Attached question

Chemical Risk Assessment: Traditional vs Public Health Perspectives

Preventing adverse health ef-

fects of environmental chemical

exposure is fundamental to pro-

tecting individual and public he-

alth. When done efficiently and

properly, chemical risk assess-

ment enables risk management

actions that minimize the in-

cidence and effects of environ-

mentally induced diseases related

to chemical exposure. However,

traditional chemical risk assess-

ment is faced with multiple chal-

lenges with respect to predicting

and preventing disease in human

populations, and epidemiological

studies increasingly report obser-

vations of adverse health effects

at exposure levels predicted

from animal studies to be safe

for humans. This discordance

reinforces concerns about the

adequacy of contemporary risk

assessment practices for pro-

tecting public health.

It is becoming clear that to

protect public health more effec-

tively, future risk assessments will

need to use the full range of

available data, draw on innovative

methods to integrate diverse data

streams, and consider health

endpoints that also reflect the

range of subtle effects and mor-

bidities observed in human pop-

ulations.

Considering these factors,

there is a need to reframe

chemical risk assessment to be

more clearly aligned with the

public health goal of minimizing

environmental exposures asso-

ciated with disease. (Am J Public

Health. 2017;107:1032–1039.

doi:10.2105/AJPH.2017.303771)

Maureen R. Gwinn, PhD, Daniel A. Axelrad, MPP, Tina Bahadori, ScD, David Bussard, BA, Wayne E. Cascio, MD, Kacee Deener, MPH, David Dix, PhD, Russell S. Thomas, PhD, Robert J. Kavlock, PhD, and Thomas A. Burke, PhD, MPH

See also Greenberg, p. 1020.

For the past several decades,human health risk assessment has been a pillar of environmental health protection. In general, the products of risk assessment have been numerical risk values derived from animal toxicology studies of observable effects at high doses of individual chem- icals. Although this approach has contributed to our understanding of overt health outcomes from chemical exposures, it does not always match our understanding from epidemiology studies of the consequences of real-world ex- posures in human populations, which are characterized by expo- sure to multiple pollutants, often chronically, at concentrations that can fluctuate over wide ranges; susceptible populations and life stages; potential interactions be- tween chemicals and nonchemical stressors and background disease states; and lifestyle factors that modify exposures (e.g., airtight houses).1 Theseandotherissuesare particularly important when de- termining risk of complex diseases, such as cardiovascular disease.

Ten years ago, the National Research Council offered a new paradigm for evaluating the safety of chemicals on the basis of chemical characterization, testing using a toxicity pathway ap- proach, and modeling and ex- trapolating the dose–response relationship from in vitro testing, all embedded in a risk context

and considering population- based data and exposure.2 Efforts such as the Tox21 Consortium3,4

and ToxCast program5 have helped us better understand the biological interactions of large numbers of chemicals using high-throughput assay systems, and we are witnessing early adoption of new technologies and approaches for screening chemicals for integrated testing.6

Several other factors are also changing the way environmental health professionals think about chemical risks and how to most effectively protect public health, especiallyforcomplexdiseaseslike cardiovascular disease. It is esti- mated that intrinsic factors (e.g., those that result in mutations stemming from random errors in DNA replication) account for only 10% to 30% of many com- mon cancers.7 Similarly, only 30% to 40% of birth defects can be attributed to known causes such as genetics, fetal alcohol syndrome, maternal smoking, and folate in- sufficiency.8 Other studies have concluded that nongenetic envi- ronmental factors and gene by

environment interactions are the primary causes of chronic dis- eases.9 The ability to evaluate and quantifytheroleofenvironmental factors on public health is a clear opportunity, but it is limited by thelack ofreadilyavailablemodels for prominent clinical outcomes.

CURRENT CHALLENGES

Understanding public health risk from environmental chem- ical exposures is complicated by many factors, such as population variability and susceptibility, long latencies between critical exposures and disease manifesta- tions, and background environ- mental exposures. Issues of population variability and sus- ceptibility are poorly understood and difficult to characterize and incorporate into risk assessments. For example, a person’s unique microbiome may modulate his or her response to environmental exposures.10,11 Although studies are limited in this emerging area, knowledge about the

ABOUT THE AUTHORS At the time of the writing of this article, all of the authors were with the US Environmental Protection Agency, Washington, DC.

Correspondence should be sent to Maureen R. Gwinn, PhD DABT ATS, Office of Research and Development, US Environmental Protection Agency, 1300 Pennsylvania Ave NW, Ronald Reagan Building, Room 41205, MC 8101R, Washington, DC 20460 (e-mail: gwinn. [email protected]). Reprints can be ordered at http://www.ajph.org by clicking the “Reprints” link.

This article was accepted March 2, 2017. doi: 10.2105/AJPH.2017.303771

1032 Analytic Essay Peer Reviewed Gwinn et al. AJPH July 2017, Vol 107, No. 7

AJPH RISK ASSESSMENT

microbiome may inform inter- individual variability and un- explained susceptibility observed in populations. Scientists have begun to appreciate the role of the microbiome in the lack of reproducibility and in- terpretability of animal studies.12

Another example is the effects of early life environmental exposures on health outcomes later in life. Advances in the field of epigenetics have revealed that developmental exposure to endocrine disrupting chemicals can alter epigenetic program- ming of gene regulation and thus may play a role in the risk of obesity later in life.13 Similar to microbiome research, studies in this area are limited, and a better understanding of the link between chemical exposure, epigenetic gene regulation, and health outcomes through epide- miological research can help us

better address factors that are currently difficult to account for in traditional risk assessment. Finally, there are also methodo- logical challenges in determining attributable risks in populations with background environmental exposures, as these background exposures may change the populationhealthbaselinesoraffect the response of the target chemical. Other examples of important fac- tors to incorporate in risk assess- ments can be found in Table 1.

OPPORTUNITIES FOR USING MULTIPLE DATA TYPES

Concurrent with these chal- lenges, science and technology are advancing rapidly and in ways that create opportunities for risk assessment. Public health

disciplines help us understand how baseline health status can influence the effect of population-level chemical ex- posures. We also need to consider how environmental pollutants may contribute to overall disease burden for endpoints not tradi- tionally considered in chemical risk assessment (e.g., metabolic disorders, autism). New methods in epidemiological re- search help us evaluate complex interactions among multifacto- rial causes of disease ranging from macro (societal, neigh- borhood) to micro (molecular) factors, relevance of exposures during sensitive life stages, and a better understanding of in- terrelatedness of disease across the life span.14

Advances in high-throughput technologies and computational modeling (e.g., ToxCast, Tox21, and ExpoCast efforts) are

providing data on hazard and exposure potential for a large number of data-poor chemicals. The increased generation of data for both hazard and exposure from these advances can be used to better understand the bi- ological pathways that lead to adverse health effects in ways that were not possible in the past. But linking these observations to specific disease endpoints is challenging because the trans- lation of effects across levels of biological organization is not well understood. One approach with the potential to advance our understanding of how chemical exposures can affect health is the use of adverse outcome path- ways, which integrate various types of biological information to link molecular initiating events to downstream key events and ultimately unwanted health outcomes.15,16

TABLE 1—Examples of Current Risk Assessment Challenges and Opportunities

Risk Assessment Challenge Description Impact on Risk Assessment Public Health Opportunity

Molecular initiating events and subsequent

key events in adverse outcome pathways

Earlybiologicalchangesor precursoreffects

in response to chemical exposures may be

identified by in vitro, animal, or

epidemiological studies

Useful for qualitative and quantitative

understanding of ultimate health effect of

early biological changes

Improved public health protection without

need for long-term toxicology or

epidemiology studies

Background exposures Population exposures to a myriad of

environmental chemicals at low

concentrations

Exposures to background chemicals may

affect response to target chemical

exposures and may change population

health baselines

Increased public health protection if

baseline exposures are taken into account

when determining prevention strategies

Nonchemical stressors Physical and psychosocial stressors,

including noise, temperature,

socioeconomic status, social stress, and

limited resources

Impact on baseline susceptibility and

potential effect modification

Potential role in cumulative assessment,

improved identification of vulnerable

populations, potential target for public

health interventions (e.g., stress

management)

Early life determinants of health Biological characteristics and exposures

that can determine chronic and lifelong

health outcomes

Effect of exposures during early life may

play a role in later disease states (e.g.,

endocrine disruptors, epigenetic changes)

Potential for early life interventions for

prevention and management of later

disease

Baseline health status Individual health status, with a focus on

potential health susceptibilities

Baseline health status may affect response

to additional environmental chemical

exposures

Increased public health protection if

baseline health status is taken into

account

Microbiome Microorganisms that reside within and on

our bodies and interact with the

environment

Exposure modification, susceptibility and

resilience to environmental pollutants,

important as an early life determinant of

health

Potential targets for prevention and

intervention, management of allergic

responses, and precision risk management

AJPH RISK ASSESSMENT

July 2017, Vol 107, No. 7 AJPH Gwinn et al. Peer Reviewed Analytic Essay 1033

To fully realize the potential of adverse outcome pathway– based approaches and to in- tegrate biological findings across disciplines, we must strengthen our ability to detect precursor events in human populations and to identify biologically rel- evant exposure metrics, ideally measurable in individuals. An- other advancement that has a great potential to advance our

understanding of data-poor che- micals is the use of nontesting approaches (e.g., quantitative structure–activity relationship) that allow us to predict toxicity when adequate testing data are absent—especially when we combine knowledge of chemical structural features and in vitro bioactivity determinations. Ad- vances in the development of chemical libraries,

cheminformatics, and read-across predictions and in- tegration with molecular data and adverse outcome pathways have significantly improved their application and predictive capacity, which will allow more comprehensive assessment of the health effects of exposures.17,18

Effectively predicting population risk by integrating a variety of data streams (e.g.,

epidemiology, toxicology, high-throughput testing) and considering multiple sources and pathways of exposure can better inform environmental public health decisions. Advances in technology and computational capabilities have fostered new opportunities for generating and analyzing molecular, animal, and human data on effects and ex- posures, which can be integrated

TABLE 2—Data Streams and Opportunities and Challenges for Informing Risk Assessment

Data Type Description Opportunity Challenge

Nontesting

data

Nontesting approaches, such as quantitative structure–

activity relationship models and read-across allow us

to predict toxicity when adequate testing data are

absent

Advances in the field have significantly improved

their application and predictive capacity

Developing principles for acceptance, for

characterizing and incorporating uncertainties into

predictions, and for developing objective metrics of

performance

Molecular Biochemical and cell-based bioactivity data and “omics-

based” data on thousands of chemicals

Can help inform our understanding of the health

outcomes of environmental exposures, using data

that are potentially more human relevant

Lack of scientific consensus on inferring hazard from

bioactivity in vitro assay and omics-based data and

providing quantitative dose–response information

on exposure metrics

Animal Traditional animal testing provides a hazard based

point of departure for risk assessments

Targeted animal testing can be performed on the

basis of the results of bioactivity data to focus on

key health outcomes

Potential uncertainties with using traditional animal

testing to estimate human risk (e.g., extrapolating

from animal to human or high to low doses and

accounting for human population variability and

life stage susceptibility)

Human Epidemiological and other human data support holistic

assessment of the effects of chemical exposures on

public health

Newer exposure science and statistical techniques

advance the understanding of human variability

that can be obtained from epidemiology and

individual sequencing; understanding effect

modification by nonchemical stressors and baseline

health status

Often limited mechanistic and dose–response data,

and exposure misclassification can bias results to

the null; possibility of unmeasured confounders

often undermines confidence in observed

associations, and it may require multiple studies

and many years to rule out chance, bias, and

confounding as possible explanations for observed

associations

Exposure Exposure characterization that captured the variability

in time, space, and within and across populations;

better toxicokinetic data link external to internal

dosimetry and relevant environmental exposure

concentrations with biological significance

Targeted and nontargeted biomonitoring,

application of sensors, and other new technologies

are greatly advancing population exposure

characterization; high-throughput exposure

models allow exposure predictions on thousands of

chemicals with associated uncertainty

Estimating and incorporating the inter- and

intraindividual variability in exposures into current

designs of toxicity testing and risk assessments;

extrapolating relevant target tissue and organ dose

information from external exposures and in vitro

assays; accounting for multiple exposures; sample

collection, data management, and analysis; and

covering or extrapolating to a broader chemical

space

Digital data The ongoing revolution in social media use and

communication has provided a new source of data

used in exposure science and environmental

epidemiology for local and timely information about

disease and health dynamics

A significant source of untapped data The collection and application of these data have

significant ethical implications that need to be

understood and managed, particularly taking

into account personal identifiable information;

methods to evaluate the quality of the data and

build confidence in the applications are needed

AJPH RISK ASSESSMENT

1034 Analytic Essay Peer Reviewed Gwinn et al. AJPH July 2017, Vol 107, No. 7

into chemical risk assessments. At the same time, probabilistic and high-throughput ap- proaches for risk assessment have been advancing. Table 2 high- lights various data types available and challenges in applying these data types to inform risk assessment.

A PUBLIC HEALTH PERSPECTIVE

A public health perspective for chemical risk assessment would approach risk assessment from a new lens. It would address population health with a focus on the health and societal burden of disease; use and integrate all available types of data—including traditional toxicology, human epidemiological findings, and

newer and emerging data streams and information, such as digital epidemiology,19 high- throughput and high-content data, and adverse outcome path- ways; and draw on public health approaches, such as attributable risk or relative risk. This new perspective may be especially important for some historically challenging aspects of risk assess- ment, such as understanding cumulative risks of exposures to multiple chemical and non- chemical stressors. Internationally, scientists have raised concerns about the large number of ubiq- uitous chemicals people are exposed to and called for re- thinking approaches to evalu- ating the health effects of chemicals.16 Figure 1 presents a conceptual model for a public health perspective for risk assessment.

Although approaching assess- ments from the perspective of health outcomes may be chal- lenging, it provides the oppor- tunity to evaluate exposures and effects across the life span that are relevant to population health. Advances in science and tech- nology, such as adverse outcome pathway development, the broader availability of chemical and biological data, and the applications of statistical and bioinformatics tools, bring this previously aspirational approach well within reach.20

EXAMPLE: CARDIOVASCULAR DISEASE

A public health approach may inform the challenge of

cardiovascular disease. Cardio- vascular disease is the number 1 cause of mortality worldwide and is a major US public health burden.21,22 Annual costs of cardiovascular disease in the United States were estimated to be $317 billion in 2011 and 2012, considering direct medical costs and lost productivity because of premature mortality.22 This es- timate is likely to substantially underestimate the social cost of cardiovascular disease because of limitations in the es- timation of indirect costs associ- ated with morbidity and premature mortality.23

Although much is known about the biochemical and be- havioral risk factors associated with cardiovascular disease, par- ticularly compared with other diseases and health conditions, the traditional risk factors fail to account for 10% to 25% of its prevalence.24 Environmental factors, including air pollution25

and chemical exposures26 are thought to contribute to the unexplained fraction. Although mortality stemming from car- diovascular disease has decreased over the past few decades in the developed world as a result of reductions in behavioral risk factors, the rising prevalence of obesity and diabetes might ac- count for the deceleration in the rate of improvement in annual cardiovascular mortality in the United States over the past few years.27

There is an urgent need to better understand the biological pathways through which envi- ronmental exposures to chemical and nonchemical stressors act to stimulate and accelerate athero- sclerosis and promote adverse cardiovascular health effects. Applying the adverse outcome pathway framework,28 the initial molecular response to a chemical exposure will often be receptor

Improved

public

health

Starting Point

• Adverse health outcome of

concern

Data Sources (along with those

used in traditional assessment)

• Clinical data on baseline

population health status

• Molecular epidemiology

• Exposure information in the

population

• Behavioral data

Synthesis

• Chemical/nonchemical

stressors contributing to the

adverse outcome

• Prevention strategies

Public Health Perspective

Starting Point

In context of a statutory authority

• Chemical or class of concern

• Route(s) of exposure

Data Sources

• Epidemiology studies

• Laboratory animal studies

• Mechanistic data

Synthesis

• Multiple health outcomes of

concern

• Toxicity values for specific

chemical/endpoint

• Output/risk metric: absolute

estimate of risk in population, or

safety assessment (e.g., hazard

index)

Traditional Risk Assessment

Note. This conceptual model illustrates how the starting point in a public health–focused risk assessment would differ from that of traditional risk assessment. In traditional risk assessment, the starting point is focused on specific chemicals or classes of chemicals of concern, with multiple data streams saying what the critical effects from that chemical are. A public health perspective would focus on the adverse health outcome of concern with multiple data streams, informing our understanding of hazard and exposure in the context of public health decisions related to that outcome and not necessarily focused on just 1 chemical or class of chemicals.

FIGURE 1—Conceptual Model for a Public Health Perspective for Chemical Risk Assessment

AJPH RISK ASSESSMENT

July 2017, Vol 107, No. 7 AJPH Gwinn et al. Peer Reviewed Analytic Essay 1035

activation and changes in meta- bolism and, ultimately, changes in tissue and organ function. Such changes can be modified by both intrinsic (e.g., gender, age, genetic, and epigenetic back- ground) and extrinsic factors (e.g., coexposures to other chemical and nonchemical stressors; Figure 2). Over time, these changes produce subclinical effects, such as changes in elec- trical and mechanical cardiac function, vascular function, and nonobstructive atherosclerotic

vascular changes. With the per- sistence of metabolic changes that stimulate the progression of vascular disease, clinical cardio- vascular events such as heart at- tacks, strokes, heart failure, and abnormal heart rhythms follow.

To date, the most compre- hensive application of this ap- proach has been in the study of population-level health effects of air pollution exposure.28

Epidemiological data at the population level has provided support that air pollutant

exposure (e.g., ambient particular matter and NO2) accelerates the development and progression of coronary atherosclerosis.25

Xenobiotic metals such as arse- nic, cadmium, lead, and mer- cury are also associated with atherosclerosis.29 Gene– environment interaction alters the risk of vascular disease30; for example, the residential proximity to highways (representing ex- posure to a mixture of traffic- related air pollutants) is associated with peripheral vascular disease,

which is modified by the gene encoding bone morphogenic protein.7,31

Because of the complexity of the drivers of atherosclerosis, a medical model treating blood pressure and high cholesterol and advising dietary modification and exercise will be inadequate to fully address this disease. Like- wise, identifying the chemicals that increase risk on an individual basis will be inadequate to pre- vent vascular disease. Instead an integrated systems approach is

Atherosclerosis

As TCDD

BaP Phthalate

PCB PM2.5

Cd

PCBs

TCDD

InflammationPlaque Growth

Chemical

Exposures

z

Oxidative Stress

Adipokine

Dysregulation

Dyslipidemia

Insulin

Resistance

Hyperglycemia

Myocardial Infarction Stroke

As

Cd

PCB

TCDD

BPA

BaP

Endothelial

Dysfunction

As

Cd

BPA

PM2.5 PCBs

TCDD

DEHP

PFOS

As

Cd

PM2.5 TCDD

Phthalates PCBs As

TCDD Cd

PFOS BPA

PM2.5

Age

Gender

Family

History

Diet

Smoking

Physical

Inactivity

Emotional

State

High BP

High LDL

Low HDL

Diabetes

Obesity

R is

k F

a c

to rs

f o

r A

th e

ro sc

le ro

si s,

C a

rd io

v a

sc u

la r

D is

e a

se , a

n d

C a

rd io

v a

sc u

la r

E v

e n

ts

Clinical

Intrinsic

Behavioral

A d

v e

rs e

O u

tc o

m e

P a

th w

a y

( A

O P

)

Clinical Events (AO)

Biochemical and Physiologic Responses (IKE)

Cellular Response (IKE)

Environmental Exposure

Subclinical and

Clinical Responses (IKE)

Molecular Initiating Event

Public Health Burden Mortality – Morbidity – Disability – Frailty

As PFOS

PCBs PM2.5

Individual Health Burden

Source. Action of specific chemicals and metals adapted from Kirkley and Sargis.26

Note. As = Arsenic; AO = adverse outcome; BaP = benzo[a]pyrene; BPA = bisphenol A; Cd = Cadmium; DEHP = di(2-ethylhexyl) phthlate; DES = diethylstilbestrol; HDL = high- density lipoprotein; IKE = intermediate key event; LDL = low-density lipoprotein; PCB = polychlorinated biphenyl; PFOS = perfluorooctane sulfonic acid; PM2.5 = particulate matter £ 2.5 mm; TCDD = tetrachlorodibenzo-p-dioxin.This figure illustrates the biological pathway leading from exposure to adverse cardiovascular outcomes for a variety of chemicals. On the left-hand side of the figure these pathways are linked to the adverse outcome pathway, and on the right-hand side of the figure we see the traditional risk factors for adverse cardiovascular outcomes.

FIGURE 2—Adverse Outcome Pathway for Cardiovascular Outcomes

AJPH RISK ASSESSMENT

1036 Analytic Essay Peer Reviewed Gwinn et al. AJPH July 2017, Vol 107, No. 7

needed to fully account for all known risk factors and formulate the problem to define the most effective strategy to decrease in- dividual risk and societal burden. Accomplishing this will require clinical data that fully reflect a population under consideration as well as exposures to traditional risk factors, biomonitoring data documenting exposures to mul- tiple chemicals, and molecular responses from in vitro and in vivo studies indicative of the activation of biochemical pathways that accelerate atherosclerosis.

Although this approach might not be practical currently, it is not unrealistic to think about future states where it could become standard practice. Our proposed innovative approach to chemical risk assessment is occurring contemporaneously during the formative stages of the National Institutes of Health–sponsored Precision Medicine Initiative, which will drive integration of genomics, data sciences, and bioinformatics as the basis for improved individual health care, disease prevention, and public health. The Affordable Care Act has accelerated electronic medical record adoption in health care practices and hospital systems, potentially offering a valuable source of information for population-level health monitoring. Recent research has used big data to study the early stages of disease and better classify and predict disease progression and could be used to inform personalized medicine to optimize wellness in healthy populations.32–34

Moreover, the anticipated integration and development of technologies and analytical tools have the potential to improve public health and increase the spatial and temporal resolution of environmental health

surveillance. The establishment of a long-term representative precision medicine cohort, if integrated with the proposed National Biomonitoring Net- work,35 could have enormous benefit in helping us understand the relationship between chem- ical exposures and disease and in managing some of the most challenging clinical problems more effectively.

Applying this framework would potentially expand our understanding of the origins of vascular disease and its progres- sion, helping define strategies for primary prevention to thwart the initiation of the process we ultimately call atherosclerosis. Thus, such a framework would provide new and ongoing insights into the associations between environmental expo- sures that contribute the greatest burden to public health. This approach would facilitate ac- counting for sensitive pop- ulations and could inform suggested individual health or behavioral measures in which there have been past exposures or in which current exposure cannot be reduced enough to protect those most at risk.

CONCLUSIONS The proposed conceptual

model is grounded in public health principles and focused on identifying the greatest oppor- tunity to reduce environmental exposures to improve health outcomes. Along with traditional risk assessment, this perspective can better inform public health decision-making. Although there are clear benefits to operating within a public health– focused framework and moving away from individual chemicals and apical endpoints, there are also challenges.

Informing Decision- Making

Since the 1980s, the Envi- ronmental Protection Agency’s decision-making has been grounded on traditional risk as- sessments that are conducted within the constraints of the Environmental Protection Agency’s statutes and programs. Although program-targeted risk assessments will remain an im- portant component, the disease- based approach draws on information in a holistic fashion that cuts across organizational and legal boundaries, integrating traditional inputs and newer data streams. These assessments will provide decision-makers with critical information to inform exposure-reduction efforts to affect the selected health out- comes and, ultimately, improve public health. Because those exposure-reduction efforts would take place within the existing statutory construct, an important implementation step would be to move from findings of disease-based risk assessments to assessments of specific risk management actions under the relevant statutory authorities.

Priorities for Screening and Testing

A health outcome–focused framework can inform priorities for screening and testing the toxicity of chemicals. Efforts to develop and synthesize ap- proaches for screening large numbers of chemicals using high-throughput toxicity testing and exposure prediction should continue to provide data for data-poor chemicals. For example, in the recently an- nounced Cancer Moonshot,36

high-throughput approaches could screen a large set of chemicals for potential carcino- genicity and identify a suite of

chemicals for additional animal toxicity testing.

Examining noncancer end- points will also be challenging, which is why developing adverse outcome pathways and networks to contextualize and interpret nonapical hazard data in relation to population health is of in- creasing value. Epidemiology studies can be designed to inform and validate high-throughput testing approaches by identifying both chemical stressors and nonchemical stressors that mod- ify responses to chemical expo- sures; they can also be designed to test relationships between disease and early markers of exposure and biological response (e.g., epigenetic changes).

The Impact of Cumulative Exposures

Although cumulative risk assessment has been of high in- terest for the past few decades, putting cumulative assessment approaches into practice has been challenging. This framework provides a new construct for considering cumulative risk. By focusing on a health endpoint of concern, one could consider the multiple exposures that may contribute to a health outcome. Past National Research Council recommendations have encour- aged assessors to evaluate the combined effects of exposures to all chemicals that affect a common adverse outcome, for example, male reproductive development.37 Challenges include gaining adequate un- derstanding of individual chemical effects to group chemicals by health outcome. Increased research into the biological pathways by which chemicals affect health status can help inform approaches for estimating the joint effect of chemicals without

AJPH RISK ASSESSMENT

July 2017, Vol 107, No. 7 AJPH Gwinn et al. Peer Reviewed Analytic Essay 1037

testing all permutations or combinations.

One example of an alternative approach is health impact assess- ment, which uses a systems ap- proach to array data sources and analytic methods and considers input from stakeholders to de- termine potential effects of a proposed action or decision on the health of a population and the distribution of those effects in the population.38 Using health impact assessment approaches for chemical risk assessments made through this framework can offer a method to organize various data streams that can influence our understanding of a health effect, inform potential multiple con- tributors to adverse health outcomes, and provide recom- mendations to decision-makers for monitoring and managing these outcomes.

Consider Public Health Concepts

This new approach takes a systematic view of collective factors that contribute to a health outcome or disease state, in- cluding those that are not reg- ulated by a single federal entity. Any single health outcome may be influenced by multiple factors beyond chemical exposures, such as nutrition, genetics, and social stressors. Because those factors are not regulated, it is important for environmental regulatory agencies to un- derstand what fraction of the disease burden is influenced by the regulated environmental exposure.

Public health approaches, such as attributable risk, can help inform this understanding. Challenges may include in- corporating these approaches, which are typically used in epi- demiology, to animal and ad- vanced toxicity testing data;

ensuring adequate training with the approaches; and communi- cating risk in a way that ac- knowledges the influence of nonregulated factors.

Public Health Implications

Understanding the health effects of chemicals has real im- plications for public health. This proposed approach for chemical risk assessment starts at the health endpoint and incorporates mul- tiple data streams, including data developed using newer tech- nologies such as high-throughput screening. In parallel with more traditional risk assessment ap- proaches, this will lead to a better understanding of mechanisms of single chemicals as well as cu- mulative exposures that lead to specific disease endpoints.

This new lens will need to be applied to the complete risk as- sessment process—problem for- mulation, data considerations, and data synthesis through mul- tipathway methods, including cumulative assessment and health impact assessment—with an eye to the prevention of adverse ef- fects. This approach draws on the best available science to improve our understanding of the health effects of environmental chem- icals and informs decision- making to prevent, reduce, or mitigate exposure and ultimately improve public health.

CONTRIBUTORS M. R. Gwinn led the compilation of text and revisions of the article. M. R. Gwinn and K. Deener led creation of Figure 1. M. R. Gwinn, K. Deener, and R. J. Kavlock led the creation of Table 1. D. A. Axelrad, T. Bahadori, D. Bussard, R. S. Thomas, and R. J. Kavlock led the drafting of text on challenges and op- portunities. T. Bahadori, R. S. Thomas, and R. J. Kavlock led the creation of Table 2. W. E. Cascio led the creation of Figure 2. W. E. Cascio and K. Deener led the drafting of text on public health. R. J. Kavlock and T. A. Burke led the discussions and outlined the article. All

authors contributed to discussions of the concepts behind this article and contrib- uted text to the various sections.

ACKNOWLEDGMENTS The authors would like to acknowledge the assistance of Laura Romano in pre- paring the article.

Note. The views expressed in this article are those of the authors and do not necessarily reflect the views or policies of the US Environmental Protection Agency.

REFERENCES 1. Birnbaum LS, Burke TA, Jones JJ. Informing 21st-century risk assessments with 21st-century science. Environ Health Perspect. 2016;124(4):A60–A63.

2. National Research Council. Toxicity Testing in the 21st Century: A Vision and a Strategy. Washington, DC: National Academies Press; 2007.

3. Kavlock RJ, Austin CP, Tice RR. Toxicity testing in the 21st century: im- plications for human health risk assess- ment. Risk Anal. 2009;29(4):485–487; discussion 492–497. [Comment on Toxicity testing in the 21st century: im- plications for human health risk assess- ment. Risk Anal. 2009]

4. Tice RR, Austin CP, Kavlock RJ, Bucher JR. Improving the human hazard characterization of chemicals: a Tox21 update. Environ Health Perspect. 2013; 121(7):756–765.

5. Kavlock R, Chandler K, Houck K, et al. Update on EPA’s ToxCast program: providing high throughput decision sup- port tools for chemical risk management. Chem Res Toxicol. 2012;25(7):1287–1302.

6. Browne P, Judson RS, Casey WM, Kleinstreuer NC, Thomas RS. Screening chemicals for estrogen receptor bioactivity using a computational model. Environ Sci Technol. 2015;49(14):8804–8814.

7. Wu S, Powers S, Zhu W, Hannun YA. Substantial contribution of extrinsic risk factors to cancer development. Nature. 2016;529(7584):43–47.

8. Weinhold B. Environmental factors in birth defects: what we need to know. Environ Health Perspect. 2009;117(10): A440–A447.

9. Rappaport SM, Barupal DK, Wishart D, Vineis P, Scalbert A. The blood exposome and its role in discovering causes of disease. Environ Health Perspect. 2014;122(8):769–774.

10. Dietert RR, Silbergeld EK. Bio- markers for the 21st century: listening to the microbiome. Toxicol Sci. 2015;144(2): 208–216.

11. Patterson AD, Turnbaugh PJ. Mi- crobial determinants of biochemical in- dividuality and their impact on toxicology and pharmacology. Cell Metab. 2014; 20(5):761–768.

12. Servick K. Of mice and microbes. Science. 2016;353(6301):741–743.

13. Stel J, Legler J. The role of epigenetics in the latent effects of early life exposure to obesogenic endocrine disrupting chem- icals. Endocrinology. 2015;156(10): 3466–3472.

14. Buck Louis GM, Bloom MS, Gatto NM, Hogue CR, Westreich DJ, Zhang C. Epidemiology’s continuing contribution to public health: the power of “then and now.” Am J Epidemiol. 2015;181(8): e1–e8.

15. Ankley GT, Bennett RS, Erickson RJ, et al. Adverse outcome pathways: a con- ceptual framework to support ecotoxicol- ogy research and risk assessment. Environ Toxicol Chem. 2010;29(3):730–741.

16. Goodson WH III, Lowe L, Carpenter DO, et al. Assessing the carcinogenic potential of low-dose exposures to chemical mixtures in the environment: the challenge ahead. Carcinogenesis. 2015; 36(suppl 1):S254–S296.

17. Mansouri K, Abdelaziz A, Rybacka A, et al. CERAPP: Collaborative Estrogen Receptor Activity Prediction Project. Environ Health Perspect. 2016;124(7): 1023–1033.

18. Judson R, Houck K, Martin M, et al. Analysis of the effects of cell stress and cytotoxicity on in vitro assay activity across a diverse chemical and assay space. Toxicol Sci. 2016;153(2):409. [Erratum in Analysis of the effects of cell stress and cytotoxicity on in vitro assay activity across a diverse chemical and assay space. Toxicol Sci. 2016]

19. Bakker KM, Martinez-Bakker ME, Helm B, Stevenson TJ, et al. Digital ep- idemiology reveals global childhood disease seasonality and the effects of im- munization. Proc Natl Acad Sci USA. 2016; 113(24):6689–6694.

20. National Academies of Sciences, En- gineering, and Medicine. Using 21st Century Science to Improve Risk-Based Evaluations. Washington, DC: National Academies Press; 2017.

21. Moran AE, Forouzanfar MH, Roth GA, et al. Temporal trends in ischemic heart disease mortality in 21 world regions, 1980 to 2010: the Global Burden of Disease 2010 study. Circulation. 2014; 129(14):1483–1492.

22. Mozaffarian D, Benjamin EJ, Go AS, et al. Heart disease and stroke statistics— 2016 update: a report from the American Heart Association. Circulation. 2016; 133(4):e38–e360.

23. US Environmental Protection Agency. Guidelines for Preparing Economic Analyses. Washington, DC; 2010.

24. Kannel WB, Vasan RS. Adverse consequences of the 50% misconception. Am J Cardiol. 2009;103(3):426–427.

25. Kaufman JD, Adar SD, Barr RG, et al. Association between air pollution and

AJPH RISK ASSESSMENT

1038 Analytic Essay Peer Reviewed Gwinn et al. AJPH July 2017, Vol 107, No. 7

coronary artery calcification within six metropolitan areas in the USA (the Multi-Ethnic Study of Atherosclerosis and Air Pollution): a longitudinal cohort study. Lancet. 2016;388(10045): 696–704.

26. Kirkley AG, Sargis RM. Environ- mental endocrine disruption of energy metabolism and cardiovascular risk. Curr Diab Rep. 2014;14(6):494.

27. Sidney S, Quesenberry CP Jr, Jaffe MG, et al. Recent trends in cardiovas- cular mortality in the United States and public health goals. JAMA Cardiol. 2016; 1(5):594–599.

28. Cosselman KE, Navas-Acien A, Kaufman JD. Environmental factors in cardiovascular disease. Nat Rev Cardiol. 2015;12(11):627–642.

29. Solenkova NV, Newman JD, Berger JS, et al. Metal pollutants and cardiovas- cular disease: mechanisms and conse- quences of exposure. Am Heart J. 2014; 168(6):812–822.

30. Zanobetti A, Baccarelli A, Schwartz J. Gene-air pollution interaction and car- diovascular disease: a review. Prog Cardiovasc Dis. 2011;53(5):344–352.

31. Ward-Caviness CK, Neas LM, Blach C, et al. Genetic variants in the bone morphogenic protein gene family modify the association between residential ex- posure to traffic and peripheral arterial disease. PLoS One. 2016;11(4):e0152670.

32. Dinov ID, Heavner B, Tang M, et al. Predictive big data analytics: a study of Parkinson’s disease using large, complex, heterogeneous, incongruent, multi- source and incomplete observations. PLoS One. 2016;11(8):e0157077.

33. Hood L, Lovejoy JC, Price ND. In- tegrating big data and actionable health coaching to optimize wellness. BMC Med. 2015;13:4.

34. Khoury MJ, Evans JP. A public health perspective on a national precision med- icine cohort: balancing long-term knowledge generation with early health benefit. JAMA. 2015;313(21): 2117–2118.

35. Association of Public Health Labora- tories. National Biomonitoring Plan. Silver Spring, MD; 2015.

36. Mitchell E. Moonshot toward a cure for cancer. J Natl Med Assoc. 2016;108(2): 104–105.

37. National Research Council. Phthalates and Cumulative Risk Assessment: The Tasks Ahead. Washington, DC: National Academies Press; 2008.

38. National Research Council. Improving Health in the United States: The Role of Health Impact Assessment. Washington, DC: National Academies Press; 2011.

AJPH RISK ASSESSMENT

July 2017, Vol 107, No. 7 AJPH Gwinn et al. Peer Reviewed Analytic Essay 1039

Copyright of American Journal of Public Health is the property of American Public Health Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.