AHA
United States
Environmental Protection
Agency
A Framework for Assessing
Health Risk of Environmental
Exposures to Children
-------
-------
EPA/600/R-05/093F
September 2006
A Framework for Assessing Health Risks of Environmental
Exposures to Children
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free
-------
DISCLAIMER
This document has been reviewed in accordance with U.S. Environmental Protection
Agency policy and approved for publication. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
Preferred Citation:
U.S. Environmental Protection Agency (EPA). (2006) A framework for assessing health risks of
environmental exposures to children. National Center for Environmental Assessment,
Washington, DC; EPA/600/R-05/093F. Available from the National Technical Information
Service, Springfield, VA, and online at http://www.epa.gov/ncea.
-------
CONTENTS
LIST OF TABLES vi
LIST OF FIGURES vii
LIST OF ABBREVIATIONS AND ACRONYMS viii
PREFACE ix
AUTHORS, CONTRIBUTORS, AND REVIEWERS xii
1. EXECUTIVE SUMMARY 1-1
2. INTRODUCTION AND PURPOSE 2-1
3. LIFESTAGE-SPECIFIC PROBLEM FORMULATION 3-1
3.1. PLANNING AND SCOPING 3-1
3.2. CONCEPTUAL MODEL 3-5
3.2.1. Exposure Considerations 3-5
3.2.2. Outcome Considerations 3-8
3.2.3. Integrating Exposure and Outcome Considerations 3-9
3.3. ANALYSIS PLAN 3-10
4. LIFESTAGE-SPECIFIC ANALYSIS 4-1
4.1. LIFESTAGE-SPECIFIC HAZARD CHARACTERIZATION 4-2
4.1.1. Introduction 4-2
4.1.2. Qualitative Evaluation of Individual Studies 4-3
4.1.2.1. Study Purpose 4-5
4.1.2.2. Study Design 4-5
4.1.2.3. Identifying Critical Windows of Exposures 4-6
4.1.2.4. Outcomes Related to Developmental Lifestage Exposure 4-6
4.1.2.5. Toxicokinetic Data 4-7
4.1.2.6. Toxicodynamic Data 4-7
4.1.2.7. Mode of Action Information 4-8
4.1.2.8. Qualitative Evaluation of Dose-Response 4-8
4.1.2.9. Variability Analysis 4-9
4.1.2.10. Uncertainty Analysis 4-10
4.1.3. Summarization of the Hazard Database 4-12
4.1.3.1. Evaluation of the Weight-of-Evidence of the Hazard
Database 4-13
4.1.3.1.1. Temporality 4-14
4.1.3.1.2. Strength of the association 4-15
4.1.3.1.2.1. Variability analysis 4-16
4.1.3.1.2.2. Uncertainty analysis 4-16
4.1.3.1.3. Qualitative dose-response relationship 4-17
4.1.3.1.4. Experimental evidence 4-18
4.1.3.1.5. Reproducibility 4-18
4.1.3.1.6. Biological plausibility 4-19
4.1.3.1.7. Alternative or multiple explanations 4-20
in
-------
CONTENTS (continued)
4.1.3.1.8. Specificity 4-20
4.1.3.1.9. Coherence 4-21
4.1.4. Iteration with Dose-Response and Exposure Characterization 4-23
4.1.5. Lifestage-Specific Hazard Characterization Narrative 4-24
4.2. LIFESTAGE-SPECIFIC DOSE-RESPONSE CHARACTERIZATION 4-25
4.2.1. Introduction 4-25
4.2.2. Mode of Action Conceptualization 4-28
4.2.2.1. Summarizing the Available Dose-Response Data 4-28
4.2.2.2. Mechanistic Data and Mode of Action (MOA) 4-30
4.2.2.3. Selection of Dose Metric Informed by MOA 4-30
4.2.3. Analysis in the Range of Observation and Dose-Response Models 4-32
4.2.4. Extrapolations and Risk Derivation from a Lifestage Approach 4-38
4.2.4.1. Duration and Route Adjustments 4-38
4.2.4.2. Interspecies and Intraspecies Adjustments 4-39
4.2.4.3. Low-Dose Extrapolation 4-43
4.2AA. Reference and Risk Value Derivation 4-44
4.2.5. Variability Analysis 4-45
4.2.6. Sensitivity Analysis 4-46
4.2.7. Uncertainty Analysis 4-46
4.2.8. Iteration with Hazard and Exposure Characterization 4-47
4.2.9. Lifestage-Specific Dose-Response Characterization Narrative 4-48
4.3. LIFESTAGE-SPECIFIC EXPOSURE CHARACTERIZATION 4-48
4.3.1. Introduction 4-48
4.3.2. Evaluation of Available Exposure Data 4-49
4.3.2.1. Chemical Properties, Environmental Sources, Fate, and Transport ....4-54
4.3.2.2. Environmental Media Concentrations 4-55
4.3.2.3. Lifestage-Specific Exposure Measurement Data 4-55
4.3.2.4. Lifestage-Specific Exposure Factors 4-57
4.3.2.5. Cumulative Evaluation of Environmental Stressors 4-59
4.3.3. Lifestage-Specific Exposure Analysis 4-60
4.3.3.1. Exposure Measurement and Estimation Approach 4-61
4.3.3.2. Analysis Level or Tiered Assessment 4-62
4.3.3.2.1. Screening assessment 4-63
4.3.3.2.2. Refined assessment 4-64
4.3.3.2.3. Supplemental data collection 4-65
4.3.4. Variability Analysis 4-66
4.3.5. Sensitivity Analysis 4-67
4.3.6. Uncertainty Analysis 4-67
4.3.7. Iteration with Hazard and Dose-Response Characterization 4-68
4.3.8. Lifestage-Specific Exposure Characterization Narrative 4-69
IV
-------
CONTENTS (continued)
5. LIFESTAGE-SPECIFIC RISK CHARACTERIZATION 5-1
5.1. LIFESTAGE-SPECIFIC RISK CHARACTERIZATION SUMMARY 5-3
5.1.1. Key Information from the Analysis Phase 5-3
5.1.2. Scientific Assumptions 5-4
5.1.3. Risk Drivers 5-4
5.1.4. Strengths and Weaknesses 5-4
5.1.4.1. Variability 5-5
5.1.4.2. Sensitivity 5-5
5.1.4.3. Uncertainty 5-6
5.1.5. Key Conclusions 5-6
5.1.6. Alternative Risk Estimates Considered 5-7
5.1.7. Research Needs 5-7
5.2. RISK CONTEXT 5-8
6. SUMMARY AND IDENTIFICATION OF GAPS IN APPROACHES FOR
CHILDREN'S HEALTH RISK ASSESSMENT 6-1
GLOSSARY G-1
REFERENCES R-l
-------
LIST OF TABLES
3-1. Developmental lifestages and age groups for exposure assessments 3-7
3-2. Lifestage-specific database inventory sheet 3-11
4-1. Examples of lifestage-specific questions for evaluation of individual studies
within hazard characterization 4-11
4-2. Examples of lifestage-specific questions for evaluation of the WOE of the
hazard database 4-21
4-3. Examples of lifestage-specific questions for MOA conceptualization 4-31
4-4. Examples of lifestage-specific questions for analysis in the range of observation 4-38
4-5. Examples of lifestage-specific questions for extrapolations and risk derivation 4-45
4-6. Examples of lifestage-specific questions for dose-response variability,
sensitivity, and uncertainty analyses 4-47
4-7. Examples of lifestage-specific questions for scenario development 4-53
4-8. Examples of lifestage-specific questions for evaluation of the available
exposure data 4-59
4-9. Examples of lifestage-specific questions for exposure analysis level or tiered
assessment 4-65
4-10. Examples of lifestage-specific questions for exposure variability, sensitivity,
and uncertainty analyses 4-68
VI
-------
LIST OF FIGURES
2-1. Flow diagram for a lifestage-specific risk assessment framework 2-2
2-2. Lifestages of outcomes after developmental exposure 2-4
3-1. Flow diagrams for lifestage-specific problem formulation 3-2
4-1. Exposure to risk continuum 4-1
4-2. Flow diagram for lifestage-specific analysis 4-2
4-3. Flow diagram for lifestage-specific hazard characterization 4-4
4-4. Conceptual view of a WOE evaluation 4-14
4-5. Flow diagram for lifestage-specific dose-response characterization 4-26
4-6. Use of BBDR modeling 4-37
4-7. Interspecies and intraspecies adjustments with lifestage considerations 4-41
4-8. Flow diagram for lifestage-specific exposure characterization 4-50
4-9. Exposure routes during developmental lifestages 4-52
5-1. Flow diagram for lifestage-specific risk characterization 5-1
vn
-------
LIST OF ABBREVIATIONS AND ACRONYMS
ADAF
AUC
BBDR
BMD
BMDL
CAA
CatReg
^rnax
CHAD
CSF
CYP
DAF
EPA
FIFRA
FQPA
GLP
HEC
HED
HEDS
LOAEL
MOA
MOE
NHEXAS
NOAEL
PBTK
POD
QRE
RfC
RfD
RfV
SAR
TD
TK
TSCA
UF
'max
WOE
Age-dependent adjustment factors
Area under the curve
Biologically based dose-response
Benchmark dose
Benchmark dose lower confidence level
Clean Air Act
Categorical regression
Maximum concentration
Consolidated Human Activity Database
Cancer slope factor
Cytochrome P450
Dosimetric adjustment factor
U.S. Environmental Protection Agency
Federal Insecticide, Fungicide, and Rodenticide Act
Food Quality Protection Act
Good laboratory practice
Human equivalent concentration
Human equivalent dose
Human Exposure Database System
Lowest-observed-adverse-effect level
Mode of action
Margin of exposure
National Human Exposure Assessment Survey
No-observed-adverse-effect level
Physiologically based toxicokinetic
Point of departure
Quantitative risk estimation
Reference concentration
Reference dose
Reference value
Structure-activity relationship
Toxicodynamic
Toxicokinetic
Toxic Substances Control Act
Uncertainty factor
Maximum velocity
Weight-of-evidence
Vlll
-------
PREFACE
The mission of the U.S. Environmental Protection Agency (EPA) is to protect human
health and the environment. In the early 1990s, the National Research Council (NRC) released a
watershed report, Pesticides in the Diets of Infants and Children, regarding evaluation of risk to
environmental exposures (NRC, 1993). Increased emphasis on protecting children from
environmental exposures has evolved since this report due to mounting scientific evidence to
support the vulnerability of the developing fetus and child. Legislative and administrative
mandates have been enacted since this NRC report. In 1995, the EPA Administrator issued
Policy on Evaluating Health Risks to Children (U.S. EPA, 1995a), which states that EPA will
consider risks to infants and children consistently and explicitly as a part of risk assessments
generated during its decision-making process, including the setting of standards to protect public
health and the environment. Subsequent provisions in the Food Quality Protection Act (FQPA)
(U.S. 104th Congress, 1996a) and the Safe Drinking Water Act (SDWA) Amendments (U.S.
104th Congress, 1996b) underscored this policy by focusing on the evaluation of children's
exposures and toxicities in the context of risk assessment. Evaluation of environmental risks to
children is an implicit consideration in human health risk assessment in other EPA legislative
mandates (Clean Air Act [CAA] [U.S. 101st Congress, 1990]; Comprehensive Environmental
Response, Compensation, and Liability Act [CERCLA] [U.S. 96th Congress, 1980], Toxic
Substances Control Act [TSCA] [ U.S. 94th Congress, 1976], Federal Insecticide, Fungicide, and
Rodenticide Act [FIFRA] [U.S. 104th Congress, 1996c]). In 1997, Presidential Executive Order
13045, Protection of Children from Environmental Health Risks and Safety Risks (Executive
Order, 1997), gave further emphasis to the need for establishing potential risks from
environmental exposures during childhood. The EPA subsequently published Strategy for
Research on Environmental Risks to Children in 2000 (U.S. EPA, 2000f).
EPA risk assessment guidelines relevant to children's health issues have been published
(U.S. EPA, 1991, 1996, 1998b, 2002a, 2005a,b,e), and other guidelines, policies, and
recommendations are under development (U.S. EPA, 2002c). Implementation of the FQPA and
the SDWA amendments required additional development of guidance and policy for protecting
children's health. In response, the application of the FQPA 10-fold safety factor was discussed
in the Determination of the Appropriate FQPA Safety Factors(s) in Tolerance Assessment (U.S.
EPA, 2002d). Thus, there are a number of guidelines and policies related to children's health,
IX
-------
but no single, comprehensive document that can serve as a resource of information on children's
health risk assessment.
In 1999, a draft report that collected information on current EPA guidance and practices
was developed for the Office of Children's Health Protection (ICF Consulting, 1999). This
report was a compendium of information on child-related risk assessment policy and
methodology guidance at the time. This Framework document builds on that report and others
referred to above by updating the information and linking to reference documents and other
published information that can be used as a resource for those interested in children's health risk
assessment.
Another major effort sponsored by EPA and others that serves as background for this
document was a workshop held in Stowe, VT, July 30-August 2, 2001, organized by the
International Life Sciences Institute (ILSI). The report of that workshop (ILSI, 2003) and
subsequent publications (Daston et al., 2004; Ginsberg et al., 2004c; Landrigan et al., 2004;
Morford et al., 2004; Olin and Sonawane, 2003) proposed a framework for children's exposures
and health risk assessment and laid out a number of issues of concern. The Framework
presented in this document builds on the efforts of the experts and participants at that workshop.
Parallel activities have been or are being developed at other agencies such as the U.S.
Food and Drug Administration (FDA), which regulates Pharmaceuticals, medical devices,
biologies, food, animal feed and drugs, cosmetics, radiation-emitting devices, and combination
products. For example, under the Best Pharmaceuticals for Children Act (U.S. FDA, 2002), an
amendment to Section 11 of the Food and Drug Modernization Act (U.S. FDA, 1997), FDA's
Office of Pediatric Therapeutics coordinates and facilitates all activities affecting the pediatric
population, the practice of pediatrics, or pediatric issues within the FDA. Assessment of risks
and benefits to children is conducted in compliance with the Pediatric Research Equity Act (U.S.
108th Congress, 2003), which requires that all applications for new active ingredients indications,
dosage forms, dosing regimens, and routes of administration contain a pediatric assessment
unless a waiver or deferral has been granted. Although the draft guidance document Guidance
for Industry - How to Comply with the Pediatric Research Equity Act (U.S. FDA, 2005) may
apply specifically to pharmaceutical testing and regulation, there can be significant overlap with
assessments conducted to determine risk to children from environmental exposures. For
example, Guidance to Industry - Nonclinical Safety Evaluation of Pediatric Drug Products (U.S.
-------
FDA, 2006) addresses considerations on the evaluation of Pharmaceuticals in juveniles, one of
the lifestages discussed in this Framework.
Additionally, the International Programme for Chemical Safety of the World Health
Organization recently developed a draft Environmental Health Criteria document entitled
Principles for Evaluating Health Risks Associated with Chemical Exposures to Children. This
EHC draft document serves as useful background information for using this EPA Framework.
Finally, EPA's Risk Assessment Forum has been working for several years to harmonize
approaches to cancer and noncancer risk assessment (U.S. EPA, 1997c, 1998c). Efforts to
develop a framework for a harmonized approach to human health risk assessment are underway,
and the intent is for this Framework on health risks from environmental exposures to children to
be incorporated into the overall framework.
XI
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
AUTHORS
Stan Barone Jr., National Center for Environmental Assessment, Office of Research and
Development, U.S. Environmental Protection Agency, Washington, DC.
Rebecca C. Brown, National Center for Environmental Assessment, Office of Research and
Development, U.S. Environmental Protection Agency, Washington, DC.
Susan Y. Euling, National Center for Environmental Assessment, Office of Research and
Development, U.S. Environmental Protection Agency, Washington, DC.
Elaine Cohen Hubal, National Center for Computational Toxicology, Office of Research and
Development, U.S. Environmental Protection Agency, Research Triangle Park, NC.
Carole A. Kimmel, formerly of National Center for Environmental Assessment, Office of
Research and Development, U.S. Environmental Protection Agency, Washington, DC.
Susan Makris, National Center for Environmental Assessment, Office of Research and
Development, U.S. Environmental Protection Agency, Washington, DC.
Hisham El-Masri, National Health and Environmental Effects Research Laboratory, Office of
Research and Development, U.S. Environmental Protection Agency, Research Triangle Park,
NC.
Jacqueline Moya, National Center for Environmental Assessment, Office of Research and
Development, U.S. Environmental Protection Agency, Washington, DC.
Sherry G. Selevan, formerly of National Center for Environmental Assessment, Office of
Research and Development, U.S. Environmental Protection Agency, Washington, DC.
Babasaheb R. Sonawane, National Center for Environmental Assessment, Office of Research
and Development, U.S. Environmental Protection Agency, Washington, DC.
Tracey Thomas, formerly of American Association for the Advancement of Science Fellow,
National Center for Environmental Assessment, Office of Research and Development, U.S.
Environmental Protection Agency, Washington, DC.
Chad Thompson, National Center for Environmental Assessment, Office of Research and
Development, U.S. Environmental Protection Agency, Washington, DC.
INVITED INTERNAL AGENCY PANEL REVIEWERS
Framework for Children's Health Risk Assessment (CHRA)1 and Harmonization of Human
Health Risk Assessment Colloquium, October 5-6, 2004, Arlington, VA
Hugh Barton, ORD, NCCT
Jerry Blancato, ORD, NCCT
Vicki Dellarco, OPPTS, OPP, HED
Framework for Children's Health Risk Assessment was the previous title of this document.
xn
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)
Elizabeth Doyle, OW, OST, HECD
Brenda Foos Perkovich, OA, OCHP
Greg Miller, OA, NCEE, OPEI
Deirdre Murphy, OAR, OAQPS, ESD
Marian Olsen, Region 2
Jennifer Seed, OPPTS, OPPT, RAD
Linda Sheldon, ORD, NERL, HEASD
Daniel Stralka, Region 9
Vanessa Vu, OA, SAB
ADDITIONAL INTERNAL AGENCY REVIEWERS
Thomas Bateson, ORD, NCEA
Ila Cote, ORD, NCEA
Michael Firestone, OA, OCHP
Matt Heberling, ORD, NCEA
Sarah Levinson, Region 1
Kelly Maguire, OA, NCEE, OPEI
Lanelle Wiggins, OA, NCEE, OPEI
Tracey Woodruff, OA, NCEE, OPEI
William Wood, ORD, NCEA
INTERNAL AGENCY REVIEW PROJECT OFFICER
Marilyn Brower, formerly RAF
EXTERNAL PEER PANEL REVIEWERS
External Peer Review Workshop for Draft Framework for Assessing Health Risks of
Environmental Exposures to Children, June 6-7, 2006, Washington, DC
James V. Bruckner, College of Pharmacy, University of Georgia, Athens, GA
Gary Ginsberg, Connecticut Department of Public Health, Hartford, CT
Lynn R. Goldman, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD
Melanie Marty, California Environmental Protection Agency, Oakland, CA
P. Barry Ryan, Rollins School of Public Health, Emory University, Atlanta, GA
Robin Whyatt, Columbia University School of Public Health, New York, NY
PUBLIC COMMENTERS
Richard A. Becker, American Chemistry Council, Arlington, VA
Shannon Cunniff, Department of Defense, Arlington, VA
Maxene R. Dwyer and Allison Jenkins, Tetra Tech EM Inc., Houston, TX
B. Sachau, Florham Park, NJ
Scott Slaughter, Center for Regulatory Effectiveness, Washington, DC
2
Chair of the external peer review panel.
Xlll
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)
EXTERNAL AGENCY REVIEW PROJECT OFFICER
Stan Barone Jr., ORD, NCEA
ACKNOWLEDGEMENTS
Jonathan Francis, formerly ORD, NCEA
Elizabeth Fryer, ECFlex, Contractor to NCEA-Cin
Maureen Johnson, ORD, NCEA
Terri Konoza, ORD, NCEA
Lana Wood, ECFlex, Contractor to NCEA-Cin
Bette Zwayer, ORD, NCEA-Cin
xiv
-------
1. EXECUTIVE SUMMARY
The purpose of this document is to provide an overarching framework for a more
complete assessment of children's exposure to environmental agents and the resulting potential
health risks within the U.S. Environmental Protection Agency's (EPA's) risk assessment
paradigm. This Framework examines the impact of potential exposures during developmental
lifestages and subsequent lifestages, while emphasizing the iterative nature of the analysis phase
with a multidisciplinary team. In addition to outlining the risk assessment process from a
lifestage perspective, the document points to published sources for more detailed information.
Guidance, policies, and other relevant materials are referenced in the document and linked
electronically (when copyright allows) to the actual reference documents for easy access. In
addition, many terms are included in a glossary at the end of this document. This Framework is
a conceptual overview of the considerations for evaluation of early-life exposures and
subsequent outcomes and does not constitute EPA guidance defined as a step-by-step process or
standard operating procedure.
The term "children" as used in this document is shorthand to include the stages of
development from conception through adolescence. EPA is concerned about health risks that
result from exposure to all lifestages; however, this document focuses on preconceptional
exposure and exposure throughout development to adulthood. Developmental exposure is used
throughout this document to define developmental lifestage exposures (preconception through
adolescence). Health risks may be identified during the same lifestage as when the exposure
occurred, or they may not become apparent until much later in life.
Lifestages are defined in this document as temporal stages of life that have distinct
anatomical, physiological, and behavioral or functional characteristics that contribute to potential
differences in vulnerability to environmental exposures. A lifestage approach to risk assessment
considers the relevant periods of exposure in developmental lifestages and subsequent outcomes
that may not be expressed until later lifestages. This approach explicitly considers existing data
as well as data gaps for both exposure and health outcomes at various lifestages.
Information on mode(s) of action (MOA) that may inform lifestages is another main
emphasis of this approach. Risk assessment using a lifestage approach is a shift in perspective
1-1
-------
from the current methodology that focuses primarily on adults, and then, secondarily, looks for
information that may suggest greater susceptibility from exposures at other lifestages.
The added value of using a lifestage approach to risk assessment is a more
comprehensive evaluation of the potential for vulnerability of the population at various
lifestages. Children may be more or less vulnerable than adults, but without data on exposure
and response and without systematic evaluation of these data, determining which lifestage may
be more vulnerable is challenging. The approach outlined here encourages evaluation of the
potential for toxicity and any adverse health outcomes during all developmental lifestages, based
on knowledge of external exposure, critical windows of development for different organ systems,
MO As, anatomy, physiology, and behavior that can affect external exposure and internal dose
metrics (units of measurement for dose). The use of MOA information is integral to this
Framework and is employed in a consistent manner to the Guidelines for Carcinogen Risk
Assessment (U.S. EPA, 2005a) and the Supplemental Guidance for Assessing Susceptibility from
Early-Life Exposure to Carcinogens (2005b). The MOA information is extended to the
evaluation of all outcomes.
It is important to consider whether anything known about developmental lifestages would
indicate particular vulnerability and incorporate that information into an assessment. This
framework addresses the difficult issue of integrating toxicity data and exposure information,
which is especially challenging when data are limited for particular lifestages (e.g., pregnancy
and early childhood development).
The conceptual framework used in this document follows the basic components
developed for other areas of risk assessment (U.S. EPA, 1997a, 1998a, 2003a) and includes
problem formulation, analysis, and risk characterization as the three major phases in the process.
Within this structure, questions for consideration in the process of scoping the problem to be
addressed, reviewing the toxicity and exposure data, and characterizing the risks are posed as a
way of prompting and refining the assessment process. Gaps in guidance needed for various
aspects of assessing risk from children's exposure are also discussed. In particular, guidance is
lacking for lifestage-specific evaluation of several system- and disease-specific areas, related
biomarkers and outcomes, MOA(s), dose-response assessment, and exposure assessment. Also,
guidance on the use of specific developmental or latent outcomes for application to risk
assessments for various timing (exposure windows) and durations of exposure has not been
1-2
-------
defined even though this issue is considered in many of the risk assessments currently being
generated across EPA. Implementation of this Framework will necessitate development of
guidance for children's health risk assessment.
1-3
-------
-------
2. INTRODUCTION AND PURPOSE
The purpose of this document is to provide an overarching framework for a more
complete assessment of children's exposure to environmental agents and the resulting potential
health risks. The term "children" as used in this document is shorthand to include the stages of
development from conception through adolescence. EPA is concerned about health risks that
result from exposure to all lifestages; however, this document focuses on exposures during
preconception through adolescence. Developmental exposure, as used in this document, means
developmental lifestage exposures (preconception through adolescence). Health risks may be
identified during the same lifestage as when the exposure occurred, or they may not become
apparent until much later in life.
The major encompassing question to be addressed by using this document is, What is the
potential risk of environmental exposure during developmental lifestages? This Framework
outlines the phases for assessing the risks of exposure to environmental agents during childhood,
singly or in combination. This information can be used in various situations, depending on the
problem to be addressed. For example, if an overall assessment of health risks is needed, the
information on risks from developmental lifestage exposures can be incorporated into the overall
assessment. If, on the other hand, the major concern is about health risks to children as a result
of environmental exposure, the information derived from this process could be used directly to
assess risk, set standards, and mitigate exposures.
In addition to outlining the process of assessing health risks as a result of environmental
exposure during childhood, this framework uses existing sources for more detailed information
which are referenced and linked to the actual reference documents (when copyright allows).
These sources include guidelines, guidance documents, policies, and other relevant published
materials that currently exist. This document incorporates this information while focusing on
inherent and acquired susceptibility at different lifestages (e.g., children and adults), as well as
the potential for greater exposure of environmental agents to children than adults.
The outline of this document follows the basic framework developed for other areas of
risk assessment (U.S. EPA, 1997a, 1998a, 2003a) and includes problem formulation, analysis,
and risk characterization as the three major phases in the process, each with a focus on lifestages
(Figure 2-1, adapted from Daston et al., 2004; Olin and Sonawane, 2003). Each phase of the
2-1
-------
Lifestage-Specific Problem Formulation
(Chapter 3)
Planning and Scoping
(Section 3.1)
/ Conceptual Model
I (Section 3.2)
v^
Analysis Plan \
(Section 3.3) J
Lifestage-Specific Analysis
(Chapter 4)
/^Lifestage-Specific
( Hazard
I Characterization
\. (Section 4.1)
Lifestage-Specific
Dose-Response
Characterization
(Section 4.2)
Lifestage-Specific ^\.
Exposure A
Characterization J
(Section 4.3) S
Lifestage-Specific Risk Characterization
(Chapter 5)
Lifestage-Specific Risk
Characterization
Summary
(Section 5.1)
Risk Context
(Section 5.2)
o
n
en
CO
03
O
o
8
03
O
P
O.
o
-
Risk Communication/Management
Figure 2-1. Flow diagram for a lifestage-specific risk assessment framework. This diagram
presents the framework for lifestage-specific risk assessment used in this document. It is based on
a number of documents on children's health risk assessment, including the ILSl workshop (Daston
et al, 2004; Olin and Sonawane, 2003). It includes three phases also identified in Guidelines for
Ecological Risk Assessment (U.S. EPA, 1998a) and Framework for Cumulative Risk Assessment
(U.S. EPA, 2003a).
2-2
-------
process raises questions to consider when assessing potential health risks to children from
environmental exposure. Assessing potential health risks to children as a result of their
environmental exposure to toxicants includes considering risk from exposure before conception,
during the prenatal period, and through childhood and adolescence (Figure 2-2). Lifestages are
defined in this document as periods of life with distinct anatomical, physiological, and
behavioral or functional characteristics that contribute to potential differences in vulnerability to
environmental exposures. Preconception is any time before conception; the prenatal stage
includes the embryonic and fetal stages from conception to birth; infancy is the period from birth
through the first birthday; child encompasses all early postnatal lifestages from birth until
adolescence, which occurs approximately between 12 and 21 years of age (with difference
between genders). The continuum between the reproductive-age adult and aged adult begins at
approximately 21 years of age and reaches aged adulthood at approximately 65 years. Broad
exposure interval categories (e.g., child) are shown in Figure 2-2 for illustration, and divisions
between lifestages are not precise (e.g., there is some reproductive age overlap between the
adolescent and the adult periods) (U.S. EPA, 2005c, 2002a, Table 3-1). The lifestages from
conception through adolescence comprise the period of development; adverse outcomes may
occur during that same lifestage or later in life. Neither the outcomes nor the risks from these
developmental exposures will necessarily be the same for all lifestages. Rather, the outcomes
will depend on the underlying developmental processes that determine susceptibility at the time
of exposure. A lifestage approach for evaluating potential risks to children is a hypothesis-
driven approach that takes into account all relevant periods of exposure explicitly considering
where data do and do not exist for exposure and health outcomes. It focuses attention on
considerations of early-life exposure and potential outcomes, which may be latent in their
manifestation. This is predicated on considerations of MOA(s) for all lifestages of exposures.
MOA is defined in this Framework as the sequence of key events and processes, starting with
interaction of a toxic agent with a cell, proceeding through functional and anatomical changes,
and resulting in the adverse health outcomes. "A key event is an empirically observable
precursor step that is itself a necessary element of the MOA or is a biologically based marker for
such an element" (U.S. EPA, 2005a,b). Both toxicokinetic (TK) and toxicodynamic (TD) steps
are part of the mechanism and MOA leading to the toxic response (Andersen et al., 2000;
Clewell et al., 2002a). As stated in the latest cancer guidelines, "MOA is contrasted with
2-3
-------
Exposure
Period
Preconception
Prenatal
Infant
^ ,
Toxicokinetics Toxicodynamics
_X A
c
Internal
Dose
^
Biologically
Effective
Dose
f ^\
Precursor
Events/Early
Biological
Effects
Altered
Structure or
Function
Clinical
Manifestation
or Outcome
Risk:
Prognostic
Significance or
Adversity
Preconception
Prenatal
Infant
O
H
Reproductive-
Age Adult
Aged Adult
Reproductive-
Age Adult
Risk:
Prognostic
Significance or
Adversity
Figure 2-2. Lifestages of outcomes after developmental exposure. Panel A: In this figure A illustrates
the developmental lifestages of exposure considered in this document (shown in the shaded boxes on the
left) and lifestages of potential outcomes considered in this document (shown in the shaded boxes on the
right). The exposure to risk continuum is discussed across the top of the figure, and expanded upon in
Figure 4-1. Panel B: Exposure during the preconception and prenatal stages may result in outcomes
occurring in any lifestage beginning prenatally.
2-4
-------
Exposure
Period
Internal
Dose
Biologically
Effective
Dose
Precursor
Events/Early
Biological
Effects
Altered
Structure or
Function
Clinical
Manifestation
or Outcome
Risk:
Prognostic
Significance or
Adversity
Preconception
Prenatal
Preconception
Prenatal
Exposure
Period
Internal
Dose
Biologically
Effective
Dose
Precursor
Events/Early
Biological
Effects
Altered
Structure or
Function
Clinical
Manifestation
or Outcome
Risk:
Prognostic
Significance or
Adversity
Preconception
Preconception
Prenatal
Prenatal
Infant
Infant
Panel C: Exposure during infancy and childhood may result in outcomes occurring in any lifestage
beginning in infancy. Panel D: Exposure during the adolescent stage may result in outcomes occurring in
any lifestage beginning in adolescence.
2-5
-------
mechanism of action, which implies a more detailed understanding and description of events,
often at the molecular level" (U.S. EPA, 2005a,b). Risk assessments may require a more refined
definition of exposure intervals (e.g., bins) than the lifestages shown in Figure 2-2 because of
rapid changes during development, even within a lifestage. For example, gestational exposure is
typically evaluated for each trimester; however, specific periods of vulnerability (also known as
critical windows) for particular outcomes might be much shorter period of time as discussed in a
series of publications that resulted from an EPA-sponsored workshop (Selevan et al, 2000).
This report synthesizes the information currently available at EPA on assessing health
risks as a result of children's exposures and is based in part on existing risk assessment
guidelines, guidance, and science policies. Also, areas and topics are identified where further
guidance and research is needed. Within this structure, questions to be considered in the process
of reviewing data are posed as a way of prompting the data evaluation. This Framework is not a
guideline or science policy paper but rather describes an overall vision of the structure, process,
and the components considered important for assessing risks as a result of children's exposure.
This document intends to provide documentation of the approaches for assessing risk to children.
It is not intended to be prescriptive or to define a step-by-step process or standard operating
procedure.
The primary intended users of this approach are risk assessors involved in hazard, dose-
response, and exposure characterization. The central focus of this Framework is the prenatal
stage, infancy, childhood, and adolescence, thus extending and expanding the approach in
Guidelines for Developmental Toxicity Risk Assessment (U.S. EPA, 1991), which only focuses
on prenatal outcomes. The Framework also takes a child-protective approach to assessing risk
(Landrigan et al., 2004) by putting the child, rather than an environmental agent, at the focus of
the evaluation. Children are not a unique population but rather childhood is a series of lifestages
through which all individuals pass; therefore, a child-protective approach is inherently public
health-oriented.
The added value of using a lifestage approach to assess risks to children from
environmental exposure is a comprehensive evaluation of the potential for vulnerability of
various lifestages. In contrast, assessments that use only available chemical-specific data, which
are often limited to data from adults, do not necessarily account for the lack of data at other
lifestages. The approach outlined here encourages evaluation of the potential for toxicity during
2-6
-------
all developmental lifestages, based on what is known about critical windows of development for
different organ systems and differences in anatomy, physiology, and behavior that can impact
external exposure and internal dose metrics. In developing an assessment, the lack of data for
certain lifestages is not meant to imply susceptibility and/or greater uncertainty in the assessment
of risk from childhood exposure. Rather, the intent is to consider whether anything is known
about lifestages that would indicate particular vulnerability during that stage and incorporate that
information into the assessment. This document also addresses the difficult issue of integrating
animal toxicity or adverse health outcome data and exposure information relevant for assessing
risks to humans. This integration is especially challenging because of data limitations for
particular periods during pregnancy and early childhood development. One result of using this
framework will be more transparent and scientifically justifiable risk characterizations, while
documenting data gaps and identifying priority data needs for children's risk.
The approach outlined here encourages evaluation of the potential for toxicity during all
developmental lifestages, based on what is known about critical windows of development for
different organ systems, MOAs, anatomy, physiology, and behavior that can affect external
exposure and internal dose metrics.
Because of the complex issues to be considered for assessing risks from children's
exposures, it is impossible for any one person to be an expert in all areas of this process. Thus,
consultation and collaboration with appropriate experts in hazard, dose-response, and exposure
assessment is recommended in all phases of the process.
2-7
-------
-------
3. LIFESTAGE-SPECIFIC PROBLEM FORMULATION
Problem formulation is a systematic planning phase that defines the problem to be
addressed in the assessment. The purpose of a problem formulation phase is to aid in efficiency
and transparency of the assessment. A general discussion of problem formulation can be found
in the Framework for Cumulative Risk Assessment (U.S. EPA, 2003a). The major components
of problem formulation are no different whether applied to broad assessment (e.g., National
Ambient Air Quality Standards, U.S. EPA, 2005d) of all lifestages of exposure or to a narrow
assessment of specific lifestages of exposure (e.g., Superfund site). However, some of the
specific considerations are different in a risk assessment for developmental exposures.
The lifestage-specific problem formulation phase establishes the context of the risk
assessment and feeds into the lifestage-specific analysis phase (Chapter 4) and ultimately to
lifestage-specific risk characterization (Chapter 5). A planning and scoping step (Section 3.1)
initially characterizes exposures and outcomes during all developmental lifestages. The problem
formulation results in two products. First, a conceptual model (Section 3.2) is developed which
considers exposures (e.g., sources, receptors, stressors, pathways, individual characteristics) and
outcomes. Second, an analysis plan (Section 3.3) is developed, where preliminary consideration
of study methods, dose-response models, data gaps, and uncertainty and variability is used to
inform hazard characterization, dose-response characterization, and exposure characterization
(Figure 3-1).
These products are then used in the lifestage-specific analysis (Chapter 4), which
comprises hazard characterization (Section 4.1), dose-response characterization (Section 4.2),
and exposure characterization (Section 4.3). Iteration between each of the three analysis steps
may lead to further refinement of the conceptual model and analysis plan.
3.1. PLANNING AND SCOPING
In the planning and scoping step, the assessment goals, breadth, and focus are
established, and regulatory and policy factors are identified. This step includes defining and
identifying the purpose, scope, participants/stakeholders, approaches, resources, and relevant
past assessments available.
3-1
-------
§
I; a
«
a 53
£" s "
ill
^ -C J^
I •- -S
5
Lifestage-Specific Problem Formulation
(Chapter 3)
Planning and Scoping
(Section 3.1)
Purpose
Scope
Participants/stakeholders
Approach
Resources
Past assessments
Conceptual Model
(Section 3.2)
Analysis Plan
(Section 3.3)
Methods
Models
Data gaps for life stages
Uncertainties
Variabilities
Lifestage-specific:
* Exposures
• Individual characteristics
• Outcomes
L ifesfage-Specific Analysis
Lifestage-Specific Risk Characterization
Risk Communication/Management
Figure 3-1. Flow diagram for lifestage-specific problem formulation. Problem formulation
includes a planning and scoping step that initially characterizes exposures and outcomes during all
developmental lifestages, and the development of two products: a conceptual model and an
analysis plan.
Source: Adapted from U.S. EPA, 2003a, Figure 1-3.
3-2
-------
A clear purpose of the assessment is defined in order to guide the lifestage-specific risk
assessment strategy. Risk assessments are often conducted within the context of a regulatory
requirement (e.g., CAA, U.S. 101st Congress, 1990; FQPA, U.S. 104th Congress, 1996a; SDWA,
U.S. 104th Congress, 1996b), a community need, a health concern, or some other driving force
(U.S. EPA, 2003a), and they require varying levels of scope or depth (U.S. EPA, 2005a, Section
1.2.2). For example, there may be judicial and societal considerations that may influence the
timing and breadth of the assessment (e.g., consent agreement on soil contamination for a site-
specific cleanup). These factors may influence the risk management options, management goals,
key participants, data sources, selection of assessment outcomes, or the schedule for developing
the assessment. The risk management and assessment planning teams need to develop dialogue
on the regulatory basis for the risk assessment and determine what kind of information is needed
to satisfy such requirements.
The scope sets the parameters of the assessment, allowing for decisions to include or
exclude various elements. Screening level analyses of hazard and exposure may help refine the
scope of the assessment. The scope can be narrow (e.g., at a site where soil screening levels are
developed with lifestage-specific data) or broad (e.g., national rule-making, tolerance setting),
depending upon the problem. Age-specific information on factors related to exposure and
response are considered in the analysis plan (Section 3.3).
Choosing the appropriate participants for problem formulation will depend on the
problem being addressed. The participants who have information, expertise, or a stake in the
assessment process and conclusion(s) of the assessment are identified in this planning and
scoping step. Stakeholders are broadly defined as the interested parties who are concerned with
the decisions made about how a risk may be avoided, mitigated, or eliminated, and as those who
may be affected by regulatory decisions. This process can include specialized expertise and a
basic understanding of critical windows of exposure and optimum timing for evaluating
outcomes. The risk assessment team (which may include epidemiologists, toxicologists, public
health specialists, child behavior specialists, exposure assessors, chemists, and other technical
experts) and the risk management team (which may include economists, policy analysts,
engineers, and public health specialists) work together, informed by stakeholder input (which
may include parents, pediatricians, community groups, non-governmental organizations, etc.) to
develop the rationale, scope, and relevant outputs for the risk assessment and characterization
3-3
-------
(U.S. EPA, 200la). The conceptual model and analysis plan, including the possible outputs of
the assessment, may require negotiation among the members of the risk assessment team. The
Framework for Cumulative Risk Assessment (U.S. EPA, 2003a, p. 21) provides guidelines for
stakeholder involvement, which are based on the recommendations in Science and Judgment in
Risk Assessment (NRC, 1994) and by the Presidential/Congressional Commission on Risk
Assessment and Risk Management (1997a,b).
Methods used for risk assessment of health outcomes can have an impact on the
economic evaluation in benefits analysis (Griffiths et al., 2002; U.S. EPA, 2000a, 2003b, 2005e)
(Section 5.1.7). Bringing economists into the discussion at the problem formulation stage will
help clarify the approaches needed for data evaluation and quantification that may be most useful
for assessing benefits. Another key consideration here is the selection of outcomes for which
economic valuation will be considered in the assessment, because this includes dialogue between
risk assessors and economists.
Identifying available resources to achieve assessment goals within the time frame of the
assessment involves a qualitative screening evaluation of resources, which may or may not
identify whether children have a greater potential for higher exposures or greater intrinsic
susceptibility. The evaluation includes a preliminary examination of the quality and quantity of
the available data on exposure and outcomes. More detailed evaluations (refined assessment)
may or may not be necessary or may not be possible, depending on the available data. Where
adequate data exist (particularly on potential critical windows of exposure, level of exposure,
individual and community characteristics, optimum timing of outcome evaluation, and the
magnitude of concerns about the public health outcome), a more detailed approach can be
employed to address important questions for the exposure and health effects characterization.
These include identifying past assessments that relate to the purpose and scope of the assessment
and that may assist the process with existing tools, methods, or models.
• Why is the risk assessment being done? What are the needs of the assessment? Is
there a regulatory driver(s)?
• What is the public health concern? Is there a specific concern for developmental
lifestage exposure?
3-4
-------
• What is the risk question(s) being asked, and is it lifestage-specific? Will the
assessment consider exposure at all lifestages or exposure at specific
developmental lifestages?
• Which lifestage(s) (age bin[s]) is likely to have the greatest external exposure, the
greatest internal dose, and the greatest inherent vulnerability?
Have other risk assessments included consideration of health risks from children's
exposures on this chemical (e.g., EPA, other federal agencies, other
organizations)?
3.2. CONCEPTUAL MODEL
Within the conceptual model, the risk assessment team develops preliminary hypotheses
about why adverse effects have occurred or may occur in the future. A conceptual model is
developed, keeping in mind the relationships among the individual characteristics, exposures,
and outcomes. The relationships are informed by the initial identification of lifestage-specific
exposure scenarios, the lifestage of exposure, the optimum times for evaluation of outcomes that
will be addressed and the identified characteristics and toxicologic outcomes of the chemical(s)
that may contribute to latent effects from early exposure and children's risk.
A qualitative characterization of hazard and exposure for specific lifestages results in the
accumulation of the information needed to develop a conceptual model that aids the segue from
the problem formulation stage to the analysis phase. The conceptual model is the starting point
for the lifestage-specific analysis phase (Chapter 4) and can be presented as a diagram, a flow
chart, or a narrative description of the predicted key relationships (U.S. EPA, 2002b).
The following provides an approach to a preliminary evaluation of the available exposure
data (Section 3.2.1), outcome data (Section 3.2.2), and the integration of the two (Section 3.2.3)
to help define the conceptual model and aid in the development of a problem-driven analysis
plan with a focus on lifestages.
3.2.1. Exposure Considerations
The exposure considerations include performing a preliminary examination of the data to
determine the lifestages likely to be exposed, given the chemical properties and uses of the
environmental agent(s) in the defined scope of the assessment. The preliminary examination
involves a qualitative characterization of the sources, pathways of exposures (including exposure
3-5
-------
media and routes), exposure scenarios (lifestages, time frames, locations, and activities), and
pattern of exposures (magnitude and duration) to parents or children, as appropriate, including
the potential for dietary, drinking water, soil and air exposures, and other exposure media (e.g.,
Pharmaceuticals) (U.S. EPA, 1992, 2002c).
An issue to consider is whether all lifestages are at the same risk from a given exposure
or whether a specific developmental lifestage is more vulnerable because of higher exposures or
intrinsic susceptibility. This includes a qualitative understanding of lifestage-specific activity
patterns to identify potentially highly exposed lifestages. Currently, EPA's Guidance on
Selecting Age Groups for Monitoring and Assessing Childhood Exposures to Environmental
Contaminants (U.S. EPA, 2005e) is to be used as a starting point for identifying and selecting
age bins for analysis (see Table 3-1). This guidance includes expert analysis of existing generic
exposure data. This guidance provides a detailed discussion of how these age groups were
developed and how to implement them in an assessment. In brief, the recommended age groups
are based on the current understanding of differences in behavior and physiology that may
impact exposures to children. Information on critical windows of susceptibility also is factored
into these age bin considerations for potential vulnerability at different lifestages.
Typically, the conceptual model will consider human exposure in the context of the
source-to-effects paradigm (U.S. EPA, 2003b, Figure 1-3). When formulating an exposure
assessment, it is useful to qualitatively evaluate this model from the "effects" back to the
"source." In this way, potentially important time periods of exposure, exposure pathways, and
vulnerable individuals or populations can be identified. However, as the risk assessment
becomes more complex, some limitations in the source-to-effect model become apparent.
Exposure assessments using a source-to-effect model are based on the characteristics of the
specific source of the exposure (e.g., geographical location, release rate, point source) and not
the characteristics of the lifestage being exposed. As a result, only individuals or populations
with exposure to this specific source are included in the model. Yet, exposure may result from
multiple independent sources, all of which could contribute toward total exposure to a chemical
or mixture of chemicals. In this case, a person-oriented exposure assessment better characterizes
the person and lifestage of interest along with the applicable sources than a population-oriented
exposure assessment.
3-6
-------
Table 3-1. Developmental lifestages and age groups for exposure
assessments.
Lifestaqes
Preconception
Prenatal
Infant
Child
Adolescent
Aae Groups3
reproductive aqe adult
conception to birth
birth to <1 month
1 to <3 months
3 to <6 months
6 to <12 months
1 to <2 years
2 to <3 years
3 to <6 years
6 to < 11 years
11 to <16 years
16 to <18 years
18to<21 years
a The age groupings from birth to adulthood are from U.S. EPA (2005e). These standard age
groups were developed based on the results of a peer involvement workshop (U.S. EPA, 2000b)
focused on developmental changes in behavior and physiology impacting exposures to children.
Below are some questions that are useful in framing the examination of exposure
considerations.
What data are available that characterize children's exposure?
• Will the risk assessment consider all possible sources, media, pathways, and
routes of exposure (aggregate and cumulative), or is it confined to specific
scenarios (e.g., children living near a specific Superfund site and potentially
exposed via air, soil, and groundwater)?
• Is it suspected that individuals in developmental lifestages are actually being
exposed to the compound?
• What are the potential exposure sources, media (e.g., breast milk, indoor air),
pathways, and routes of exposure?
3-7
-------
What are the human lifestage behaviors (e.g., mouthing, crawling), activities (e.g.,
bathing, sleeping), and locations (e.g., indoors, outdoors, daycare) that may
impact exposure?
What other individual or community characteristics may be present that could put
children at higher risk of exposure and thus make them more vulnerable (e.g., pre-
existing diseases or disorders, belonging to a farm worker family, socio-economic
status, poor nutrition, sanitation conditions, cultural practices)?
What are elements of the physical environment that may impact exposure (e.g.,
altitude, climate, urban vs. rural)?
3.2.2. Outcome Considerations
In this screening approach, a preliminary identification of toxic effects is performed,
including TK and TD profiles, including to what degree these data support a hypothesized
MOA(s). Evaluating critical windows of susceptibility and number of critical effects that have
been observed relevant to the problem or scenario of concern for the risk assessment can be used
to qualitatively assess the database. This qualitative assessment assures that the risk assessment
team is appropriately staffed and has the essential resources to meet the timetables established in
the analysis plan.
Below are some questions that are useful in framing the examination of hazard and dose-
response considerations.
• What toxicology, epidemiology, or other data are available that examine outcomes
following exposure to the chemical(s) of interest?
• Are there any suspected MO As and other factors to be considered for relevant child
health outcomes?
• Are there TK (e.g., metabolic activation/conjugation) or TD (e.g., MOA)
considerations during certain developmental lifestages that may make the chemical
more or less toxic?
• What do we know about the properties of the chemical being evaluated that may be
important for considering lifestage-specific risk?
• Does the chemical cause known organ-specific toxicity? How might these organs
be differentially susceptible during development?
• What is known about critical windows of exposure (e.g., developmental windows of
susceptibility) for humans? For the experimental animal species and strain?
What is known about critical windows of effect (e.g., latent expression of
developmental toxicity) for the experimental animal species and strain?
3-*
-------
Are there any toxicologic outcomes noted in animal or human studies that are
signals of possible increased susceptibility of developmental lifestages (e.g.,
carcinogenicity, neurotoxicity, immunotoxicity, endocrine disruption)?
What are the background rates for outcomes of concern in the general population?
What dose metrics (AUC or Cmax) are being considered for the lifestage-specific
assessment?
3.2.3. Integrating Exposure and Outcome Considerations
The concepts of timing and dosimetry are incorporated as unifying factors for both
exposure and hazard components of the analysis. In a child-centered approach, multiple
stressors may need to be considered for a particular outcome of interest due to convergence on a
common MOA, as well as possible confounding, effect modification, or bias present in some
studies. Additional stressors may have an impact on behavior. For example, a person with
asthma may be less active or spend less time outside where an exposure may occur. Dialogue
between experts such as exposure scientists, health scientists, epidemiologists, and toxicologists
will ensure that the critical windows of exposure and critical effects are sufficiently identified, at
least at a qualitative level, for the development of a conceptual model and an adequate analysis
plan (Section 3.3). Below are some questions that are useful when integrating exposure and
response considerations.
How do chemical sources, fate, and transport influence target outcomes for various
lifestages?
• How do magnitude, patterns, and pathways of exposure influence target outcomes
for various lifestages?
• How does lifestage-specific dosimetry impact the temporal resolution required for
exposure assessment?
• Based on the transport and fate of the chemical under evaluation, do the available
exposure and hazard data address the compound(s) to which children may actually
be exposed?
Can exposure to multiple stressors during a critical window of development lead to
modification of a health outcome of interest (e.g., additivity, synergism,
antagonism)?
3-9
-------
3.3. ANALYSIS PLAN
The analysis plan identifies the methods, models, critical data gaps, major variabilities
and uncertainties, and key assumptions to be considered as the problem-driven assessment
moves forward to a more in-depth lifestage-specific analysis (Chapter 4). The analysis plan is a
working outline that provides the rationale for the resources (expertise, time, and finances)
required to complete the assessment. Examination of the most vulnerable age groups and key
risk drivers relevant to the problem identified will help conscribe the assessment and shape the
decision points and decision tree in the analysis plan.
A database inventory may be useful for identifying data gaps (Table 3-2). This table
presents an example of a database inventory method. After assessing the available information
on lifestages of exposure, the assessor can note whether there are the various types of
information for each lifestage. For example, are there human studies assessing outcomes after in
utero exposure? In many instances, few of these fields will have data. Input from the relevant
risk managers may be needed on the scope of the conceptual model and analysis plan,
particularly with respect to the questions the assessment is meant to answer. This exercise can
facilitate identification of strengths and weaknesses in the database, especially with regard to a
lifestage-specific assessment. Many of these boxes will be blank for most chemicals; these data
gaps do not necessarily represent research needs, but the data gaps may be useful in identifying
where more information would be helpful and communicate this need to conduct research. For
example, if the problem formulation suggests that infants have a potentially high risk due to
biological susceptibility or probability of increased exposure, then absence of data for that
lifestage may affect the relevancy of the risk assessment to address the identified problem or
question of the assessment.
Planning and scoping (Section 3.1), the conceptual model (Section 3.2), and the analysis
plan (Section 3.3) are then used in the lifestage-specific analysis (Chapter 4), which comprises
hazard characterization (Section 4.1), dose-response characterization (Section 4.2), and exposure
characterization (Section 4.3). Further scoping may be considered in each of the three analysis
steps, thus leading to further refinement of the conceptual model and analysis plan.
3-10
-------
Table 3-2. Lifestage-specific database inventory sheet. Types of information
are described in the left-hand column, and lifestages of exposure are shown in the
top row.
Developmental lifestages
WO
II"
I
0)
V « v
z f
f §
Human studies
Animal studies
Toxicokinetic data
Toxicodynamic data
Mode of Action
Chemical properties,
environmental sources,
fate and transport
Environmental media
concentrations
Lifestage-specific
exposure measurement
data
Lifestage-specific
exposure factors
Preconception
In Utero
Infant
Child
Adolescent
Adult
Does the analysis plan focus on what are likely to be the most vulnerable age
groups?
Does the analysis plan focus on the key risk drivers?
What decision points are needed in the analysis plan for the specific problem
identified?
3-11
-------
-------
4. LIFESTAGE-SPECIFIC ANALYSIS
The analysis phase of risk assessment includes hazard characterization (Section 4.1),
dose-response characterization (Section 4.2), and exposure characterization (Section 4.3), where
data are analyzed, both qualitatively and quantitatively. Iterations among all three steps provide
communication among the risk assessment team members and refine the focus on the key
assessment questions identified in the problem formulation phase (Chapter 3). These iterations
are performed to enhance, but not effectively delay, the final assessment.
Focusing on data with outcomes after exposure during developmental lifestages of
greatest susceptibility (i.e., critical windows) is key to the lifestage-specific evaluation of hazard,
dose-response, and exposure data. These data may identify critical windows of exposure and
data gaps for particular lifestages of exposure. MOA information based on TK and TD data may
inform the lifestage-specific analysis (Figure 4-1).
Mode of Action
Toxicokinetics
A
Toxicodynamics
Exposure
Period
Internal
Dose
Biologically
Effective
Dose
Precursor
Events/Early
Biological
Effects
Altered
Structure or
Function
Clinical
Manifestation
or Outcome
Risk:
Prognostic
Significance or
Adversity
Figure 4-1. Exposure to risk continuum. This figure identifies the major elements in Figure
2-2a. This includes specific elements of TK and TD that may be lifestage-specific. This TK and
TD information (MOA) can lead to increased characterization of the altered structural and
functional outcomes
Source: Adapted from Schulte, 1989.
The next three Sections (4.1, 4.2, and 4.3) discuss the three steps of the analysis phase
and provide information to guide the assessor through the process (Figure 4-2). In order to link
exposures and outcomes appropriately, an iterative process among all steps of the analysis is
suggested for a robust risk characterization, the final phase in the risk assessment process
(Chapter 5).
4-1
-------
Lifestage-Specific Problem
Lifestage-Specific Analysis
(Chapter 4)
stage-Specific
Hazard
iracterization
section 4.1)
V ^
Lifestage-Specific
Dose-Response
Characterization
(Section 4.2)
Exposure
Characterization
(Section 4.3)
r
i
Lifestage-Spectfh Risk (
Risk Communication' Vlanagement
Figure 4-2. Flow diagram for lifestage-specific analysis. Following the problem formulation
stage, the three steps of the analysis phase include hazard characterization (Section 4.1), dose-
response characterization (Section 4.2), and exposure characterization (Section 4.3). This is
followed by the risk characterization (Chapter 5) and risk communications/management phases.
4.1. LIFESTAGE-SPECIFIC HAZARD CHARACTERIZATION
4.1.1. Introduction
Hazard characterization is the analysis step in which the data are evaluated for potential
adverse health effects. Hazard characterization begins with the identification of the human and
animal toxicology studies to be included in the database. It includes the identification of any
outcomes associated with exposure at specific doses. The primary purpose of a lifestage-specific
hazard characterization is to develop a detailed description of the potential for health outcomes
after exposure to the agent of interest during preconception or developmental lifestages. This
begins with a description of each of the available studies (Section 4.1.2), considering critical
windows of exposure and susceptibility, TK, TD, MOA, and dose-response information as well
as the variability and uncertainty present in each study. The database is then synthesized from
4-2
-------
the individual study evaluations, and the quality and quantity (i.e., the comprehensiveness) are
characterized using a weight-of-evidence (WOE) evaluation (Section 4.1.3.1). This includes
information about differences and similarities in experimental animal species versus humans
regarding lifestage-specific TK and TD, the extent of the database for different lifestages, and
lifestage-specific susceptibilities. The results of the hazard characterization are iterated with the
dose-response and exposure analyses (Section 4.1.4) if indicated by the conclusions from
summarizing the hazard database.
Finally, the lifestage-specific hazard characterization is summarized including a scientific
rationale for the identification of relevant outcomes and susceptible lifestages based upon the
data (Section 4.1.5). The identified outcomes and susceptible lifestages are further evaluated in
the subsequent dose-response characterization step (Section 4.2). This information feeds into the
comprehensive lifestage-specific risk characterization (Chapter 5).
Throughout the hazard characterization, relevance of the information to the overall goals
of the assessment is considered. It may be appropriate to refine the conceptual model (Section
3.2) or analysis plan (Section 3.3) after thoroughly evaluating the available hazard data. For
example, a conceptual model may focus on an exposure to a chemical or chemical class that
results in thyroid tumors. Thyroid hormone is critical to development of the nervous system
(Farwell et al., 2006; Pals et al., 2006; Ramos and Weiss, 2006; Santisteban and Bernal, 2005)
and immune system (Bossowski et al., 2003; Lam et al., 2005). If development of these organ
systems were not considered in the conceptual model for analysis of the chemical(s), then the
conceptual model will need to be refined to consider the relevant critical windows of
development.
Figure 4-3 illustrates a detailed approach to characterizing hazard from environmental
exposures during development. More specific information on hazard characterization for
developmental lifestage exposures can be found in the existing EPA risk assessment guidelines
for developmental toxicity (U.S. EPA, 1991), reproductive toxicity (U.S. EPA, 1996),
neurotoxicity (U.S. EPA, 1998b), and cancer (U.S. EPA, 2005b).
4.1.2. Qualitative Evaluation of Individual Studies
The objectives and scope of the risk assessment, defined in the problem formulation
phase (Chapter 3), provide focus and a plan for identifying and examining all the relevant
4-3
-------
Lifi'sfuge-Specifii• I*rohicin F<> rmu 1 ation
Lifestage-Specific Analysis
(Chapter 4)
Lifestage-Specific
Hazard Characterization
(Section 4.1)
Qualitative Evaluation of Individual Studies
(Section 4.1.2)
Summarization of the Hazard Database (Section 4.1.3) and
Evaluation of the Weight-of-Evidence (WOE) (Section 4.1.3.1)
Iteration with Dose-Response and Exposure Characterization
(Section 4.1.4)
Lifestage-Specific Hazard Characterization Narrative
(Section 4.1.5)
I. ift's fag c-Sp et ijlc
Dose-Response
Characterization
(Section 4,21
t''.\|)i)sure
Characterization
(Seei'on -1.3 <
Risk ('haracteri/afion
Risk Communication Management
Figure 4-3. Flow diagram for lifestage-specific hazard characterization. The steps in hazard
characterization include the evaluation of individual studies (Section 4.1.2), summarization of the
hazard database (Section 4.1.3), an evaluation of the weight-of-evidence (Section 4.1.3.1),
potential iteration with the other analysis steps (Section 4.1.4), and the hazard characterization
narrative (Section 4.1.5). The dashed lines indicate where iterations may occur with other parts of
the risk assessment process.
4-4
-------
published human and experimental animal studies. A thorough qualitative evaluation of each
study includes a complete description of the findings, an assessment of the study conduct and
data quality, and a determination of sufficiency of data. To focus on risk from exposure to
children, the evaluation process considers lifestage-specific information (pertaining to both the
lifestage at which exposures occur and outcomes are observed) and issues within the overall
context of the risk assessment. To assess study quality, the adequacy of the methods and results
are characterized. In addition, it can be helpful to establish criteria for confidence in the
evaluation and interpretation of the study findings that can be used later in the WOE evaluation
(Section 4.1.3.1). The description of individual studies will contribute to the overall
determination of the adequacy, strength, and completeness of the database for the
characterization of hazard across lifestages. The following subsections describe topics to
consider during the qualitative evaluation of each study, and example questions are addressed in
Table 4-1.
4.1.2.1. Study Purpose
Describing the purpose of each study may provide information to evaluate the study as it
relates to lifestages. For example, the study may be conducted in response to general risk
evaluation issues, to explore an aspect of basic toxicology or biology, or to investigate a specific
public health concern. The purpose of the study can range from hypothesis generation to
hypothesis testing.
4.1.2.2. Study Design
A clear, concise description of the study design includes the number of subjects in each
exposure group; descriptions of the study participants (e.g., gender, age); route, timing, and
duration of exposure; and outcomes assessed. The timing of exposure and outcome assessment
is important in relation to identifying and characterizing lifestage-specific risk. All of these are
related to statistical power, which is further discussed in the WOE evaluation (Section 4.1.3.1).
It is helpful to highlight strengths and weaknesses in the study design, particularly in relation to
lifestage-specific assessments and how they may illuminate questions identified in the problem
formulation (Chapter 3). For example, statistical power is a limitation that is often not discussed
when studies are concluded to be "negative."
4-5
-------
4.1.2.3. Identifying Critical Windows of Exposure
An evaluation of the exposures (or dosing/treatment to experimental animals) to the study
participants involves characterizing the timing and duration of the exposures (e.g., exposure
during preconception and critical windows of pre- or postnatal development) that have occurred
across the lifestages of the study individuals. The timing and the duration of exposure to test
substance in experimental animal studies could be informed by data on the critical windows of
development of organ systems.
A useful source of information is the proceedings of a workshop on critical windows of
exposure for children (Selevan et al., 2000), which addresses the respiratory and immune
systems (Dietert et al., 2000; Holladay and Smialowicz, 2000; Peden, 2000; Pinkerton and Joad,
2000), the reproductive system (Lemasters et al., 2000; Pryor et al., 2000), the nervous system
(Adams et al., 2000; Rice and Barone, 2000), the cardiovascular and endocrine systems (Barr et
al., 2000; Hoet et al., 2000; Osmond and Barker, 2000; Sadler, 2000), and cancer/neoplasms
(Anderson et al., 2000; Olshan et al., 2000). The WHO draft document, Principles for
Evaluating Health Risks in Children Associated with Exposure to Chemicals (WHO, 2006) also
reviews critical windows of development by organ systems.
4.1.2.4. Outcomes Related to Developmental Lifestage Exposure
A description of study findings, including the relationship of the outcome (both the
outcome itself and timing of the outcome assessment) to the time of exposure, is a primary goal
of hazard characterization. This includes an explicit consideration of outcomes at various
lifestages due to exposure occurring during developmental lifestage(s). Developmental lifestage
exposures may result in early or latent effects (Selevan et al., 2000; WHO, 2006). The
evaluation of each study includes whether and how study outcomes address questions raised
during the problem formulation phase (Chapter 3). For example, if the problem formulation
specifically identifies a potential for risk after exposure to pregnant women in a residential
setting, then it is important to carefully evaluate any available human and experimental animal
data that examines outcomes following gestational exposures. Toxicities resulting from
alteration of precursor events may be expected to be different depending on lifestage. Alteration
of a precursor event in a mature animal or adult human may not have any significant health
consequence, where the same precursor event alteration in a developing organism may have
significant health consequences.
4-6
-------
4.1.2.5. Toxicokinetic Data
All available lifestage-specific TK data are included and described in order to determine
the relevance and impact of the TK data in evaluating the study and to determine the impact of
exposure on response across lifestages. TK data can be used to verify that indirect exposure of
the fetus or neonate (e.g., via maternal circulation or breast milk) occurred without relying on
observable outcomes. In some situations, internal dose can be measured, providing greater
confidence in derivation of the dose metrics (Section 4.2.2.3). If TK data are available across
lifestages, this information can aid in highlighting key lifestages for the assessment. For
example, immaturity of specific metabolic enzymes or renal capabilities (e.g., elimination) can
result in a more or less toxic response in the young. Therefore, information on the
developmental profiles of enzymes or organ systems can help identify particularly susceptible
age groups.
Studies may find increased susceptibility of immature individuals but lack TK data to
assist in the interpretation of these findings. In that case, default assumptions are generally
applied. Three typical examples are (1) internal dose is equivalent to dose at the portal of entry,
(2) the dose to the fetus is equivalent to the dose administered to the maternal animal, or (3) the
internal dose to the immature individual is equivalent to that of adults. However, these default
assumptions may not be health protective; therefore, the availability and use of TK data will
likely decrease uncertainty in the risk assessment.
4.1.2.6. Toxicodynamic Data
TD data include information about the steps between the toxicant's first interaction with
the target organ and the subsequent toxic outcome. Describing TD data for specific lifestages
may provide corroborative evidence of potentially susceptible lifestages for a given chemical.
For example, if TD information for a chemical suggests effects on the nervous system via
decreasing luteinizing hormone and disruption of the hypothalamic-pituitary-gonadal axis, then
greater concern would be warranted in cases when there are lifestage-specific TK data. This TK
data may demonstrate that the chemical is found in the brain only during a developmental
lifestage when the blood-brain barrier is not fully formed.
4-7
-------
4.1.2.7. Mode of Action Information
Consideration of MOA information (key TK and/or TD steps) can be useful in
• understanding the susceptibility differences among various lifestages,
• determining the most appropriate experimental animal model for relevance to
humans,
• determining when human exposure or outcome data during lifestages are limited or
not available,
• predicting types of effects that might be seen during particular lifestages, and
• predicting potential susceptible lifestages.
For example, if a chemical has an anti-androgen MOA, in utero and peri-pubertal intervals might
be sensitive exposure windows for male reproductive outcomes. Further, differences in
androgen activity by lifestage can explain some of the observed differences in susceptibility; for
example, for the pesticide vinclozolin (Anway et al., 2006; Euling and Kimmel, 2001). It is also
possible that the MOA for a given chemical differs among lifestages; this is one possible
explanation for differences in outcomes after exposures during developmental lifestages versus
adulthood. For example, diethylstilbestrol (DBS) produces reproductive, developmental, and
carcinogenic outcomes after in utero exposure which are not observed following adult exposure
(Herbst, 1987; Mericskay et al., 2005; Robboy et al., 1982). Also, organophosphorous pesticides
inhibit cholinesterase throughout one's lifespan, but certain of these pesticide's inhibitory effects
on neuronal differentiation and migration, which are attributed to an alternative, noncholinergic
MOA, occur only during in utero and early postnatal neurological development (Campbell et al.,
1997; Chakraborti et al., 1993; Dam et al., 1998; Young et al., 2005). However, chemicals with
more than one MOA, such as methoxychlor, have been described (Chapin et al., 1997; Gaido et
al., 2000; Gray et al., 1999a). Therefore, it is possible that the activity of the different MOAs
may vary across lifestages.
4.1.2.8. Qualitative Evaluation of Dose-Response
A detailed qualitative evaluation of the lifestage-specific dose-response profile is useful,
but not always available, when interpreting the outcome for individual studies. A well-
characterized dose-response relationship helps support the judgment of whether an outcome is
-------
due to exposure during a specific lifestage. The shape of the dose-response curve may or may
not be monotonic in nature.
These dose-response data are carried forward into the WOE evaluation (Section
4.1.3.1.3) because determining the relationship between adverse responses and exposures is
achieved through consideration of the results in context of the other studies in the database and
may highlight the importance of borderline or suggestive findings in individual studies and,
ultimately, refine the interpretation of the data. For example, a prenatal developmental toxicity
study in rats may identify a treatment-related malformation (e.g., spina bifida) that occurs with a
demonstrable dose-response relationship; in a two-generation reproduction study, the
interpretation of incidences of spina bifida that are observed in litters from treated groups may
take on greater weight in the overall hazard characterization even in spite of the lack of
significant incidence or a clear dose-response.
4.1.2.9. Variability Analysis
There are a number of sources of variability, both intrinsic and extrinsic, in human and
animal toxicologic data. Intrinsic, or biological, variability includes heterogeneity across
lifestages and is expressed to some degree in each parameter being measured. Examples of
intrinsic variables in both human and experimental animal studies include age, gender, and
genetic factors. On the other hand, the sources of extrinsic variability are external to the study
individuals and can often be attributed to methodological considerations, to errors in study
design, or to variations in implementation. Examples of extrinsic variables for experimental
animal studies include handling techniques, ambient temperature, and noise. For epidemiologic
studies, examples include variations in recruitment or data collection procedures.
Variability can be adequately characterized by the appropriate statistical treatment of
individual study data. For example in developmental toxicologic studies, all pups in one litter
are used as the unit of measure (n=l) to address issues of between-litter variability in response.
High levels of variability may affect the ability to identify associations and make the
interpretation of study data difficult. A detailed consideration of variability with appropriate
analyses contributes to a determination of the adequacy, strength, and reliability of a study and
its conclusions. Variability can be a source of uncertainty in the evaluation and interpretation of
individual studies (Section 4.1.3.1.2). High variability can sometimes render a study
4-9
-------
uninterpretable within the context of the rest of the data or result in reduced confidence in the
veracity of the study findings, thereby decreasing the confidence placed in the study and its value
for use in the WOE evaluation (Section 4.1.3.1).
4.1.2.10. Uncertainty Analysis
Uncertainty from a variety of sources in lifestage-specific data can affect the assessment
of risk. Uncertainties can result from data gaps (i.e., missing information), inadequacies in the
study protocol or methodologies, inaccuracies in the reporting of study findings, or inconclusive
results. After a thorough consideration and description of the uncertainties for each study, any
resulting assumptions, extrapolations, or speculative interpretations are described and utilized in
the risk characterization (Chapter 5). Detailing data gaps helps provide an adequate
characterization of the uncertainties of the risk from developmental lifestage exposure. For
example, in laboratory animal studies, if the toxicologic evaluation characterizes adverse
outcomes following exposures that traditionally occur throughout all developmental lifestages,
then future study exposure methods may need to incorporate direct dosing techniques during
specific lifestages (e.g., in pre-weaning or juvenile experimental animals) (Bruckner and Weil,
1999; Zoetis and Walls, 2003). In particular, experimental animal studies of exposure during the
juvenile period specifically are rare, although they are increasingly becoming more common as
they gain greater prominence in regulatory hazard characterization (Hurtt et al., 2004; U.S. FDA,
2006). Developmental (in utero) studies are more common but are not done for all chemicals
and are limited because they do not involve direct dosing in postnatal life. One- and two-
generation reproduction studies are also not conducted for all chemicals and are often limited in
having postnatal dosing only via nursing and involve a limited number of outcomes (e.g.,
reproductive outcomes). Developmental neurotoxicity, developmental immunotoxicity, and
other organ system-specific developmental studies also are not commonly performed and have
limitations regarding the exposure route and apical outcomes/organ systems assessed. Due to the
iterative nature of the evaluation process and the consideration of information from multiple
sources, data from other human or experimental animal studies, data on structure-activity
relationships (SARs), or TK or TD information, may be used to address uncertainties identified
in a given study.
4-10
-------
Table 4-1. Examples of lifestage-specific questions for evaluation of
individual studies within hazard characterization.
Topic
Lifestage-Specific Question(s)
Study Purpose
(Section 4.1.2.1)
Was the purpose of the study to address a lifestage-specific hypothesis or public health
concern.'
Study Design
(Section 4.1.2.2)
Did the study design and methods address specific lifestages of exposure and their
outcomes? What lifestages were assessed? How was lifestage/age measured?
What were the strengths and limitations of the study design in assessing lifestage-
specific exposure and outcome?
For human studies, how did the methods impact the validity and reliability to determine
children's exposure and outcome? Were lifestage factors (potential confounders)
examined and accounted for, where appropriate?
In experimental animal studies, was an appropriate route and matrix (e.g., vehicle,
formulation, duration) of exposure employed across various lifestages? Were the dose
range and levels appropriate across lifestages evaluated?
Was the power of the study adequate to detect an effect after exposure during a specific
lifestage? Were sample sizes, inclusion of both sexes, and animal litter numbers
considered?
Identifying
Critical Windows
of Exposure
(Section 4.1.2.3)
What is known about critical windows of exposure for the outcome and chemical?
Were the routes of exposure relevant to the age-related exposure pathways for the age
groups of interest? Did the exposure interval cover different lifestages, partially or
completely?
• What exposure/dose levels were assessed during the lifestage(s) of development?
Were they the same across all the lifestage(s) identified in the study?
• Were lifestage-specific behaviors discussed that could influence the exposure (e.g.,
maternal nurturing behaviors, offspring nursing or weaning activities, or
exploratory/play behaviors in the immature individual)? If so, in what direction
would the dose likely be affected?
Was exposure verified for critical lifestages?
• In animals, what was the route of exposure and was it the same throughout all
lifestages? Did exposure occur across more than one developmental lifestage(s) in
the study?
• For humans, was there more likely to be exposure(s) from this source during
certain lifestages than others? If so, would this be expected to affect the results of
the study and was this accounted for in the study? Were other possible sources of
exposure considered for various lifestages?
Outcomes Related
to Developmental
Lifestage
Exposure
(Section 4.1.2.4)
What was the timing of assessment of outcomes? Were outcomes dependent upon the
exposures during critical stages of development? How were latent effects assessed?
Were lifestage-specific outcomes assessed in the study (e.g., different outcomes during
different developmental stages vs. adult stages)?
What methods were used to assess lifestage-specific outcomes after developmental
lifestage exposures? Were they appropriate? What were their limitations (e.g., were
relevant lifestage-specific outcomes not assessed)?
Were biological plausibility and internal consistency of findings considered for
lifestage-specific data?
Did the authors make lifestage-specific conclusions in the study, and what were their
assumptions and interpretations?
TK Data
(Section 4.1.2.5)
Are there lifestage-specific differences in absorption, distribution, metabolism
(toxification or detoxification), or elimination assessed in the study? At varying doses?
4-11
-------
TD Data
(Section 4.1.2.6)
Are there TD data for the specific lifestage(s) that relate to outcomes?
MOA Information
(Section 4.1.2.7)
For the outcome(s) assessed in this study, what is known about the chemical's MOA
after exposure at different lifestages? Is there information suggesting similarities or
differences in MOA for different lifestages of exposure?
Have precursor events (e.g., biomarkers) been identified for a particular outcome? If so,
were precursor events similar across lifestages? Are the toxicities resulting from
precursor events expected to be different depending on lifestage of the outcome?
Are outcomes related to the MOA relevant to the lifestages of concern in this study? Are
there different MOAs suspected for different lifestages?
If there are multiple outcomes described at differing lifestages, then are these consistent
with one or more MOAs?
Qualitative
Evaluation of
Dose-Response
(Section 4.1.2.8)
Are there lifestage-specific dose-response relationships assessed in the study? What are
the similarities and differences in dose-response across lifestage of exposure?
What is the shape of the dose-response curve for lifestage-specific toxicologic
outcomes?
Variability
Analyses
(Section 4.1.2.9)
What was the variability in the control data for parameters of normal growth and
development and other outcomes for the lifestage of interest?
Was this variability in measures within expected ranges? If not, could this mask
detection of an outcome?
Uncertainty
Analyses
(Section 4.1.2.10)
Are there any lifestage data gaps or uncertainty considerations (i.e., were some lifestages
exposed and/or assessed, while others were not)?
Were critical windows of exposure and associated outcomes adequately addressed?
Were lifestage-specific studies conducted with appropriate quality laboratory practices
and standards (e.g., Good Laboratory Practice) (U.S. FDA, 1978)?
Did the conduct of the study result in uncertainties in findings that are particularly
pertinent to lifestage-specific data interpretation? Were any inadequacies in the data
lifestage specific?
4.1.3. Summarization of the Hazard Database
After summarizing the relevant studies for the lifestage-specific hazard database (Section
4.1.2), the exposure-response array is assembled and then evaluated. Not all summarized studies
judged may be useful to the risk assessment (NRC, 1994). Well-justified decisions to include or
exclude a study are provided in the hazard narrative (Section 4.1.5). The adequacy of studies
and characterization of the database are discussed in detail in A Review of the Reference Dose
and Reference Concentration Processes (U.S. EPA, 2002a, Section 4.3).
The overall hazard database includes detailed descriptions of all available studies relevant
to and critical for evaluating the hazard to children, specifically those with developmental
exposures, effects, or outcomes. The database may also include in vitro data, MOA or
mechanistic studies, and toxicity data in adults to help profile the toxicologic response in
children, or the database may provide support for assumptions made during the hazard
characterization. A careful review of the studies' exposure durations and lifestages may help to
4-12
-------
determine the relative importance (weight) of the studies when estimating potential risks to
children. Issues to consider include the pathways (including media and route), and whether they
are relevant to children; the intervals of exposure, and whether they included critical lifestages;
and issues suggestive of differential susceptibility of children or specific lifestages.
A detailed characterization of the study outcomes is also important for the
characterization of the database. Often, the structure and presentation of data summaries are
driven by the outcome data. Common links are examined across studies. For example, for one
chemical with detailed MOA information, the summary could focus on hazard in relation to that
MOA and what the MOA may predict about potential critical windows. For other chemicals, the
description might focus on specific developmental outcomes, target organs, or susceptible
lifestages. The emphasis of the hazard summary is on the relationships (i.e., patterns) across
observed outcomes, in relationship to lifestages and MOA. For some chemicals, only very
limited human or experimental animal hazard information may be available. However, detailing
the lack of information about an agent (i.e., data gaps and uncertainties) is crucial to an adequate
characterization of risk to children from environmental exposures.
4.1.3.1. Evaluation of the Weight-of-Evidence of the Hazard Database
During the evaluation of the hazard database, the major strengths and weaknesses of the
available relevant data are identified and are summarized in the WOE evaluation. The WOE
evaluation includes expert judgment of the completeness of the database. For this Framework,
key themes were adapted to meet the needs of evaluating human and toxicologic studies relevant
to children's health risk assessment. These key themes include temporality, strength of the
association, qualitative dose-response relationship, experimental evidence, reproducibility,
biological plausibility, alternative explanations, specificity, and coherence (Figure 4-4) (Hill,
1965; Gray et al., 2001; Seed et al., 2005; U.S. EPA, 2000c, Chapter 4; Vineis and Kriebel,
2006; Weed, 2005). Criteria for evaluating the key themes for the WOE may be developed
during problem formulation (Chapter 3) to address specific assessment needs. The adequacy,
strength, and completeness of the entire database are considered. The description of the database
includes a qualitative exposure-response array, data gaps, uncertainties, and assumptions that are
summarized in the hazard characterization narrative (Section 4.1.5).
4-13
-------
Weight-of-
Evidence
Assessment
of the
Hazard
Database
Strength of
Association
ose-Response
Relationship
Alternative
Explanations
Experimental
Evidence
Figure 4-4. Conceptual view of a WOE evaluation. This figure illustrates the considerations
within a WOE evaluation of toxicity data. The relative weight of each consideration will vary for
each assessment.
Source: Adapted from Hill, 1965; Gray et al, 2001.
The principles developed by Hill (1965) focused on evaluating human studies, while
Gray et al. (2001) focused on evaluating animal toxicology studies (Figure 4-4). Further details
about EPA's WOE evaluation approach can be found in the Methods for Derivation of Inhalation
Reference Concentrations and Application of Inhalation Dosimetry (U.S. EPA, 1994),
Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), Supplemental Guidance for
Assessing Susceptibility from Early-Life Exposure to Carcinogens (U.S. EPA, 2005b), A Review
of the Reference Dose and Reference Concentration Processes (U.S. EPA, 2002a, Section
4.3.2.1), and Determination of the Appropriate FQPA Safety Factor (s) in Tolerance Assessment
(U.S. EPA, 2002d, Section III). The following subsections describe the key themes that can be
considered in the WOE evaluation, and example questions are presented in Table 4-2.
4.1.3.1.1. Temporality, Temporality is the basic premise that the exposure must occur prior to
the outcome (U.S. EPA, 2002a, pp. 4-13 to 4-14). For developmental lifestage-specific data,
temporality includes consideration of the relationship between the timing of exposure and
outcome. Depending on what is known about potential critical windows of exposure (Section
4.1.2.3), more or less credence may be given to the association with the observed outcome.
When developmental lifestage-exposure data exist, the temporal relationship between the
4-14
-------
exposure and outcome(s) may be assessed. Further, when there are data that provide an accurate
characterization of the timing of the exposure and outcome, the latency time between the
exposure to the outcome may be determined. For example, exposure to dibutyl phthalate (DBF)
during late gestation leads to a number of male reproductive developmental effects (e.g.,
decreased anogenital distance, increased nipple retention, and hypospadias) that are observed at
different stages of development in the rat (Barlow and Foster, 2003; Gray et al., 1999b;
Mylchreest et al., 1999, 2000).
4.1.3.1.2. Strength of the association. Greater weight is generally given to more rigorous
studies as well as those with higher statistical power, and therefore, greater statistical precision.
Strength of the association considers both rigor and statistical power. Rigor is the degree of
proper design, conduct, and analysis of a study. It can be difficult to determine rigor because in
some cases, study methods presented in published studies lack sufficient detail. Additionally,
rigor is not simply equivalent to conduct under GLP regulations for nonclinical laboratory
studies (U.S. FDA, 1978). Many older studies showing early-lifestage sensitivity to carcinogens
were rigorously conducted, but before the GLP regulations were first published in 1978;3
similarly, many rigorous studies in academic institutions do not follow GLP regulations.
Statistical power is the ability of a study to detect effects of a relevant magnitude and relates to
the sample size, the number of data points, the stratification of findings, and the background rates
of the specific outcome(s).
For the evaluation of human studies, the strength of an observed association may be
affected by the presence of uncontrolled or unmeasured confounders, the prevalence of effect
modifiers in the study population, or bias. A confounder is a variable that can cause or prevent
the detection of a change in an outcome of interest and is not an intermediate variable on the
causal pathway between exposure and outcome but is associated with the factor under
investigation. A confounding factor can often be controlled for or accounted for in the statistical
analysis. An effect modifier is a variable that modifies the outcome of interest by a greater
(synergistic or additive) or lesser (antagonistic) effect. An effect modifier can sometimes be
identified through stratification of the data. Many effect modifiers and confounders are
3 There are different GLP citations for U.S. FDA and U.S. EPA (including for FIFRA and TSCA). These have been
updated several times over the years.
4-15
-------
potentially lifestage specific, and whether and how these have been evaluated in the data analysis
could affect the study outcomes or interpretation of study results. A lifestage-specific example
of a confounder is maternal socio-economic status (SES), which can influence or bias the
interpretation of the offspring's cognitive development. For animal toxicology studies, a
lifestage-specific example of an effect modifier is maternal health status, such as maternal or
offspring nutrition, which can influence the development and maturation of the young (Cappon
et al., 2005; Fleeman et al., 2005). Another example is the timing of heat exposure and effects
on skeletal development in the rat (Cuff et al., 1993; Kimmel et al., 1993).
4.1.3.1.2.1. Variability analysis. The sources of variability within individual studies (Section
4.1.2.9) are also important factors for the interpretation of the dataset. They can contribute to
overall uncertainties in the database, including those uncertainties that are applicable to the
lifestage-specific hazard characterization. Variability of response across studies and possible
reasons for the variability are assessed and considered when developing an exposure-response
array. For example, in animal studies the response variable could vary among studies performed
when using different strains of the same experimental animal species or when studies are
performed in different decades, possibly due to genetic drift in laboratory animal populations
(Haiti, 2001; White and Lee, 1998).
4.1.3.1.2.2. Uncertainty analysis. In the evaluation of individual studies (Section 4.1.2), data
gaps (missing information) may be identified that could impact the quality of the study, and these
are considered in total when evaluating the database. In addition, when combining the data from
all the studies, data gaps for the comprehensive database of information on the chemical can be
assessed. For example, the combined studies may have assessed outcomes after exposure during
all developmental stages except for the peri-pubertal period. If this were the case, then a data
gap in coverage of this particular developmental lifestage of exposure would be noted. For any
chemical assessment, there will be inevitable gaps in the available lifestage-specific information;
it is the relative impact of missing or inadequate information to the overall goals of the
assessment that are to be judged. In some cases, information gleaned from the toxicologic
profiles of structurally-related chemicals or chemicals with a similar MOA can assist in
interpreting the relative importance of a data insufficiency. Sometimes this information can
4-16
-------
provide a way of bridging a data gap (Julien et al., 2004). When evaluating lifestage-specific
uncertainties and data gaps, study design (e.g., measurements, exposure, and outcomes across
lifestages) is addressed (U.S. EPA, 1991, Section 3.1.2.1; U.S. EPA, 1996, Section 3.3.1.5; U.S.
EPA, 2002a, Section 4.3.1). The characterization of data gaps also includes a determination of
whether required toxicologic studies (i.e., by statute or convention) are present (e.g., a rodent and
a non-rodent prenatal developmental toxicity study, and a reproduction and fertility effects
study).
Uncertainties arising from the absence of any other relevant data identified are addressed.
The potential qualitative and quantitative impact of these missing data on the risk assessment
(e.g., on the point of departure [POD]) is considered and may be useful in determining the
magnitude of a database uncertainty factor (UF) during dose-response characterization (U.S.
EPA, 2002a) (Section 4.2.4.4). Additionally, information from the exposure characterization
(Section 4.3) could be useful when identifying any remaining uncertainties in the hazard
characterization. For example, if the exposure characterization identifies a high potential of
exposure to nursing infants, specific TK data on milk partitioning may be deemed particularly
important in the risk assessment, and absence of these data could be considered an important
source of uncertainty. Finally, the level of confidence in the final risk estimates is based on a
detailed description of the assumptions and interpretations of the uncertainties in the overall
database.
Sometimes, other types or sources of data can assist in satisfying an identified data gap or
uncertainty. For example, if for a chemical being evaluated, there are no data relevant to the
hazard characterization following exposure during a particular lifestage, data from a similar
lifestage exposed for a different chemical that has been shown to produce the same active
metabolite might be useful in informing the assessment and reducing uncertainties relevant to
this data gap.
4.1.3.1.3. Qualitative dose-response relationship. The dose-response relationship demonstrates
a predictable change in an effect as a function of exposure/dose. Studies that directly relate the
exposure/dose to the degree of the effect (i.e., increasing dose results in increasing effects) give
stronger weight to the evidence (exposure-response array). For example, an association between
increasing blood lead levels and a lower IQ in children has been reported (Canfield et al., 2003).
4-17
-------
In some cases, the failure to observe a dose-response relationship may be due to the choice of
dose levels or dose spacing in given studies, to a threshold effect, or to a more complex (e.g., a
U- or J-shaped) dose-response relationship. Also, an observed dose-response relationship may,
in fact, be related to a confounder, if that confounder has a direct response on the effect, and may
be associated with the exposure at higher doses but not at lower doses. This is further discussed
in the hazard characterization narrative (Section 4.1.5) and the dose-response characterization
(Section 4.2).
4.1.3.1.4. Experimental evidence. Experimental evidence is provided with hypothesis testing.
This hypothesis testing includes manipulation of the exposure scenario with resulting alterations
in the response or response rate of outcomes. Hill (1965) defined experimental evidence, as
evidence that removal of the exposure or supplementation with an antidote leads to a reversal of
the outcome. Experimental evidence or hypothesis testing would include manipulation of the
exposure scenario with resulting alterations in the response or response rate of outcomes. For
some agents, this concept can apply to a lifestage assessment. For example, cases of exogenous
estrogen exposure in prepubertal boys can lead to gynecomastia (male breast development) that
can be reversed after removal of the estrogenic agent (Edinin and Levitsky, 1982; Felner and
White, 2000). However, for other agents, removal of the exposure after the critical window has
passed may not result in a reversible effect. In addition, effects may not occur when the
exposure occurs outside of a given critical window. If studies exist that demonstrate a particular
exposure during a defined critical developmental window, then this could constitute
experimental evidence of the importance of that critical window of exposure. For example,
prenatal thalidomide exposure leads to altered limb bud development (Stevens and Fillmore,
2000), and prenatal alcohol exposure can result in the irreversible outcomes related to fetal
alcohol syndrome (e.g., facial dysmorphogenesis, cognitive deficits, Hubert et al., 2006; Yelin et
al., 2005). If there is a hypothesized MOA, in vitro or transgenic animal models (e.g., knock-out,
knock-in, conditional expressors) may add further weight and experimental evidence to a
hypothesized association.
4.1.3.1.5. Reproducibility. Reproducibility, also termed corroboration by Gray et al. (2001),
means that specific effects are seen under varied conditions. In the case that a lifestage-specific
4-18
-------
effect is consistently observed in similar studies, under varied conditions, in multiple
laboratories, across species, and by various routes of exposure, stronger weight can be placed on
the chemical's association with the effect since it is less likely that biases or confounding factors
are responsible for the results. However, inconsistent findings may be notable. For
developmental toxic agents, exposure occurring during only one specific lifestage, but not all
developmental exposures, may result in the outcome of concern. What may appear as a lack of
reproducibility may actually be the result of disparate study designs examining different critical
windows. Therefore, caution is warranted in dismissing seemingly inconsistent findings without
careful consideration.
4.1.3.1.6. Biological plausibility. Biological plausibility is the determination of whether an
observed outcome could be attributed to the toxicologic insult, given the currently known
science. Biological plausibility may be informed by such things as available information on the
biologic mechanism of a toxic response or on TK and TD similarities and differences across
species or strains or for various lifestages. Some differences in sensitivity between different
rodent strains have been found (Spearow et al., 1999, 2001). A toxic response observed
following developmental lifestage exposure may be different from the response after exposure to
an adult, and the response may be explained by critical windows of susceptibility. Cross-species
and cross-strain similarities or differences in developmental windows of exposures may impact
comparison for the database as a whole. For example, certain prenatal stages in humans are
comparable to certain postnatal stages in rodents. However, when intra- or interspecies lifestage-
specific data are lacking, a default assumption that exposure during any lifestage in experimental
animals causes similar effects in humans is often applied. Another default assumption is that a
response observed in experimental animals is expected to occur in humans (U.S. EPA, 1991,
1996, 1998b, 2005a,b). Defining these assumptions and the uncertainties that they address is a
key part of the WOE evaluation and the identification of data needs.
To move towards more quantitative interspecies comparisons will require a better
understanding of developmental biology and ontogeny of different organ systems. Several
relevant papers comparing organ and system development across species are available for
reference (Hattis et al., 2004, 2005; Hurtt and Sandier, 2003a,b; Selevan et al., 2000).
Comparison of specific physiological systems include the female (Beckman and Feuston, 2003)
4-19
-------
and male (Marty et al., 2003) reproductive system, the cardiac system (Hew and Keller, 2003),
the immune system (Holsapple et al., 2003), the central nervous system (Wood et al., 2003), the
gastrointestinal system (Walthall et al., 2005), the renal system (Zoetis and Hurtt, 2003a), the
respiratory system (Zoetis and Hurtt, 2003b), and osteogenesis (Zoetis et al., 2003).
4.1.3.1.7. Alternative or multiple explanations. One must consider and clearly articulate other
explanations for the observed outcome(s) after the exposure of interest. It is important to
consider whether these explanations are consistent with the database. Reasons for null findings
must also be examined. Alternative hypotheses may also explain similar findings. If other
hypotheses can be ruled out, then more weight can be given to the principal hypothesis or
alternative hypotheses defined in problem formulation (Chapter 3). For example, a non-
mutagenic MOA could be considered as an alternate explanation to the primary hypothesis of a
mutagenic MOA leading to childhood leukemia. In another example, decreased pup body
weight in a two-generation reproduction study may be the result of direct toxicity to the pups at a
susceptible lifestage, or alternatively, the toxicant may be interfering with lactation in the dams,
thereby depriving the pups of nutrition needed for normal growth. These alternative
explanations could have very different implications for judgments about children's risk. This
information is also discussed in the risk characterization when considering explanations for
alternative risk estimates (Section 5.1.6).
4.1.3.1.8. Specificity. Specificity, as discussed by Hill (1965), entails a single cause and effect
relationship resulting from exposure to an environmental agent. It may be difficult to define
such a relationship for developmental outcomes since the alteration of organizational events may
be altered during development and thus may lead to multiple outcomes, depending on the critical
window of exposure (Barker hypothesis, Lau and Rogers, 2004). Evaluating specificity of a
particular MOA, with regard to both the timing of exposure and individual outcomes, presents a
challenge in part because so much time elapses between the occurrence of exposure and latency
of expression of an outcome (Section 4.1.3.1.1). Specificity is defined within the context of this
document as a determination of the relationship between one exposure, the effect(s), and whether
each effect is mediated through a single or alternative MOAs. Exposure during a critical window
may lead to several adverse outcomes; alternatively, the MOA may be unique for developmental
4-20
-------
lifestage exposures when compared to later lifestage exposures. Similarly, the effect of one
agent may vary depending on differences in critical windows of development for the target tissue
or organ.
4.1.3.1.9. Coherence. Coherence summarizes all the principles discussed above and discusses
the extent to which the data are similar in outcome and exposure/dose and whether they support
each biologically plausible hypothesis or MOA. An observed association is given more weight
when it is consistent across the database. This relates to both reproducibility (Section 4.1.3.1.5)
and biologic plausibility (Section 4.1.3.1.6). An example of coherence is the observance of
treatment-related increases in total resorptions at cesarean section in a prenatal developmental
toxicity study and the corollary observation of reduced litter sizes at parturition in a two-
generation reproduction study. Relating the existing database to the larger toxicologic database
about structurally related chemicals or chemicals with a similar MOA can be useful to address
coherence and bridge some data gaps. For example, for one chemical with detailed MOA
information, the summary could focus on hazard in relation to that MOA and what the MOA
may predict about potential critical windows. For other chemicals, the description might focus
on specific developmental outcomes, target organs, or susceptible lifestages. SARs with other
chemicals or chemical classes may be explored to determine the extent to which these data can
inform the assessment via an MOA discussion or to help reduce uncertainties.
Table 4-2. Examples of lifestage-specific questions for evaluation of the
WOE of the hazard database.
Topic
Lifestage-Specific Question(s)
Temporality
(Section 4.1.3.1.1)
To what degree were the timing of exposures described, including the exposure level
and the lifestage of exposure?
Do time-course data exist following developmental lifestage exposures?
Within the hazard database, are exposure intervals or timing of outcome assessments
missing that are necessary in describing the relationship between the exposure and
outcome timing?
Strength of the
Association
(Section 4.1.3.1.2)
How sufficient is the database for evaluating developmental lifestage exposure?
Are the lifestage-specific data of adequate quality? Do the adequate quality studies
comprise a database of adequate quantity?
Did the relevant studies have sufficient statistical precision for confidence in the
results?
For human data, to what degree were confounding factors, effect modifiers, and other
risk factors considered? Were the major demographic and other personal/community
characteristics examined (e.g., age, sex, ethnic group, socioeconomic status, smoking
status, occupational exposure)? To what degree were biases considered?
4-21
-------
Variability Analysis
What sources of variability have been identified in the lifestage-specific database?
What effect does this variability have on the interpretation of the lifestage-specific
database?
Uncertainty Analysis
What are the significant data gaps in the database with regard to children's risk?
• Which lifestages of exposure were assessed? Did exposure occur throughout all
critical lifestages? Were relevant lifestage pathways of exposure and exposure
intervals evaluated? Were there developmental stages during which exposure
was intermittent or did not occur. What was the potential impact of any gaps in
exposure? Were lifestage-appropriate biomarkers of exposure assessed?
• Were all critical outcomes evaluated across lifestages? Have appropriate organ
systems, tissues, and outcomes been adequately assessed for all lifestages of
concern? Were lifestage-appropriate biomarkers of outcome assessed?
• Does the extent of the database for risk from children's exposure indicate the
need for follow-up studies to better define uncertainties for the specific
assessment question and issues?
What are the resulting uncertainties in the database with regard to children's risk?
• Have any uncertainties in developmental exposure been identified?
• Have any uncertainties in internal dose estimation been identified following
developmental exposures (e.g., are there TK data that support the study design
and the interpretation of the data for critical lifestages)?
• Did the conduct of the studies in the database result in uncertainties in findings
that are particularly pertinent to lifestage-specific data interpretation? Were
some studies or data excluded on the basis of poor quality?
Can information from the comparison of structurally related chemicals, or chemicals
with a similar MOA with lifestage-specific data, be used to modify the impact of
identified uncertainties or data gaps?
Qualitative Dose-
Response
Relationship
(Section 4.1.3.1.3)
What is the nature of the dose-response relationship for developmental exposures and
outcomes at all lifestages? What is the shape of the dose-response curve?
Are there differences seen in dose-response curves for the same outcome between
studies? Could confounding factors explain these differences?
Are there differences in dose-response curves for specific lifestages?
Experimental
Evidence
(Section 4.1.3.1.4)
Has the hypothesized critical window of exposure been supported by additional
epidemiologic data in humans or experimental evidence in animals?
Do alterations or differences in exposure paradigms result in alterations in outcome?
Reproducibility
(Section 4.1.3.1.5)
Were the findings examined for consistency within and across studies, laboratories,
species, and strains?
Could inconsistencies in findings be explained by differences in exposures during a
critical window of development?
Biological
Plausibility
(Section 4.1.3.1.6)
If there are lifestage-specific findings, were they examined for biologic plausibility?
Are there temporal differences between experimental animals and humans for the
lifestages when exposures or specific outcomes occur (i.e., what are the comparable
developmental events among the species and strains)?
Are there any cross-species differences in developmental windows of exposures that
impact comparison for the database as a whole?
Was dosing/exposure during potential or known critical windows of exposure
identified for both humans and experimental animals? Is the dosing route used for
animals relevant to human exposure?
4-22
-------
Is the dose-response relationship seen in experimental animals at doses that are
relevant to exposure at developmental lifestages in humans (i.e., environmental
levels)?
Are there any interspecies similarities or differences of effects for comparable
lifestages of development?
Are there any intraspecies (e.g., cross-strain) concordance of effects at lifestages of
development? If not, are there underlying biological reasons to explain these
differences?
What are the key toxicologic and epidemiologic studies that provide the basis for
health concerns following children's exposures? Do other valid studies support or
contradict these findings? Are negative studies considered?
What adverse outcomes at the lowest exposure levels were observed, and what is the
basis for these observed outcomes? Have precursor events/biomarkers or the MOA
been identified?
Besides the developmental lifestage effects observed in the key studies, are there
other health outcomes of concern?
Have the appropriate studies been performed (within the database or elsewhere) to
determine critical windows of exposure? If so, what are they? Did exposure intervals
include known or suspected critical windows?
Alternative or
Multiple
Explanations
(Section 4.1.3.1.7)
Should some data or studies be eliminated from consideration or inclusion in the
WOE evaluation?
To what degree were alternative explanations considered?
Are studies with null findings considered?
Are alternative hypotheses considered that might explain the observed lifestage-
specific outcomes? Does an alternative hypothesis better explain the data than the
primary hypothesis?
Specificity
(Section 4.1.3.1.8)
Is there a specific outcome associated with a specific lifestage exposure?
Are there multiple outcomes that manifest after developmental lifestage exposures?
Can these be explained through a common MOA?
Coherence
(Section 4.1.3.1.9)
Was a meta-analysis performed to combine epidemiologic or toxicologic studies?
Do the data provide information about lifestage susceptibility? Is the relationship
consistent across lifestages or specific to exposure during one or more lifestages?
What types of human studies are available (e.g., case-control, cohort or human
ecologic studies, or case reports or series)?
Assuming relevant exposure routes considered for specified lifestages, were study
results consistent?
Could differences in pathways and intervals of exposure explain differences in study
results for relevant lifestages?
4.1.4. Iteration with Dose-Response and Exposure Characterization
The information gathered in this hazard characterization step will subsequently be used in
the dose-response characterization step (Section 4.2). For example, if there are data from the
exposure-response array (e.g., no-observable-adverse-effect-levels [NOAELs], lowest-
observable-adverse-effect-levels [LOAELs], benchmark doses [BMDs], BMD lower confidence
limits [BMDLs]), or data supporting other quantitative approaches like quantitative risk
4-23
-------
estimation (QRE), then this information is subsequently considered in dose-response
characterization.
For human studies, consideration of and coordination with the exposure characterization
step (Section 4.3) is helpful at this point in the process and can provide important context for the
evaluation of the hazard outcomes, characterization of uncertainties, and identification of further
testing or research needs.
4.1.5. Lifestage-Specific Hazard Characterization Narrative
In this final step in the hazard characterization, a scientific rationale for the selection of
outcomes is clearly and concisely summarized. Included are considerations of lifestage-specific
outcomes, including susceptibility of individuals, the impact of interindividual variability on
response, and remaining uncertainties in the hazard evaluation. Lifestage-relevant outcomes in
the lower dose ranges are described for use in the quantitative dose-response characterization
(Section 4.2). For example, different low-dose ranges (e.g., NOAEL; LOAEL) may have been
identified for different outcomes or for different lifestages of exposure, depending upon the
routes and durations of exposure. The hazard characterization information is also combined with
the exposure characterization information (Section 4.3) to determine a risk characterization that
includes components for describing lifestage-specific risks (Chapter 5).
The report A Review of the Reference Dose and Reference Concentration Processes
recommends summarizing the extent of the database by describing it as a continuum from a
minimal to a robust database (U.S. EPA, 2002a, pp. 4-19). These terms define the continuum of
database characteristics, with minimal describing the least amount of information that would be
sufficient to conduct a risk assessment, and robust including data that fully characterize the
potential toxicity of a chemical or group of chemicals. The intent is for the assessors to
characterize and justify the extent of the database in a narrative form, including variabilities, data
gaps, and uncertainties (e.g., lifestage-specific exposures and outcomes, TK and TD data, the
types of outcomes evaluated and lifestages of assessment of outcomes, reversibility of effect, and
latency to response) that aid in determining the extent of the database.
The lifestage-specific hazard characterization narrative includes thorough assessment of
the overall variabilities (Section 4.1.3.1.2.1), uncertainties, and data gaps (Section 4.1.3.1.2.2)
that have been identified in the database, both generally and specifically for evaluation across
4-24
-------
lifestages. For example, well-justified decisions to include or exclude a given study from the
database or exposure-response array are explicitly stated. The emphasis of the hazard summary
is on the relationships (i.e., patterns) across observed outcomes in relationship to lifestages and
MOA. Often, the structure and presentation of the summaries are driven by the outcome data.
The database may also include in vitro, MOA, and exposure data and toxicity data in adults that
help profile the toxicologic response in children or provide support for assumptions made during
the hazard characterization. Following are some overarching questions to ask in the hazard
characterization narrative:
• What are the lifestage-specific outcomes from the whole database that were
identified in the lower dose range(s) (not just a single "critical effect")? What are
the lifestage-specific outcomes relevant for use in quantitative dose-response
characterization?
• What are the most susceptible lifestages for exposure (e.g., women of
childbearing age [preconception and fetuses], breast feeding infants, toddlers or
older children) from the available data? Is there justification for the most
susceptible lifestage(s) provided by the data to support the relevant outcomes of
concern?
4.2. LIFESTAGE-SPECIFIC DOSE-RESPONSE CHARACTERIZATION
4.2.1. Introduction
Ideally, the adverse health effects identified in the hazard characterization (Section 4.1)
are linked to relevant environmental exposure predictions (Section 4.3) through a dose-response
characterization. The nature and number of risk estimates is governed by the problem
formulation (Chapter 3), hazard characterization (Section 4.1), and the available data. A
lifestage-specific dose-response characterization (Figure 4-5) begins with the summary of the
available data from the hazard characterization (Section 4.1) to conceptualize a MOA, to select
dose-response models, and to apply extrapolations and derive risk values. Variability,
sensitivity, and uncertainty are analyzed, and the results of the entire analysis are iterated, if
necessary, with the hazard and exposure characterizations. The dose-response characterization
culminates with a descriptive narrative of the data, models, estimates, and uncertainties applied
in the dose-response estimate.
Consideration of differences in both human and experimental animals for routes and
durations of exposure, TK and TD processes, and outcomes helps inform the selection of the
4-25
-------
"ft
ForrmtiiUinn
Lifestage-Specific Analysis
(Chapter 4)
Lifestage-Specific
Dose-Response Characterization
(Section 4.2)
Mode of Action (MOA) Conceptualization (Section 4.2.2)
Analysis in the Range of Observation and Dose-Response
Models (Section 4.2.3)
Extrapolations and Risk Derivation from a Lifestage Approach
(Section 4.2.4)
Variability, Sensitivity, and Uncertainty Analyses
(Sections 4.2.5, 4.2.6, 4.2.7)
Iteration with Hazard and Exposure Characterization
(Section 4.2.8)
Lifestage-Specific Dose-Response Characterization Narrative
(Section 4.2.9)
,- "/ . <«,, V""'
;j,i/»i
V « ' is it,«t r ' •.-)(,
H«VK
h,\- ,n I- t s »*ir
Risk ( '
^ I
, M:
-------
most appropriate dose-response data and models for a given assessment. In the past, different
analytical approaches have been used, depending on whether the outcomes were cancer or
noncancer effects. More recently, there has been a recognition within the scientific community
that the traditional dichotomy of cancer versus noncancer dose-response characterization is
problematic, and approaches for characterizing outcomes to have either threshold (i.e. nonlinear)
or non-threshold (i.e., linear) responses based upon their MOA(s) have been proposed
(Bogdanffy et al., 2001). This harmonized approach recognizes that both cancer and noncancer
outcomes can appropriately be characterized as threshold or non-threshold when data are
available to support this selection.
Based on the problem formulation for a given risk assessment, an approach for carrying
out a dose-response characterization is developed. As described in the problem formulation
(Chapter 3), the scope and breadth of an assessment are established and generally fall into two
categories, narrow and broad. The narrow or broad focus of the problem can restrict the dose-
response characterization to more defined approaches. Regardless of the breadth of the
assessment, the exposure scenario, or the hypothesized MOA of the environmental agent, the
lifestage approach can add to the overall soundness and confidence in the assessment.
Dose-response values are typically categorized by route (oral, dermal, inhalation) and
duration of exposure (acute, short-term, chronic). For instance, reference dose (RfD) and
reference concentration (RfC) values can be calculated for various routes and durations of
exposure (U.S. EPA, 2002a). Acute, short-term, and subchronic exposures are of particular
concern because embryogenesis and prenatal, neonatal, and postnatal development provide
ample opportunities for toxicant exposures to alter the regulation of development, which may
lead to qualitatively different outcomes than equivalent exposures in adults.
Perhaps less apparent, however, is the applicability of long-term, or chronic, risk values
to children. Although reference values (RfVs) derived from adult data are thought to be health-
protective of sensitive populations (due to the application of intraspecies and database UFs),
children may be chronically exposed to environmental toxicants. Chronic exposure is defined as
exposure up to 10% of lifetime; therefore, seven years of exposure meets the EPA definition of
chronic human exposure (U.S. EPA, 2002a). Thus if data suggest that a developmental lifestage
is the most sensitive and sufficient data are available, an RiV could be derived from this
4-27
-------
lifestage. In this derivation, the magnitude of the intraspecies and database UFs may be different
than if the RfV were derived from adult data.
Unit risk estimates such as cancer slope factor (CSF) and inhalation unit risk (IUR) are
used to define the exposure concentration that yields a given level of risk during a lifetime (e.g.,
1 x 10~6). Although the latency of time to tumor may mask detection of cancer from exposures
occurring in developmental lifestages, early exposures may indeed increase the risk of tumor
development in later lifestages. In fact, there is evidence to support the notion that susceptibility
to tumor development from exposure to mutagenic chemicals during earlier lifestages is greater
relative to exposure in later lifestages (U.S. EPA, 2005b). Depending on the goals stated in the
problem formulation of a risk assessment (Chapter 3), consideration of studies that have
examined cancer in adult humans and experimental animals following early-life exposure may be
warranted.
4.2.2. Mode of Action Conceptualization
Dose-response characterization can proceed along two paths, one in which the
quantitative dose-response values are developed with little or no insight into the MOA of an
environmental toxicant, or one in which the dose-response values are informed by MOA. In the
latter case, the assessment uses a broader body of scientific literature to look for commonalities
in responses across studies, similarities to other chemicals, and mechanistic data from a wide
array of studies and fields of specialization. MOA information is increasingly recognized in the
scientific community as a foundation from which to build a dose-response characterization
(Andersen et al., 2000; Andersen and Dennison, 2001; Clewell et al., 2002a; Preston, 2004). In
order to conceptualize an MOA, the following are summarized: the available dose-response
model(s), the mechanistic data that relate the critical effect(s) of interest to a particular dose
metric, and the data supporting the choice of a likely or hypothesized dose metric. The following
subsections describe topics to consider during the MOA conceptualization and example
questions are addressed in Table 4-3.
4.2.2.1. Summarizing the Available Dose-Response Data
Quantitative assessments identify and summarize dose-response data to characterize the
potential risks from exposure scenarios identified during the problem formulation (Chapter 3).
4-28
-------
This process also interfaces with the exposure characterization (Section 4.3), where source-to-
dose modeling informs assessors about the relevant exposure scenarios (i.e., route and duration)
and likely ranges of external exposure levels for various lifestages. Because low-dose
extrapolation has inherent uncertainties regarding MOA over dose ranges (Slikker et al.,
2004a,b), the exposure characterization can help inform selection of the appropriate dose-
response model from which to obtain a POD.
An exposure-response array can help identify critical outcomes (U.S. EPA, 2002a,
Section 4.4.1) across dose ranges and aid in the conceptualization of the MOA. For instance,
different effects at similar doses may originate through common mechanisms, and thus lend
support to one or more MOAs. Alternatively, different effects across dose ranges may represent
a gradient of effects operating through common mechanisms, and thus also lend support to one
or more MOAs. It is also possible, of course, that different MOAs are operational across dose
ranges, and an exposure-response array can be useful for defining the range of effects. Multiple
responses can be described as a continuum of dose as well as continuum of lifestages when using
this array. For instance, exposure-response arrays for various toxicity outcomes across
developmental lifestages have been used to help inform outcome selection for dose-response.
For example, exposure-response arrays have been used in the assessment of dibutyl phthalate
(U.S. EPA, 2006a); where, using this approach, it becomes evident that adverse developmental
effects occur at lower exposure levels than other adverse effects (e.g., hepatotoxicity). This
approach is both important for dose-response characterization and is informative for risk
characterization (Chapter 5). An alternative approach for summarizing the dose-response data is
to use categorical regression (Section 4.2.3.1). This approach lumps different responses together
by assigning key outcomes to severity categories—perhaps irrespective of MOA.
In circumstances where data exist for multiple lifestages, it is possible that effects at
earlier lifestages pose greater risk due to the potential for irreversible changes (e.g.,
developmental neurotoxicity) or changes that confer an increase in risk to subsequent exposures
in later lifestages. For instance, it is hypothesized that acute lymphocytic leukemia (the most
prevalent childhood leukemia) results from an early (perhaps prenatal) initiation event forming a
fusion gene, followed by a subsequent key event in later childhood (Greaves, 2003). If this
initiation event could be ascribed to a particular environmental exposure, then this event could
potentially be an important precursor event to consider due to the increased risk for latent
4-29
-------
adverse effects, such as leukemia. Therefore, detailed MOA considerations can inform the
selection of lifestage data for dose-response characterization. However, in cases where dose-
response data do not exist for specific lifestages of concern (i.e., data gaps), MOA may be able to
inform dose-response characterization for these lifestages by allowing for intraspecies (e.g.,
lifestage) extrapolation using biologically based modeling techniques (Section 4.2.3). Effects
that are thought to share common key events in the proposed MOAs can give assessors
confidence in choosing dose-response models that most closely relate to the underlying biology
and adapt those models to other lifestages of interest.
4.2.2.2. Mechanistic Data and Mode of Action
The complexity of physiological development provides opportunity for toxic exposures to
create TD effects that may or may not be relevant to all lifestages within a species.
Developmental stages or age groupings (Table 3-1, U.S. EPA, 2005e) can be based on such
metrics as growth rates/spurts, behavioral traits, organ systems, or perhaps functional
development. It may be possible to plot these metrics for development throughout lifestages and
across species. This comparison can aid in identification of organ systems (e.g., respiratory,
cardiovascular, central and peripheral nervous systems, immune system) that might be at risk
during comparable windows of exposure and can inform the decision of which effects and dose-
response data are most useful. Although matching comparable lifestages across species is a
challenge (U.S. EPA, 2002a, Table 3-1), such efforts have the potential to decrease the
interspecies TD differences that influence dose-response relationships across species
(Section 4.2.4.2).
4.2.2.3. Selection of Dose Metric Informed by Mode of Action
When physiologically based toxicokinetic (PBTK) models are available for a chemical, it
may be possible to convert the external/applied dose in a study to an internal target tissue dose
(i.e., dose metric). This can be an internal measure of the chemical or its metabolite(s) but can
also be measures of adduct formations, cofactor depletion, etc. In addition to identifying the
chemical moiety (e.g., adduct) of the dose metric, it is equally important to identify the most
appropriate measure of the dose metric; frequently these are the average daily doses under the
concentration versus time curve (area under the curve, AUC), peak concentration (Cmax), or rate
4-30
-------
of production. Selection of the appropriate dose metric for a given dose-response relationship is
done in the context of what is known or hypothesized about the MOA, and thus is an inherently
iterative process between dose-response characterization and hazard characterization (Section
4.1). In practical terms, the measured substance may not be the toxic moiety at the target tissue
but rather a surrogate such as blood concentration of the parent compound or one or more of its
metabolites. Another important consideration to the selection of the dose metric is the outcome.
For example, peak concentration may be more important for some outcomes compared to others
(e.g., neurotoxicity vs. tumors). When choosing among potential dose metrics, often the
appropriate choice can be identified as the one that demonstrates a consistent relationship with
positive and negative responses observed at various dose levels and across exposure scenarios
within a single species (U.S. EPA, 2006b).
Clewell et al. (2002a) have proposed two criteria for dose metric determination:
plausibility, defined as consistency with MOA and ability to simplify a complex dose-response
relationship, and conservatism, defined as the selection of the dose metric that poses the highest
risk or the lowest acceptable exposure level. It is the environmental exposure level to humans
that is regulated as a result of risk assessment; thus, a potent dose metric is not synonymous with
a potent external dose. Therefore, when there is insufficient data with which to determine the
more appropriate dose metric, the one related to the most potent external exposure dose is often
appropriate. More detailed information on dose metric selections is in Approaches for the
Application of Physiologically Based Pharmacokinetic (PBPK) Models and Supporting Data in
Risk Assessment (U.S. EPA, 2006b).
Table 4-3. Examples of lifestage-specific questions for MOA
conceptualization.
Topic
Lifestage-Specific Question(s)
Summarizing the
Available Dose-
Response Data
(Section 4.2.2.1.)
What dose-response data/models are available for lifestages of interest (e.g., preconception,
pregnancy, infancy, childhood)?
Are the exposure scenarios in these studies the scenarios of interest? Can route and duration
extrapolations be employed using modeling techniques (Section 4.2.4)?
If data are available for a different lifestage than is of interest, are these amenable to
extrapolation to lifestages for which there is little or no data (Section 4.2.4)?
Can an exposure-response array inform the most relevant studies and outcomes?
Mechanistic Data
Are TD effects known or hypothesized?
and MOA What are the relative expression levels of the key players (e.g., receptors, metabolic enzymes,
(Section 4.2.2.2.)
DNA repair enzymes) in the known or hypothesized MOA at the lifestages of interest?
If multiple outcomes are evident, are they likely linked by MOA? Do the outcomes share
common mechanisms? Or, do the outcomes represent a gradient of the same MOA?
4-31
-------
Selection of Dose
Metric Informed ammai data available regarding the dose metric that is likely to be relevant to the human
by MOA (Section
What is the human lifestage exposure scenario of interest (route, duration, and pattern)?
lifestage of interest?
4.2.2.3)
Is the selected type of dose metric appropriate for both the outcome and the exposure (e.g.,
duration) of interest?
Are there models available which can convert the external/applied dose used in a study to an
internal delivered dose (i.e., a dose metric)?
4.2.3. Analysis in the Range of Observation and Dose-Response Models
A number of models are typically employed in order to determine PODs, which are used
for extrapolations in dose-response characterization and margin of exposure (MOE) analysis in
risk characterization (Section 5.1.3). Data for dose-response characterization in the range of
observation come in many forms: empirical PODs derived from either a NOAEL, a LOAEL, or
sophisticated models incorporating mechanistic data. The nature and amount of data required for
each type of dose-response characterization might represent a hierarchy, although the more
sophisticated dose-response models still rely on the same experimental animal studies from
which a NOAEL or LOAEL can be derived, either as a basis for curve fitting mathematical
models or a starting point from which to calculate an internal target tissue dose using other
modeling techniques.
In Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), the EPA has adopted
an approach that advocates the use of as much biologically informed dose-response data as
possible, and suggests that "default" approaches be used only in instances where little data exists
concerning an environmental toxicant of interest. PBTK modeling and BBDR modeling provide
strong biological foundations for a chemical risk assessment; their application in risk assessment
is discussed more thoroughly in Approaches for the Application of Physiologically Based
Pharmacokinetic (PBPK) Models and Supporting Data in Risk Assessment (U.S. EPA, 2006b).
Moreover, their use in conjunction with statistical modeling is perhaps the most rigorous and
scientifically based approach to dose-response modeling (U.S. EPA, 1999a).
The following brief descriptions summarize the types of analyses used in dose-response
characterization. Example questions are addressed in Table 4-4, including those based on
limited data sets and those requiring rich data sets for dose-response characterization.
Traditional approaches to dose-response modeling of a toxicant with an assumed
nonlinear MOA have relied (and continue to rely) heavily on the use of empirical data points for
determining PODs. Often these are NOAEL and LOAEL values derived from experimental
4-32
-------
dosing conditions in toxicologic studies. Two main disadvantages of using these single point
estimate values are that they do not consider the shape of the dose-response curve, and they do
not allow for estimation of risks at any exposure level of interest (Allen et al., 1998). Thus the
use of NOAEL and LOAEL values alone represents the bottom tier of dose-response models and
are used most often when limited data are available concerning the toxicant of interest.
Empirical modeling approaches, sometimes called curve fitting or statistical modeling,
represent an improvement over traditional NOAEL and LOAEL dose-response characterization
techniques. In these approaches, statistical models are fit to empirical response data (e.g.,
tumors) or precursor events (e.g., signal transduction or changes in blood hormone level). In
some instances, low-dose extrapolation beyond the observed response data can be informed by
precursor data over the low-dose range (U.S. EPA, 2005a). In other instances, linear low-dose
extrapolation may be employed for extrapolating from the range of observation down to, for
instance, background levels, rates, or incidence. This form of statistical modeling has been used
for noncancer outcomes to develop quantitative risk estimates (discussed at the end of this
section). The draft Air Quality Criteria for Lead (U.S. EPA, 2006c) contains further discussions
on implications for low-dose extrapolation using statistical modeling (i.e., linear and log linear
models).
Another form of statistical modeling for determining PODs is BMD analysis (Crump,
1984).4 The BMD is defined as the dose at which a predetermined change in response incidence
(e.g., 5% or 10% change in critical effect such as pup body weight or pup mortality) occurs; with
the 95% lower confidence limit being the BMDL (Allen et al., 1994a,b; Faustman et al., 1994;
Kavlock et al., 1995; Kimmel et al., 1995; U.S. EPA, 1995b). An advantage of this approach is
that it attempts to fit statistical models to existing dose-response data, regardless of whether the
MOA is linear or nonlinear, taking into account all of the data points in an individual dose-
response study (Brown and Strickland, 2003). Thus, unlike the NOAEL/LOAEL approach, the
BMD is influenced by the shape of the dose-response curve for developmental outcomes (Allen
et al., 1998). The selection of BMD may require studies with more dose groups and a higher
number of subjects, and therefore, can be performed only when the scientific database for an
environmental chemical is relatively large. Because the BMDL depends on the study design,
more rigorous studies generally have narrower confidence limits (Barnes et al., 1995).
4 EPA has developed software for BMD analysis, available at http://cfpub.epa.gov/ncea/cfm/bmds.cfm.
4-33
-------
Importantly, the BMD approach is less sensitive to dose spacing, and thus a BMD can be
determined in the absence of a NOAEL as well as for any increase in response level (Allen et al.,
1998; Barnes et al., 1995). For further readings on choosing studies for BMD analysis, refer to
the draft Benchmark Dose Technical Guidance Document (U.S. EPA, 2000d).
Categorical regression5 analysis is similar to BMD analysis, but whereas BMD analysis
uses a single study, categorical regression combines studies. In this method, data are pooled
from different studies (possibly with different exposure parameters and outcomes) that are
"assigned" to the same severity category (Brown and Strickland, 2003). An advantage to this
approach is that a small number of studies can essentially be combined into one larger study and
can thus narrow the confidence limits (Brown and Strickland, 2003). This methodology may be
particularly useful in a lifestage approach where it is likely that fewer studies have been
performed on the specific lifestages of interest or critical windows of susceptibility.
PBTK and BBDR modeling are perhaps the most amenable modeling techniques for
using a lifestage approach as they are designed to mimic true biological processes and model
whole organisms. Knowledge and understanding of TK differences during each lifestage
(absorption, distribution, metabolism, and elimination), as well as anatomy and behaviors, are
used in estimating delivered dose and may require modification of available adult models.
Several reviews have described the variation in TK factors between adults and children
(Besunder et al., 1988a,b; Bruckner, 2000; Clewell et al., 2002a,b, 2004; Hines and McCarver,
2002; McCarver and Hines, 2002).
Although the use of PBTK models for internal dose estimates is increasing, more effort is
needed in developing such models for children's dosimetric adjustments across lifestages and
experimental animal species. In this regard, pharmacokinetic data from pediatric pharmacologic
studies could be appropriately applied for some portions of certain risk assessments for
developmental lifestage environmental exposures. For instance, general knowledge of
differences between adults and children in metabolic clearance of CYP3A-specific
pharmaceutical substrates could be used by adjusting for these differences in activities in a TK
model when the toxicant is thought to be metabolized by CYP3A (Ginsberg et al., 2004a,b).
1 EPA has developed CatReg software, available at http://cfpub.epa.gov/ncea/cfm/recordisplay.cftn?deid=18162.
4-34
-------
Ginsberg et al. (2002) compiled a database of 45 drugs for which TK data are available across
lifestages.6
PBTK models are particularly useful for conducting extrapolations (e.g., route-to-route,
duration, interspecies, including lifestage extrapolations). Other advantages are that these
models can mimic any exposure scenario (continuous or otherwise) and changes in the
underlying biology (e.g., development). For instance, if children are likely to be exposed to an
environmental toxicant for one hour per day for five days a week (followed by 48 hours of no
exposure), these models can predict the levels of metabolites of interest under these conditions.
Similarly, numerous small exposure doses from breast milk to nursing infants could be modeled
to determine steady-state levels of a toxicant at one month and at three months after birth.
However, PBTK models are not necessarily applicable for extrapolating from short-term
exposure studies to longer-term predictions. This is because the key events leading to the
observed responses are not likely to be impervious to the effects of time and repeated exposure.
Many dose-response relationships may be dependent on temporal changes in TD processes due
to developmental- and exposure-induced changes (e.g., cell proliferation rates, DNA repair
processes, receptor tolerance and desensitization, and age-related changes in physiologic
parameters). Thus, it is feasible to predict steady-state levels of a compound in the body over
long periods of time, yet the response to these levels may differ between short- and long-term
durations of exposure. These differences due to duration of exposure highlight the importance of
having dose-response data for the exposure duration and the lifestages of interest.
Application and review of PBTK models in risk assessment can be found in Ginsberg et
al. (2004b), Pelekis et al. (2001), and U.S. EPA (2006b). There are some developmental
lifestage PBTK models, some of which include infant exposure to chemicals such as dioxin in
breast milk (Gentry et al., 2003; Lorber and Phillips, 2002), fetal exposure to ethylene glycol
monomethyl ether (Gargas et al., 2000), and neonatal exposure to compounds such as lead
(O'Flaherty, 1998) and perchlorate (Clewell et al., 2003; Clewell and Gearhart, 2002). Several
pregnancy and lactation models have been reviewed (Corley et al., 2003).
BBDR models represent the state of the art in dose-response characterization, where
mechanistic TD data are modeled in such a way that responses can be predicted, even at low
' This database can be accessed at http://www2.clarku.edu/faculty/dhattis.
4-35
-------
exposure levels. Usually, output from a PBTK model serves as the dose input to a BBDR model,
relating that dose to a response outcome (Andersen and Dennison, 2001; Ashani and Pistinner,
2004; Setzer et al., 2001). In addition to lifestage-specific TK data, the relationship between the
internal dose metric and response may require lifestage-specific TD data. Currently, relatively
few BBDR models are available due to the inherent complexity of integrating TK and TD data.
Model transparency, quality criteria, and short shelf-life of some models beyond initial
publication also limit BBDR model development (DeWoskin et al., 2001). The use of BBDR
models is expected to increase as toxicologic studies move beyond more frank effects toward
molecular precursor events (Andersen and Dennison, 2001; Faustman et al., 1999).
In instances where the dose metric of a toxicant of interest is structurally related to
another compound for which there exists a validated BBDR model, consideration of the
application of this model to the toxicant being assessed may be warranted.7 As stated in
Evaluation of BBDR Modeling for Developmental Toxicity: A Workshop Report, "the challenge
is to define.. .application of a quantitative BBDR model.. .generalizable to other compounds in a
similar class and perhaps to certain other classes of compounds" (Lau et al., 2000). For example,
two chemicals might be hypothesized to affect similar TD processes (e.g., activation of a
particular receptor), yet a BBDR model may exist for only one of the chemicals. If a PBTK
model is available (or can be developed) for the chemical that does not have a corresponding
BBDR model, it is conceivable that the existing BBDR model might be sufficient for analyzing
both chemicals.
The top line in Figure 4-6 represents a BBDR model for the dose-response of chemical A,
where TKA, TDA, and RA represent the TK, TD, and response of interest related to chemical A,
respectively. In this scenario, the TD of chemical B (TDe) is thought to be equivalent to that of
chemical A (i.e., both have the same MOA from a TD perspective). If a PBTK model (but not a
BBDR model) exists for chemical B (TKe), then the predicted internal target tissue dose of
chemical B can be integrated into the existing BBDR model for chemical A.
7 Compounds with common TD effects may not necessarily be structurally related.
4-36
-------
BBDR model exists for chemical A: {TKA-> [TDA = TDB] -> RA}
T
PBTK model exists for chemical B:
Figure 4-6. Use of BBDR modeling.
A probabilistic risk assessment approach has typically been used in exposure
characterization and is increasingly being applied for dose-response characterization as data
become available for physiological parameters such as genetic polymorphisms in TK and TD
pathways (Beck et al., 2001; Pelekis et al., 2003). When readily measurable, inputs such as
exposure dose and duration, intake rate, clearance, and body mass can be expressed as
distributions and modeled in such a way as to estimate dose for a particular population, over a
certain time frame, or at a specific location. Similarly, lifestage-specific parameters can be
employed in order to estimate the variability in dose and response among lifestages (e.g., infants
and children).
In regard to noncancer outcomes, one limitation applicable to many of the
aforementioned dose-response modeling approaches is that the analyses are based on toxicologic
outcomes as opposed to public health outcomes. QRE is a broad-based method for relating
human exposures to non-toxicologic outcomes. For example, exposure to l,2-dibromo-3-
chloropropane can be linked to increases in infertility rates through mathematical modeling
(Pease et al., 1991). In this regard, it is similar to BMD analysis, but whereas risk is typically
defined by percent change (e.g., 1% or 5%) in a biological response (e.g., sperm count), QRE
attempts to define risk (e.g., excess infertility cases) for all human exposure levels. The
advantage of this approach over BMD is that a noncancer risk can be defined for any individual
based on exposure level as is done for cancer assessments. Other examples of this type of
analysis include associations among particulate matter and daily mortality and certain measures
of morbidity (U.S. EPA, 2005d) and associations between acute ozone exposures and respiratory
morbidity and mortality (U.S. EPA, 2005f). An inherent disadvantage to this approach is that
acceptable levels of risk must be defined, whereas other approaches to noncancer dose-response
modeling arguably rely less on value judgment.
4-37
-------
Table 4-4. Examples of lifestage-specific questions for analysis in the range
of observation.
Topic
Lifestage-Specific Question(s)
Selection of Dose-
Response Models
(Section 4.2.3)
What data were used to develop the dose-response curve? Are data available from the
lifestage and exposure scenario of interest? Were there differences (e.g., in potency) in the
dose-response curves for different lifestages?
Was a model used to develop the dose-response curve and, if so, which one? What rationale
supports this choice? For example, how was the benchmark response chosen?
What modeling approaches are amenable to the available dose-response data? Is there
sufficient data to support, for example, the use of biological modeling approaches?
4.2.4. Extrapolations and Risk Derivation from a Lifestage Approach
After PODs have been established from various dose-response studies or modeling
techniques, low dose extrapolation is performed in order to derive dose-response values. Again,
this may be done for assessments of narrow or broad scope (Section 3.1), and will have
regulatory implications for various adjustments in order to extrapolate to the exposure scenarios
and lifestages of interest. As described below, these adjustments may involve sophisticated
approaches or default approaches that have developed over time. Despite the term default, many
of these approaches are informed and supported by empirical evidence. For example, empirical
analysis supports the use of body weight scaling (see below) to adjust for TK differences across
species.
However, the use of more sophisticated techniques does not necessarily result in
refinements of final reference or risk values. For instance, a recent assessment of xylenes
resulted in nearly identical RfC values using either default approaches starting from a NOAEL or
sophisticated PBTK modeling (U.S. EPA, 2003c). Despite the fact that this may be a possible
outcome, the use of sophisticated techniques, MOA information, and lifestage analyses certainly
improve the confidence that dose-response values (i.e., RfVs and risk values) are health
protective. The following subsections describe topics to consider for extrapolations and risk
derivation, and example questions are addressed in Table 4-5.
4.2.4.1. Duration and Route Adjustments
Experimental animal studies almost always employ discontinuous exposure protocols and
therefore use continuous dose adjustment. Although such adjustments are conservative from a
risk evaluation standpoint (i.e., they shift the dose-response curve leftward), mathematical
adjustments do not necessarily maintain the dose-response relationship (i.e., AUC) that likely
4-38
-------
reflects the MOA by which a response is generated. An alternative to continuous dose
adjustment is to use PBTK models to determine (in silico) an applied dose (continuous or
otherwise) that results in the same AUC or Cmax which simulates that which could have been
generated in the experimental animal under the original laboratory study conditions. This may
require parameterization with lifestage- and species-specific data. Developmental windows of
susceptibility are relatively short, thus the changing underlying biology during development
suggests that Cmax may be a more relevant dose metric in young children than AUC. Since the
minimal exposure period to elicit an increased risk is often not known, especially during a
window of vulnerability, the choice of exposure period is a critical decision point that integrates
TK, TD, and exposure information.
For route-to-route extrapolation, default equivalent dose adjustments can be used. For
example, standard mg/kg/day adjustments assume similar TK and TD processes between
experimental animals and humans. However, such assumptions are tenuous because different
cell types, enzymes, and proliferation rates exist across portals of entry. PBTK models can be
used to predict target dose across routes by incorporating route-specific TK factors. A limitation,
however, is that route extrapolations are not useful in instances where the critical effects are
specific to the portal of entry. For more on route and duration adjustments, see Approaches for
the Application of Physiologically Based Pharmac akinetic (PBPK) Models and Supporting Data
in Risk Assessment (U.S. EPA, 2006b) and A Review of the Reference Dose and Reference
Concentration Processes (U.S. EPA, 2002a).
4.2.4.2. Interspecies and Intraspecies Adjustments
The EPA RfC process describes the interspecies adjustment from experimental animals to
human equivalent concentration (HEC) via dosimetric adjustment factors (DAFs) (U.S. EPA,
2002a). For oral exposures, default interspecies extrapolation based on body weight (BW)
scaling, either BW1 or BW/4, have been employed. In particular, BW/4 scaling is typically
thought to account for TK differences among species, and therefore, often reduces the
interspecies UF from 10 to 3 (U.S. EPA, 2002a). Recent harmonization efforts at EPA advocate
the adoption of BW " scaling for RfD derivation in instances where there are limited data with
which to perform an assessment (U.S. EPA, 2006d). This has been proposed in an effort to
harmonize oral RfD methodology with RfC methodology. In addition, this effort also aims to
4-39
-------
harmonize the use of BW3/4 scaling in the application of DAFs for oral cancer assessments (U.S.
EPA, 2006a).
For inhalation exposures, DAFs are applied on the basis of physicochemical, anatomical,
and physiological parameters. These parameters include such factors as species-to-species ratios
of surface area: ventilation rate, blood:gas partition coefficients, and regional deposition dose
ratios for particulate matter. In the case of children, it is currently recommended that HECs and
human equivalent doses (HEDs) be determined experimentally and theoretically (U.S. EPA,
2002a). In the absence of DAFs, simple ventilation rate adjustments can be made for HECs.
Finally, it is worth noting that DAFs are thought to be most appropriately applied for chronic
exposures, where the dose metric is likely best represented by AUC; discussion of adjustments
for acute exposures can be found elsewhere (U.S. EPA, 2002a).
In addition to interspecies adjustments, BW% scaling may also be useful for intraspecies
adjustments based on lifestage (U.S. EPA, 2006d). Pharmaceutical data indicate that TK
processes (e.g., chemical half life) in children may also scale to BW%, particularly in children
over two months of age (Ginsberg et al., 2002, 2004a,b; Hattis et al., 2004). Under two months
of age, however, the immaturity of such processes likely precludes scalability.
When more data are available for carrying out an assessment, lifestage considerations can
be incorporated using either intraspecies adjustments or interspecies extrapolation (Figure 4-7).
Adjustments across human lifestages from adult to earlier developmental stages includes
exposure, TK, and TD considerations (Barton, 2005), and this process can be qualitative or
quantitative (Ginsberg et al., 2002). Qualitatively, adultchild ratios for TK processes
representing various metabolic pathways can be used to predict the relative difference in TK
processes between children and adults for a toxicant that is metabolized by the same pathway.
For example, the mean half-lives of several Pharmaceuticals metabolized by C YP3 A can be
compared in adults and children; this ratio could then be used to adjust the intraspecies UF for an
environmental toxicant that is known to be metabolized by CYP3A. Quantitatively, adult PBTK
models (if available) could be parameterized in order to predict the dose metric in children. The
left panel in Figure 4-7 depicts the frequent case where adult animal toxicity data is used to
extrapolate to humans. If sufficient data and models are available, a subsequent intraspecies (or
lifestage) extrapolation could be performed. The right panel depicts a less-frequent (but
preferred) case where toxicity data in a lifestage of interest is used for interspecies extrapolation
4-40
-------
to the corresponding lifestage of interest in humans. Importantly, this approach requires a
qualitative or quantitative evaluation of how homologous the animal lifestage is relative to the
lifestage of interest in humans. In the former case, such TK changes might increase the
intraspecies UF with respect to TK consideration; it has been shown, for example, that such
differences between adults and infants can exceed 3.2-fold (Hattis et al., 2004). In the latter
case, the intraspecies UF may be reduced due to the improved characterization of TK. The
advantage to this approach is that assessors may have greater confidence in extrapolating within
the human species; on the other hand, this approach requires that the underlying toxic response
and MOA are concordant across lifestages. This assumption may add additional uncertainty to
the dose-response characterization.
Adult
Animal
Response
t
TD
t
TK
t
Exposure
V
>
Adult
Human
— *• Response
t
-»• TD
t
-•> TK
t
— »• Exposure .
j \^
Child
+ Response
t
* TD
t
> TK
t
>• Exposure
J
~~Y
interspecies intraspecies
Lifestage
in Animal
Response
t
TD
t
TK
t
Exposure -
V
Corresponding3
Human Lifestage
*• Response
t
> TD
t
* TK
t
>• Exposure
J
V
interspecies
"Importantly, this approach requires a qualitative or quantitative evaluation of how homologous the animal
lifestage is relative to the lifestage of interest in humans.
Figure 4-7. Interspecies and intraspecies adjustments with lifestage considerations.
More often, however, the data needed for lifestage extrapolation will be available only in
experimental animals and thus will often require both qualitative and quantitative adjustments
4-41
-------
(Barton, 2005) (Figure 4-7, right panel). Qualitative adjustments include determining the
developmental stages in experimental animals and humans that exhibit the same window of
susceptibility related to the critical outcome of interest. This may require both empirical
evidence and expert judgment. Several articles have examined the relative development of organ
systems across species (reviewed in Hurtt and Sandier, 2003a,b; Selevan et al., 2000; WHO,
2006). Quantitative adjustments are then needed to account for the TK differences that exist
across species at the equivalent (with respect to the window of susceptibility) lifestages. For
instance, rodents are born at an overall developmental stage roughly equivalent to the end of the
second human trimester. Thus, if equivalent windows of susceptibility exist at these two
different lifestages across species, then altogether different PBTK models and TK data would be
needed to calculate the equivalent internal dose, i.e., a lactational model for the rodent and a
pregnancy model for the human.
An advantage of this approach is that the assessor starts with age-relevant developmental
effects (e.g., two-generation reproduction studies) as opposed to assuming concordance of effects
across lifestages. This will likely have the effect of reducing the interspecies UF due to TK
adjustments and due to a general increase in confidence that TD differences (if they exist) have
been minimized. One caveat is that human data (from controlled exposures or epidemiologic
studies) with which to test the predictive capability of the model is often nonexistent.
Additionally, if extrapolation requires the use of different model structures (e.g., perinatal
exposure in rats and fetal exposure in humans), then each model, with its own inherent
uncertainties, may add to the overall uncertainty in the extrapolation (U.S. EPA, 2006b).
Because the majority of data concerning a chemical will pertain to nonhuman species, TK
and TD data are important elements for lifestage-specific dose-response characterization. It is
for this reason that PBTK and BBDR models have been emphasized for dose-response modeling
under the lifestages approach. There are several examples where existing adult models have
been adapted to developmental lifestages. Gentry et al. (2003) incorporated new tissue
compartments and parameters into a previously published PBTK model for modeling
isopropanol and acetone metabolism in adult humans and rats (Clewell et al., 2001). These
additions include compartments for the uterus, mammary tissue, placenta, and fetus (Gentry et
al., 2003), some of which are modeled to account for growth throughout gestation. Physiological
parameter values were derived from numerous previous publications; currently, the EPA is
4-42
-------
developing relational databases for human and rodent physiological parameters so that these
values become more standardized and reduce some of the variability and uncertainty in PBTK
models.8 Pelekis et al. (2003) demonstrated the use of a lifestage approach by applying
probabilistic analysis to a previously published PBTK model. Briefly, this study modeled daily
exposure of individuals to dichloromethane from birth to 70 years of age using age-specific
physiological parameters, partition coefficients, and CYP2E1 age-specific metabolic data. This
model does not take into account age-related differences in exposure, nor are TD factors
addressed (Figure 4-7, left panel). Lifestage data have also been used in BBDR modeling. For
instance, a BBDR model has been developed for modeling the developmental effects on fetuses
following maternal exposure to 5-fluorouracil on gestational day 14 resulting in birth defects and
birth weight in rats (Figure 4-7, right panel) (Lau et al., 2000). This model employs a PBTK
component that describes the formation of the metabolite, relates the metabolite levels to
deoxyribonucleotide pool perturbation, and relates this perturbation to low fetal birth weight
(Shuey et al., 1994) and fetal malformation (Lau et al., 2001; Setzer et al., 2001; Shuey et al.,
1994).
4.2.4.3. Low-Dose Extrapolation
Ideally, extrapolation beyond the range of observation is informed by MO A. When
MOA is not known, it is possible that the shape of the dose-response curve can be informative
for low-dose extrapolation; however, Lutz et al. (2005) have demonstrated that these shapes can
sometimes be misleading. For instance, the linearity of the dose-response curve often seen in
epidemiologic studies may be due in part to interindividual genetic and life style differences as
well as other issues related to epidemiologic studies such as difficulties in dose reconstruction.
Conversely, Lutz et al. (2005) also demonstrated that animal bioassay studies that suggest a
threshold effect may be misleading. For instance, in silico simulations of dose-response
relationships can result in threshold (or J-shaped) relationships by chance; thus animal bioassays,
often unrepeated, may suggest a relationship that does not exist in reality. Conolly et al. (2005),
also using in silico methods, demonstrated that modeling of adaptive responses to DNA damage
can result in both linear and threshold dose-response relationships depending upon model
8 Conversely, these databases can be used to incorporate variability in physiological parameters into probabilistic
modeling techniques.
4-43
-------
assumptions. Taken together, these studies highlight the importance of a strong understanding of
MOA for choosing the most appropriate low-dose extrapolation approach.
The Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a) advocate an MOA
approach to low-dose extrapolation of cancer outcomes, where low-dose linear extrapolation is
performed when a carcinogen is thought to act through a linear MOA (e.g., mutagenesis) or
when the MOA for a carcinogen is not understood. This is based, in part, on the concept of
additivity (Crump et al, 1976), where any amount of a carcinogen adds to the underlying
biological processes that are responsible for the background incidence of a particular cancer.
Nonlinear extrapolation is used when the MOA can be demonstrated to result from a
threshold (i.e., nonlinear) MOA and can be used for both cancer and noncancer outcomes.
Although nonlinear extrapolation approaches are frequently used for noncancer outcomes, risk
based approaches to noncancer outcome low-dose extrapolation, with potential relevance to cost-
benefit analysis, have been proposed (Clewell and Crump, 2005; Gaylor and Kodell, 2002).
There may also be biological support for low-dose linear extrapolation for certain noncancer
outcomes. For example, l,2-dibromo-3-chloropropane is thought to reduce sperm count by
interaction with DNA (Pease et al., 1991); thus, like for mutagens, there may be a scientific
rationale for using low-dose linear extrapolation for this compound.
4.2 A A. Reference and Risk Value Derivation
Lifestage extrapolations for RfV and risk value derivations can affect the magnitude of
the UFs applied in the final risk value derivation. Current practices for RfC and RfD derivation
and the application of UFs are outlined in A Review of the Reference Dose and Reference
Concentration Processes (U.S. EPA, 2002a). New guidance on CSF derivation from early-life
exposure to environmental agents can be found in the Supplemental Guidance for Assessing
Susceptibility from Early-Life Exposure to Carcinogens (U.S. EPA, 2005b). In brief, the new
guidance states that for toxicants acting through a mutagenic MOA where data concerning early
life susceptibility are lacking, early life susceptibility should be assumed and the following age-
dependent adjustment factors (ADAFs) should be applied to the CSF:
• 10-fold for exposure occurring before 2 years of age
• 3-fold for exposure occurring between the ages of 2 and 16
• no adjustment after 16 years of age
4-44
-------
No such adjustments are advocated for toxicants with either an unknown or non-mutagenic
MOA. These adjustments are based, in part, on analyses indicating an increased incidence of
tumor formation from early-life exposure as compared to adult exposure.
Historically, lifestage-related uncertainties have been folded into the database UF when
the MOA is nonlinear. Lifestage-specific data gaps do not necessarily imply a greater database
UF; rather, the method helps focus attention on the most critical data gaps deserving of
additional uncertainty weighting. This additional weighting following uncertainty analysis
(Section 4.2.7) would support prioritization of data needs. Indeed, the rationale for using the
lifestage approach is to better characterize individual risk and thus decrease uncertainty in risk
assessment.
Table 4-5. Examples of lifestage-specific questions for extrapolations and
risk derivation.
Topic
Lifestage-Specific Question(s)
Duration and
Route
Adjustments
(Section 4.2.4.1)
Do default duration adjustments maintain the relationship between exposure and response?
Are the effects specific to the portal of entry?
Can existing models be used to extrapolate to the lifestage-specific exposure scenario of
interest using PBTK models?
Interspecies and
Intraspecies
Adjustments
(Section 4.2.4.2)
Should the same interspecies factors (e.g., DAFs) be applied in deriving HECs and human
equivalent doses for all lifestages?
Can developmental lifestage dose-response characterization be conducted based on adult
animal or human data (Figure 4-7)?
Can developmental lifestage dose-response characterization be conducted based on
developmental lifestage animal data (Figure 4-7)?
What data are available to perform extrapolations for developmental lifestages?
Low-Dose
Extrapolation
(Section 4.2.4.3)
Is the MOA known?
Is the chemical a known mutagen?
Do statistical modeling approaches result in reasonable results in the low-dose range?
Are PBTK models available? Can, for instance, a BMD;0 be based on internal dose metric
rather than applied dose?
Reference and
Risk Value
Derivation
(Section 4.2.4.4)
Is the toxicant of interest mutagenic? If so, is there sufficient data to argue against using an
ADAF?
Have inter- and intra-species TK and TD differences been addressed through modeling?
Are there significant concerns about a missing lifestage? What impact will this have on the
database UF?
4.2.5. Variability Analysis
Variability analysis evaluates the range of values for a parameter in a population. This is
particularly useful when sensitivity analysis has identified a key parameter as having a
significant impact on model output. When an outcome is predicted to be sensitive to certain
4-45
-------
parameters, probabilistic approaches (e.g., Monte Carlo simulation) can be incorporated into
models (U.S. EPA, 2006b). This type of analysis, for instance, allows assessors to predict upper
and lower bounds on a dose metric level in an experimental species; thus multiple calculations of
the relevant exposure concentration for humans could be calculated and perhaps used for
subsequent risk derivation.
Model evaluation may not be the final step in the dose-response process. Sensitive
parameters provide red flags that are examined carefully for variability of these parameters
within the population. Alternatively, the sensitivity might suggest the need for careful
examination and consideration of susceptible lifestages. Example questions regarding dose-
response variability are in Table 4-6.
4.2.6. Sensitivity Analysis
Sensitivity analysis allows risk assessors to examine which parameters in a model are
most important to the outcome of concern. This analysis is a key evaluation technique for PBTK
models. This analysis can identify the key parameters that can be further examined for accuracy,
either through available data or estimation. In addition, selection of sensitive parameters could
help in identifying more susceptible lifestages. For instance, model sensitivity to ventilation rate
provides a starting point for addressing lifestage differences. Example questions regarding dose-
response sensitivity are in Table 4-6.
4.2.7. Uncertainty Analysis
Uncertainty analysis can have both quantitative and qualitative components. Model
uncertainty comprises that which is unknown about how well a model reflects the underlying
biology. Models are approximations of biological processes, and therefore, have inherent
shortfalls. Quantitative elements include model structure, choice of dose metric, and
extrapolation procedures. Often these elements can be altered in order to compare model results.
Results from this type of analysis, together with reasons supporting the various choices used in
each model, can be expressed as subjective probabilities that each model is correct. Qualitative
elements of uncertainty analysis include such things as choice of experimental animal species or
the applicability of experimental animal species to the human lifestage of interest (Section
4.2.4.2). These particular efforts enhance the scientific underpinnings of the dose-response
4-46
-------
characterization and are explicitly carried forward in the dose-response narrative (Section 4.2.9)
through to the risk characterization (Chapter 5). See the final Approaches for the Application of
Physiologically Based Pharmacokinetic (PBPK) Models and Supporting Data in Risk
Assessment (U.S. EPA, 2006b) for an in-depth treatment of PBTK model evaluation. Example
questions regarding dose-response uncertainty are in Table 4-6.
Table 4-6. Examples of lifestage-specific questions for dose-response
variability, sensitivity, and uncertainty analyses.
Topic
Lifestage-Specific Question(s)
Variability Analysis
(Section 4.2.5)
How has variability been incorporated into a dose-response model? Have "average"
individuals for one or more age groups been modeled, or has population variability (across
age groups) been incorporated using probabilistic approaches?
Are the differences in model outcomes among different age groups or the entire population
greater or lesser than the typical intraspecies UF of 10?
Sensitivity Analysis
(Section 4.2.6)
What lifestage-specific parameters (inputs) have been included in the dose-response model?
What parameters have the greatest influence on the dose-response model outcome?
Are the parameters to which a model is most sensitive likely to vary across lifestages? What
is the likely impact of such differences on model predictions?
Uncertainty
Analysis
(Section 4.2.7)
Can the outcomes of multiple dose-response models and/or multiple variations (e.g.,
structures or curve fits) of such models be compared? How much do these outcomes differ?
Can variability and uncertainty in a parameter be distinguished from one another? Is the
variability true variation or is it a large component uncertainty that can be reduced through
more lifestage-specific data collection or research?
4.2.8. Iteration with Hazard and Exposure Characterization
During the dose-response characterization, situations may arise where information
obtained can lead to iteration with hazard characterization (Section 4.1). For instance, it is
conceivable that evaluation of a PBTK model could lead to the conclusion that the model
inadequately predicts empirical data. While this could be due to deficiencies in the model, it
could also suggest that the dose metric previously hypothesized to be associated with a response
may not be correct and thus may require a re-evaluation of the MO A. Such a situation may arise
when the dose-response relationship between exposure and response does not become clearer
when based on an internal dose metric.
Analysis of dose-response data could also warrant re-examination of the exposure
characterization (Section 4.3). For example, data that indicate a sensitive dose-response
relationship at environmentally relevant low-exposure levels, particularly in the context of
precursor events, may suggest that certain exposure scenarios are more important than initially
4-47
-------
thought and perhaps be an impetus for further characterization and refinement of exposure
models employed for predicting external doses.
4.2.9. Lifestage-Specific Dose-Response Characterization Narrative
The dose-response narrative summarizes recommended estimates, data supporting those
estimates, modeling approaches, a POD narrative, key default assumptions, uncertainty,
sensitivity, and variability. The narrative also provides identification of susceptible lifestages
and quantification of their susceptibility. A discussion of the strengths and limitations of the
dose-response characterization are presented, highlighting significant issues in developing risk
values, including alternative approaches considered equally plausible, and how these issues were
resolved. Dose-response estimates may be accompanied by the descriptors used in the WOE
evaluation (Section 4.1.3.1). For instance, a toxicant may be described as "likely to be
carcinogenic to humans" when exposed by "oral route" (U.S. EPA, 2005a). In this regard, risk
managers will be able to put each estimate into context. Questions to ask during the dose-
response characterization narrative include the following:
• What were the results of variability, sensitivity, and uncertainty analyses?
Are there data needs that should be highlighted to direct future research (by
various scientific bodies and processes)?
• Are there lessons/implications for past, current, or future assessments?
4.3. LIFESTAGE-SPECIFIC EXPOSURE CHARACTERIZATION
4.3.1. Introduction
Exposure characterization is the analysis step in which human interaction with the
environmental agent of concern is evaluated. Exposure (sometimes referred to as potential dose)
is the pattern of contact of an individual with a toxic agent. To characterize exposure, an
assessor needs information on the concentrations of a pollutant in exposure media, the activities
that result in contact, and the transfer rates from the exposure media to the individual. Exposure
results in an internal dose when the agent is transferred into and taken up by the body. Clearly,
not all exposures will result in a significant dose (e.g., contaminated hands may be washed
before dermal absorption or oral transfer can occur). Yet, it is the dose at the target tissue that
will ultimately cause health effects. The primary purpose of a lifestage-specific exposure
4-48
-------
characterization is to get a detailed description of the potential for exposure during preconception
or developmental lifestages.
Exposure characterization (Figure 4-8) begins in the problem formulation phase
(Chapter 3) with identification of potential sources, pathways, and scenarios. The resulting
conceptual model (Section 3.2) is used to guide collection of available exposure data and other
required information for exposure characterization. The assessor identifies and evaluates
potentially significant exposure scenarios in order to conduct a lifestage-specific exposure
characterization. Variability, sensitivity, and uncertainty analyses are conducted to determine
impact of the available exposure data on the resulting analysis. The results of the exposure
characterization are iterated with the hazard and dose-response characterizations if a critical
window of susceptibility is identified that was not considered in the initial exposure
characterization or if an important exposure period is identified that was not considered in the
hazard or dose-response characterization. Finally, the assessor writes a summary of the exposure
characterization, which includes a discussion of the confidence in the analysis results based on
available data. This information feeds into the comprehensive lifestage-specific risk
characterization (Chapter 5).
Throughout the exposure characterization, the assessor keeps in mind the relevance of the
information to the overall goals of the assessment. It may be appropriate to refine the conceptual
model (Section 3.2) or analysis plan (Section 3.3) after more thoroughly evaluating the available
exposure data. For example, a conceptual model may focus on exposure to a chemical that is
transformed in the environment before there is potential for a child to contact the agent. If the
final form of the compound relevant for exposure was not considered in the conceptual model,
then the conceptual model will need to be refined to consider all relevant agents.
4.3.2. Evaluation of Available Exposure Data
The objectives and scope of the risk assessment, defined in the problem formulation
phase (Chapter 3), provide focus for identifying all the relevant human exposure data and other
required information. To characterize exposure for a broad (e.g., national-scale) risk assessment
will require distributional exposure factor data for all relevant lifestages. A narrow (e.g., site-
specific) assessment will require measured or modeled environmental concentrations to estimate
potentially significant exposures for all relevant lifestages.
4-49
-------
Lifestagt'-SfH'cific Proh 1 em For afion
Lifestage-Specific Analysis
(Chapter 4)
Lifestage-Specific
Exposure Characterization
(Section 4.3)
Evaluation of Available Exposure Data (Section 4.3.2)
Lifestage-Specific Exposure Analysis (Section 4.3.3)
Variability, Sensitivity, and Uncertainty Analyses
(Sections 4.3.4, 4.3.5,4.3.6)
Iteration with Hazard and Dose-Response Characterization
(Section 4.3.7)
Lifestage-Specific Exposure Characterization Narrative
(Section 4.3.8)
Lifestagc-Spt'cific
Hazard
Characteri/ation
(Section 4.1)
Lifesfiige-Specific
Dose-Response
C'hariicteri/ation
(Section 4.2)
Lifestage-Specifh Risk C haracterization
Risk Communication/Management
Figure 4-8. Flow diagram for lifestage-specific exposure characterization. Using the
lifestage-specific exposure information identified in problem formulation (Chapter 3), exposure is
estimated using a tiered approach. The lifestage-specific exposure is characterized by discussing
the variability and uncertainty in the results. Key sources of variability and uncertainty can be
assessed using sensitivity analysis. Iteration with hazard characterization (Section 4.1) and dose-
response characterization (Section 4.2) (illustrated by dashed arrows) occurs throughout this
process to ensure that critical windows of exposure are considered.
4-50
-------
To focus on risk from exposure to children, the most appropriate data will be on sources
and exposure media concentrations that have been identified in the locations where children
spend time, which may change by developmental lifestage. For example, sources may be
identified in:
• residence and workplace for pregnant and lactating women;
• residence, daycare, and outdoor play areas for infants and toddlers;
• residence, school, and locations of after-school activities for school-age children; and
• residence, school, and locations of after-school activities and workplace for
adolescents.
For a given source, exposure media (e.g., water, soil/dust/sediments, food, and
objects/surfaces) and exposure routes (i.e., inhalation, ingestion, dermal absorption, and indirect
ingestion) define the pathway of exposure (U.S. EPA, 2002c, 2003a). Figure 4-9 highlights the
stages of development and their relevant exposure routes. The result of evaluating the exposure
data would be a table in which potential exposure routes are identified for each exposure medium
(direct and indirect) (Hubal et al., 2000).
Exposure media may also change with lifestage. For example, the fetus will be exposed
to cord blood and amniotic fluid, the infant to breast milk, the teething child to many objects for
mouthing, the school-age child to pesticides used in the classroom, and the adolescent to
vocational or recreational hazards.
For any given pathway, a set of associated exposure scenarios describes how an exposure
takes place and is used to estimate distribution of exposure. An exposure scenario is defined by
the combination of all the discussed details (Hubal et al., 2000). Example questions for refining
life-stage specific scenarios to facilitate exposure analysis are presented in Table 4-7.
Children may experience unique exposure patterns that are important to consider in
relation to their kinetic development and critical windows for effects. Therefore, the assessor
must carefully consider the temporal scale for estimating exposures and doses in children.
Exposure estimates may be presented as:
• peak doses;
• exposures occurring over a very short period of time (e.g., minutes);
• time weighted averages (e.g., TWA over 8 hours);
• single day doses (representing the sum over 24 hours);
4-51
-------
Prenatal
All exposures to the fetus occur
transplacentally or via physical
factors. The mother's exposure to
environmental media can be a
significant source of exposure for
environmental media for the fetus.
Infant/Young Child
Exposures for the infant and
young child can occur through all
environmental media. When
breastfed, the mother's exposure
to environmental media can be an
additional source of exposure to
the infant.
Older Child/Adolescent
Exposures to the child and
adolescent can occur through all
environmental media. The
mother's exposure is no longer a
factor for the child.
Figure 4-9. Exposure routes during developmental lifestages. The solid lines represent relevant
exposure, while dotted lines represent exposures that are not relevant to the specific lifestage. During
gestation, the majority of exposures (except for physical factors) occur transplacentally through
exposure to the mother. After birth, exposures may either be directly to the child, with an additional
route from the mother for those agents that may be present in human milk.
4-52
-------
Table 4-7. Examples of lifestage-specific questions for scenario development.
Sources of Exposure
Pathways of
Exposure
Lifestages of
Exposure
Exposure Patterns
Locations of
Exposure
Activities and
Behaviors
What are the sources of exposure to chemicals or agents that are of special concern for
children?
Where in the environment can the child come into contact with the chemical? In what
quantities? If it is a consumer product, how is it used by children?
What are all the potential exposure media (e.g., breast milk)?
What are all the potential exposure routes (hand-to-mouth ingestion)?
What are the specific pathways that may be of concern for children (e.g., absorption from
amniotic fluid, ingesting breast milk, ingestion of food eaten off contaminated floor)?
How are parents and/or children being exposed, from the source to the absorbed dose, for all
pathways of exposure?
What are the potentially exposed lifestages?
Are there any community factors that may put a subgroup of children at higher risk (e.g.,
ethnic, cultural, racial, or socioeconomic groups)?
Are there any individual characteristics that may put an individual child at higher risk (e.g.,
health status, nutritional status, genetic susceptibility)?
What is the relevant time frame of exposure (e.g., acute, short term, chronic, intermittent)?
What are the potential locations of exposure (e.g., in utero, residence, school, outdoors,
indoors)?
Are there other relevant factors that may be relevant for identifying exposure scenarios for
specific lifestages? Geographical location? Urban, rural? Near water bodies? Near parks?
Near industrial sites?
What are the potential activities (e.g., mouthing, playing soccer, mowing lawns) at the
lifestages of concern that may lead to exposure?
What developmental stage-specific behaviors may lead to contact with the chemicals? How
do the behaviors vary among children of various ages?
• short-term average daily doses (e.g., averaged over a month or a year); or
• lifetime average daily doses.
A potential problem with the time integration of exposure estimates is that the pattern of
exposure can be obscured. If the exposure pattern is relatively continuous and at a constant
level, the time averaged doses will be close in magnitude to single-day dose estimates and will
match actual human experience. However, when infrequent exposure events of high magnitude
and short duration are averaged, they are equated with continuous, lower-level exposures that do
not match human experience.
The following subsections describe information required to characterize exposure. Data
and other information used to assess exposure include chemical properties, environmental
sources, fate and transport (Section 4.3.2.1), environmental media concentrations (Section
4.3.2.2), lifestage-specific exposure measurement data (Section 4.3.2.3), lifestage-specific
exposure factors (Section 4.3.2.4), and cumulative evaluation of environmental stressors (Section
4-53
-------
4.3.2.5). Although there have been several large human exposure studies conducted to collect
integrated data on environmental concentrations, personal exposure measurements, and time-
activity data (e.g., Human Exposure Measurements: National Human Exposure Assessment
Survey [NHEXAS] ), these studies have focused on the adult lifestage. As such, these data may
be useful for characterizing preconception exposures to the parent but less so for accurately
characterizing exposures to pregnant women and to very young children. Nevertheless, these
provide a significant data source that should be evaluated with respect to the utility for
addressing the significant life-stage exposures of a given assessment. The Human Exposure
Database System (HEDS)10 is an integrated database system that contains information related to
many of these EPA human exposure research studies. Some additional life-stage specific
resources are described in the following subsections and example questions for each section are
presented in Table 4-8.
4.3.2.1. Chemical Properties, Environmental Sources, Fate, and Transport
An agent of concern may be released into the ambient environment from multiple sources
(e.g., industrial, agricultural, mobile, household, and natural sources). Also, the agent of concern
may be released directly into exposure media (e.g., via occupational activities, residential use of
consumer products, and cooking activities) of direct concern for a lifestage-specific assessment.
Once a chemical is released into the environment, it may be chemically modified or
transported, in its original or transformed state, into an exposure medium of concern for children
(e.g., outdoor air, residential water, food, and/or breast milk). Scientists and engineers can
predict the environmental movement of a chemical using information on chemical properties
(e.g., volatilization rate, water solubility, soil/water partitioning coefficients, chemical state, and
bioavailability) and environmental conditions (e.g., soil characteristics, amount of rainfall, wind
direction, and presence of water bodies). Information on the form, fate and transport of the
agents in the residential environment is also required for exposure characterization and can be
predicted based on properties of the chemicals and residential environment (e.g., size of rooms,
surface types, air exchange).
9 The NHEXAS database is available online at http://www.epa.gov/heasd/edrb/nhexas.htm.
10 The HEDS database is available online at http://www.epa.gov/heds.
4-54
-------
Information on these types of releases and associated fate and transport may be generally
required for risk assessment and is not lifestage specific. However, information on chemical
properties is required to identify potential lifestage concerns as well as particular scenarios and
pathways that may be of particular concern for children. For example, if a chemical is lipid
soluble, an infant's ingestion of breast milk may be an important route of exposure to consider.
As another example, if a chemical is highly volatile, the inhalation pathway will be of particular
concern because on a body-mass basis, young children have higher ventilation rates than adults.
4.3.2.2. Environmental Media Concentrations
Exposure characterization requires information on contaminant concentrations in the
exposure media in the environment where the individual spends time. Contaminant
concentrations can be measured directly in the exposure medium of interest or predicted by using
information on the release of the contaminant and subsequent fate and transport in the
environment. Site-specific assessments will require measured and/or model information on
concentrations of an agent in the relevant media (e.g., soil, water, indoor air). For broad (e.g.,
population-based) assessments, information may be available in the literature.
The largest exposure study conducted to collect exposure media concentration data for
children is A Pilot Study of Children's Total Exposure to Persistent Pesticides and Other
Persistent Organic Pollutants (CTEPP) (Morgan et al., 2006). In this study, concentrations of a
wide range of environmental contaminants were measured in multiple media in the homes and
daycares of children from ages 3 to 5 years. Exposure media concentration data was also
collected for children in the Minnesota NHEXAS study (Adgate et al., 2004). Recently there has
also been considerable research conducted to develop residential models for several
environmental contaminants including pesticides (Stout and Mason, 2003) and pthalates (Xu and
Little, 2006). These models use data collected in controlled laboratory settings and test house
situations and may provide insight into potential pathways for lifestage-specific exposures.
4.3.2.3. Lifestage-Specific Exposure Measurement Data
Additional data may also be available that provide a more direct measure of exposure.
Personal monitoring techniques, such as the collection of personal air or duplicate diet samples,
are used to directly measure exposure to an individual during particular time intervals. In
4-55
-------
children especially, different factors might affect the child's dose. It is important to give
consideration to measurement techniques at the physical locations where the child spends his/her
time (e.g., home, school, daycare) as well as the child's characteristics and behaviors. For
example, the breathing zone of a child is closer to the floor than the breathing zone of an adult,
and concentrations of chemicals that are heavier than air may be higher in areas closer to the
ground. Some of these types of data are available in the CTEPP study (Morgan et al., 2006) and
in other smaller studies that have been published in the literature (Adgate et al., 2004; Cohen
Hubal et al., 2006; Liu et al., 2003; Macintosh et al., 2001).
For some environmental contaminants, biomarkers can serve as a useful measure of
direct exposure aggregated over all sources and pathways. However, few studies using
biomarkers have collected all the information required to accurately estimate exposure. The
most significant source of biomonitoring information is the Third National Report on Human
Exposure to Environmental Chemicals (CDC, 2005), collected as part of the National Health and
Nutrition Examination Survey (NHANES) study." This study measured a wide range of
chemicals in the blood and urine of a representative sample of the U.S. population. However,
young children (under 6 years) are only monitored for a select group of chemicals in this study
(lead, mercury, pthalates, and organophosphates). Other lifestage-specific biomonitoring data
have been collected in studies conducted by the National Institute of Environmental Health
Sciences (NIEHS)/EPA Centers for Children's Environmental Health and Disease Prevention
Research (Kimmel et al., 2005). Several of these centers have collected data from a variety of
biological media from both pregnant mothers and their infants. Additional information on
collection and interpretation of biomonitoring data for lifestage-specific exposure
characterization is presented by Barr et al. (2005).
It is important to note that biomonitoring data may demonstrate exposure, although it
may be difficult to translate into estimates of exposure. Biomonitoring may be useful for
quantifying exposures at the population level, if the relationship between the substance found in
the body and the amount of substance the child was in contact with can be established.
Currently, there are significant research efforts associated with interpreting biomonitoring data
for assessing human exposure to environmental agents (Albertini et al., 2006; NRC, 2006).
Details on the NHANES study are available online at http://www.cdc.gov/nchs/nhanes.htm.
4-56
-------
4.3.2.4. Lifestage-Specific Exposure Factors
In addition to information on sources, exposure media concentrations, and human
exposure measurements, exposure factor data (time-activity data; product use; and air, fluid, and
dietary intake rates) are required to characterize exposure. Information is required on activities
and behaviors that result in significant exposures (e.g., breast feeding, mouthing, sports, after-
school employment) for each lifestage. The most current version of Child-Specific Exposure
Factors Handbook (U.S. EPA, 2002c) could be the starting point for identifying these values.
The purposes of the Child-Specific Exposure Factors Handbook are to summarize key data on
human behaviors and characteristics that affect children's exposure to environmental
contaminants and to recommend values to use for these factors. Data contained in the handbook
includes drinking water consumption; soil ingestion; inhalation rates; dermal factors including
skin surface area and soil adherence factors; consumption of produce, fish, meats, dairy products,
homegrown foods, and breast milk; activity patterns; body weight; and consumer products. Age-
specific activity data are also available from the Consolidated Human Activity Database
(CHAD).12
Within each lifestage there may be a series of critical developmental periods for which
exposure could be characterized. These periods may be defined on the basis of exposures that
can affect development (e.g., parental preconception exposures, U.S. EPA, 1991, 1996), or
windows of potentially high exposure due to age-specific behaviors (e.g., crawling, teething),
activities (e.g., types of sport/other activities, length of sport seasons, physical education
requirements), and physiology. Behavior varies by developmental stage, and this may have a
significant impact on exposure.
EPA has recommended a standard set of age groups (Table 3-1) for exposure assessors to
consider when assessing childhood exposure and potential dose to environmental contaminants
and for purposes of designing exposure monitoring studies (U.S. EPA, 2005e). These age groups
reflect a consideration of developmental changes in various behavioral, anatomical, and
physiologic characteristics that impact exposure and potential dose. Data from the Child-
Specific Exposure Factors Handbook emphasize the value of independently assessing the
relevant age group where sufficient data are available. In the case of vegetable intake, data
1 The CHAD database is available online at http://www.epa.gov/chadnetl/.
4-57
-------
indicate that biases are introduced when combining age groups, especially for the <1 -year-olds
because children 6 to < 12 months eat three times as many vegetables than children 3 to <6
months old.
Guidance on Selecting Age Groups for Monitoring and Assessing Childhood Exposures
to Environmental Contaminants (U.S EPA, 2005e) also recognizes that exposure factors data
may not be available for many of the recommended age groupings or that a specific age group
may not need to be the subject of a particular assessment; therefore, flexibility and professional
judgment is essential in applying these generic age groupings. There may be instances where
combining some of these age groups (e.g., combining the first three groups into one representing
birth to < 6 months) could be considered when estimating exposure or potential dose, especially
if little variation is expected. For example, there is little variation in ventilation rates for children
between 11 and 18 years. Therefore, these age categories can be combined into one age group
representing 11 to <18 years. In addition, there may be instances where it is not necessary to
address every age group in Table 3-1 because the focus of a risk assessment may be on toxicity
data that indicate a health effect for which only one or two of the age groups represent a critical
window.
Exposure factors and resulting effects during developmental stages may be a function of
additional individual and population characteristics. These factors may be characteristics of the
communities in which children live and include, for example, SES, family size, ethnicity,
cultural setting, geographical location, and seasonal considerations (e.g., temperature, humidity,
rainfall, sun exposure). Other factors specific to the individual child include genetic
susceptibility, nutritional status, and health status. Mechanisms of vulnerabilities associated with
individual and community characteristics include differences in susceptibility, differential
exposure, differential preparedness, and differential ability to recover. These mechanisms are
defined and discussed in the Framework for Cumulative Risk Assessment (U.S. EPA, 2003a, pp.
39^12). Discussion on other risk factors, effect modifiers, and confounders is detailed in
Guidelines for Developmental Toxicity Risk Assessment (U.S. EPA, 1991, Section 3.1.2.1.1.C, pp.
24-25) and Guidelines for Reproductive Toxicity Risk Assessment (U.S. EPA, 1996, Section
3.3.1.5.3, pp. 60-61).
4-58
-------
4.3.2.5. Cumulative Evaluation of Environmental Stressors
The focus of this section is the examination of vulnerability associated with differential
exposure due to lifestage. It is difficult to separate consideration of vulnerability due to lifestage
from consideration of vulnerability due to other key individual (e.g., ethnicity, dietary
preferences) and community characteristics (e.g., social and physical home environment,
religious/cultural practices) that may influence or modify exposures. In order to fully
characterize risk to children, consideration could include environmental heath disparities (e.g.,
residential segregation) (Gee and Payne-Sturges, 2004) and the built environment (e.g., design
and integrity of housing, land use and planning) (Cummins and Jackson, 2001).
EPA is examining the full range of issues related to characterizing risks to children
through a variety of initiatives, including development of Framework for Cumulative Risk
Assessment (U.S. EPA, 2003a). As EPA develops further guidance for cumulative risk
assessment, the full range of vulnerabilities will be considered more consistently in both hazard
characterization (Section 4.1) and exposure characterization. A child-centered approach (Section
3.2.3) to cumulative risk assessment may be useful in moving these issues forward (WHO, 2006,
Chapter 5).
Table 4-8. Examples of lifestage-specific questions for evaluation of the
available exposure data.
Topic
Chemical
Properties,
Environmental
Sources, Fate, and
Transport
(Section 4. 3.2.1)
Environmental
Media
Concentrations
Lifestage-Specific Question(s)
What are the physical and chemical properties of the chemicals or agents? What is known
about their fate and transport?
What are the environmental conditions (e.g., wind direction, rainfall) that may affect the
fate and transport of the chemical(s)? In the case of a release of an air pollutant, are there
areas highly populated by children that are downwind from the release (e.g., schools, play
grounds)?
What are potential chemical sources (industrial, agricultural, occupational, residential,
consumer product) of the compound?
What are the release rates from the chemical source? What is known about the
manufacturing processes that may lead to information about where the chemical can be
found (e.g., children's toys, play ground equipment, certain foods)?
Are there data on the temporal and spatial patterns of compound release and transport
relevant for specific lifestages? Is the release from the source continuous, periodic, or
intermittent?
What does the fate of the compound imply for exposure? Is the exposure to the released
compound a byproduct created in the manufacturing process or a degradation product? If
not, what are the compounds that should be assessed?
What are the concentrations of the chemicals in various media (e.g., air, water, food, breast
milk, on surfaces, in consumer products) that the child may come into contact with during
an exposure?
4-59
-------
(Section 4.3.2.2)
Are the ranges and distributions of environmental media concentration data relevant for
children's exposure?
What bioavailability data are there for the chemical(s) from the various exposure media?
How are the concentrations in environmental media changing over time? Are these
intermittent?
If environmental monitoring data are not available, are there models that can be used to
predict the concentration at the exposure point?
Lifestage-Specific
Exposure
Measurement
Data
(Section 4.3.2.3)
Are relevant exposure measurements available for various lifestages (parents, infants, and
children)? Are these direct or indirect measurements of exposure (e.g., personal air,
handwipes, duplicate diet, biomarkers of exposure)?
Are there biomonitoring data that demonstrate exposure potential? Is additional
information available to use the biomonitoring data to estimate a population's exposure?
Are there lifestage-specific data in biological media (e.g., maternal cord blood, placenta,
meconium)? Can these be used to estimate exposure or to indicate potentially critical
windows of exposure?
Lifestage-Specific
Exposure Factors
(Section 4.3.2.4)
What are the child-specific exposure factors (U.S. EPA, 2002c) that characterize the
exposure scenarios?
What are the ranges or distributions of exposure factors for relevant lifestages?
Are time-activity data available for all relevant lifestages?
Are dietary data available for all relevant lifestages? How do differences in diet during
specific lifestages impact exposure?
Are product-use data available for all relevant lifestages (e.g., pregnant women, children)?
Are the products used by children or in proximity of children?
Are data available for other children's exposure factors (e.g., contact rates for the
individual with exposure media, contaminant transfer efficiency from the contaminated
medium to the individual)?
Do children's physiological parameters influence exposure to the specific agent (e.g., body
weight, uptake rates - inhalation, dermal absorption, gastrointestinal absorption)? If so, are
there data available (Hattis, 2004)?
Cumulative
Evaluation of
Environmental
Stressors
(Section 4.3.2.5)
Are there data indicating potentially important co-exposures with chemicals that may
interact to increase health risk for a sensitive lifestage?
Are there data on relevant non-chemical stressors that may impact exposure and/or increase
vulnerability of specific lifestages (e.g. SES, health status)?
Are there any community factors that may put a subgroup of children at higher risk (e.g.,
ethnic, cultural, racial, or socioeconomic groups)?
Are there any individual characteristics that may put an individual child at higher risk (e.g.,
health status, nutritional status, genetic susceptibility)?
4.3.3. Lifestage-Specific Exposure Analysis
Based on the data and information identified for exposure characterization (Section
4.3.2), the scenarios developed during problem formulation (Chapter 3) could be refined to
facilitate exposure analysis. Exposure estimates may be developed for all relevant lifestage-
specific scenarios. At this point in the assessment, patterns of exposure will be characterized
(intermittent, continuous, acute, or chronic) and exposure levels will be quantified. Because
children may have higher exposures (Section 4.3.2.4) or because they may experience unique
4-60
-------
exposure patterns (Section 4.3.2.5), exposures may be significant during critical windows, which
can then affect the outcomes observed.
The health effect of concern is considered when selecting the appropriate temporal scale
for estimating exposure/dose. Depending upon the problem, it may be important to consider
peak exposures as well as exposures that have been averaged over a specified period of time
(U.S. EPA, 2005e). Assessments of agents with multiple sources or in multiple media may
require additional analysis to estimate children's exposure patterns. This would indicate that
even for a screening-level analysis (Section 4.3.3.2.1), a large number of factors may need to be
collected and tracked, along with their associated variabilities and uncertainties. Thus to
efficiently and effectively assess children's exposures, a person/population-oriented approach
(Section 3.2.1) may be needed for all but the most basic assessments.
To conduct the lifestage-specific exposure characterization, a calculation approach
described in Section 4.3.3.1 is selected on the basis of available data and the risk assessment
questions that were defined during the problem formulation phase (Chapter 3). Typically, an
exposure characterization will begin with a screening-level assessment (Section 4.3.3.2.1) and
then, if there appear to be significant exposures or an unacceptable level of uncertainty, a second,
more refined level of analysis will be conducted (Section 4.3.3.2.2). This type of tiered level
analysis is often used to facilitate efficient allocation of resources. Often, two or more
calculation approaches will be used and the results compared in the exposure characterization
narrative (Section 4.3.8). The following subsections describe each tier, and example questions
are presented in Table 4-9.
4.3.3.1. Exposure Measurement and Estimation Approach
Three approaches may be used to calculate exposures: (1) the point-of-contact approach,
(2) the scenario evaluation approach, and (3) the dose reconstruction approach. Each approach
has advantages and disadvantages over another.
The point-of-contact approach, sometimes referred to as the direct approach, involves
measurements of chemical concentrations at the point where exposure occurs (at the interface
between the person and the environment) and records of the length of contact with each
chemical. This approach does not take into account an individual's characteristics or behaviors.
4-61
-------
The scenario evaluation approach, sometimes referred to as the indirect approach, utilizes
data on chemical concentration, frequency, and duration of exposure as well as information on
the exposed lifestage. Child-specific behaviors and physiologic characteristics may be assumed
on the basis of exposure factor data (U.S. EPA, 2002c) or from exposure study databases (the
Consolidated Human Activity Database [CHAD]; the Human Exposure Database System
[HEDS]),13 or they can be obtained specifically for the assessment (e.g., by questionnaire, diary,
videotaping). Chemical concentration may be determined by sampling and analysis or by use of
fate and transport models (including simple dilution models). Models can be particularly helpful
when resources for additional sampling are limited but some analytical data are available.
Finally, the dose reconstruction approach allows exposure to be estimated from dose,
which can be reconstructed through internal indicators (e.g., biomarkers, body burden, excretion
levels) after the exposure has taken place. The use of biomarkers of exposure or effect may
provide a more detailed analysis; however, only a few examples currently exist for applying this
approach successfully. At the present time, much of biomarker data are difficult to interpret,
either because the presence of a biomarker may not be unique (e.g., many stressors result in a
change in the same biomarker) or there may not be adequate exposure pathway information to
link the biomarker to the exposure. Currently, this approach is most successful for persistent
compounds.
4.3.3.2. Analysis Level or Tiered Assessment
Typically, an exposure characterization will begin with a screening-level assessment and
then, if there appears to be significant exposures or an unacceptable level of uncertainty, a
second, more refined level of analysis will be conducted. Probabilistic techniques may be used
at either level of analysis depending on the types of scenarios being evaluated. The major
difference between the levels of assessment described below is related to the assumptions that
are used.
The first tier screening assessment (Section 4.3.3.2.1) is used to identify and prioritize
potentially important exposures. After results of the screening assessment are compared with
results of the hazard characterization (Section 4.1), a more refined assessment (Section 4.3.3.2.2)
13 The CHAD database is available online at http://www.epa.gov/chadnetl; the HEDS database is available online at
http://www.epa.gov/heds.
4-62
-------
may be required, using more realistic estimates of exposure for selected scenarios to reduce the
uncertainty. This second tier is generally more resource-intensive than the first tier and is used
to refine estimates for exposure scenarios that were identified as potentially significant in the
screening assessment. Finally, if a high level of uncertainty remains around estimates of
exposure following a refined assessment, supplemental data collection may be needed.
4.3.3.2.1. Screening assessment. The purpose of a screening tier is to identify probable
pathways and scenarios and to rule out insignificant ones. Bounding values for exposure factors
and conservative simplifying assumptions are used at this level of analysis. As a result, the
output may have a high level of uncertainty. Historically, deterministic calculations were used in
most screening-level exposure analyses. However, exposure assessments have become
increasingly complex, and probabilistic techniques may be useful when, for example, exposure
parameters have large variability or when multiple sources exist (U.S. EPA, 2001b).
Based on the bounding assumptions used in this level of analysis and comparison with
the hazard characterization (Section 4.1), a set of potentially significant exposure scenarios for
relevant age groups will be identified. In the screening-level analysis, differences in exposure
between children of different developmental stages are identified. For some specific exposure
scenarios and compounds, combining or subdividing some of the age groups may be appropriate,
for example, where variation in exposure factors and resulting exposures is insignificant (U.S.
EPA, 2005e).
Limited data may be an impediment in conducting accurate lifestage-specific
assessments, and for making decisions regarding combining or eliminating age groups. When
making an assessment and limited data is available, the assessor should use the recommended
age groups (U.S. EPA, 2005e) as a starting point. Then, based on qualitative information, the
assessor can determine if little variability is expected among some age groups, in which case the
age bins can be combined. If data are not available to make this determination, then this can be
described as an area of uncertainty and identified as an area for future research. A possible
approach to estimating exposure factors and dose when data are not available uses age-dependent
curve fitting to help fill in the data gaps. Any assumptions used in assessing exposure for a
particular age bin should be discussed in the assessment.
4-63
-------
Once screening-level estimates of exposure are developed for each scenario and each age
group, the questions in Table 4-9 could be considered. In order to identify and understand the
importance of parameters and uncertainties in these exposure estimates, a sensitivity analysis is
generally conducted on the potentially significant scenarios. For a screening assessment to have
value, the potential range of parameter values is considered when conducting the sensitivity
analysis (e.g., some parameters can vary only between 0 and 1; others can vary by three orders of
magnitude). In addition, the uncertainty associated with assumptions that are based on little or
no data would need to be evaluated before any conclusions about the level of "conservatism" can
be made. Methods for conducting a sensitivity analysis are discussed further in Section 4.3.5.
4.3.3.2.2. Refined assessment. This tier of the analysis level provides more detail for
potentially relevant scenarios and potentially vulnerable age groups. The goal of this tier is often
to estimate the distribution of exposure for the relevant lifestages. Based on results of the
sensitivity analysis conducted for the screening-level assessment, significant exposure factors
and important assumptions are revisited to develop more realistic estimates of exposure.
This more advanced analysis may include the application of sophisticated modeling tools
to develop exposure estimates for use in regulatory decisions. A variety of modeling tools have
been developed over the years to facilitate exposure assessment (Price et al., 2003, and
references therein for review of available tools). Some of the types of models available include
total source models (e.g., aggregate and cumulative models developed to meet requirements of
FQPA); multi-route models of exposure (e.g., local waste site models, tap-water exposure
models), models of exposures to specific sources or routes (e.g., dietary models, consumer
product models), indoor air models, and occupational models. Few of these models are designed
currently to specifically address lifestage exposures. As a result, data on the age bins used in the
models and outputs produced by the models may not address the specific age groups of interest
for a complete lifestage-specific assessment. This issue is discussed further in Guidance on
Selecting Age Groups for Monitoring and Assessing Childhood Exposures to Environmental
Contaminants (U.S. EPA, 2005e).
Limitations of the data, model results, and associated uncertainties remaining in the
refined tier are considered and addressed in this analysis. Available exposure data sets may not
4-64
-------
allow modelers or risk assessors to directly extract data from the underlying sources to conduct
lifestage-specific analyses. Potential approaches to address this issue include the following:
• reorganizing the exposure input data set to conform to the age groupings;
• using probabilistic sampling techniques to go beyond the categorical limits of the
underlying database to utilize all the data, and then formatting the probabilistic model
output into the desired age groupings to represent exposure doses; and
• developing a weighting scheme for the underlying data set to align it with the desired
age groupings.
The exposure data may need to be statistically weighted so that equal weight is given to all ages
within the group when estimating the group mean and variability statistics.
4.3.3.2.3. Supplemental data collection. Based on results of the refined assessment and the
associated sensitivity and uncertainty analyses, specific data needs may be identified. If the
objectives of the risk assessment indicate that any specific uncertainties in the exposure
characterization be addressed, collection of new data to address them may be needed and
additional analyses conducted.
Table 4-9. Examples of lifestage-specific questions for exposure analysis
level or tiered assessment.
Topic
Lifestage-Specific Question(s)
Screening
Assessment
(Section 4.3.3.2.1)
Do these results address the questions posed in the problem definition phase of the risk
assessment?
What are the bounding assumptions used to identify relevant sources, pathways, and scenarios?
What is the potential magnitude of exposures?
How do potentially relevant scenarios and potentially vulnerable age groups compare with
critical windows identified in the hazard characterization?
How do these lifestages compare to the critical windows identified based on the TK and TD
vulnerabilities (Section 4.1)?
How do potential exposure levels compare with hazard levels (e.g., MOE)?
Which exposure factors drive the results of the screening assessment and why?
What is the potential variability of exposure factors (e.g., orders of magnitude vs. factor of 2 or
3)'?
Is the available exposure information adequate? What criteria are used to determine
adequacy? What are the significant exposure data needs that may require additional exposure
data?
Refined
Assessment
(Section 4.3.3.2.2)
Were the exposure data adequate to sufficiently investigate and identify relevant differences
across age groups?
What is the central tendency of the distribution of the exposure when compared with the high-
end exposures?
How do potential exposure levels compare with dose-response characterization results?
4-65
-------
Which developmental stage of children (age bins) represent the highest exposures?
Is the available exposure information adequate for a more refined assessment? What
additional criteria are used to determine adequacy? What are the significant exposure data
needs that may require additional exposure data?
What are the available exposure assessment models?
Are distributions available for exposures of interest (e.g., by media, source, pathway)? If not,
do they need to be developed? Are there sufficient data for their development?
How will variability and uncertainty be addressed?
What are the time patterns of exposure?
How will exposure monitoring data, PBTK modeling, and biomonitoring data be incorporated?
What are the additional stressors and their cumulative impact?
Supplemental
Data Collection
(Section 4.3.3.2.3)
Have any critical data needs been identified?
4.3.4. Variability Analysis
Variability refers to the inherent lack of uniformity in a population that cannot be reduced
with additional data but can be presented by providing ranges or distributions of the exposure.
Differences among individuals in a population are referred to as inter-individual variability.
Differences associated with an individual over time are referred to as intra-individual variability.
Among children, inter-individual variability is due to rapid physiologic and behavioral
changes. Even within a relatively narrow age group, variability may be large. For oral and
dermal exposures, variability in exposure/dose is due to factors such as gross motor
development, fine motor development, cognitive development, and social development. For
inhalation exposures, relevant factors influencing variability in exposure/dose include, for
example, activity level and breathing behavior (e.g., the transition from mouth to nasal
breathing) (U.S. EPA, 2005e). Infants may be breast-fed or bottle-fed. Young children may
have higher contact with surfaces than do older children and they explore their environment by
mouthing objects. Physiologic characteristics affecting variability in exposure/dose include
anatomical characteristics (e.g., body weight and proportion of body fat) and specific organ and
physiologic systems. For example, infants have immature immune systems, and renal functions
are less than those predicted by surface area (U.S. EPA, 2005e).
This variability affects the determination of upper percentiles of exposure and its
associated risk. That is, given a high-quality/high-quantity set of data for each age group, there
may still be significant variability for a particular exposure factor, set of factors, or exposure
pathway. The better the data and the characterization of this variability, the better the basis for
4-66
-------
final selection of age groups for a specific assessment. Example questions are presented in
Table 4-10.
4.3.5. Sensitivity Analysis
Sensitivity analysis is defined as the assessment of the impact of changes in input values
on model outputs. Its main purpose in any exposure characterization is to determine which
variables in the model equations and what pathways or scenarios most affect the exposure
estimate. These techniques can also be used to assess key sources of variability and uncertainty
for the purpose of prioritizing additional data collection or research. This is particularly relevant
in children's assessments because they are often based on limited data. Because the variables of
particular interest are those that have an impact on lifestage-specific estimates, the sensitivity
analysis may need to focus considerable attention on the impact of exposure factors related to
children's behavior. These factors affect the exposure patterns in space and time and are also
typically the most uncertain. Example questions are presented in Table 4-10.
4.3.6. Uncertainty Analysis
Uncertainty is described as a lack of knowledge about factors affecting exposure or risk.
Uncertainty in the exposure estimates may be a result of limited data for significant exposure
factors for a particular age group. Uncertainty may also be due to assumptions made in
development of the model. For example, soil ingestion studies in the literature have focused on
children between 2 and 7 years of age, resulting in a lack of data for children less than 2 years of
age. Uncertainties are acknowledged and characterized to the extent possible.
Probabilistic assessments can be useful statistical tools for analyzing variability and
uncertainty in risk assessments, given that adequate data are available. The Monte Carlo analysis
can be used to better characterize variability and uncertainty across the population, and to
compare one lifestage (e.g., infants) to another (e.g., adults). General issues to consider when
applying these quantitative methods are described in EPA's Guiding Principles for Monte Carlo
Analysis (U.S. EPA, 1997b). The EPA sponsored workshop in 1998 discussed issues regarding
the selection of input distributions for probabilistic assessments (U.S. EPA, 1999b).
Methodologies for selecting parametric distributions to be used in probabilistic assessments are
4-67
-------
described in Options for Developing Parametric Probability Distributions for Exposure Factors
(U.S. EPA, 2000b). Example questions are presented in Table 4-10.
Table 4-10. Examples of lifestage-specific questions for exposure variability,
sensitivity, and uncertainty analyses.
Topic
Lifestage-Specific Question(s)
Variability Analysis
(Section 4.3.4.)
If different approaches were used to estimate exposure for different lifestages or within a
lifestage, what were the results? Can they be compared and, if so, how do they compare?
Which approach is more appropriate?
Does the lifestage-specific assessment capture the variability in the exposed groups? What
are the ranges or distributions of exposure?
What are the route, level, timing (i.e., lifestage), and duration of exposure used in the
experimental animal studies as compared with expected human exposures?
Are the available data from the same route of exposure as the expected human exposures? If
not, are TK data available to extrapolate across routes of exposure?
Are experimental animal data available from the same lifestages as the expected exposed
human lifestage? If not, are TK data available to extrapolate across species and lifestages?
What information was used to support duration adjustment and to calculate the human
equivalent concentration or dose?
How far does one need to extrapolate from the observed data to environmental exposures
(i.e.. MOE)? One, two or multiple orders of magnitude? What is the impact of such an
extrapolation?
Sensitivity Analysis
(Section 4.3.5)
What parameters have the greatest influence on the exposure model outputs?
What is the adequacy of the data for the parameters that are identified in the sensitivity
analysis as the most important parameters?
Uncertainty
Analysis
(Section 4.3.6.)
What are the uncertainties in the estimates, both within and across lifestages?
What are the data limitations and how do they compare across lifestages?
What data gaps exist, both within and across lifestages? How significant are these data
gaps? How sensitive are the results to these data gaps?
Is it feasible or desirable to collect more data pertaining to particular lifestages? Could the
exposure estimates be refined if more data were available?
4.3.7. Iteration with Hazard and Dose-Response Characterization
Following exposure characterization, coordination, and communication with the hazard
and dose-response assessors (Sections 4.1 and 4.2) may be useful. For example, if a screening-
level analysis revealed that the 0-1 year age bin was more highly exposed due to nursing
ingestion than was any other lifestage, an assessor may be prompted to re-evaluate hazard and
dose-response characterization to make sure that potential vulnerabilities during this age window
are well understood or if further data needs could be identified.
4-68
-------
4.3.8. Lifestage-Specific Exposure Characterization Narrative
The results of the exposure characterization are summarized in a narrative that includes a
discussion of the results, analysis, and conclusions. The narrative includes a discussion of the
key assumptions, limitations, and uncertainties associated with the exposure estimates and any
potential bias in the results. Variability analysis (Section 4.3.4), sensitivity analysis (Section
4.3.5), and uncertainty analysis (Section 4.3.6) are summarized. It is useful to also include a
description of how the exposure characterization can be improved and uncertainties be reduced
by additional research or collection of data. Through this narrative, the results of the exposure
characterization are communicated in a clear and concise manner to the risk manager. These
results include considerations of childhood variability and uncertainty within the exposure
characterization.
The focus of the exposure characterization is to identify age groups and address
vulnerability resulting from differential exposure. It is impossible to completely separate
consideration of exposure and potential dose from consideration of internal dosimetry and
response; therefore hazard characterization (Section 4.1), dose-response characterization (Section
4.2), and exposure characterization are intimately linked. For example, information on exposure
scenarios of a compound to humans ensures that hazard information is relevant to the measured
exposure. Also, understanding the dosimetry of an absorbed agent can inform the temporal
resolution needed in the exposure data and characterization. Some questions to consider when
summarizing the exposure characterization narrative include the following:
• What is the basis for the exposure characterization (i.e., monitoring, modeling, or
other analyses of exposure distributions)?
• How was the central tendency estimate developed? What factors or methods were
used in developing this estimate?
How was the high-end estimate developed? What factors or methods were used
in developing this estimate?
• How do the adverse health effects identified in the hazard characterization phase
(Section 4.1) inform the identification of exposures of greatest relevance for the
observed outcomes?
How do patterns of exposure (continuous vs. intermittent) and half-life in the
body influence the health outcome? What are the exposures during critical
windows in development?
4-69
-------
Are there particular developmental stages during which children are highly
exposed? Do health outcomes vary during different developmental periods?
How does this inform identification of the exposures of greatest biological
significance for the observed outcomes?
How does information on dosimetry indicate the level of temporal resolution
needed in exposure data and modeling? What dose metrics are being considered
for child-related assessments?
How does the fate of the agent being evaluated affect exposure in children? Are
children exposed to other agents with a similar MOA to the one being assessed?
Is sufficient MOA information available to consider a cumulative exposure
assessment?
4-70
-------
5. LIFESTAGE-SPECIFIC RISK CHARACTERIZATION
Risk characterization is the final phase of the risk assessment process (Figure 5-1). This
final phase of the risk assessment utilizes the information from the problem formulation
(Chapter 3) and analysis (Chapter 4) phases. After risk characterization is put into context
(Section 5.2), the information is utilized in risk communication and risk management.
f'f
i
Lifestage-Specific Risk Characterization
(Chapter 5)
Lifestage-Specific
Risk Characterization
Summary
(Section 5.1)
Risk Context
(Section 5.2)
Figure 5-1. Flow diagram for lifestage-specific risk characterization.
The risk characterization describes the overall picture of health risks resulting from
children's exposures, in which the hazard characterization (Section 4.1), dose-response
characterization (Section 4.2), and exposure characterization (Section 4.3) components of the
analysis phase are integrated and summarized. Major non-technical conclusions are drawn that
inform the risk managers, who will make risk decisions in context with the problem identified in
problem formulation (Chapter 3).
5-1
-------
During hazard characterization (Section 4.1), an assessor evaluates and describes the
information on the capacity of an environmental agent's exposure during developmental
lifestages to cause outcomes at any lifestage in both laboratory animals and humans. The
qualitative WOE evaluation is based both on the type and quality of data derived from humans
and laboratory animals and on the integration of ancillary data (SAR, genetic toxicity, TK, TD,
and MOA).
The dose-response characterization (Section 4.2) focuses on quantitative relationships
between exposure during developmental lifestages of concern and critical outcomes during
lifestages of concern identified in the hazard characterization (Section 4.1). Methods for
assessing dose-response relationships often depend on assumptions used in the absence of data.
Thus, assumptions are clearly articulated in the risk characterization section.
The exposure characterization (Section 4.3) describes the basis for values used in
exposure scenarios. Exposure estimates are based on a combination of available data and
assumptions. In exposure characterizations, the quality and representativeness of the available
data are discussed. Then, in turn, the assumptions made, the general logic to develop these
assumptions, and the effect that they may have on the results are also discussed. The major
factors considered to contribute to the greatest uncertainty in the exposure characterization are
described and linked to information from sensitivity analyses. Lack of exposure data or
limitations of specific types of data are described.
Detailed guidance on integration of these analysis steps into a risk characterization is
provided in EPA's Science Policy Handbook: Risk Characterization (U.S. EPA, 2000e). Other
sources of information that provide guidance regarding children's health risk assessment include
the EPA guidelines for developmental toxicity (U.S. EPA, 1991), reproductive toxicity (U.S.
EPA, 1996), neurotoxicity (U.S. EPA, 1998b), and cancer risk assessment (U.S. EPA, 2005a,b).
In addition, the NRC report Science and Judgment in Risk Assessment (NRC, 1994) provides
additional information about the risk characterization process.
The issues to be addressed in risk characterization are provided in example questions,
with an emphasis on lifestage-specific issues, to guide the assessor through this process. The
information to answer these questions is derived from the analysis phase (Chapter 4) and used in
the risk characterization. The questions that follow are a modification of those presented in
EPA's Science Policy Handbook: Risk Characterization (U.S. EPA, 2000e) and those developed
5-2
-------
for risk characterization within the Guidelines for Reproductive Toxicity Risk Assessment (U.S.
EPA, 1996).
5.1. LIFESTAGE-SPECIFIC RISK CHARACTERIZATION SUMMARY
A lifestage-specific risk characterization includes a concise description of the key
qualitative and quantitative aspects of the analysis. This includes a discussion of the critical
windows for duration and timing of exposure and outcome. Then, the assessor identifies and
describes the assumptions, uncertainties, and significant data gaps that could affect the major
conclusions. Finally, the summary includes a qualitative and quantitative justification for the
application of lifestage-specific adjustments for duration-specific health values (e.g., use of
lifestage-specific RfV for a specific duration of exposure) if the assessment warrants it. Three
basic questions this Framework highlights are (U.S. EPA, 2000e, p.39)
• Have the potential hazards to children been adequately characterized?
• Were the potential hazards incorporated into dose-response characterization
(Section 4.2)?
• Have the exposures to children been adequately characterized?
5.1.1. Key Information from the Analysis Phase
The assessor reviews the narratives for the three analysis steps of the risk assessment
(Chapter 4) in order to determine the key information relevant to children's risk. In the narrative,
the assessor identifies the key studies, summarizes the WOE, presents the justification for the
calculated major risk estimates, and articulates the defaults and assumptions. The assessor
considers how the key information from the analysis phase relates back to the purpose and scope
of the assessment. The following are sample questions to ask when considering the key
information from the analysis phase of the assessment:
• What lifestages were assessed? Are there any highly exposed subgroups?
• What are the most significant lifestage-specific exposure scenarios? What are the
ranges of exposures?
• What are the critical effects observed following developmental lifestage
exposures? Do they differ qualitatively and/or quantitatively from adults who are
exposed?
5-3
-------
How were the exposure scenarios and lifestage(s) accounted for in dose-response
characterization (Section 4.2)?
What are the key studies for TK, TD, and MOA? Does available MOA
information aid in the interpretation of the hazard data for different lifestages?
What are the implications of the hypothesized MOAs for potential adverse effects
and their relationship to risk?
5.1.2. Scientific Assumptions
During risk characterization, scientific assumptions and defaults used in the analysis
phase (Chapter 4) are described. An example of an assumption is using a % body weight scaling
for inhalation dosimetry in children (U.S. EPA, 2006d). It is important to transparently
document these assumptions and rationale for decisions made in the assessment.
• What are the major scientific assumptions related to children's risks and how are
they addressed?
• Was SAR information or MOA information used to bridge chemical-specific data
gaps for specific lifestages of concern?
5.1.3. Risk Drivers
The development of MOE or hazard quotients for critical effects that might occur during
specified exposures scenarios for certain lifestages may provide worst case scenarios and provide
some appreciation of relative risk for different adverse outcomes for different exposure
scenarios.
• What are the risk drivers, and what are the policy implications?
• Are specific exposure scenarios during specific lifestages major risk drivers?
• Are specific critical windows of exposures contributing to the critical outcomes
that are the major risk drivers?
5.1.4. Strengths and Weaknesses
Characterizing the strengths and weaknesses of the database is central to a lifestage-
specific risk characterization. In many cases, the information on outcomes following exposure
during developmental lifestages will be very limited but substantial enough to invoke concern or
consideration of the strengths of the database. Weaknesses in the database will influence the
lifestage-characterization of the variability (Section 5.1.4.1), sensitivity (Section 5.1.4.2), and
uncertainty (Section 5.1.4.3). Integration of the WOE evaluation (Section 4.1.3.1) with the
5-4
-------
variability, sensitivity, and uncertainty analyses for dose-response (Sections 4.2.5, 4.2.6, and
4.2.7) and exposure (Sections 4.3.4, 4.3.5, and 4.3.6) provide further characterization and
integration of the strengths and weaknesses of the overall assessment. This summary strives for
balance by describing the areas of confidence and uncertainty in the assessment.
5.1.4.1. Variability
Explicit acknowledgment of sources of variability is considered in the risk
characterization phase. By summarizing the findings from the variability analyses conducted in
the analysis phase (Sections 4.1.2.9, 4.1.3.1.2.1, 4.2.5, and 4.3.4), it may be possible to determine
whether different approaches provide similar risk estimates. Answers to the following questions
may be helpful to describe the overall variability of the assessment:
• Does the assessment capture the variability in the exposed population? How is
variability addressed?
• Who is most at risk (e.g., physiologically, genetically, highly exposed)?
• What is the relevance of experimental animal studies to humans at particular
lifestages?
• What are the limitations of the data available regarding variability? What data
gaps related to variability exist?
• Are there biological, behavioral, ethnic, racial, or socioeconomic factors that may
affect variability in human exposure or response?
5.1.4.2. Sensitivity
The findings from the sensitivity analyses conducted in the analysis phase (Sections 4.2.6
and 4.3.5) are summarized in the risk characterization phase in order to underscore the strengths
and the weaknesses related to the derivations of health values and exposure values in the
assessment. Answers to the following questions may help describe the overall sensitivity of the
assessment:
• What parameters have the greatest influence on the dose-response and exposure
model outputs?
• Are the parameters to which a model is most sensitive likely to vary across
lifestages? What is the likely impact of such differences on model predictions on
defining variability or uncertainty in the assessment?
What are the limitations of the data available regarding sensitivity? What data
gaps related to sensitivity exist?
5-5
-------
5.1.4.3. Uncertainty
Uncertainty originating from various data sources can have an impact on risk analysis.
Explicit acknowledgment of sources of uncertainty described in the analysis phase (Sections
4.1.2.10, 4.1.3.1.2.2, 4.2.7, and 4.3.6) is considered when integrating the uncertainties in the risk
characterization. This summary includes clear and concise statements about the limitations of
the data from the analysis phase for this lifestage-specific assessment and may include discussion
of uncertainties in other related assessments. Critical data gaps, defined by the impact they have
on the risk assessment, are identified and described. These critical data gaps may require
consideration and application of uncertainty factors (e.g., database UF). In addition, uncertainty
or critical data gaps may suggest further studies that may provide new information or insight to
reduce uncertainties in a future risk assessment. Answers to the following questions may prove
helpful in describing the overall uncertainty of the assessment:
• What are the uncertainties in the assessment for different lifestages of
development? How are these uncertainties addressed?
• How are the limitations of the available data related to uncertainty? What
significant data gaps exist relevant to uncertainty? How do these impact the
magnitude of uncertainty in the assessment?
• What are the priority data-needs studies that could produce information that may
reduce uncertainties in lifestage-specific risk assessment?
• What are the degrees of confidence in the dose-response and exposure model(s)
that are used to derive risk values?
5.1.5. Key Conclusions
A description of critical effects and the supporting evidence for these conclusions is
included in this section. Attendant risk numbers or a range of risk values for the critical effects
can illustrate some degree of certainty for the key conclusions. For outputs of this analysis to be
most useful in benefits analysis (Chapter 3), the outcomes that are quantified are expressed as
changes in adverse outcomes or precursor effect (e.g., change in incidence of illness or
symptoms) that are readily understood by the public. Reliance on single point risk estimates for
key conclusions may not be very useful for benefits analysis.
What are the major qualitative conclusions regarding risk from developmental
exposure? What is the degree of confidence in the conclusions?
5-6
-------
• What are the quantitative estimates of the risk from developmental exposure?
How do risks compare across lifestages? What is the degree of confidence in the
risk estimates?
• Are there any broad risk implications for classes of compounds (e.g., SAR-
related, same MOA)? Lifestages (e.g., in male fetuses, the period of sexual
differentiation in utero is sensitive to exposure to anti-androgens)?
5.1.6. Alternative Risk Estimates Considered
Consideration of alternative hypotheses to explain lifestage-specific outcomes and the
related exposures (Section 4.1.3.1.7) is part of transparency. Principles of parsimony (economy
or simplicity of assumptions in logical formulation) should be considered in the presentation of
alternatives and related to the lifestage-specific data that exist. The following examples are
questions to consider regarding alternative risk estimates:
• What are the results of different analysis approaches (i.e., modeling, monitoring,
and probability distributions)?
• Were adults considered to be more or less sensitive than other lifestages?
• What is the relative difference in the final risk value when using adult versus
developmental lifestages of exposure? What is the relative difference in the final
risk value when using a default versus a data rich approach?
• Are alternative hypotheses considered that might explain the observed lifestage-
specific outcomes? Does an alternative hypothesis provide different risk
estimates than the primary hypothesis?
5.1.7. Research Needs
The characterization of risk in many cases reveals lifestage-specific data gaps, but not all
of these data gaps may translate into critical research needs. Research needs may be based upon
qualitative or quantitative considerations in the database and the prioritization of research needs
helps determine whether specific new data could potentially reduce uncertainty in the
assessment. Questions to consider when assessing research needs for characterizing variability
and uncertainty in risk estimates include the following:
• What are the priority lifestage-specific research needs? Are these chemical-
specific, chemical class-specific, or basic research needs?
Can priorities be assigned if more than one lifestage-specific research need is
identified?
5-7
-------
• Can the impact of the research be estimated (e.g., reduction of uncertainty in the
assessment)?
• What are the key sources of variability, sensitivity, and uncertainty for the
purpose of prioritizing additional data collection or research?
5.2. RISK CONTEXT
The risk characterization is anticipated to provide an answer to the problem formulation
(Chapter 3), which may have included an initial screening of risk for prioritization and a
preliminary estimation of risk). If the statement of the problem evolved during the analysis
phase (Chapter 4), then this process is summarized in the risk characterization phase.
The risk estimates in this lifestage-specific assessment are described in the context of
other similar or related risk assessments. The science policy assumptions employed in this
assessment are clearly articulated in order to compare with previous decisions. Discussion of
alternative hypotheses, alternative MOAs, and alternative risk estimates can be included to
provide context to other previous risk decisions. The risk context could include discussion of
cumulative and multiple exposures and their potential impact on a common MOA(s).
The risk context can also provide background for developing risk communication
materials, which could include risk perception in light of related or prior risk decisions.
Questions regarding risk context include the following:
• Where appropriate, can this risk be compared with other risks characterized by EPA
or by other federal or state agencies? Have these other previous assessments
reached similar or significantly different conclusions? What are the limitations of
making these comparisons?
• What science policy (default) assumptions were employed in each of the three steps
of the analysis phase?
• What were the scientific assumptions in each of the three steps of the analysis phase
that may have policy implications?
• What alternative hypotheses were evaluated? What is the justification for the
decision to choose one hypothesis over another?
• Is there reason to be concerned about cumulative or multiple exposures to classes of
agents with a similar mechanism or MOA?
• Are there significant community concerns or common risks with which people may
be familiar that may influence public perception of risk to children?
• Is the risk characterization information presented in a way that could be used for
benefit analysis?
5-*
-------
6. SUMMARY AND IDENTIFICATION OF GAPS IN APPROACHES FOR
CHILDREN'S HEALTH RISK ASSESSMENT
This Framework summarizes the process for assessing health risks resulting from
children's exposure to environmental agents using a phased approach that includes problem
formulation (Chapter 3), analysis (Chapter 4), and risk characterization (Chapter 5). It uses
many EPA documents that have outlined similar risk assessment approaches (U.S. EPA, 1998a,
2003a) and a workshop report that identified the need for and began the development of an
approach to assessing children's risk from environmental exposures (ILSI, 2003).
This Framework is a conceptual overview of the considerations for evaluation of early
life exposures and subsequent outcomes and does not constitute EPA guidance defined as a step-
by-step process or standard operating procedure. This overview is accomplished by posing
targeted questions to address each phase of the process and by referencing appropriate
guidelines, guidance documents, and other relevant reports and literature. These references,
including several EPA risk assessment guidelines related to health risks from children's
exposures, can be drawn upon for more detailed information. One of the most relevant
references is the Guidelines for Developmental Toxicity Risk Assessment (U.S. EPA, 1991) that
focuses primarily on the effects of prenatal exposures and, to a limited extent, on postnatal
exposures and outcomes. Other EPA guidelines or guidance are focused on system- or disease-
specific issues that include the effects of developmental exposures, specifically reproductive
toxicity (U.S. EPA, 1996), neurotoxicity (U.S. EPA, 1998b), and cancer (U.S. EPA, 2005a,b).
Guidelines or guidance on the effects of developmental exposures on other systems (e.g.,
respiratory, immune, renal, hepatic, cardiovascular, and, to some extent, endocrine) or outcomes
(e.g., biomarkers of exposure or effect, toxicogenomics data) are lacking.
The relevance of specific developmental exposures on latent outcomes for application to
risk assessments for various durations of exposure (i.e., acute, short term, and subchronic) is
considered in many of the risk assessments currently being generated across EPA, although this
issue has not been thoroughly explored to date. The document A Review of the Reference Dose
and Reference Concentration Processes previously identified data needs and alternative
approaches and strategies for developing testing guidelines; these have not yet been addressed
and are reiterated below (U.S. EPA, 2002a, Section 5). In addition, there is a need for focused
guidance on dose-response assessment after developmental exposures, despite the fact that a
6-1
-------
good deal of research and methods development on HMD (U.S. EPA, 2000d) and biological
modeling (Clewell et al., 2002a; Ginsberg et al, 2004b; Lau et al., 2000, 2001; Setzer et al.,
2001) has been done using developmental data in experimental animals and humans. With
regard to exposure assessment, there is limited EPA guidance on approaches specific to children
at different lifestages, with the exception of the interim document Child- Specific Exposure
Factors Handbook (U.S. EPA, 2002c) and the Guidance on Selecting Age Groups for
Monitoring and Assessing Childhood Exposures to Environmental Contaminants (U.S. EPA,
2005e). Methods for both screening level and more detailed quantitative estimates of children's
exposures are needed. Data for the recommended age groups (U.S. EPA, 2005e) are limited or
nonexistent for some exposure factor determinations. The Framework for Cumulative Risk
Assessment (U.S. EPA, 2003a) addresses generic concepts and approaches to evaluate
cumulative risk; however, there is no specific guidance on developmental lifestage risk from
cumulative exposures.
The integration of toxicity data and children's exposure estimates is an area for which no
guidance exists but is needed. This integration is important because one exposure can lead to
multiple outcomes, particularly for developmental exposures. In addition, the characteristics for
each age group of concern to environmental agents can differ significantly for exposure and
susceptibility. Therefore, guidance is also needed on using information on biological processes
underlying development, MOA information, chemical-specific mechanisms, and anatomical,
physiological, and behavioral characteristics at different developmental lifestages to determine
critical times for exposure and the corresponding outcomes of concern.
At this time, significant research questions remain unanswered on the use of available
exposure data to assess children's risk, such as:
• How can biomonitoring data be interpreted to characterize exposure? How can
available adult biomonitoring data be applied to children?
• How can available data from children be interpreted across developmental stages
for which there are limited data?
How can activity pattern data be used to classify children for exposure
characterization?
• What resources or approaches can one use to address risk methodology for
extrapolating inhalation dose to developmental lifestages?
6-2
-------
• Can guidance be developed on incorporating critical window of vulnerability to
reduce uncertainty, specifically for the time frame over which exposure should be
averaged?
• How can risks be extrapolated to developmental exposure to non-genotoxic
carcinogens?
• How can developmental lifestage-specific MO As influence latent expression of
adverse outcomes?
• Since TK and TD in children can rarely be studied, how can model variability in
internal dose and sensitivity to toxicant action be better characterized?
Many of these questions are actively being investigated. These efforts will likely
contribute to future guidance and policy papers on specific issues related to children's exposure
and subsequent outcomes.
6-3
-------
-------
GLOSSARY
Activity Pattern Data - Information on human activities used in exposure assessments. The
information may include a description of the activity, frequency of activity, duration spent
performing the activity, and the microenvironment in which the activity occurs.
Adverse Effect - A biochemical change, functional impairment, or pathologic lesion that affects
the performance of the whole organism or reduces an organism's ability to respond to an
additional environmental challenge.
Age-Dependent Adjustment Factors (ADAF) - Adjustments to cancer slope factors that
recognize the increased susceptibility to cancer from early life exposures to mutagens in the
absence of chemical-specific data.
Area Under the Curve (AUC) - The area of the time x concentration curve that helps to define
the internal dose.
Benchmark Dose (BMD) - A dose that produces a predetermined change in response rate of an
adverse effect (called the benchmark response or BMR) compared to background.
Benchmark Dose Lower Confidence Level (BMDL) - A statistical lower confidence limit on
the dose at the BMD.
Benefits Analysis - A method that develops monetary values comparing costs and benefits to
inform the policy making process (U.S. EPA, 2000a).
Bias - A trend in methodology or analysis that can lead to systematic deviations from the true
data.
Biologically Based Dose-Response (BBDR) Model - A predictive model that describes
biological processes at the cellular and molecular level linking the target organ dose to the
adverse effect.
Biomarker - A biological molecule or biochemical indicator of exposure or biological changes
resulting from exposures, or markers of risk or susceptibility.
Biomonitoring - The assessment of human exposure to chemicals by the measurement of the
chemicals or their metabolites (breakdown products) in human tissues or fluids such as blood or
urine. Blood and urine levels reflect the amount of the chemical in the environment that actually
gets into the body.
Body Burden - The amount of a particular chemical, especially a potentially toxic chemical,
stored in the body at a particular time as a result of exposure. Body burdens can be the result of
long-term or short-term storage, e.g., the amount of a metal in bone, the amount of a lipophilic
G-l
-------
substance such as PCB in adipose tissue, or the amount of carbon monoxide (as
carboxyhemoglobin) in the blood.
Bounding Estimate - An estimate of exposure, dose, or risk that is higher than that incurred by
the person in the population with the highest exposure, dose, or risk. Bounding estimates are
useful in developing statements that exposures, doses, or risks are "not greather than" the
estimated value.
Cancer - A disease of heritable, somatic mutations affecting cell growth and differentiation and
characterized by an abnormal, uncontrolled growth of cells.
Case-Control Study - An epidemiologic study that compares subjects with the disease of
interest (cases) to subjects without the disease (controls). The groups are compared with respect
to exposure history to ascertain whether they differ in the proportion exposed to the chemical(s)
under investigation.
Case Report - A description of a person in a population or study group identified as having a
particular disease, health disorder, or condition under investigation, without a comparison made
to a control.
Child - Conception to maturation of all organ systems, approximately 21 years of age.
Concentration - The ratio of the mass or volume of a solute to the mass or volume of the
solution or solvent.
Conceptual Model - A written description or a visual representation of actual or predicted
relationships between humans or ecological entities and the chemicals or other stressors to which
they may be exposed.
Confounder (or Confounding Factor) - A condition or variable that is both a risk factor for
disease and is associated with an exposure or outcome of interest. This association between the
exposure of interest and the confounder may make it falsely appear that the exposure of interest
is associated with the outcome.
Critical Effect - The first adverse effect, or its known precursor, that occurs to the most
sensitive species as the dose rate of an agent increases.
Critical Window of Exposure - Developmental period when vulnerability to exposures is
increased and can result in developmental effects.
Cumulative Impact - The combination of aggregate exposures to multiple agents or stressors.
Detoxification - Process of chemical modification that make a toxic molecule less toxic.
Dose - The amount of a substance available for interaction with metabolic processes or
biologically significant receptors after crossing the outer boundary of an organism.
G-2
-------
• Absorbed Dose is the amount crossing a specific absorption barrier (e.g., the exchange
boundaries of skin, lung, and digestive tract) through uptake processes.
• Biologically Effective Dose is the amount of the chemical available for interaction by any
particular organ or cell.
• Internal Dose is a more general term denoting the amount absorbed without respect to
specific absorption barriers or exchange boundaries.
• Potential Dose is the amount ingested, inhaled, or applied to the skin.
Dose Metric - The target tissue dose that is closely related to ensuing adverse response. Dose
metrics reflect the biologically active form of the chemical, its level, and duration of exposure,
and its intensity. Examples of units of measurement for dose are AUC, maximum concentration.
Dose-Response Assessment - The determination of the relationship between the magnitude of
administered, applied, or internal dose and a specific biological response. Response can be
expressed as measured or observed incidence, percent response in groups of subjects (or
populations), or the probability of occurrence of a response in a population.
Dose-Response Curve - A graphical representation of the quantitative relationship between
administered, applied, or internal dose of a chemical or agent, and a specific biological response
to that chemical or agent.
Dosimetric Adjustment Factor (DAF) - A multiplicative factor used to adjust observed
experimental or epidemiological data to human equivalent concentration for assumed ambient
scenario.
Dosimetry - A process of measuring or estimating dose.
Effect Modifier - A variable that modifies the outcome of interest by a greater (synergistic) or
lesser (antagonistic) effect. An effect modifier can be identified through stratification of the
data.
Environmental Fate - The destiny of a chemical or biological pollutant after release into the
environment. Environmental fate involves temporal and spatial considerations of transport,
transfer, storage, and transformation.
Epidemiology - The study of the distribution and determinants of health-related states or events
in specified populations.
Exposure - Contact made between a chemical, physical, or biological agent and the outer
boundary of an organism. Exposure is quantified as the amount of an agent available at the
exchange boundaries of the organism (e.g., skin, lungs, gut).
• Acute Exposure is exposure by the oral, dermal, or inhalation route for 24 hours or less.
• Chronic Exposure is repeated exposure by the oral, dermal, or inhalation route for more
than approximately 10% of the life span in humans (more than approximately 90 days to
2 years in typically used laboratory animal species).
G-3
-------
• Intermittent Exposure is a repeated exposure in which there is no effect of one exposure
on the effect of the next; this definition implies sufficient time for the chemical and its
metabolites to subchronic clear the biological system before the subsequent exposure
(i.e., non-cumulative toxicokinetics).
• Longer-Term Exposure is repeated exposure by the oral, dermal, or inhalation route for
more than 30 days, up to approximately 10% of the life span in humans (more than 30
days up to approximately 90 days in typically used laboratory animal species).
• Short-Term Exposure is multiple or continuous exposure to an agent for a short period of
time, usually 1 week.
Exposure Assessment - An identification and evaluation of the human population exposed to a
toxic agent that describes its composition and size and the type, magnitude, frequency, route, and
duration of exposure.
Exposure Concentration - The concentration of a chemical in its transport or carrier medium at
the point of contact.
Exposure Factor - Variables that define how exposure to a chemical or agent takes place (e.g.,
concentration, intake, body weight).
Exposure Media - Major environmental categories that surround or contact humans, animals,
plants, and other organisms (surface water, ground water, soil, or air) and through which
chemicals or pollutants move.
Exposure Pathway - The physical course a chemical or pollutant takes from its source to the
organism exposed.
Exposure Route - The way a chemical or pollutant enters an organism after contact, e.g., by
ingestion, inhalation, or dermal absorption.
Exposure Scenario - A combination of facts, assumptions, and inferences that define a discrete
situation where potential exposures may occur. These may include the source, the exposed
population, the time frame of exposure, microenvironment(s), and activities. Scenarios are often
created to aid exposure assessors in estimating exposure.
Database (Extent of) - Minimal Database is a database in which no human data are available,
and route-specific toxicity data are limited to dose-response data applicable to the duration in
question with assessment of outcomes other than mortality. A study showing only effect levels
for mortality or other extremely severe toxicity would not be sufficient to set a reference value.
Robust Database is a database that includes extensive human and/or animal toxicology data that
cover route-specific information on many health outcomes, durations of exposure, timing of
exposure, lifestages, and susceptible subpopulations (see U.S. EPA, 2000b, pages 4-19).
Hazard Assessment - The process of determining whether exposure to an agent can cause an
increase in the incidence of a particular adverse health effect (e.g., cancer, birth defect) and
whether the adverse health effect is likely to occur in humans.
G-4
-------
Hazard Characterization - A description of the potential adverse health effects attributable to a
specific environmental agent, the mechanisms by which agents exert their toxic effects, and the
associated dose, route, duration, and timing of exposure.
Human Equivalent Concentration (HEC) or Dose (HED) - The human concentration (for
inhalation exposure) or dose (for other routes of exposure) of an agent that is believed to induce
the same magnitude of toxic effect as the experimental animal species concentration or dose.
This adjustment may incorporate toxicokinetic information on the particular agent, if available,
or use a default procedure, such as assuming that daily oral doses experienced for a lifetime are
proportional to body weight raised to the 0.75 power.
Intake Rate - Rate of inhalation, ingestion, and dermal contact, depending on the route of
exposure. For ingestion, the intake rate is simply the amount of food containing the contaminant
of interest that an individual ingests during some specific time period (units of mass/time). For
inhalation, the intake rate is the rate at which contaminated air is inhaled. Factors that affect
dermal exposure are the amount of material that comes into contact with the skin and the rate at
which the contaminant is absorbed.
Key Event - A key event is an empirically observable precursor step that is itself a necessary
element of the mode of action (U.S. EPA, 2005a,b). Toxicokinetic and toxicodynamic steps that
lead to a toxic response can be considered as key event(s).
Lifestage Approach - The comparison of exposure and effect data across different lifestages
from conception to old age. This approach provides a temporal context in which to evaluate data
for risk assessment.
Longitudinal Study - An epidemiologic study comparing subject with an exposure of interest to
those without the exposure. These two cohorts are then followed over time to determine the
differences in the rates of disease between the exposure subjects.
Low-Dose Extrapolation - An estimate of the response at a point below the range of the
experimental data, generally through the use of a mathematical model.
Lowest-Observed-Adverse-Effect Level (LOAEL) - The lowest exposure level at which there
are biologically significant increases in frequency or severity of adverse effects among the
exposed population when compared with an appropriate control group.
Margin of Exposure (MOE) - The ratio of the point of departure (POD) over an exposure
estimate (MOE = POD/Exposure).
Mechanism of Action - The complete sequence of biological events (i.e., including
toxicokinetic and toxicodynamic events) from exposure to the chemical to the ultimate cellular
and molecular consequences of chemical exposure that are required in order to produce the toxic
effect. However, events that are coincident but not required to produce the toxic outcome are not
included.
G-5
-------
Media - see Exposure Media.
Meta-Analysis - Any systematic method that uses statistical analysis to integrate the data from a
number of independent studies.
Mode of Action - The sequence of key event(s) (i.e., toxicokinetics and toxicodynamics) after
chemical exposure upon which the toxic outcome depend.
Model - A mathematical function with parameters that can be adjusted so that the function
closely describes a set of empirical data. A mechanistic model usually reflects observed or
hypothesized biological or physical mechanisms and has model parameters with real world
interpretation. In contrast, statistical or empirical models selected for particular numerical
properties are fitted to data; model parameters may or may not have real world interpretation.
When data quality is otherwise equivalent, extrapolation from mechanistic models (e.g.,
biologically based dose-response models) often carries higher confidence than extrapolation
using empirical models (e.g., logistic model).
No-Observed-Adverse-Effect Level (NOAEL) - The highest exposure level at which there are
no biologically significant increases in the frequency or severity of adverse effect between the
exposed population and its appropriate control; some effects may be produced at this level, but
they are not considered adverse or precursors of adverse effects.
Outcome - A clinical manifestation of biological effects that results from an exposure.
Pathway - see Exposure Pathway.
Person-Oriented Model - An approach in which the individual's exposure-related
characteristics are defined first and then used to determine the probability of the individuals'
being exposed to a specific source and the resulting dose.
Physiologically based Toxicokinetic (PBTK) Model - A model that estimates the dose to a
target tissue or organ by taking into account the rate of absorption into the body, distribution
among target organs and tissues, metabolism, and excretion. (Also referred to as physiologically
based pharmacokinetic model.)
Point-of-Contact Approach - An approach to quantifying exposure by taking measurements of
concentration over time at or near the point of contact between the chemical and an organism
while the exposure is taking place.
Point of Departure (POD) - The dose-response point that marks the beginning of a low-dose
extrapolation. This point can be the lower bound on dose for an estimated incidence or a change
in response level from a dose-response model (HMD) or a NOAEL or LOAEL for an observed
incidence, or change in level of response.
Portal of Entry - The point at which the contaminant enters the body (e.g., mouth, nose, skin).
G-6
-------
Precursor Event - An early condition or state preceding the pathological onset of a disease.
Reference Concentration (RfC) - An estimate (with uncertainty spanning perhaps an order of
magnitude) of a continuous inhalation exposure to the human population (including sensitive
subgroups) that is likely to be without an appreciable risk of deleterious effects during a lifetime.
It can be derived from a NOAEL, a LOAEL, or a benchmark concentration, with uncertainty
factors generally applied to reflect limitations of the data used. It is generally used in EPA's
noncancer health assessments.
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. It can be derived from a
NOAEL, a LOAEL, or a benchmark dose, with uncertainty factors generally applied to reflect
limitations of the data used. It is generally used in U.S. EPA's noncancer health assessments.
Reference Value (RfV) - An estimation of an exposure for (a given duration) to the human
population (including susceptible subgroups) that is likely to be without an appreciable risk of
adverse effects over a lifetime. It is derived from a BMDL, a NOAEL, a LOAEL, or another
suitable POD, with uncertainty/variability factors applied to reflect limitations of the data used.
Risk (in the context of human health) - The probability of adverse effects resulting from
exposure to an environmental agent or mixture of agents.
Risk Assessment (in the context of human health) - The evaluation of scientific information
on the hazardous properties of environmental agents (hazard characterization), the dose-response
relationship (dose-response assessment), and the extent of human exposure to those agents
(exposure assessment). The product of the risk assessment is a statement regarding the
probability that populations or individuals so exposed will be harmed and to what degree (risk
characterization).
Risk Characterization - The integration of information on hazard, exposure, and dose-response
to provide an estimate of the likelihood that any of the identified adverse effects will occur in
exposed people.
Risk Management (in the context of human health) - A decision-making process that
accounts for political, social, economic, and engineering implications together with risk-related
information in order to develop, analyze, and compare management options and select the
appropriate managerial response to a potential chronic health hazard.
Route - see Exposure Route.
Structure-Activity Relationship (SAR) approach to toxicology screening - This approach
elucidates the relationship between features of chemical structure and biological activity. It is
based on the premise that the biological fate and activity of a chemical (i.e., whether it is
absorbed, metabolized, or bioaccumulated and whether it interacts at a molecular level to exert a
response) is ultimately determined by chemical structure.
G-7
-------
Scenario Evaluation Approach - An approach to quantifying exposure by measurement or
estimation of both the amount of a substance contacted and the frequency/duration of contact and
subsequently linking these together to estimate exposure or dose.
Sensitivity Analysis - Refers to the variation in output of a model with respect to changes in the
values of the model input(s). Sensitivity analysis can provide a quantitative ranking of the model
inputs based on their relative contributions to model output variability and uncertainty (U.S. EPA
2001b).
Short-Term Exposure - Repeated exposure by the oral, dermal, or inhalation route for more
than 24 hours, up to 30 days.
Slope Factor - An upper bound, approximating a 95% confidence limit, on the increased cancer
risk from a lifetime exposure to an agent. This estimate, usually expressed in units of proportion
(of a population) affected per mg/kg/day, is generally reserved for use in the low-dose region of
the dose-response relationship, i.e., for exposures corresponding to risks less than 1 in 100.
Source - The origin of an agent for the purposes of an exposure assessment.
Source-to-Dose Model - An approach where an environmental agent is followed from its source
to the resulting dose.
Stakeholder - An interested party who is concerned with the decisions made about how a risk
may be mitigated, avoided, reduced, or eliminated, and the communities that may be impacted by
regulatory decisions.
Stressor - Any entity, stimulus, or condition that can modulate normal functions of the organism
or induce an adverse response (e.g., agent, lack of food, drought).
Superfund — Federal authority, established by the Comprehensive Environmental Response,
Compensation, and Liability Act (CERCLA) (U.S. 96th Congress, 1980) to respond directly to
releases or threatened releases of hazardous substances that may endanger health or welfare.
Susceptibility - Increased likelihood of an adverse effect or an exposure, often discussed in
terms of relationship to a factor, that can be used to describe a human subpopulation (e.g.,
lifestage, demographic feature, or genetic characteristic).
Susceptible Subgroups - May refer to lifestages (e.g., children or the elderly), or to other
segments of the population (e.g., asthmatics, the immune-compromised, or the highly exposed).
The term is likely to be chemical-specific, and may not be consistently defined in all cases.
Target Organ - The biological organ most adversely affected by exposure to a chemical,
physical, or biological agent.
Toxicity - Deleterious or adverse biological effects elicited by a chemical, physical, or
biological agent.
G-8
-------
Toxicodynamic (TD) - The determination and quantification of the sequence of events at the
cellular and molecular levels leading to a toxic response to an environmental agent (sometimes
referred to as pharmacodynamics, also MOA.
Toxicokinetic (TK) - The determination and quantification of the time course of absorption,
distribution, metabolism, and excretion of chemicals (sometimes referred to as
pharmacokinetics).
Toxification - Metabolic conversion of a potentially toxic substance to a product that is more
toxic.
Uncertainty - Uncertainty occurs because of a lack of knowledge. It is not the same as
variability. For example, a risk assessor may be very certain that different people drink different
amounts of water but may be uncertain about how much variability there is in water intakes
within the population. Uncertainty can often be reduced by collecting more and better data,
whereas variability is an inherent property of the population being evaluated. Variability can be
better characterized with more data but it cannot be reduced or eliminated. Efforts to clearly
distinguish between variability and uncertainty are important for both risk assessment and risk
characterization.
Uncertainty Factor (UF) - One of several, generally 10-fold, default factors used in
operationally deriving the RfD and RfC from experimental data. The factors are intended to
account for
• variation in susceptibility among the members of the human population (i.e.,
interindividual or intraspecies variability)
• uncertainty in extrapolating experimental animal data to humans (i.e., interspecies
uncertainty);
• uncertainty in extrapolating from data obtained in a study with less-than-lifetime
exposure (i.e., extrapolating from subchronic to chronic exposure)
• uncertainty in extrapolating from a LOAEL rather than from a NOAEL and
• uncertainty associated with extrapolation when the database is incomplete.
Variability — Variability refers to true heterogeneity or diversity. For example, among a
population that drinks water from the same source and with the same contaminant concentration,
the risks from consuming the water may vary. This may be due to differences in exposure (i.e.,
different people drinking different amounts of water and having different body weights, different
exposure frequencies, and different exposure durations) as well as differences in response (e.g.,
genetic differences in resistance to a chemical dose). Those inherent differences are referred to
as variability. Differences among individuals in a population are referred to as interindividual
variability; differences for one individual over time is referred to as intraindividual variability.
Vulnerability - A matrix of physical, chemical, biological, social, and cultural factors which
result in certain communities and subpopulations being more susceptible to environmental
toxins, being more exposed to toxins, or having compromised ability to cope with and/or
recover from such exposure. Four types of vulnerability are considered with regard to a
G-9
-------
lifestage approach: susceptibility or sensitivity, differential exposure, differential
preparedness, and differential ability to recover (NEJAC, 2004).
Weight-of-Evidence (WOE) - An approach requiring a critical evaluation of the entire body of
available data for consistency and biological plausibility. Potentially relevant studies are judged
for quality and studies of high quality given much more weight than those of lower quality (see
U.S. EPA, 2000b, pages 4-11-12).
G-10
-------
REFERENCES14
Adams, J; Barone, S, Jr; LaMantia, A; et al. (2000) Workshop to identify critical windows of exposure for children's
health: neurobehavioral workgroup summary. Environ Health Perspect 108(Suppl 3):535-544. Available online at
http://ehp.niehs.nih.gov/members/2000/suppl-3/535-544adams/adams-full.html.
Adgate, JL; Church, TR; Ryan, AD; et al. (2004) Outdoor, indoor, and personal exposure to VOCs in children.
Environ Health Perspect 112:1386-1392. Available online at
http://www.ehponline.org/members/2004/7107/7107.html.
Albertini, R; Bird, M; Doerrer, N; et al. (2006) The use of biomonitoring data in exposure and human health risk
assessments. Environ Health Perspect [e-pub before print]. Available online at
http://www.ehponline.org/docs/2006/9056/abstract.html.
Allen, BC; Kavlock, RJ; Kimmel, CA; et al. (1994a) Dose-response assessment for developmental toxicity. II.
Comparison of generic benchmark dose estimates with no observed adverse effect levels. Fundam Appl Toxicol
23(4):487-495. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=786790
0&query_hl=2&itool=pubmed_docsum.
Allen, BC; Kavlock, RJ; Kimmel, CA; et al. (1994b) Dose-response assessment for developmental toxicity. III.
Statistical models. Fundam Appl Toxicol 23(4):496-509. Available online at
http://www.ncbi. nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=786790
l&query_hl=2&itool=pubmed_docsum.
Allen, B; Gentry, R; Shipp, A; et al. (1998) Calculation of benchmark doses for reproductive and developmental
toxicity observed after exposure to isopropanol. Reg Toxicol Pharmacol 28(l):38-44. Available online at:
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Retrieve&list_uids=9784431&dopt=Citation.
Andersen, ME; Dennison, JE. (2001) Mode of action and tissue dosimetry in current and future risk assessments. Sci
Total Environ 274(1-3):3-14. Available online at
http://www.ncbi. nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Retrieve&listjaids=l 1453304&dopt=Citation.
Andersen, ME; Meek, ME; Boorman, GA; et al. (2000) Lessons learned in applying the U.S. EPA proposed cancer
guidelines to specific compounds. Toxicol Sci 53:159-172. Available online at
http://toxsci.oxfordjournals.Org/cgi/reprint/53/2/159.
Anderson, LM; Diwan, BA; Fear, NT; et al. (2000) Critical windows of exposure for children's health: cancer in
human epidemiological studies and neoplasms in experimental animal models. Environ Health Perspect 108(Suppl
3):573-594. Available online at http://ehp.niehs.nih.gov/members/2000/suppl-3/573-594anderson/anderson-
full.html.
Anway, MD; Leathers, C; Skinner, MK. (2006) Endocrine disruptor vinclozolin induced epigenetic
transgenerational adult onset disease. Endocrinology [Epub ahead of print]. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=l 69737
26&query_hl=2&itool=pubmed docsum.
Ashani, Y; Pistinner, S. (2004) Estimation of the upper limit of human butyrylcholinesterase dose required for
protection against organophosphates toxicity: a mathematically based toxicokinetic model. Toxicol Sci
77(2):358-367. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=14600276&
query_hl=13.
14 All references are hyperlinked to the online document when available. Due to copyright regulations, there may be
links to sites where the document is available for purchase. Note that links may change over time.
R-l
-------
Barlow, NJ; Foster, PM. (2003) Pathogenesis of male reproductive tract lesions from gestation through adulthood
following in utero exposure to di(n-butyl) phthalate. Toxicol Pathol 31(4):397-410. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=128511
05&query hl=4&itool=pubmed_docsum.
Barnes, DG; Daston, GP; Evans, JS; et al. (1995) Benchmark dose workshop: criteria for use of benchmark dose to
estimate a reference dose. Reg Toxicol Pharmacol 21(2):296-306. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list uids=7644719&q
ueiyjil=25.
Barr, M, Jr; DeSesso, JM; Lau, CS; et al. (2000) Workshop to identify critical windows of exposure for children's
health: cardiovascular and endocrine work group summary. Environ Health Perspect 108(Suppl 3):569-571.
Available online at http://ehp.niehs.nih.gov/members/2000/suppl-3/569-571barr/barr-full.html.
Barr, DB; Wang, RY; Needham, LL. (2005) Biologic monitoring of exposure to environmental chemicals
throughout the life stages: requirements and issues for consideration for the National Children's Study. Environ
Health Perspect 113(8): 1083-1091. Available online at http://www.ehponline.org/members/2005/7617/7617.html.
Barton, H. (2005) Computational pharmacokinetics during developmental windows of susceptibility. J Toxicol
Environ Health A: 68(11-12):889-900. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=16020183&
query _hl=8.
Beck, BD; Mattuck, RL; Bowers, TS; et al. (2001) The development of a stochastic physiologically based
pharmacokinetic model for lead. Sci Total Environ 274(1-3):15-19. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=11453291&
query Jil=27.
Beckman, DA; Feuston, M. (2003) Landmarks in the development of the female reproductive system. Birth Defects
Res B Dev Reprod Toxicol 68(2):137-143. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=12866705&
query_hl=23.
Besunder, JB; Reed, MD; Blumer, JL. (1988a) Principles of drug biodisposition in the neonate. A critical evaluation
of the pharmacokinetic-pharmacodynamic interface (Part I). Clin Pharmacokinet 14:189-216. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=3292100&q
uery_hl=25.
Besunder, JB; Reed, MD; Blumer, JL. (1988b) Principles of drug biodisposition in the neonate. A critical evaluation
of the pharmacokinetic-pharmacodynamic interface (Part II). Clin Pharmacokinet 14(5):261-286. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=3293867&q
uery_hl=25.
Bogdanffy, MS; Daston, G; Faustman, EM; et al. (2001) Harmonization of cancer and noncancer risk assessment:
proceedings of a consensus-building workshop. Toxicol Sci 61(1): 18-31. Available online at
http://toxsci.oxfordjournals.0rg/cgi/content/abstract/61/l/18.
Bossowski, A; Urban, M; Stasiak-Barmuta, A. (2003) Analysis of changes in the percentage of B (CD19) and T
(CD3) lymphocytes, subsets CD4, CDS and their memory (CD45RO), and naive (CD45RA) T cells in children with
immune and non-immune thyroid disease. J Pediatr Endocrinol Metab 16(1):63-70. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=125853
42&query_hl=3&itool=pubmed_docsum.
R-2
-------
Brown, KG; Strickland, JA. (2003) Utilizing data from multiple studies (meta-analysis) to determine effective dose-
duration levels. Example: rats and mice exposed to hydrogen sulfide. Reg Toxicol Pharmacol 37(2):305-317.
Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=12726759&
query_hl=5.
Bruckner, JV. (2000) Differences in sensitivity of children and adults in chemical toxicity. The NAS panel report.
Reg Toxicol Pharmacol 31(3):280-285. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=10915586&
query_hl=32.
Bruckner, JV; Weil WB. (1999) Biological factors which may influence an older child's or adolescents responses to
toxic chemicals. Reg Toxicol Pharmacol 29(2 Pt. 1): 158-164. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=Abstract&list_uids=l 0341146&
query _hl= 1 &itool=pubmed_docsum.
Campbell, CG; Seidler, FJ; Slotkin, TA. (1997) Chlorpyrifos interferes with cell development in rat brain regions,
Brain Res Bull 179-189. Available online at
http://www.sciencedirect.com/science?_ob=MImg&_imagekey=B6SYT-3S9C423-8-
G&_cdi=4843&_user= 14684&_orig=search&_coverDate= 12%2F31 %2F 1997& qd= 1 &_sk=999569997&view=c&
_alid=451503315&_rdoc= 1 &wchp=dGLbVlb-
zSkWW&md5=2b06d49bfa93589b!23d65117b20dl 7c&ie=/sdarticle.pdf.
Canfield, RL; Henderson, CR, Jr; Cory-Slechta, DA; et al. (2003) Intellectual impairment in children with blood
lead concentrations below 10 microg per deciliter. N Engl J Med 348(16):1517-1526. Available online at
http://content.nejm.org/cgi/content/abstract/348/! 6/1517.
Cappon, GD; Fleeman, TL; Chapin, RE; et al. (2005) Effects of feed restriction during organogenesis on embryo-
fetal development in rabbit. Birth Defects Res B Dev Reprod Toxicol 74(5):424-430. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=l 62499
98&query_hl= 11 &itool=pubmed_docsum.
CDC (Centers for Disease Control and Prevention). (2005) Third national report on human exposure to
environmental chemicals. Department of Health and Human Services, Atlanta, GA. NCEH Pub. No. 05-0570.
Available online at http://www.cdc.gov/exposurereport/3rd/pdf/thirdreport.pdf.
Chapin, RE; Harris, MW; Davis, BJ; et al. (1997) The effects of perinatal/juvenile methoxychlor exposure on adult
rat nervous, immune, and reproductive system function. Fundam Appl Toxicol 40(1):138-157. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=939849
6&query_hl=5&itool=pubmed_docsum.
Chakraborti, TK; Farrar, JD; Pope, CN. (1993) Comparative neurochemical and neurobehavioral effects of repeated
chlorpyrifos exposures in young and adult rats, Pharmacol Biochem Behav 46:219-224. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=750482
1 &query_hl= 17&itool=pubmed_docsum.
Clewell, HJ; Crump, KS. (2005) Quantitative estimates of risk for noncancer endpoints. Risk Analysis
25(2):285-289. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=15876204&
query_hl=17.
Clewell, RA; Gearhart, JM. (2002) Pharmacokinetics of toxic chemicals in breast milk: use of PBPK models to
predict infant exposure. Environ Health Perspect 110(6):A333-337. Available online at
http://ehp.niehs.nih.gov/members/2002/110pA333-A337clewell/clewell-full.html.
R-3
-------
Clewell, HJ, III; Gentry, PR; Gearhart, JM; et al. (2001) Development of a physiologically based pharmacokinetic
model of isopropanol and its metabolite acetone. Toxicol Sci 63(2): 160-172. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=l 1568359&
query_hl=36.
Clewell, HJ, III; Andersen, ME; Barton, HA. (2002a) A consistent approach for the application of pharmacokinetic
modeling in cancer and noncancer risk assessment. Environ Health Perspect 110(l):85-93. Available online at
http://ehp.niehs.nih.gov/members/2002/110p85-93clewell/clewell-full.html.
Clewell, HJ; Teeguarden, J; McDonald, T; et al. (2002b) Review and evaluation of the potential impact of age- and
gender-specific pharmacokinetic differences on tissue dosimetry. Crit Rev Toxicol 32(5):329-389. Available online
at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=12389868&
query hl=38.
Clewell, RA; Merrill, EA; Yu, KO; et al. (2003) Predicting neonatal perchlorate dose and inhibition of iodide uptake
in the rat during lactation using physiologically-based pharmacokinetic modeling. Toxicol Sci 74(2):416-436.
Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=12805655&
query_hl=42.
Clewell, HJ; Gentry, PR; Covington, TR; et al. (2004) Evaluation of the potential impact of age- and gender-specific
pharmacokinetic differences on tissue dosimetry. Toxicol Sci 79(2):381-393. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=l 5056818&
query_hl=34.
Cohen Hubal, EA; Egeghy, P; Leovic, K; et al. (2006) Measuring potential dermal transfer of a pesticide to children
in a child care center. Environ Health Perspect 114(2)264-269. Available online at
http://www.ehponline.org/members/2005/8283/8283.html.
Conolly, RB; Gaylor, DW; Lutz, WK. (2005) Population variability in biological adaptive responses to DNA
damage and the shape of carcinogen dose-response curves. Toxicol Appl Pharmacol 207(Suppl. 2):S570-S575.
Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=15996697&
query_hl=19.
Corley, RA; Mast, TJ; Carney, EW; et al. (2003) Evaluation of physiologically based models of pregnancy and
lactation for their application in children's health risk assessments. Crit Rev Toxicol 33(2): 137-211. Available online
at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=l 2708613&
query_hl=51.
Crump, KS. (1984) A new method for determining allowable daily intakes. Fundam Appl Toxicol 4(5):854-871.
Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list uids=6510615&q
ueryjil=2.
Crump, KS; Hoel, DG; Langley, CH; et al. (1976) Fundamental carcinogenic processes and their implications for
low dose risk assessment. Cancer Res 36(9 Pt. l):2973-2979. Available online at
http://www.ncbi. nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=975067&qu
ery_hl=12.
R-4
-------
Cuff, JM; Kimmel, GL; Kimmel, CA; et al. (1993) Relationship between abnormal somite development and axial
skeletal defects in rats following heat exposure. Teratology 48(3):259-266. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=824886
3&query_hl=15&itool=pubmed_docsum.
Cummins, SK; Jackson, RJ. (2001) The built environment and children's health. Pediatr Clin North Am
48(5):1241-1252. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=l 15796
72&query_hl= 19&itool=pubmed_docsum.
Dam, K; Seidler, FJ; Slotkin, TA. (1998) Developmental neurotoxicity of chlorpyrifos: delayed targeting of DNA
synthesis after repeated administration. Brain Res Dev Brain Res 108:39-45. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=969378
2&query_hl=19&itool=pubmed_docsum.
Daston, G; Faustman, E; Ginsberg, G; et al. (2004) A framework for assessing risks to children from exposure to
environmental agents. Environ Health Perspect 112(2):238-56. Available online at
http://ehp.niehs.nih.gov/members/2003/6182/6182.html.
DeWoskin, RS; Barone, S, Jr; Clewell, HJ; et al. (2001) Improving the development and use of biologically based
dose response models (BBDR) in risk assessment. Hum Ecol Risk Assess 7(5): 1091-1120. Available online at
http://www.crcjournals.com/eJournals/articles/article_synopsis.asp?id=29393.
Dietert, RR; Etzel, RA; Chen, D; et al. (2000) Workshop to identify critical windows of exposure for children's
health: immune and respiratory systems work group summary. Environ Health Perspect 108(Suppl 3):483-490.
Available online at http://ehp.niehs.nih.gov/members/2000/suppl-3/483-490dietert/dietert-full.html.
Edidin, DV; Levitsky, LL. (1982) Prepubertal gynecomastia associated with estrogen-containing hair cream. Am J
Dis Child 136(7):587-588. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=709108
4&query_hl=37&itool=pubmed_docsum.
Euling, SY; Kimmel, CA. (2001) Developmental stage sensitivity and mode of action information for androgen
agonists and antagonists. Sci Total Environ 2:274(1-3):103-113. Available online at
http://www.sciencedirect.com/science?_ob=MImg&_imagekey=B6V78-43B2999-9-
4&_cdi=5836&_user=14684&_orig=search&_coverDate=07%2F02%2F2001&_qd=l&_sk=997259998&view=c&
_alid=451506870&_rdoc=l&wchp=dGLbVzz-
zSkWA&md5=a9ct7b77fb2e20fffb99e54884fba331&ie=/sdarticle.pdf.
Executive Order. (1997) Executive Order 13045-Protection of children from environmental health risks and safety
risks. Fed Reg 62(78):19883-19888. Available online at http://www.epa.gov/fedrgstr/eo/eol3045.htm.
Farwell, AP; Dubord-Tomasetti, SA; Pietrzykowski, AZ; et al. (2006) Dynamic nongenomic actions of thyroid
hormone in the developing rat brain. Endocirnology 147(5):2567-2574. Available online at
http://endo.endojournals.Org/cgi/reprint/147/5/2567.
Faustman, EM; Allen, BC; Kavlock, RJ; et al. (1994) Dose-response assessment for developmental toxicity. I.
Characterization of database and determination of no observed adverse effect levels. Fundam Appl Toxicol
23(4):478-486. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=786789
9&query_hl=2&itool=pubmed_docsum.
R-5
-------
Faustman, EM; Lewandowski, TA; Ponce, RA; et al. (1999). Biologically based dose response models for
developmental toxicants: lessons from methylmercury. Inhal Toxicol 11(6-7):559-572. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=11202998&
query_hl=65.
Felner, El; White, PC. (2000) Prepubertal gynecomastia: indirect exposure to estrogen cream. Pediatrics 105(4):E55.
Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=l 07423
76&queryjil=40&itool=pubmed_docsum.
Fleeman ,TL; Cappon, GD; Chapin, RE; et al. (2005) The effects of feed restriction during organogenesis on
embryo-fetal development in the rat. Birth Defects Res B Dev Reprod Toxicol 74(5):442-449. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=161935
01 &query_hl= 11 &itool=pubmed_docsum.
Gaido, KW; Maness, SC; McDonnell, DP; et al. (2000) Interaction of methoxychlor and related compounds with
estrogen receptor alpha and beta, and androgen receptor: structure-activity studies. Mol Pharmacol 58(4):852-858.
Available online at http://molpharm.aspetjournals.Org/cgi/content/full/58/4/852.
Gargas, ML; Tyler, TR; Sweeney, LM; et al. (2000) A toxicokinetic study of inhaled ethylene glycol monomethyl
ether (2-ME) and validation of a physiologically based pharmacokinetic model for the pregnant rat and human.
Toxicol Appl Pharmacol 165(l):53-62. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=10814553&
query_hl=69.
Gaylor, DW; Kodell, RL. (2002) A procedure for developing risk-based reference doses. Reg Toxicol Pharmacol
35(2 Pt. 1):137-141. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=12051999&
query_hl=22.
Gee, GC; Payne-Sturges, DC. (2004) Environmental health disparities: a framework integrating psychosocial and
environmental concepts. Environ Health Perspect 112(17): 1645-1653. Available online at
http://www.pubmedcentral.gov/articlerender.fcgi?tool=pubmed&pubmedid=l 5579407.
Gentry, PR; Covington, TR; Clewell, HJ. (2003) Evaluation of the potential impact of pharmacokinetic differences
on tissue dosimetry in offspring during pregnancy and lactation. Reg Toxicol Pharmacol 38(1): 1-16. Available
online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=12878049&
query hl=73.
Ginsberg, G; Hattis, D; Sonawane, BR; et al. (2002) Evaluation of child/adult pharmacokinetic differences from a
database derived from the therapeutic drug literature. Toxicol Sci 66(2): 185-200. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=11896285&
query_hl=77.
Ginsberg, G; Hattis, D; Miller, R; et al. (2004a) Pediatric pharmacokinetic data: implications for environmental risk
assessment for children. Pediatrics 113(4 Suppl):973-983. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=l 5060190&
query_hl=79.
Ginsberg, G; Hattis, D; Sonawane, BR. (2004b) Incorporating pharmacokinetic differences between children and
adults in assessing children's risks to environmental toxicants. Toxicol Appl Pharmacol 198(2):164-183. Available
online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=15236952&
query_hl=83.
R-6
-------
Ginsberg, Gl; Slikker, W, Jr; Bruckner, J; et al. (2004c) Incorporating children's toxicokinetics into a risk
framework. Environ Health Perspect 112(2):272-283. Available online at
http://ehp.niehs.nih.gov/members/2003/6013/6013.html.
Gray, LE, Jr; Ostby, J; Cooper, RL; et al. (1999a) The estrogenic and antiandrogenic pesticide methoxychlor alters
the reproductive tract and behavior without affecting pituitary size or LH and prolactin secretion in male rats.
Toxicol Ind Health 15(l-2):37-47. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=101881
90&query_hl=7&itool=pubmed_docsum.
Gray, LE, Jr; Wolf, C; Lambright, C; et al. (1999b) Administration of potentially antiandrogenic pesticides
(procymidone, linuron, iprodione, chlozolinate, p,p'-DDE, and ketoconazole) and toxic substances (dibutyl- and
diethylhexyl phthalate, PCS 169, and ethane dimethane sulphonate) during sexual differentiation produces diverse
profiles of reproductive malformations in the male rat. Toxicol Ind Health 15(l-2):94-l 18. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=101881
94&query hl=28&itool=pubmed_docsum.
Gray, GM; Baskin, SI; Charnley, G. (2001) The Annapolis accords on the use of toxicology in risk assessment and
decision-making: an Annapolis center workshop report. Toxicol Methods 11(3):225-231. Available online at
http://taylorandfrancis.metapress.com/openurl.asp?genre=article&id=doi: 10.1080/105172301316871626.
Greaves, M. (2003). Pre-natal origins of childhood leukemia. Rev Clin Exp Hematol 7(3):233-245. Available online
at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=15024967&
query_hl=7.
Griffiths, CW; Dockins, C; Owens, N; et al. (2002) What to do at low doses: a bounding approach for economic
analysis. Risk Anal 22(4):679-88. Available online at
http://www.ncbi.nlm. nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=12224742&
query_hl=86.
Haiti, DL. (2001) Genetic management of outbred laboratory rodent populations. Charles River Laboratories.
Research Models. Available online at
http://www.criver.com/flex_contentarea/documents/rm_gt_r_genetic_management outbred rodent.pdf.
Hattis, D. (2004) Human interindividual variability in parameters related to susceptibility for toxic effects. Available
online at http://www2.clarku.edu/faculty/dhattis.
Hattis, D; Goble, R; Russ, A; et al. (2004) Age-related differences in susceptibility to carcinogenesis: a quantitative
analysis of empirical animal bioassay data. Environ Health Perspect 112(4):1152-1158. Available online at
http://ehp.niehs.nih.gov/members/2004/6871/6871.html.
Hattis, D; Goble, R; Chu, M. (2005) Age-related differences in susceptibility to carcinogenesis. II. Approaches for
application and uncertainty analyses for individual genetically acting carcinogens. Environ Health Perspect
113(4):509-516. Available online at http://ehp.niehs.nih.gov/members/2005/7564/7564.html.
Herbst, AL. (1987) The effects in the human of diethylstilbestrol (DES) use during pregnancy. Princess Takamatsu
Symp 18:67-75. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=350654
6&query_hl=l 0&itool=pubmed DocSum.
Hew, KW; Keller, KA. (2003) Postnatal anatomical and functional development of the heart: a species comparison.
Birth Defects Res B Dev Reprod Toxicol 68(4):309-320. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list uids=14666994&
query_hl=93.
R-7
-------
Hill, AB. (1965) The environment and disease: association or causation? Proc R Soc Med 58:295-300. Available
online at http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids= 14283879.
Hines, RN; McCarver, DG. (2002) The ontogeny of human drug-metabolizing enzymes: phase 1 oxidative enzymes.
J Pharmacol Exp Ther 300(2):355-360. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=l 1805191 &
query _hl=95.
Hoet, JJ; Ozanne, S; Reusens, B. (2000) Influences of pre- and postnatal nutritional exposures on vascular/endocrine
systems in animals. Environ Health Perspect 108(Suppl 3):563-568. Available online at
http://ehp.niehs.nih.gov/members/2000/suppl-3/563-568hoet/hoet-full.html.
Holladay, SD; Smialowicz, RJ. (2000) Development of the murine and human immune system: differential effects
of immunotoxicants depend on time of exposure. Environ Health Perspect 108(Suppl 3):463-474. Available online
at http://ehp.niehs.nih.gov/members/2000/suppl-3/463-473holladay/holladay-full.html.
Holsapple, MP; West, LJ; Landreth, KS. (2003) Species comparison of anatomical and functional immune system
development. Birth Defects Res B Dev Reprod Toxicol 68(4):321-334. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=14666995&
query_hl=97.
Hubal, EA; Sheldon, LS; Zufall, MJ; et al. (2000) The challenge of assessing children's residential exposure to
pesticides. J Exposure Anal Environ Epidemiol 10(6 Pt. 2):638-649. Available online at
http://www.ncbi. nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=l 1138656&
query _hl=99.
Hurtt, ME; Sandier, JD. (2003a) Comparative organ system development: introduction. Birth Defects Res B Dev
Reprod Toxicol 68(2):85. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=12866700&
query_hl=102.
Hurtt, ME; Sandier, JD. (2003b) Comparative organ system development: continuing the series. Birth Defects Res B
Dev Reprod Toxicol 68(4):307-308. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=14666993&
query_hl=104.
Hurtt, ME; Daston, G; Davis-Bruno, K; et al. (2004) Workshop Summary - Juvenile animal studies: testing
strategies and design. Birth Defects Res (Part B) 71(4):281-288. Available online at
http://www3.interscience.wiley.com/cgi-bin/abstract/109595570/ABSTRACT.
ICF Consulting. (1999) Child-related risk assessment policy and methodology guidance document survey (draft).
Office of the Administrator, Office of Children's Health Protection, Meg Kelly. Washington, DC: ICF Consulting.
ILSI (International Life Sciences Institute). (2003) Workshop to develop a framework for assessing risks to children
from exposures to environmental agents. Washington, DC: ILSI Press, Risk Science Institute. Available online at
http://rsi.ilsi.Org/NR/rdonlyres/lB183267-ADF3-4FEC-9CA2-339DD1482FC2/0/WorkshopReportFinal.pdf.
Julien, E; Willhite, CC; Richard, AM; et al. (2004) Challenges in constructing statistically based structure-activity
relationship models for developmental toxicity. Birth Defects Res A Clin Mol Teratol 70(12):902-911. Available
online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=15558547&
query _hl= 106.
Kavlock, RJ; Allen, BC; Faustman, EM; et al. (1995) Dose-response assessments for developmental toxicity. IV.
Benchmark doses for fetal weight changes. Fundam Appl Toxicol 26(2):211-222. Available online at
R-8
-------
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list uids=758991
0&queryjil=2&itool=pubmed_docsum.
Kimmel, CA; Cuff, JM; Kimmel, GL; et al. (1993) Skeletal development following heat exposure in the rat.
Teratology 47(3):229-242. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&listuids=847546
6&query_hl= 15&itool=pubmed_docsum.
Kimmel, CA; Kavlock, RJ; Allen, BC; et al. (1995) The application of benchmark dose methodology to data from
prenatal developmental toxicity studies. Toxicol Lett 82-83:549-554. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list uids=859710
8&query_hl=2&itool=pubmed_docsum.
Kimmel, CA; Collman, GW; Fields, N; et al. (2005) Lessons learned for the National Children's Study from the
National Institute of Environmental Health Sciences/U.S. Environmental Protection Agency Centers for Children's
Environmental Health and Disease Prevention Research. Environ Health Perspect 113(10): 1414-1418. Available
online at http://www.ehponline.org/docs/2005/7669/abstract.html.
Lam, SH; Sin, YM; Gong, Z; et al. (2005) Effects of thyroid hormone on the development of immune system in
zebrafish. Gen Comp Endocrinol 142(3):325-335. Available online at
http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6WGO-4FWKDR3-
l&_coverDate=07%2F31%2F2005&_alid=456156585&_rdoc=l&_fmt=&_orig=search&_qd=l&_cdi=6808&_sort
=d&view=c&_acct=COOOOO 1678&_version= 1 & urlVersion=0&jaserid= 14684&md5=ab62ec8973d3f8d8b342c 1 d
b911841a7.
Landrigan, P; Kimmel, CA; Correa, A; et al. (2004) Children's health and the environment: public health issues and
challenges for risk assessment. Environ Health Perspect 112(2):257-265. Available online at
http://ehp.niehs.nih.gov/members/2003/6115/6115.html.
Lau, C; Rogers, JM. (2004) Embryonic and fetal programming of physiological disorders in adulthood. Birth
Defects Res C Embryo Today 72(4):300-312. Available online at http://www3.interscience.wiley.com/cgi-
bin/abstract/10986721 I/ABSTRACT.
Lau, C; Andersen, ME; Crawford-Brown, DJ; et al. (2000) Evaluation of biologically based dose-response modeling
for developmental toxicity: a workshop report. Reg Toxicol Pharmacol 31(2 Pt. 1): 190-199. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=10854125&
query_hl=l 11.
Lau, C; Mole, ML; Copeland, MF; et al. (2001) Toward a biologically based dose-response model for
developmental toxicity of 5-fluorouracil in the rat: acquisition of experimental data. Toxicol Sci 59(l):37-48.
Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=l 1134542&
query_hl=114.
Lemasters, GK; Perreault, SD; Hales, BF; et al. (2000) Workshop to identify critical windows of exposure for
children's health: reproductive health in children and adolescents work group summary. Environ Health Perspect
108(Suppl 3):505-509. Available online at http://ehp.niehs.nih.gov/members/2000/suppl-3/505-
5091emasters/lemasters-full.html.
Liu, LJS; Box, M; Kalman, D; et al. (2003) Exposure assessment of particulate matter for susceptible populations in
Seattle. Environ Health Perspect 111(7):909-918. Available online at
http://www.ehponline.org/members/2003/6011/601 l.html.
R-9
-------
Lorber, M; Phillips, L. (2002) Infant exposure to dioxin-like compounds in breast milk. Environ Health Perspect
110(6):A325-332. Available online at http://ehp.niehs.nih.gov/members/2002/! 10pA325-A3321orber/lorber-
full.html.
Lutz, WK; Gaylor, DW; Conolly, RB; et al. (2005) Nonlinearity and thresholds in dose-response relationships for
carcinogenicity due to sampling variation, logarithmic dose scaling, or small differences in individual susceptibility.
Toxicol Appl Pharmacol 207(Suppl. 2):S565-S569. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=15982698&
query_hl=24.
Macintosh, DL; Kabiru, CW; Ryan, PB. (2001) longitudinal investigation of dietary exposure to selected pesticides.
Environ Health Perspect 109(2):145-150. Available online at http://www.ehponline.org/members/2001/109pl45-
ISOmacintosh/macintosh-full.html.
Marty, MS; Chapin, RE; Parks, LG; et al. (2003) Development and maturation of the male reproductive system.
Birth Defects Res B Dev Reprod Toxicol 68(2): 125-136. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=12866704&
query hl=l 16.
McCarver, DG; Hines, RN. (2002) The ontogeny of human drug-metabolizing enzymes: phase II conjugation
enzymes and regulatory mechanisms. J Pharmacol Exp Ther 300(2):361-366. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=l 1805192&
query_hl=118.
Mericskay, M; Carta, L; Sassoon, D. (2005) Diethylstilbestrol exposure in utero: a paradigm for mechanisms
leading to adult disease. Birth Defects Res A Clin Mol Teratol 73(3):133-135. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&listuids=l 57510
30&query_hl=8&itool=pubmed_docsum.
Morford, LL; Henck, JW; Breslin, WJ; et al. (2004) Hazard identification and predictability of children's health risk
from animal data. Environ Health Perspect 112(2):266-271. Available online at
http://ehp.niehs.nih.gov/members/2003/6014/6014.html.
Morgan, MK; Sheldon, LS; Croghan, CW; et al. (2006) A pilot study of children's total exposure to persistent
pesticides and other persistent organic pollutants (CTEPP). Volume I: Final Report. Contract Number 68-D-99-011,
U.S. Environmental Protection Agency, Office of Research and Development, Research Triangle Park, NC.
Available online at http://www.epa.gov/heasd/ctepp/ctepp report.pdf.
Mylchreest, E; Sar, M; Cattley, RC; et al. (1999) Disruption of androgen-regulated male reproductive development
by di(n-butyl) phthalate during late gestation in rats is different from flutamide. Toxicol Appl Pharmacol
156(2):81-95. Available online at http://www.sciencedirect.com/science?_ob=MImg&_imagekey=B6WXH-
45FKOYO-67-
1 &_cdi=7159&_user= 14684&_orig=search&_coverDate=04%2F 15%2F 1999&_qd= 1 & sk=998439997&view=c&
_alid=451508681 & rdoc= 1 &wchp=dGLbVtb-
zSkzk&md5=ad20cd594f6ddeb5a209898flb8f77b5&ie=/sdarticle.pdf.
Mylchreest, E; Wallace, DG; Cattley, RC; et al. (2000) Dose-dependent alterations in androgen-regulated male
reproductive development in rats exposed to Di(n-butyl) phthalate during late gestation. Toxicol Sci 55(1):143-151.
Available online at http://toxsci.oxfordjournals.Org/cgi/reprint/55/l/143.
NEJAC (National Environmental Justice Advisory Council). (2004) Ensuring risk reduction in communities with
multiple stressors: environmental justice and cumulative risks/impacts. Cumulative Risks/Impacts Work Group.
Available online at http://www.epa.gov/compliance/resources/publications/ej/nejac/nejac-cum-risk-rpt-122104.pdf.
R-10
-------
NRC (National Research Council). (1993) Pesticides in the diets of infants and children. Washington, DC: National
Press Academy. Available online at http://www.nap.edu/catalog/2126.html.
NRC. (1994) Science and judgment in risk assessment. Committee on Risk Assessment of Hazardous Air Pollutants,
Board on Environmental Studies and Toxicology, Commission on Life Sciences. Washington, DC: National Press
Academy. Available online at http://www.nap.edu/catalog/2125.html.
NRC. (2006) Human biomonitoring for environmental chemicals. Committee on Human Biomonitoring for
Environmental Toxicants. Washington, DC: The National Academies Press. Available online at
http://www.nap.edu/catalog/! 1700.html.
O'Flaherty, EJ. (1998) A physiologically based kinetic model for lead in children and adults. Environ Health
Perspect 106(Suppl. 6):1495-1503. Available online at http://ehp.niehs.nih.gov/members/1998/Suppl-6/1495-
1503oflaherty/oflaherty- full.html.
Olin, SS; Sonawane, BR. (2003) Workshop to develop a framework for assessing risks to children from exposure to
environmental agents. Environ Health Perspect 111(12): 1524-1526. Available online at
http://ehp.niehs.nih.gov/members/2003/6183/6183.html.
Olshan, AF; Anderson, L; Roman, E; et al. (2000) Workshop to identify critical windows of exposure for children's
health: cancer work group summary. Environ Health Perspect 108(Suppl 3):595-597. Available online at
http://ehp.niehs.nih.gov/members/2000/suppl-3/595-597olshan/olshan-full.html.
Osmond, C; Barker, DJP. (2000) Fetal, infant, and childhood growth are predictors of coronary heart disease,
diabetes, and hypertension in adult men and women. Environ Health Perspect 108(Suppl 3):545-553. Available
online at http://ehp.niehs.nih.gov/members/2000/suppl-3/545-553osmond/osmond-full.html.
Pals, K; Vankelecom, H; Denef, C. (2006) Triiodothyronine expands the lactotroph and maintains the
lactosomatrotroph population, whereas thyrotrophin-releasing hormone augments thyrotroph abundance in
aggregate cell cultures of postnatal rat pituitary gland. J Neuroendocrinol 18(3):203-216. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt:=AbstractPlus&list_uids=l 64548
04&queryjil=5&itool=pubmeddocsum.
Pease, W; Vandenberg, J; Hooper, K. (1991) Comparing alternative approaches to establishing regulatory levels for
reproductive toxicants: DBCP as a case study. Environ Health Perspect 91:141-155. Available online at
http://ehp.niehs.nih.gov/members/1991/091/91015.PDF.
Peden, DB. (2000) Development of atopy and asthma: candidate environmental influences and important periods of
exposure. Environ Health Perspect 108(Suppl 3):475-482. Available online at
http://ehp.niehs.nih.gov/members/2000/suppl-3/475-482peden/peden-full.html.
Pelekis, M; Gephart, LA; Lerman, SE. (2001) Physiological-model-based derivation of the adult and child
pharmacokinetic intraspecies uncertainty factors for volatile organic compounds. Reg Tox Pharmacol 33(1):12-20.
Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=l 1259175&
query_hl=132.
Pelekis, M; Nicolich, MJ; Gauthier, JS. (2003) Probabilistic framework for the estimation of the adult and child
toxicokinetic intraspecies uncertainty factors. Risk Anal 23(6):1239-1255. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=l 4641898&
query_hl=134.
Pinkerton, KE; Joad, JP. (2000) The mammalian respiratory system and critical windows of exposure for children's
health. Environ Health Perspect 108(Suppl 3):457-462. Available online at
http://ehp.niehs.nih.gov/members/2000/suppl-3/457-462pinkerton/pinkerton-full.html.
R-ll
-------
Presidential/Congressional Commission on Risk Assessment and Risk Management. (1997a) Final Report Volume
1: Framework for Environmental Health Risk Management. Available online at
http://www.riskworld.com/Nreports/1997/risk-rpt/pdf/EPAJAN.PDF.
Presidential/Congressional Commission on Risk Assessment and Risk Management. (1997b) Final Report Volume
2: Risk Assessment and Risk Management in Regulatory Decision-Making. Available online at
http://www.riskworld.com/Nreports/1997/risk-rpt/volume2/pdf/v2epa.PDF.
Preston, RJ. (2004) Children as a sensitive subpopulation for the risk assessment process. Toxicol Appl Pharmacol
199(2):132-141. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=15313585&
queryjil=140.
Price, PS; Chaisson, CF; Koontz, M; et al. (2003) Construction of a comprehensive chemical exposure framework
using person oriented modeling. Developed for The Exposure Technical Implementation Panel, American
Chemistry Council, Contract #1388. Available online at
http://www.thelifelinegroup.org/lifeline/documents/comprehensive_chemical_exposure_framework.pdf.
Pryor, JL; Hughes, C; Foster, W; et al. (2000) Critical windows of exposure for children's health: the reproductive
system in animals and humans. Environ Health Perspect 108(Suppl 3):491-503. Available online at
http://ehp.niehs.nih.gov/members/2000/suppl-3/491-503pryor/pryor-full.html.
Ramos, HE; Weiss, RE. (2006) Regulation of nuclear coactivator and corepressor expression in mouse cerebellum
by thyroid hormone. Thyroid 16(3):211-216. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dept=AbstractPlus&list_uids=1657108
2&query_hl=5&itool=pubmed_docsum.
Rice, D; Barone, S, Jr. (2000) Critical periods of vulnerability for the developing nervous system: evidence from
humans and animal models. Environ Health Perspect 108 (Suppl 3): 511-533. Available online at
http://ehp.niehs.nih.gOv/members/2000/suppl-3/511 -533rice/rice-full.html.
Robboy, SJ; Taguchi, O; Cunha, GR. (1982) Normal development of the human female reproductive tract and
alterations resulting from experimental exposure to diethylstilbestrol. Hum Pathol 13(3): 190-198. Available online
at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=707620
7&query_hl= 10&itool=pubmed DocSum.
Rubert, G; Minana, R; Pascual, M; et al. (2006) Ethanol exposure during embryogenesis decreases the radial glial
progenitorpool and affects the generation of neurons and astrocytes. J Neurosci Res 84(3):483-496. Available online
at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list uids=167707
75&query hl=43&itool=pubmed_DocSum.
Sadler, TW. (2000) Susceptible periods during embryogenesis of the heart and endocrine glands. Environ Health
Perspect 108 (Suppl 3):555-561. Available online at http://ehp.niehs.nih.gov/members/2000/suppl-3/555-
561 sadler/sadler- full. html.
Santisteban, P; Bernal, J. (2005) Thyroid development and effect on the nervous system. Rev Endocr Metab Disord
6(3):217-228. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=161516
26&query_hl=5&itool=pubmed_docsum.
R-12
-------
Schulte, PA. (1989) A conceptual framework for the validation and use of biologic markers. Environ Res
48(2): 129-144. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=264748
8&query_hl= 11 &itool=pubmed_docsum.
Seed, J; Carney, EW; Corley, RA; et al. (2005) Overview: Using mode of action and life stage information to
evaluate the human relevance of animal toxicity data. Critical Reviews in Toxicology 35(8-9):663-672. Available
online at http://journalsonline.tandf.co.uk/openurl.asp?genre=article&eissn=l 547-
6898&volume=35&issue=8&spage=663.
Selevan, SG; Kimmel, CA; Mendola, P. (2000) Identifying critical windows of exposure for children's health.
Environ Health Perspect 108(SuppI 3):451-555. Available online at http://ehp.niehs.nih.gov/docs/2000/suppl-
3/toc.html.
Setzer, RW; Lau, C; Mole, ML; et al. (2001) Toward a biologically based dose-response model for developmental
toxicity of 5-fluorouracil in the rat: a mathematical construct. Toxicol Sci 59(l):49-58. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids= 11134S43&
query _hl= 142.
Shuey, DL; Lau, C; Logsdon, TR; et al. (1994) Biologically based dose-response modeling in developmental
toxicology: biochemical and cellular sequelae of 5-fluorouracil exposure in the developing rat. Toxicol Appl
Pharmacol 126(1): 129-44. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=8184423&q
uery_hl=144.
Slikker, W, Jr; Andersen, ME; Bogdanffy, MS; et al. (2004a) Dose-dependent transitions in mechanisms of toxicity.
Toxicol Appl Pharmacol 201(3):203-225. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=15582645&
query_hl=146.
Slikker, W, Jr; Andersen, ME; Bogdanffy, MS; et al. (2004b) Dose-dependent transitions in mechanisms of toxicity:
case studies. Toxicol Appl Pharmacol 201: 226-294. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list uids=15582646&
query_hl=148.
Spearow, JL; Doemeny, P; Sera, R; et al. (1999) Genetic variation in susceptibility to endocrine disruption by
estrogen in mice. Science 285(5431):1259-61. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=l 0455051&
query_hl=150.
Spearow, JL; O'Henley, P; Doemeny, P; et al. (2001) Genetic variation in physiological sensitivity to estrogen in
mice. APMIS 109(5):356-64. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=l 1478683&
query_hl=152.
Stephens, TD; Fillmore, BJ. (2000) Hypothesis: thalidomide embryopathy-proposed mechanism of action.
Teratology 61(3): 189-195. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=106619
08&query_hl=30&itool=pubmed_docsum.
Stout, DM; Mason, MA. (2003) The distribution of chlorpyrifos following a crack and crevice type application in
the US EPA Indoor Air Quality Research House. Atmos Environ 37(38):5539-5549. Available online at
http://www.ingentaconnect.com/content/els/! 3522310/2003/00000037/00000039/art00771;jsessionid=bc6yzy6451w
j.alice
R-13
-------
U.S. 94th Congress. (1976) Toxic Substances Control Act (TSCA). Public Law (PL) 94-469. Available online at
http://www.epa.gov/region5/defs/html/tsca.htm.
U.S. 96th Congress. (1980) Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA).
P.L. 96-510. Available online at http://www.epa.gov/superfund/action/law/cercla.htm.
U.S. 101st Congress. (1990) Clean Air Act (CAA) Amendments. PL 101-549. Available online at
http://www.epa.gov/air/oaq caa.html/.
U.S. 104th Congress. (1996a) Food Quality Protection Act (FQPA). PL 104-170. Available online at
http://www.epa.gov/pesticides/regulating/laws/fqpa/.
U.S. 104th Congress. (1996b) Safe Drinking Water Act (SDWA) Amendments. PL 104-182. Available online at
http://www.epa.gov/safewater/sdwa/text.html.
U.S. 104th Congress. (1996c) Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA). PL 104-170. Available
online at http://www.epa.gov/region5/defs/html/fifra.htm.
U.S. 108th Congress (2003) Pediatric Research Equity Act. PL 108-155. Available online at
http://frwebgate.access.gpo.gov/cgi-bin/getdoc.cgi?dbname=108_cong_public_laws&docid=f:publl 55.108.
U.S. EPA (Environmental Protection Agency). (1991) Guidelines for developmental toxicity risk assessment. Fed
Reg 56(234):63798-63826. Available online at http://www.epa.gov/ncea/raf/pdfs/devtox.pdf.
U.S. EPA. (1992) Guidelines for exposure assessment. Risk Assessment Forum, Washington, DC;
EPA/600/Z-92/001. Available online athttp://cfpub.epa.gov/ncea/raf/recordisplay.cfm?deid=15263.
U.S. EPA. (1994) Methods for derivation of inhalation reference concentrations and application of inhalation
dosimetry. Office of Research and Development, Office of Health and Environmental Assessment, Environmental
Criteria and Assessment Office, Research Triangle Park, NC; EPA/600/8-90/066F. Available online at
http://cfpub2.epa.gov/ncea/cfm/recordisplay.cfm?deid=71993.
U.S. EPA. (1995a) Policy on evaluating health risks to children. Office of the Administrator, Washington, DC.
Available online at http://www.epa.gov/osa/spc/2poleval.htm.
U.S. EPA. (1995b) Application of benchmark dose methodology to data from prenatal developmental toxicity
studies. Washington, DC; EPA/600/A-95/135. Available online at
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=40957.
U.S. EPA. (1996) Guidelines for reproductive toxicity risk assessment. Risk Assessment Forum, Washington, DC;
EPA/630/R-96/009. Available online at http://cfpub.epa.gov/ncea/raf/pdfs/repro51.pdf.
U.S. EPA (1997a) Guidance on cumulative risk assessment. Part 1. Planning and scoping. Office of the
Administrator, Washington, DC. Available online at http://www.epa.gov/osa/spc/pdfs/cumrisk2.pdf
U.S. EPA. (1997b) Guiding principles for Monte Carlo analysis. Office of Research and Development, Risk
Assessment Forum. Washington, DC; EPA/630/R-97/001. Available online at
http://www.epa.gov/ncea/raf/montecar.pdf.
U.S. EPA. (1997c) Summary of the U.S. EPA colloquia on a framework for human health risk assessment: Volume
1. Risk Assessment Forum, Washington, DC; EPA/600/R-99/001. Available online at
http://cfpub.epa.gov/ncea/raf/recordisplay.cfm?deid= 12186.
U.S. EPA. (1998a) Guidelines for ecological risk assessment. Fed Reg 63(93):26846-26924. EPA/630/R-95/002F.
Available online at http://oaspub.epa.gov/eims/eimscomm.getfile?p_download_jd=36512.
R-14
-------
U.S. EPA. (1998b) Guidelines for neurotoxicity risk assessment. Federal Register 63(93):26926-26954.
EPA/630/R-95/001F. Available online at http://oaspub.epa.gov/eims/eimscomm.getfile?p_download_id=4555.
U.S. EPA. (1998c) Summary of the U.S. EPA colloquium on a framework for human health risk assessment:
Volume 2. Risk Assessment Forum, Washington, DC. Available online at
http://cfpub.epa.gov/ncea/raf/recordisplay.cfm?deid=55007.
U.S. EPA (1999a) Toxicological review for ethylene glycol monobutyl ether. Available online at
http://www.epa.gov/iris/toxreviews/0500-tr.pdf.
U.S. EPA. (1999b) Report of the workshop for selecting input distributions for probabilistic assessments. Office of
Research and Development, Washington, DC; EPA/630/R-98/004. Available online at
http://cfpub2.epa.gov/ncea/cfm/recordisplay.cfm? deid= 12487.
U.S. EPA. (2000a) Guidelines for preparing economic analyses. Office of the Administrator, Washington DC;
EPA/240-R-00-003. Available online at http://yosemite.epa.gov/ee/epa/eed.nsf/webpages/Guidelines.html.
U.S. EPA. (2000b) Options for development of parametric probability distributions for exposure factors. National
Center for Environmental Assessment, Office of Research and Development, Washington DC; EPA/600/R-00/058.
Available online at http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=20867.
U.S. EPA. (2000c) Stressor identification guidance document. Office of Research and Development, Washington,
DC. EPA 822-B-00-0025. Available online at http://www.epa.gov/waterscience/biocriteria/stressors/.
U.S. EPA. (2000d) Benchmark dose technical guidance document (external review draft). Risk Assessment Forum,
Washington, DC; EPA/630/R-00/001. Available online at
http://oaspub.epa.gov/eims/eimscomm.getfile?p_download_id=4727.
U.S. EPA. (2000e) Science policy handbook: risk characterization. Science Policy Council, Washington, DC;
EPA/100/B/00/002. Available online at http://www.epa.gov/osa/spc/pdfs/rchandbk.pdf.
U.S. EPA. (2000f) Strategy for research on environmental risks to children. Office of Research and Development,
Washington, DC; EPA/600/R-00/068. Available online at http://www.epa.gov/ncea/pdfs/strat4resrch.pdf.
U.S. EPA. (2001a) Risk assessment guidance for Superfund (RAGS): Volume I - human health evaluation manual
(Part D, standardized planning, reporting and review of Superfund risk assessments). Office of Emergency and
Remedial Response, Washington, DC; 9285.7-01D. Available online at
http://www.epa.gov/oswer/riskassessment/ragsd/index.htm.
U.S. EPA. (2001b) Risk assessment guidance for Superfund (RAGS): process for conducting probabilistic risk
assessments. Volume 3, Part A. Office of Solid Waste and Emergency Response, Washington, DC; EPA/540/R-
02/002. Available online at http://www.epa.gov/oswer/riskassessment/rags3a/index.htm.
U.S. EPA. (2002a) A review of the reference dose and reference concentration processes. Risk Assessment Forum,
Washington, DC; EPA/630/P-02/002F. Available online at
http://oaspub.epa.gov/eims/eimscomm.getfile?p_download_id=36836.
U.S. EPA. (2002b) Lessons learned on planning and scoping for environmental risk assessments. Planning and
Scoping Workgroup of the Science Policy Council Steering Committee. Available online at
http://www.epa.gov/osa/spc/pdfs/handbook.pdf.
U.S. EPA. (2002c) Child-specific exposure factors handbook (interim report). Office of Research and Development,
National Center for Environmental Assessment, Washington, DC; EPA/600/P/00/002B. Available online at
http://oaspub.epa.gov/eims/eimscomm.getfile?p_download_id=36528.
R-15
-------
U.S. EPA. (2002d) Determination of the appropriate FQPA safety factor(s) in tolerance assessment. Office of
Pesticide Programs, Office of Pesticide Programs, Washington, DC. Available online at
http://www.epa.gov/oppfead 1 /trac/science/determ.pdf.
U.S. EPA. (2003a) Framework for cumulative risk assessment. Risk Assessment Forum, Washington, DC;
EPA/630/P-02/001F. Available online at http://cfpub.epa.gov/ncea/raf/recordisplay.cfm?deid=54944.
U.S. EPA. (2003b) Children's health valuation handbook. Office of Children's Health Protection and National
Center for Environmental Economics. Office of Policy, Economics and Innovation, Washington, DC;
EPA-100-R-03-003. Available online at
http://yosemite.epa.gov/ee/epa/eed.nsf/webpages/HandbookChildrensHealthValuation.html.
U.S. EPA. (2003c) Toxicological review of xylenes. Office of Research and Development, National Center for
Environmental Assessment, Washington, DC; EPA 635/R-03/001. Available online at
http://www.epa.gov/iris/toxreviews/0270-tr.pdf.
U.S. EPA. (2005a) Guidelines for carcinogen risk assessment. Risk Assessment Forum, Washington, DC;
EPA/630/P-03/001F. Available online at http://oaspub.epa.gov/eims/eimscomm.getfile?p_download_id=439797.
U.S. EPA. (2005b) Supplemental guidance for assessing susceptibility from early-life exposure to carcinogens. Risk
Assessment Forum, Washington, DC; EPA/630/R-03/003F. Available online at
http://oaspub.epa.gov/eims/eimscomm.getfile?p_downloadjd=439798.
U.S. EPA. (2005c) Aging and toxic response: issues relevant to risk assessment. Risk Assessment Forum,
Washington, DC; EPA/600/P-03/004A. Available online at
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=l 56648.
U.S. EPA. (2005d) Review of the national ambient air quality standards for particulate matter: policy assessment of
scientific and technical information, OAQPS Staff Paper (EPA-452/R-05-005). Available online at
http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_cr_sp.html.
U.S. EPA. (2005e) Guidance on selecting age groups for monitoring and assessing childhood exposures to
environmental contaminants. Risk Assessment Forum, Washington, DC; EPA/630/P-03/003F. Available online at
http://cfpub2.epa.gov/ncea/cfm/recordisplay.cfm?deid= 146583.
U.S. EPA. (2005f) Air quality criteria for ozone and related photochemical oxidants (Second external review draft)
(EPA/600/R-05/004aB). Available online at http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=137307.
U.S. EPA. (2006a) Toxicological review of dibutyl phthalate(di-n-butyl phthalate) (CAS No. 84-74-2) (External
peer review draft). Office of Research and Development, National Center for Environmental Assessment,
Washington, DC; NCEA-S-1755. Available online at
http://oaspub.epa.gov/eims/eimscomm.getfile?p_download_id=457421.
U.S. EPA. (2006b) Approaches for the application of physiologically based pharmacokinetic (PBPK) models and
supporting data in risk assessment. Office of Research and Development, National Center for Environmental
Assessment, Washington, DC. Available online at http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=157668.
U.S. EPA. (2006c) Air quality criteria for lead (Second external review draft). Office of Research and Development,
National Center for Environmental Assessment, Washington, DC; EPA/600/R-05/144aB-bB, 2006. Available online
at http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=l 54041.
U.S. EPA. (2006d) Harmonization in interspecies extrapolation: use of BW3/4 as default method in derivation of the
oral RfD (External Review Draft). Risk Assessment Forum, Washington, DC; EPA/630/R-06/001. Available online
at http://cfpub.epa. go v/ncea/raf/recordisplay.cfm?deid= 148525.
R-16
-------
U.S. FDA (Food and Drug Administration). (1978) Good laboratory practice for nonclinical laboratory studies. 21
CFR Part 58. Available online at
http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?CFRPart=58.
U.S. FDA. (1997) Food and drug administration modernization act. PL 105-115. Available online at
http://www.fda.gov/cder/guidance/105-115.htm.
U.S. FDA. (2002) Best Pharmaceuticals for children act. PL 107-109. Available online at
http://www.fda.gov/opacom/laws/pharmkids/contents.html.
U.S. FDA (2005) Draft Guidance for industry - how to comply with the pediatric research equity act. Available
online at http://www.fda.gov/cder/guidance/6215dft.pdf.
U.S. FDA. (2006) Guidance for industry- nonclinical safety evaluation of pediatric drug products. U.S. Department
of Health and Human Services. Food and Drug Administration. Center for Drug Evaluation and Research (CDER).
Available online at http://www.fda.gov/cder/guidance/5671fnl.pdf.
Vineis, P; Kriebel, D. (2006) Causal models in epidemiology: Past inheritance and genetic future. Environ Health
5(1):21. [e-pub ahead of print] Available online at
http://www.pubmedcentral.nih.gov/articlerender.fcgi?tool=pubmed&pubmedid=l 6859537.
Walthall, K; Cappon, GD; Hurtt, ME; et al. (2005) Postnatal development of the gastrointestinal system: a species
comparison. Birth Defects Res B Dev Reprod Toxicol 74(2): 132-156. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids=l 58349
02&query_hl=45&itool=pubmed_docsum.
Weed, DL. (2005) Weight of evidence: a review of concept and methods. Risk Anal 25(6):1545-1557. Available
online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=pubmed&cmd=Retrieve&dopt=AbstractPlus&list_uids= 165069
81 &query_hl=5&itool=pubmed_docsum.
White, WJ; Lee, CS. (1998) The development and maintenance of the Crl:CD®(SD)IGS BR breeding system.
Charles River Laboratories. Research Models. Available online at
http://www.criver.com/flex_content_area/documents/rm_rm_a_igs_rat_breeding_system.pdf.
WHO (World Health Organization). (2006) Principles for evaluating health risks in children associated with
exposure to chemicals (draft). International Programme on Chemical Safety (IPCS) Environmental Health Criteria
document (EHC-237). Available online at http://www.who.int/ipcs/features/ehc/en/index.html.
Wood, SL; Beyer, BK; Cappon, GD. (2003) Species comparison of postnatal CNS development: functional
measures. Birth Defects Res B Dev Reprod Toxicol 68(5):391-407. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=14745989&
query_hl=163.
Xu, Y; Little, JC. (2006) Predicting emissions of SVOCs from polymeric materials and their interaction with
airborne particles. Environ Sci Technol 40(2):456-461. Available online at
http://www.ncbi. nlm.nih.gov/entrez/query.fcgi?CMD=search&DB=pubmed.
Yelin, R; Schyr, RB; Kot, H; et al. (2005) Ethanol exposure affects gene expression in the embryonic organizer and
reduces retinoic acid levels. Dev Biol 279(1): 193-204. Available online at
http://www.sciencedirect.com/science? ob=MImg&_imagekey=B6WDG-4F7Y93H-2-
l&_cdi=6766&_user=14684&_orig=search&_coverDate=03%2F01%2F2005&_qd=l&_sk=997209998&view=c&
_alid=451515301 &_rdoc= 1 &wchp=dGLbVlb-
zSkWA&md5=e2aOa807b351 ce612e88e5893d2f 16cd&ie=/sdarticle.pdf.
R-17
-------
Young, JG; Eskenazi, B; Gladstone, EA; et al. (2005) Association between in utero organophosphate pesticide
exposure and abnormal reflexes in neonates. Neurotoxicology 26:199-209. Available online at
http://www.sciencedirect.com/science?_ob=MImg&_imagekey=B6W81-4F7SOVD-l-
l&_cdi=6641&_user=14684&_orig=search&_coverDate=03%2F01%2F2005&_qd=l&_sk=999739997&view=c&
_alid=4515 07612&_rdoc= 1 & wchp=dGLb Vtz-
zSkzk&md5=b02f3248dbl3b6a40b8ddled9d003d3e&ie=/sdarticle.pdf.
Zoetis, T; Hurtt, ME. (2003a) Species comparison of anatomical and functional renal development. Birth Defects
Res B Dev Reprod Toxicol 68(2): 111-120. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=12866702&
query_hl=160.
Zoetis, T; Hurtt, ME. (2003b) Species comparison of lung development. Birth Defects Res B Dev Reprod Toxicol
68(2): 121-124. Available online at
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=12866703&
query_hl=158.
Zoetis, T; Walls, I. (ed.) (2003) Principles and practices for direct dosing of pre-weaning mammals in toxicity
testing and research. Risk Science Institute, Washington, DC: ILS1 Press. Available online at
http://rsi.ilsi.org/Publications/directdosingpreweaning.htm.
Zoetis, T; Tassinari, MS; Bagi, C; et al. (2003) Species comparison of postnatal bone growth and development.
Birth Defects Res B Dev Reprod Toxicol 68(2):86-l 10. Available online at
http://www.ncbi. nlm.nih.gov/entrez/query.fcgi?CMD=search&DB=pubmed.
R-18
-------
-------
xvEPA
United States
Environmental Protection
Agency
National Center for
Environmental Assessment
Washington, DC 20460
Official Business
Penalty for Private Use $300
EPA/600/R-05/093F
September 2006
I
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
i
,
------- |