala, Director, Wake Forest Institute for Regenerative Medicine

Wake Forest School of Medicine

[email protected]

Anthony Atala, MD, is the G. Link Professor and Director of the Wake Forest Institute for Regenerative

Medicine, and the W. Boyce Professor and Chair of the Department of Urology at Wake Forest University.

His work focuses on growing human cells, tissues and organs. Fifteen applications of technologies

developed in Dr. Atala’s laboratory have been used clinically. Dr. Atala was named by Scientic American

as one of the world’s most inuential people in biotechnology, by U.S. News and World Report as one

of 14 Pioneers of Medical Progress in the 21st Century, by Life Sciences Intellectual Property Review as

one of 50 key inuencers in the life sciences intellectual property arena, and by the journal Nature Biotechnology as one of the top 10

translational researchers in the world.

Weihsueh A. Chiu, Professor

Texas A&M University

[email protected]

Weihsueh A. Chiu, Ph.D. is a professor in the Department of Veterinary Physiology and Pharmacology

at Texas A&M University. Before joining the university in 2015, he worked at the U.S. Environmental

Protection Agency (EPA) for more than 14 years, most recently as branch chief in the Ofce or Research

and Development. His research in human health risk assessment includes toxicokinetics, physiologically-

based pharmacokinetic modeling, dose-response assessment, characterizing uncertainty and variability,

systematic review, and meta-analysis, with particular interest in Bayesian and probabilistic methods.

Dr. Chiu has participated or chaired expert review panels for multiple government agencies, including NTP, CalEPA, the FDA, and

ATSDR. He has also served on numerous national and international committees and workgroups for Health Canada, the World

Health Organization, the Organisation for Economic Cooperation and Development, and the U.S. National Academies of Sciences,

Engineering and Medicine.

Dana Dolinoy, Chair and Professor

University of Michigan

[email protected]

Dana C. Dolinoy is Professor of Environmental Health Sciences and Nutritional Sciences and NSF

International Chair of Environmental Health Sciences at the University of Michigan School of Public Health

as well as Faculty Director of the Epigenomics Core at Michigan Medicine. Her research focuses on

how nutritional and environmental factors interact with epigenetic gene regulation to shape health and

disease. In 2015, she received the 2015 NIH Director’s Transformative Research Award to develop piRNA

epigenetic editing technologies and in 2018 received the Society of Toxicology Achievement Award and

has recently co-edited the book ToxicoEpigenetics: Core Principles and Applications. She has authored

>130 manuscripts and 10 book chapters, and served as Chair of the Gordon Conference in Cellular and

Molecular Mechanisms of Toxicity. She has mentored 14 doctoral students, one of whom recently received

a F31 award, and 6 post-doctoral fellows, one of whom recently received a NIEHS K99/R00 award, as well as several masters and

undergraduate students.

Thomas Hartung, Professor and Chair

Johns Hopkins University

https://www.jhsph.edu/faculty/directory/prole/2308/thomas-hartung

[email protected]

Thomas Hartung is the Doerenkamp-Zbinden Chair for Evidence-based Toxicology in the Department of

Environmental Health and Engineering at Johns Hopkins Bloomberg School of Public Health, Baltimore,

with a joint appointment at the Whiting School of Engineering. He also holds a joint appointment for

Molecular Microbiology and Immunology at the Bloomberg School. He is adjunct afliate professor

at Georgetown University, Washington D.C.. In addition, he holds a joint appointment as Professor for

Pharmacology and Toxicology at the University of Konstanz, Germany; he also is Director of Centers for Alternatives to Animal Testing

(CAAT, http://caat.jhsph.edu) of both universities.

CAAT hosts the secretariat of the Evidence-based Toxicology Collaboration (http://www.ebtox.org), the Good Read-Across Practice

Collaboration, the Good Cell Culture Practice Collaboration, the Green Toxicology Collaboration and the Industry Renement Working

Group. As PI, Dr. Hartung headed the Human Toxome project funded as an NIH Transformative Research Grant and the series of World

Summits for Microphysiological Systems started in 2022. He is Field Chief Editor of Frontiers in Articial Intelligence. He is the former

Head of the European Commission’s Center for the Validation of Alternative Methods (ECVAM), Ispra, Italy, and has authored more than

625 scientic publications (h-index 105).

Lauren Heine, Director of Science and Data Integrity

ChemFORWARD

www.chemforward.org

lauren@chemforward.org

Lauren Heine applies green chemistry, green engineering, alternatives assessment and multi-stakeholder

collaboration to develop tools that result in safer and more sustainable chemical products and processes.

Her work with ChemFORWARD builds on prior experience developing GreenScreen’s for Safer

Chemicals, a pioneering method for chemical hazard assessment to enable informed substitution; and

CleanGredients, a web-based information platform for identifying greener chemicals for use in cleaning

products; both tools were designed to scale access to information needed to develop materials and products that are safe and circular.

Lauren worked closely with the US EPA Safer Choice Program to facilitate development of ingredient and hazard criteria for the Safer

Choice Program.

Salman Khetani, Associate Professor and Director of Graduate Studies

University of Illinois at Chicago

mtm.uic.edu

[email protected]

Salman Khetani is an associate professor of Biomedical Engineering at the University of Illinois at Chicago

where he directs the Microfabricated Tissue Models (MTM) laboratory that is engaged in developing in

vitro models of various tissues (liver, cardiac, intestine, brain, and placenta) for drug screening, disease

modeling, and regenerative medicine. Prior to academia, Dr. Khetani co-founded and directed research

at Hepregen Corporation, which launched engineered models of the human liver that continue to serve

the pharmaceutical industry for elucidating drug metabolism, toxicity, and efcacy for liver diseases. Dr.

Khetani’s research is currently funded by the US National Science Foundation and National Institutes of Health. His laboratory focuses

on engineering the microenvironmental cues around mammalian cells towards stabilizing their long-term phenotype in vitro for

applications in drug development and regenerative medicine. He has developed model systems to mimic key aspects of diseases in

vitro and elucidate underlying molecular mechanisms of disease progression as a function of cell-cell, cell-ECM, and cell-soluble factor

interactions. His liver models have been translated to the commercial realm through licensing of issued patents and patent applications

to companies. Recent work on iPSC-derived atrial cardiomyocytes are used to study the underlying genetic determinants of atrial

brillation in close collaboration with leading cardiologists.

Marianthi-Anna Kioumourtzoglou, Assistant Professor

Columbia University

https://www.publichealth.columbia.edu/people/our-faculty/mk3961

[email protected]

Marianthi-Anna Kioumourtzoglou is an environmental engineer and epidemiologist. She holds a Master

of Science in Public Health (MSPH) from the Environmental Sciences and Engineering Department at the

University of North Carolina at Chapel Hill and a Doctor of Science (ScD) in Environmental Health from

the Harvard TH Chan School of Public Health, where she also conducted her post-doctoral fellowship.

She is currently an Assistant Professor at the Department of Environmental Health Sciences at Columbia

University’s Mailman School of Public Health. Her research focuses on applied statistical issues related

to environmental epidemiology, including quantifying and correcting for exposure measurement

error, exposure prediction uncertainty propagation, and assessment of high-dimensional and complex

exposures in health analyses. Her studies mainly (albeit not exclusively) focus on air pollution exposures

and, additionally, on identifying vulnerable sub-populations and characterizing how risks may vary across neighborhood-level and other

urban characteristics, as well as in a changing climate.

Nicole Kleinstreuer, Director

NICEATM

https://www.publichealth.columbia.edu/people/our-faculty/mk3961

[email protected]

Nicole Kleinstreuer is the acting director of the NTP Interagency Center for the Evaluation of Alternative

Toxicological Methods (NICEATM), the US federal resource for alternatives to animal testing. At NICEATM,

she leads domestic and international efforts to develop novel testing and analysis strategies that provide

more rapid, mechanistic, and human-relevant predictions of potential environmental chemical hazards.

Kleinstreuer’s research focuses on mathematical and computational modeling of biological systems and

their susceptibility to perturbations that result in adverse health outcomes. She has a secondary appointment in the NIEHS Division

of Intramural Research Biostatistics and Computational Biology Branch, and adjunct faculty positions in the Yale University School of

Public Health and the Eshelman School of Pharmacy at UNC Chapel Hill.

Koren Mann, Professor and Chair

McGill University, Department of Pharmacology and Therapeutics

[email protected]

Koren Mann is a Professor and Chair of the Department of Pharmacology and Therapeutics at McGill

University, and a Senior Investigator at the Lady Davis Institute for Medical Research in Montreal, Quebec,

Canada. She received her doctorate in Pathology and Immunology from Boston University in 1999 and was

a postdoctoral fellow at McGill University from 1999-2004. Her laboratory focuses on studying the health

effects of metal exposure, although recent studies include other environmental pollutants. She focuses on

how modulation of the immune system results in pathology, especially cardiovascular toxicity. The overarching theme of her lab is to

integrate toxicology questions within the framework of the epidemiology, providing relevance and feed-forward questions to further

interrogate in human cohorts.

Uwe Marx, MD

TissUse GmbH

www.tissuse.com

[email protected]

Uwe Marx is the founder and CSO of TissUse, a 2010 spin-out from the Technische Universitat Berlin

dedicated to the development of human organ and body-on-a-chip systems for drug testing and precision

medicine approaches. The solutions aim to shorten the drug development process and to reduce animal

experiments. With more than 30 years of experience in protein drug development and tissue engineering

Dr. Marx has published about 150 scientic papers and numerous reviews and book chapters. He is an inventor in more than 30 patent

families. Uwe Marx received his doctorate degree in immunology from the Charite of the Humboldt-University in Berlin in 1991 after

nishing his medical and biochemistry training. His academic research at the Charite Berlin, the University of Leipzig and the Technische

Universitat Berlin focused on human monoclonal antibodies, tissue engineering and human multi-organ chip solutions respectively.

Between 2000 and 2010, Uwe Marx joined ProBioGen, a biotech Company he founded in 1994 - as CSO. He served as a reviewer for

various German governmental biotech programmes and received several awards for the development of animal-free technologies. Dr.

Marx is a serial entrepreneur and co-founder of numerous German biotech companies.

Patrick McMullen, Director of Computational Toxicology

ScitoVation

www.scitovation.com

[email protected]

Patrick McMullen is the Director of Computational Toxicology at ScitoVation. Dr. McMullen works with

diverse stakeholders spanning government, non-prot, and industry groups to bring new approaches

to toxicology, with the goal of improving chemical safety decision-making processes. His research

and consulting work combine high-content biological experiments with statistical and computational

approaches to advance the understanding of biological fundamentals that underlie chemical safety

challenges. Dr. McMullen’s background in molecular biology, engineering, and computational science has been instrumental in

interpreting and communicating complex data problems in diverse applications. Dr. McMullen manages a diverse computational

biology team that uses modeling and cell-based experiments to deepen our understanding of how chemicals interact with biological

systems. Dr. McMullen earned his Ph.D. in Chemical and Biological Engineering from Northwestern.

Gary Miller, Vice Dean for Research Strategy and Innovation

Columbia University

https://www.publichealth.columbia.edu/people/our-faculty/gm2815

[email protected]

Gary Miller serves as Vice Dean for Research Strategy and Innovation and Professor of Environmental

Health Sciences at the Columbia University Mailman School of Public Health. He also has appointments

in the Department of Molecular Pharmacology and Therapeutics and the Department of Neurology in

the Vagelos College of Physicians and Surgeons. He is an international leader on the exposome, the

environmental analogue to the genome. Dr. Miller founded the rst exposome center in the U.S. and

wrote the rst book on the topic. He has helped develop high-resolution mass spectrometry methods

to provide an omic-scale analysis of the human exposome. He served as Editor-in-Chief of Toxicological

Sciences, the ofcial journal of the Society of Toxicology from 2013-2019 and is now Editor-in-Chief of

Exposome, the rst journal in the eld. He is on the Scientic Advisory Board of NIHs All of Us Research Program, the NIH Human

Health Exposure Analysis Resource (HHEAR), and the Human Biomonitoring for the European Union (HBM4EU) project.

Ana Navas-Acien, Professor

Columbia University

https://www.publichealth.columbia.edu/people/our-faculty/an2737

[email protected]

Ana Navas-Acien is a physician-epidemiologist (MD, University of Granada, Spain ‘96) with a specialty in

Preventive Medicine and Public Health (Hospital La Paz, Madrid ‘01) and a PhD in Epidemiology (Johns

Hopkins University ‘05). Her research investigates the long-term health effects of environmental exposures

(arsenic and other metals, tobacco smoke, e-cigarettes, air pollution), relevant molecular pathways, and

effective interventions for reducing involuntary exposures. She collaborates with major cohort studies such

as the Strong Heart Study, a study of cardiovascular disease and its risk factors in American Indian communities, and the Multi-Ethnic

Study of Atherosclerosis (MESA), a study of cardiovascular, metabolic and lung disease in urban settings across the US, and with the

TACT2 study (a clinical trial assessing the cardiovascular benets of metal chelation). Both in the US and internationally, she evaluates

exposure to tobacco smoke including e-cigarettes through toxicological and epidemiological research strategies. Her goals are to

contribute to the reduction of environmental health disparities in underserved and disproportionately exposed populations.

Katie Paul-Friedman, Toxicologist

US EPA

[email protected]

Dr. Katie Paul Friedman joined the Center for Computational Toxicology and Exposure in the Ofce of Research

and Development at the US EPA in August 2016, where she is currently focused on application of new approach

methodologies to chemical safety assessment, with additional interests in uncertainty in alternative and traditional

toxicity information, endocrine bioactivity and developmental neurotoxicity prediction, and in vitro kinetics. One

of her roles in the Center is to run the ToxCast program. Previously, Dr. Paul Friedman worked as a regulatory

toxicologist at Bayer CropScience with specialties in neuro-, developmental and endocrine toxicity, and predictive

toxicology. She has been actively involved in multi-stakeholder projects to develop adverse outcome pathways,

alternative testing approaches, and the regulatory acceptance of new approach methodologies. Her laboratory background includes

development of high-throughput screening assays, the combined use of myriad in vitro and in vivo approaches, including receptor-

reporter and biochemical assays, primary hepatocyte cultures, and targeted animal testing paradigms, to investigate the human

relevance of thyroid and metabolic adverse outcome pathways using probe chemicals. Dr. Paul Friedman received a Ph.D. in Toxicology

from the University of North Carolina at Chapel Hill.

Jennifer Sass, Senior Scientist

Natural Resources Defense Council

[email protected]g

Jennifer Sass is a Senior Scientist at the Natural Resources Defense Council (2001-2021) and part-time

faculty at George Washington University Milken School of Public Health (2008-2021). She has published

over 50 articles. She holds BSc, MSc, and PhD (1998) degrees from the University of Saskatchewan,

College of Medicine, Department of Anatomy and Cell Biology, and a Post-Doctoral Certicate (2000)

from the University of Maryland, College of Medicine, Program in Human Health and the Environment.

Kristina Thayer, Director

US EPA, Chemical and Pollutant Assessment Division (CPAD)

thayer[email protected]

Kristina Thayer is Director of the Chemical and Pollutant Assessment Division (CPAD) at the U.S.

Environmental Protection Agency (https://www.epa.gov/aboutepa/about-chemical-and-pollutant-

assessment-division-cpad). CPAD occupies an essential position in EPA’s Ofce of Research and

Development between researchers generating scientic data and EPA’s program and regional ofces

that make decisions regarding the protection of public health and the environment. CPAD scientists

develop a range of t-for-purpose human health risk assessment products based on the evaluation,

synthesis, and analysis of the most up-to-date scientic information. Products include the Integrated Risk

Information System (IRIS) and Provisionally Peer Reviewed Toxicity Values (PPRTV) assessments. These

products are developed through interactions with EPA’s program and regional ofces, other agencies,

the scientic community, industry, policymakers, and the public. Once nalized, they serve as a major

scientic component supporting EPA’s regulations, advisories, policies, enforcement, and remedial action

decisions. CPAD also conducts cutting-edge research to develop innovative human health risk assessment methods (e.g., systematic

review) that facilitate careful evaluation of scientic evidence, as well as tools and models (e.g., benchmark dose modeling software).

Cheryl Walker, Director

Baylor College of Medicine, Center for Precision Environmental Health

[email protected]

Cheryl Walker holds the Alkek Presidential Chair in Environmental Health and is the founder and Director

of the Center for Precision Environmental Health at Baylor College of Medicine. She also directs the

NIEHS P30 Gulf Coast Center for Precision Environmental Health. Dr. Walker has over 200 publications in

the scientic literature and is an elected member of the National Academy of Medicine. Her research on

gene:environment interactions and environmental epigenomics has been continuously funded by the NIH,

DOD, and Foundations and advocacy groups for over 25 years Her laboratory actively investigates gene x

environment interactions and their role in diseases such as cancer, broids and NAFLD. The TSC2 tumor suppressor, and its role in cell

signaling, has been one of the areas of interest for her lab, in addition to the study of TSC2-linked pathways that regulate key cellular

functions.

Cavin Ward-Caviness, Computational Biologist

US EPA

[email protected]

Cavin Ward-Caviness is a Principal Investigator in the Public Health and Integrated Toxicology Division of the US

Environmental Protection Agency. With a background in computational biology and environmental epidemiology,

Dr. Ward-Caviness seeks to understand the environmental factors which inuence health in vulnerable populations

and the molecular mechanisms that inuence environmental health risks. He is the PI of the EPA CARES research

resource, which allows researchers to study environmental health effects in vulnerable patient populations, using

large electronic health record databases. Ward-Caviness also leads the Environmental Health Domain Team for the

National Covid Cohort Consortium. He is also interested in how epigenetics and metabolomics can serve as an

early indicator of adverse health effects from chemical and social environmental exposures and in particular how molecular biomarkers

can give us insight into how the environment may accelerate the aging process and thus contribute to chronic disease.

Katrina Waters, Director

Pacic Northwest National Laboratory

https://www.pnnl.gov/people/katrina-m-waters

[email protected]

Katrina Waters is a Laboratory Fellow and Director for Biological Sciences Research at the Pacic

Northwest National Laboratory (PNNL). Her research interests are focused at the intersection of

environmental exposures and infectious disease on human health. Her current programs include the study

of health effects of chemicals at Superfund sites and personal environmental exposure assessment for

epidemiological studies in disadvantaged communities. She recently completed a Department of Energy

research program focused on airborne and environmental transmission of COVID-19. She has also led

numerous research efforts in Computational Modeling, Bioinformatics, and Data Management for a NIAID

Center for Predictive Modeling of Infectious Diseases and a Department of Homeland Security program for Predictive Modeling of Viral

Infections. Dr. Waters holds joint faculty appointments with OSU and the University of Washington.

Hao Zhu, Professor

Rutgers University-Camden

[email protected]

Hao Zhu is a Professor of Chemistry at the Rutgers University-Camden. His major research interest is to

use cheminformatics tools to develop predictive models. All resulted models can be used to directly

predict the chemical toxicity based on the public big data and molecular structure information. His current

research interests also include data-driven modeling, articial intelligence algorithm development and

computer-aided nanomedicine design. He is the Principal Investigator of several prestigious research

grants (NIH R01, R15 and etc). Dr. Zhu is author/co-author of 81 peer-reviewed journal articles and 7 book chapters with over 5,600

citations. His research was recognized with different awards, such as Rutgers Chancellor’s Award for Outstanding Research and Creative

Activity, National Institute of Environmental Health Sciences (NIEHS) Extramural Paper of the Month (two times, 2019 and 2020) and

Drug Discovery Today top citation paper of the year.

Appendix II—Workshop Agenda and Prospectus

Basic Research Innovation Collaboration Center

4100 N. Fairfax Rd. | Fourth Floor| Suite 450

Arlington, VA 22203

DAY 1—THURSDAY, APRIL 28, 2022

Time Title Speaker

8:00—8:15 Check-in and Connental Breakfast

8:15 - 8:20 Welcome and Introducons and Expectaons Thomas Hartung, JHU

8:20 -8:45 Workshop Framing Talk Co-chairs

8:45—9:00

Breakout Instrucons and Morning Break

9:00—10:45

Working Group I: Dene the Problem

Small group discussions to frame a vision for toxicology as a predicve science and idenfy the

greatest hurdles to achieving it.

Group A—Exposure-driven Toxicology

Group B—Technical Advances

Group C—Evidence Integraon

10:45—11:00 BREAK - Transion to main conference room and leads prepare outbrieng

11:00 –12:00 Working Group 1: Outbrieng

12:00—1:00 LUNCH (provided for parcipants)

1:00—3:45

Working Group II: Technical Capabilies and Opportunies

What are the promising research direcons for moving to a more predicve

toxicology? What are the potenal capabilies in the 10- to 20-year horizon?

Group A—Exposure-driven Toxicology

Group B—Technical Advances

Group C—Evidence Integraon

3:45—4:00 BREAK - Transion to main room and leads prepare outbrieng

4:00—4:45 Working Group II: Outbrieng

4:45—5:00 Summary of Day Co-chairs

5:00 MEETING ADJOURNED FOR THE DAY

DAY 2—FRIDAY, APRIL 29

, 2022

Time

Title Speaker

8:00—8:15 Check-in and Connental Breakfast

8:15—8:30 Welcome and Day 1 Recap Co-chairs

8:30 -9:30

‘White Space’ Discussion I

Discussion of topics which did not t into

the framework of day 1 but need to be

discussed.

9:30—10:30

‘White Space’ Discussion II

Discussion of parcularly far-out (or long-term),

high-risk, high-impact ideas.

10:30—10:45 BREAK

10:45—11:45 Discussion of Key Ideas/Components for Report

11:45—12:00 Closing Remarks Co-chairs

12:00 DEPARTURE

Future Directions Workshop: Advancing the Next Scientic Revolution in Toxicology

Basic Research Ofce, Ofce of the Under Secretary of Defense (RandE)

28-29 April 2022

Basic Research Innovative Collaboration Center

4100 N. Fairfax Road, Suite 450 Arlington, VA 22203

Co-Chairs: Thomas Hartung (Johns Hopkins), Ana Navas-Acien (Columbia), Weihsueh Chiu (Texas A&M)

In the nearly two decades since the human genome was sequenced, the eld of toxicology has undergone a transformation, taking

advantage of the explosion in biomedical knowledge and technologies to move from a largely empirical science aimed at ensuring

the absence of harmful effects to a mechanistic endeavor aimed at elucidating disease etiology. However, a substantial gap remains

between the promise of mechanistic toxicology and the actualization of the eld as a predictive science. For instance, high-throughput

in vitro and in silico toxicity testing remains largely focused on prioritization of individual chemicals for future investigation. Moreover,

efforts to translate such data into hazard or risk have been hampered by inadequate coverage of important biological targets,

inadequate consideration of population heterogeneity, and aiming still to provide assurances of safety rather than quantication of

effects across the population. Furthermore, there has been little progress on understanding the complex interactions among chemicals

and between chemicals and other intrinsic and extrinsic factors that affect population health, such as genetics and non-chemical

stressors, including marginalization and other social determinants of health.

This Future Directions Workshop on Advancing the Next Scientic Revolution in Toxicology aims to establish a new overarching vision

for toxicology as a predictive science. This vision entails a major paradigm shift in how toxicology is both conceived and practiced,

recognizing the multi-factorial, multi-causal nature of toxicity. Specically, this vision involves two critical steps:

• Moving away from reductionist interrogation of single chemicals, individual model systems, and discrete biological targets, which

ultimately cover only a minute sliver of relevant human experiences.

• Striving for a holistic understanding of the interactions among chemicals, non-chemical stressors, heterogeneous populations, and

life-stages, in order to prospectively identify and quantify their impacts on the incidence and severity of human disease.

A key outcome of this Workshop will be a roadmap of key basic science research needs that, if addressed in the next 10-20 years,

can substantially advance this transformational vision. The discussions and ensuing distributed report will provide valuable long-term

guidance to the DoD community, as well as the broader federal funding community, federal labs, and other stakeholders. Workshop

attendees will emerge with a better ability to identify and seize potential opportunities in the different elds addressed. This workshop

is sponsored by the Basic Research Ofce within the Ofce of Secretary of Defense, along with input and interest from the Services and

other DoD components.

Agenda

Rather than a standard conference format, the workshop design emphasizes interactive dialogue with primarily small-group breakout

sessions followed by whole-group synthesis of ideas.

Day One: The majority of the rst day will be spent in small-group breakout sessions on fundamental challenges to progress and

technical capabilities. The three breakout themes include:

1. Exposure-driven Toxicology

Populations are exposed to multiple environmental agents, including chemical agents through air, water, food and soil, and non-

chemical agents such as noise, light, and social stressors (e.g., racism, socioeconomic deprivation, climate). Toxicological research

that embraces an exposure-driven approach, characterizing real-life exposure scenarios including exposure mixtures and how these

agents work together affecting multiple mechanistic pathways and health outcomes is needed. A key opportunity is the expansion

of exposomic approaches to include this broader landscape of exposures. The interplay of environmental and social stressors

with genetic and molecular variants, and the contributions of this research towards the identication and evaluation of effective

interventions will be critical elements for discussion.

2. Technical Advances and Challenges

Predictive toxicology requires expanding the “toolbox” in several directions. First, because adverse outcomes involve interactions

of environment (see above), genes, and lifestage, we need our “model systems” to cover “gene” and “lifestage” more broadly.

Example technologies include genetically diverse population-based in vitro and in vivo resources, and expansion of experimental

designs to cover different stages of development, as well as developmental origins of health and disease. Additionally, our

approaches currently cluster at the beginning (e.g., high throughput assays) and the end (e.g., in vivo apical endpoints) of the

pathophysiological process, neglecting the modulating and stochastic factors that inuence outcomes that lie between. Thus,

approaches that provide access to intermediate states, perturbations, and outcomes are needed to better understand the

progression to disease. Example technologies include novel biomarkers, microphysiological systems (e.g., organ on a chip), and

in silico models (e.g., systems toxicology/virtual experiments, AI/Machine Learning). Finally, a key challenge is characterizing the

predictive accuracy, precision, and relevance of new approaches, as well as understanding their domains of applicability.

3. Evidence Integration

Toxicology is currently transitioning from a data-poor to a data-rich science with the curation of legacy databases, “grey”

information in the internet, mining of scientic literature, sensor technologies, ~omics, robotized testing, high-content imaging

and others. Key questions include how to handle these new types of information sources, which may be incomplete, how to weigh

(evidence strength, risk of bias, quality scoring etc.), and how to integrate this evidence. For instance, probabilistic risk assessment

integrates across sources of evidence resulting in a more holistic probability of risk/hazard, though challenges include how to

validate (real-life, t for purpose, ground truthing, qualication, triangulation) and communicate these probabilities. Additional

challenges, such as data curation and storage, mining, analysis and visualization will be discussed.

Day Two: The second day of the workshop is a half-day consisting of white-space, whole group discussions on topics that did not

fall into the Day 1 framework or were especially ambitious and/or high-risk. Participants will also discuss cross cutting themes and

the trajectory of the eld over the next 10-20 years. At the end of the day, the whole group will discuss the overarching themes of the

workshop that should be included in the nal workshop report.

Cross Cutting themes to discuss

• What disease endpoints are the most promising in terms of developing the knowledge (e.g., availability of human biomarkers,

understanding of genetic/non-genetic risk factors) and technologies (e.g., microphysiological systems, computational models)

needed to enable predictive toxicology?

• Where are the greatest opportunities for synergies between toxicology and other disciplines including the social sciences?

• How to ensure standards (e.g., reporting standards, systematic review, meta-analysis, risk of bias analysis, etc.) that retain quality

assurance (best practices, validation and other aspects of QA and QC) and public health protection in a mechanistic toxicology?