EPA/635/R-20/131
^ssessments Protocol
www.epa.gov/iris
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and
PFDA (anionic and acid forms) IRIS Assessments
CASRN 335-76-2 [PFDA]
CASRN 375-95-1 [PFNA]
CASRN 307-24-4 (PFHxA]
CASRN 355-46-4 [PFHxS]
CASRN 375-22-4 [PFBA]
Supplemental Information—Appendix A
October 2019
Updated February 2020 (in response to public comments)
This document was posted for public comment on November 8, 2019 ("link to more information"),
and subsequently updated in response to those comments (updates are outlined in Section 12). It
does not represent and should not be construed to represent any Agency determination or policy.
This document will serve as Appendix A of the Supplemental Materials for all five IRIS PFAS
assessment
Integrated Risk Information System
Center for Public Health and Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
DISCLAIMER
This document was posted for public comment on November 8, 2019 (link to more information),
and subsequently updated in response to those comments (updates are outlined in Section 12). It
does not represent and should not be construed to represent any Agency determination or policy.
This document is a draft for review purposes only and does not constitute Agency policy.
ii DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
CONTENTS
AUTHORS| CONTRIBUTORS| REVIEWERS ii
1. INTRODUCTION 1-1
2. SCOPING AND PROBLEM FORMULATION SUMMARY 2-2
2.1. SUMMARY OF BACKGROUND INFORMATION 2-2
2.1.1. Chemical and Physical Properties 2-2
2.1.2. Sources, Production, and Use 2-5
2.1.3. Environmental Fate and Transport 2-7
2.1.4. Environmental Concentrations 2-8
2.1.5. Potential for Human Exposure 2-11
2.1.6. Populations and Lifestages with Potentially Greater Exposures 2-12
2.1.7. Other Environmental Protection Agency (EPA) Assessments of Per- and
Polyfluoroalkyl Substances (PFAS) 2-13
2.1.8. Assessments and Toxicity Values from Other Sources 2-13
2.2. SCOPING SUMMARY 2-16
2.3. PROBLEM FORMULATION 2-20
2.3.1. Preliminary Literature Inventory for the Five Per- and Polyfluoroalkyl Substances
(PFAS) Being Assessed 2-20
2.4. KEY SCIENCE ISSUES 2-22
2.4.1. Toxicokinetic Differences across Species and Sexes 2-23
2.4.2. Human Relevance of Effects in Animals that Involve Peroxisome
Proliferator-Activated Receptor Alpha (PPARa) Receptors 2-25
2.4.3. Potential Confounding by Other Per- and Polyfluoroalkyl Substances (PFAS)
Exposures in Epidemiology Studies 2-27
2.4.4. Toxicological Relevance of Changes in Certain Urinary and Hepatic Endpoints in
Rodents 2-27
2.4.5. Characterizing Uncertainty Due to Missing Chemical-Specific Information 2-27
3. OVERALL OBJECTIVES, SPECIFIC AIMS, AND POPULATIONS, EXPOSURES, COMPARATORS,
AND OUTCOMES (PECO) CRITERIA 3-1
3.1. SPECIFIC AIMS 3-2
3.2. POPULATIONS, EXPOSURES, COMPARATORS, AND OUTCOMES (PECO) CRITERIA 3-3
4. LITERATURE SEARCH AND SCREENING STRATEGIES 4-1
This document is a draft for review purposes only and does not constitute Agency policy.
iii DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
4.1. LITERATURE SEARCH STRATEGIES 4-1
4.1.1. Non-Peer-Reviewed Data 4-4
4.2.SCREENING PROCESS 4-4
4.2.1. Multiple Publications of the Same Data 4-7
4.2.2. Literature Flow Diagrams 4-8
4.3. SUMMARY-LEVEL LITERATURE INVENTORIES 4-13
5. REFINED EVALUATION PLAN 5-1
6. STUDY EVALUATION (REPORTING, RISK OF BIAS, AND SENSITIVITY) STRATEGY 6-1
6.1. STUDY EVALUATION OVERVIEW FOR HEALTH EFFECT STUDIES 6-1
6.2. EPIDEMIOLOGY STUDY EVALUATION 6-6
6.2.1. Epidemiology Study Evaluation Criteria Specific to These Five Per- and
Polyfluoroalkyl Substances (PFAS) 6-16
6.3. EXPERIMENTAL ANIMAL STUDY EVALUATION 6-19
6.3.1. Animal Toxicological Study Evaluation Considerations Specific to These Five
Per-and Polyfluoroalkyl Substances (PFAS) 6-29
6.4. PHARMACOKINETIC MODEL EVALUATION 6-30
6.5. MECHANISTIC STUDY EVALUATION 6-30
7. ORGANIZING THE HAZARD REVIEW 7-1
8. DATA EXTRACTION OF STUDY METHODS AND RESULTS 8-1
8.1. STANDARDIZING REPORTING OF EFFECT SIZES 8-2
8.2. STANDARDIZING ADMINISTERED DOSE LEVELS/CONCENTRATIONS 8-3
9. SYNTHESIS OF EVIDENCE 9-1
9.1. SYNTHESES OF HUMAN AND ANIMAL HEALTH EFFECTS EVIDENCE 9-6
9.2. MECHANISTIC INFORMATION 9-7
9.2.1. Toxicokinetic Information and Pharmacokinetic (PK)/Physiologically Based
Pharmacokinetic (PBPK) Models 9-8
9.2.2. Peroxisome Proliferator-Activated Receptor Alpha (PPARa) Dependence for
Health Effect(s) Observed in Animals 9-11
9.2.3. Toxicological Relevance of Select Outcomes Observed in Animals 9-14
9.2.4. Other Focused Mechanistic Analyses 9-16
10. EVIDENCE INTEGRATION 10-1
10.1. EVALUATING THE STRENGTH OF THE HUMAN AND ANIMAL EVIDENCE STREAMS 10-6
10.2. OVERALL EVIDENCE INTEGRATION JUDGMENTS 10-10
10.3. HAZARD CONSIDERATIONS FOR DOSE-RESPONSE 10-17
This document is a draft for review purposes only and does not constitute Agency policy.
iv DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
11. DOSE-RESPONSE ASSESSMENT: SELECTING STUDIES AND QUANTITATIVE ANALYSIS 11-1
11.1. SELECTING STUDIES FOR DOSE-RESPONSE ASSESSMENT 11-2
11.2. CONDUCTING DOSE-RESPONSE ASSESSMENTS 11-6
11.2.1. Dose-Response Analysis in the Range of Observation 11-6
11.2.2. Extrapolation: Slope Factors and Unit Risk 11-10
11.2.3. Extrapolation: Reference Values 11-10
12. PROTOCOL HISTORY 12-14
REFERENCES R-l
ADDENDUM A. SUMMARY OF EXISTING TOXICITY VALUE INFORMATION FOR
PERFLUOROBUTANOIC ACID (PFBA), PERFLUOROHEXANOIC ACID (PFHXA),
PERFLUOROHEXANESULFONATE (PFHXS), PERFLUORONONANOIC ACID (PFNA), AND
PERFLUORODECANOIC ACID (PFDA) A-l
ADDENDUM B. SEARCH AND SCREENING STRATEGIES B-l
This document is a draft for review purposes only and does not constitute Agency policy.
v DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
TABLES
Table 2-1. Predicted or experimental physiochemical property values for the per- and
polyfluoroalkyl substances (PFAS) being assessed (see
https://comptox.epa.gov/dashboard/) 2-4
Table 2-2. Levels of the per- and polyfluoroalkyl substances (PFAS) being assessed in
environmental media at 10 military installations 2-10
Table 2-3. Levels of the per- and polyfluoroalkyl substances (PFAS) being assessed in water, soil,
and air at National Priorities List sites 2-10
Table 2-4. Serum concentrations of the per- and polyfluoroalkyl substances (PFAS) being
assessed based on National Health and Nutrition Examination Survey (NHANES)
2013-2014 data (jig/L) 2-11
Table 2-5. Environmental Protection Agency (EPA) considerations for the selection of per- and
polyfluoroalkyl substances (PFAS) for evaluation 2-17
Table 2-6. Potential Environmental Protection Agency (EPA) needs and applications for five
per- and polyfluoroalkyl substances (PFAS) 2-19
Table 2-7. Preliminary serum half-life estimates of five per- and polyfluoroalkyl substances
(PFAS) across species and sexes 2-25
Table 3-1. Populations, exposures, comparators, and outcomes (PECO) criteria 3-4
Table 5-1. Epidemiology outcome grouping categories 5-2
Table 5-2. Animal endpoint grouping categories 5-4
Table 6-1. Questions and criteria for evaluating each domain in epidemiology studies 6-7
Table 6-2. Criteria for evaluating exposure measurement in epidemiology studies of per- and
polyfluoroalkyl substances (PFAS) and health effects 6-17
Table 6-3. Considerations to evaluate domains from animal toxicological studies 6-20
Table 7-1. Querying the evidence to organize syntheses for human and animal evidence 7-2
Table 9-1. Information most relevant to describing primary considerations for assessing
causality during evidence syntheses 9-3
Table 9-2. Individual and social factors that may increase susceptibility to exposure-related
health effects 9-6
Table 9-3. Examples of questions and considerations that can trigger focused analysis and
synthesis of mechanistic information 9-17
Table 10-1. Evidence profile table template 10-4
Table 10-2. Considerations that inform evaluations of the strength of the human and animal
evidence 10-7
Table 10-3. Evidence integration judgments for characterizing potential human health hazards
in the evidence integration narrative 10-12
Table 11-1. Attributes used to evaluate studies for deriving toxicity values 11-4
Table 12-1. Topic areas of public comments on the protocol and how comments were
addressed in this update (generally ordered based on descending number of
comments on the topic areas) 12-14
Table A-l. Details on derivation of the available health effect reference values for inhalation
exposure to selected per- and polyfluoroalkyl substances (PFAS) (current as of
June 2019; please consult source references for up-to-date information) 1
Table A-2. Details on derivation of the available health effect reference values for oral exposure
to selected per- and polyfluoroalkyl substances (PFAS) (current as of June 2019;
please consult source references for up-to-date information) 4
This document is a draft for review purposes only and does not constitute Agency policy.
vi DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table A-3. Details on derivation of PFOA and PFOS reference values which served as the basis
for values for the five per- and polyfluoroalkyl substances (PFAS) of interest
(current as of June 2019; please consult source references for up-to-date
information) 7
Table B-l. Perfluorobutanoic acid (PFBA) database search strategy 1
Table B-2. Perfluorodecanoic acid (PFDA) database search strategy 4
Table B-3. Perfluorononanoic acid (PFNA) database search strategy 7
Table B-4. Perfluorohexanoic acid (PFHxA) database search strategy 10
Table B-5. Perfluorohexanesulfonate (PFHxS) database search strategy 12
Table B-6. Title/abstract-level screening criteria for the initial literature searches 15
Table B-7. Example DistillerSR form questions to be used for title/abstract and full text-level
screening for literature search updates from 2019 17
This document is a draft for review purposes only and does not constitute Agency policy.
vii DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
FIGURES
Figure 2-1. Chemical structures of the per- and polyfluoroalkyl substances (PFAS) being
assessed 2-4
Figure 2-2. Existing oral reference values for (a) perfluorobutanoic acid (PFBA), (b)
perfluorohexanoic acid (PFHxA), (c) perfluorohexanesulfonate (PFHxS), (d)
perfluorononanoic acid (PFNA), and (e) perfluorodecanoic acid (PFDA).
Abbreviations and additional details on the derivation of the values can be
found in Addendum A 2-16
Figure 2-3. Results of a preliminary literature inventory of five per- and polyfluoroalkyl
substances (PFAS). Data are approximated based on a cursory review of the
literature search results for studies published through 2018 (see Section 4 for
details; this includes at-the-time-unpublished reports from NTP, see
Section 4.1). Health effects are based on groupings from the EPA's Integrated
Risk Information System (IRIS) website
(https://cfpub.epa.gov/ncea/iris/search/index.cfm).a For this summary,
metabolic effects are captured under "other" and "hepatic" includes lipid and
lipoprotein measures 2-21
Figure 4-1. Literature flow diagrams for PFBA and its ammonium salt (a), PFHxA and its
ammonium and sodium salts (b), PFHxS and its potassium salt (c), PFNA and its
ammonium and sodium salts (d), and PFDA and its ammonium and sodium salts
(e) 4-13
Figure 6-1. Overview of Integrated Risk Information System (IRIS) study evaluation process.
(a) An overview of the general evaluation process (note: see Section 5 for
deviations from independent evaluation by two reviewers for some health
outcomes in epidemiology studies), (b) The evaluation domains and definitions
for ratings (i.e., domain and overall judgments, performed on an
outcome-specific basis) 6-2
Figure 6-2. Preliminary mean correlation coefficients across per- and polyfluoroalkyl substances
(PFAS) among studies in the inventory, for all media types 6-19
Figure 9-1. Preliminary proposed mechanistic pathway for per- and polyfluoroalkyl substances
(PFAS)-induced noncancer liver effects. Based on previous reviews of
perfluorooctane sulfonate (PFOS)- and perfluorooctanoic acid (PFOA)-induced
noncancer liver effects in animals (ATSDR, 2018; Li et al., 2017; Viberg and
Eriksson, 2017; U.S. EPA, 2016c, d), and proposed adverse outcome pathways
for hepatic steatosis (Mellor et al., 2016) 9-13
Figure 10-1. Process for evidence integration. Note that "sufficient evidence" could indicate a
judgment of "sufficient evidence for hazard" or "sufficient evidence to judge
that a hazard is unlikely," depending on the nature and extent of the available
evidence (see Table 10-3) 10-2
This document is a draft for review purposes only and does not constitute Agency policy.
viii DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
ABBREVIATIONS
ADME
absorption, distribution, metabolism,
HED
human equivalent dose
and excretion
HERO
Health and Environmental Research
AFFF
aqueous film-forming foam
Online
AK DEC
Alaska Department of Environmental
HFPO
hexafluoropropylene oxide
Conservation
hPPARa
humanized peroxisome
ALT
alanine aminotransferase
proliferator-activated receptor alpha
AOP
adverse outcome pathway
HRL
health risk limit
AST
aspartate aminotransferase
i.p.
intraperitoneal
ATSDR
Agency for Toxic Substances and
IARC
International Agency for Research on
Disease Registry
Cancer
BMDL
benchmark dose lower confidence limit
IPCS
International Programme on Chemical
BMI
body mass index
Safety
BMR
benchmark response
IRIS
Integrated Risk Information System
BW3/4
body-weight scaling to the 3/4 power
IUR
inhalation unit risk
CAR
constitutive androstane receptor
K
potassium
CAS
Chemical Abstracts Service
LDso
median lethal dose
CASRN
Chemical Abstracts Service registry
LOAEL
lowest-observed-adverse-effect level
number
LOD
limit of detection
CBI
confidential business information
MAC
maximum acceptable concentration
CERCLA
Comprehensive Environmental
MCL
maximum contaminant level
Response, Compensation, and Liability
MDH
Minnesota Department of Health
Act
MF
modifying factor
CLa
clearance in animals
MLR
mixed leukocyte reaction
CLh
clearance in humans
MOA
mode of action
CPAD
Chemical and Pollutant Assessment
MPPD
multiple path particle dosimetry
Division
MRL
minimum reporting level
CPHEA
Center for Public Health and
Na
sodium
Environmental Assessment
NAFLD
nonalcoholic fatty liver disease
CPN
chronic progressive nephropathy
ND
no data
CRD
chemical reporting data
NF-kB
nuclear factor kappa B pathway
CTDPH
Connecticut Department of Health
nh4+
ammonium
CTL
cytotoxic T lymphocyte
NHANES
National Health and Nutrition
CWA
Clean Water Act
Examination Survey
DNA
deoxyribonucleic acid
NH DES
New Hampshire Department of
DTH
delayed-type hypersensitivity
Environmental Services
DWEL
drinking water equivalent level
NJDEP
New Jersey Department of
ECHA
European Chemicals Agency
Environmental Protection
EFSA
European Food Safety Authority
NMD
normalized mean difference
EPA
Environmental Protection Agency
NOAEL
no-observed-adverse-effect level
FDA
Food and Drug Agency
NPDWR
National Primary Drinking Water
FIFRA
Federal Insecticide, Fungicide, and
Regulation
Rodenticide Act
NPL
National Priorities List
FOB
functional operational battery
NR
nuclear receptor
FXR
farnesoid X receptor
NTP
National Toxicology Program
GLP
good laboratory practice
OCSPP
Office of Chemical Safety and Pollution
GRADE
Grading of Recommendations
Prevention
Assessment, Development, and
OECD
Organisation for Economic
Evaluation
Co-operation and Development
HA
health advisory
OLEM
Office of Land and Emergency
HAWC
Health Assessment Workspace
Management
Collaborative
OR
odds ratio
This document is a draft for review purposes only and does not constitute Agency policy.
ix DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
ORD
Office of Research and Development
RfC
inhalation reference concentration
OSF
oral slope factor
RfD
oral reference dose
OW
Office of Water
ROBINS-I
Risk of Bias in Nonrandomized Studies
PAC
protective action criteria
of Interventions
PBPK
physiologically based pharmacokinetic
ROS
reactive oxygen species
PBTK
physiologically based toxicokinetic
RXR
retinoid X receptor
PCL
protective concentration level
SD
standard deviation
PECO
populations, exposures, comparators,
SDWA
Safe Drinking Water Act
and outcomes
tl/2A
elimination half-life in animals
PFAS
per- and polyfluoroalkyl substances
tl/2H
elimination half-life in humans
PFBA
perfluorobutanoic acid
TCEQ
Texas Commission on Environmental
PFBS
perfluorobutane sulfonate
Quality
PFCA
perfluoroalkyl carboxylic acid
TD
toxicodynamic
PFDA
perfluorodecanoic acid
TDI
tolerable daily intake
PFHxA
perfluorohexanoic acid
TEEL
temporary emergency exposure limit
PFHxS
perfluorohexanesulfonate
TNFa
tumor necrosis factor alpha
PFNA
perfluorononanoic acid
TRI
Toxics Release Inventory
PFOA
perfluorooctanoic acid
TSCA
Toxic Substances Control Act
PFOS
perfluorooctane sulfonate
TSCATS
Toxic Substances Control Act Test
PFSA
perfluoroalkane sulfonic acid
Submissions
PI3K-Akt
phosphatidylinositol-3-kinase-
UCMR
Unregulated Contaminant Monitoring
serine/threonine kinase Akt
Rule
PK
pharmacokinetic
UF
uncertainty factor
POD
point of departure
1
UFa
animal-to-human uncertainty factor
PPARa
peroxisome proliferator-activated
2
UFc
composite uncertainty factor
receptor alpha
3
UFd
database deficiencies uncertainty factor
PPRTV
Provisional Peer-Reviewed Toxicity
4
UFh
human variation uncertainty factor
Value
5
UFl
LOAEL-to-NOAEL uncertainty factor
PR
preliminary review
6
UFs
subchronic-to-chronic uncertainty
pt.
point
7
factor
PVDF
polyvinylidene fluoride
Vd
volume of distribution
PWS
public water system
WHO
World Health Organization
PXR
pregnane X receptor
wt.
weight
RCRA
Resource Conservation and Recovery
XME
xenobiotic metabolizing enzymes
Act
This document is a draft for review purposes only and does not constitute Agency policy.
ii DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
AUTHORS | CONTRIBUTORS | REVIEWERS
Assessment Teams
Michelle Angrish (PFHxA co-lead) U.S. EPA/ORD/CPHEA
Xabier Arzuaga (PFHxS co-lead)
Thomas Bateson
Johanna Congleton (PFNA co-lead)
Allen Davis (PFBA and statistics co-lead)
Laura Dishaw (PFHxA co-lead)
Ingrid L. Druwe (PFHxS co-lead)
J. Phillip Kaiser (PFDA co-lead)
Andrew Kraft (IRIS PFAS assessment team
lead)
Lucina Lizarraga (PFDA co-lead)
Pamela Noyes (PFNA co-lead)
Elizabeth Radke (epidemiology lead)
Kristen Rappazzo
Alan Sasso
Paul Schlosser (PBPK and ADME lead)
Michele Taylor (PFBA co-lead)
Michael Wright
Jay Zhao (statistics co-lead)
Technical Experts/Contributors
Audrey Galizia U.S. EPA/ORD/CPHEA
Kelly Garcia (Oak Ridge Associated
Universities [ORAU] contractor, no longer
with EPA)
Carolyn Gigot (ORAU contractor, no longer
with EPA)
Andrew Greenhalgh (ORAU contractor, no
longer with EPA)
Belinda Hawkins
Amanda Persad
Linda Phillips (retired)
Brittany Schulz (ORAU contractor)
Andre Weaver
Scott Wesselkamper (no longer with EPA)
Amina Wilkins
George Woodall
This document is a draft for review purposes only and does not constitute Agency policy.
ii DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Technical Experts/Contributors
John Bucher National Toxicology Program/Office of Health Assessment and
Translation
Andrew Rooney National Toxicology Program/Office of Health Assessment and
Translation/Director
Kyla Taylor National Toxicology Program/Office of Health Assessment and
Translation
Executive Direction
Tina Bahadori
Samantha Jones (former PFAS assessment
team lead [prior to 2019])
Andrew Kraft (PFAS IRIS assessment team
lead)
Kristina Thayer
U.S. EPA (National Center for Environmental Assessment
[NCEA1] Director during protocol development)
U.S. EPA/CPHEA/Associate Director
U.S. EPA/CPHEA/CPAD/Senior Science Advisor
U.S. EPA/CPHEA/Chemical and Pollutant Assessment Division
(CPAD)/Director
Production Team and Review
Anna Chaplin U.S. EPA/ORD/CPHEA
Madison Feshuk (ORAU contractor)
Catherine Gibbons
Hillary Hollinger (ORAU contractor)
Ryan Jones
Jennifer Nichols
Dahnish Shams
Vicki Soto
1
1NCEA was reorganized (largely into CPHEA) during the 2019 Office of Research and Development reorganization.
This document is a draft for review purposes only and does not constitute Agency policy.
iii DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
1. INTRODUCTION
Per- and polyfluoroalkyl substances (PFAS) are a large class of synthetic (man-made)
chemicals widely used in consumer products and industrial processes. The basic structure of PFAS
consists of a carbon chain surrounded by fluorine atoms, with different chemicals possessing
different end groups (see examples in Section 2.1.1); thousands of distinct PFAS exist in commerce.
To help address this complex issue, the Environmental Protection Agency (EPA) is taking a
proactive approach. Specifically, the development of human health toxicity assessments for
exposure to individual PFAS represents only one component of the broader PFAS action plan
underway at the EPA (https: //www.epa.gov/pfas/epas-pfas-action-plan). The five toxicity
assessments being developed according to the scope and methods outlined in this protocol build
upon several other PFAS assessments that have already been developed (see Section 2.1.7).
This protocol document presents the methods for conducting the systematic reviews and
dose-response analyses for assessments of perfluorodecanoic acid (PFDA), perfluorononanoic acid
(PFNA), perfluorohexanoic acid (PFHxA), perfluorohexanesulfonate (PFHxS), and
perfluorobutanoic acid (PFBA), and their related salts (see Figure 2-1). This includes a summary of
why these specific PFAS were prioritized for evaluation, description of the objectives and specific
aims of the assessments, draft populations, exposures, comparators, and outcomes (PECO) criteria,
and identification of key areas of scientific complexity. This assessment protocol will be posted on
the Integrated Risk Information System (IRIS) website fhttps: //cfpub.epa.gov/ncea/iris2 /atoz.cfm]
for a 45-day comment period. The protocol will also be published in the Zenodo data repository
fhttps://zenodo.orgA Public input received on the protocol is considered during preparation of
the draft assessments, and any adjustments made to the protocol will be reflected in an updated
version released in conjunction with the draft assessments. The literature search results for these
five PFAS will also be posted to the Health and Environmental Research Online (HERO) database2
upon public release of the protocol (the literature search results will be regularly updated during
draft development and the subsequent stages of assessment review).
2PFBA: https://hero.epa.gov/hero/index.cfm/proiect/page/proiect id/2632
PFHxA: https://hero.epa.gov/hero/index.cfm/proiect/page/proiect id/2628
PFHxS: https://hero.epa.gov/hero/index.cfm/proiect/page/proiect id/2630
PFNA: https://hero.epa.gov/hero/index.cfm/proiect/page/proiect id/2633
PFDA: https://hero.epa.gov/hero/index.cfm/proiect/page/proiect id/2614.
This document is a draft for review purposes only and does not constitute Agency policy.
1-1 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
2.SCOPING AND PROBLEM FORMULATION
SUMMARY
2.1. SUMMARY OF BACKGROUND INFORMATION
Section 2.1 provides a summary of background information for contextual purposes only.
These brief overviews emphasize reviews and other summary information (e.g., in public
databases) and are not intended to be comprehensive descriptions of the available information. In
addition, the information in this section (developed in 2019) is notupdated and thus may not
represent the current state of the science at the time of review. The reader is encouraged to refer to
the source materials and other updated information for current PFAS-specific details. The
information in this section is not recommended for use in decision making.
2.1.1. Chemical and Physical Properties
Perfluorodecanoic acid (PFDA; CASRN 335-76-2), perfluorononanoic acid (PFNA;
CASRN 375-95-1), perfluorohexanoic acid (PFHxA, CASRN 307-24-4), perfluorohexanesulfonic acid
(PFHxS, CASRN 355-46-4), and perfluorobutanoic acid (PFBA, CASRN 375-22-4), and their related
salts, are all PFAS. Section 2.2 ("Scoping Summary") outlines the rationale for why these PFAS were
prioritized for assessment No single, consensus definition of PFAS exists. Bucketal. (2011)
defined PFAS as fluorinated substances that "contain 1 or more C atoms on which all the H
substituents (present in the nonfluorinated analogues from which they are notionally derived) have
been replaced by F atoms, in such a manner that they contain the perfluoroalkyl moiety (CnF2n+r)."
The definition in the EPA Chemistry Dashboard, which (as of late 2019) yields over 6,600 PFAS
structures (https://comptox.epa.gov/dashboard/chemical lists/PFASTRUCT). includes all
substances for which "the structure contains the substructure RCF2CFR'R" (R cannot be H)"; the
Dashboard defines this substructure as "general enough to encompass the largest set of structures
having sufficient levels of fluorination to potentially impart PFAS-type properties." Regardless of
the definition used, the PFAS being assessed in association with this protocol are members of a
subset of PFAS called perfluoroalkyl acids (PFAAs; PF0A and PF0S are also members), which
consist of a carbon backbone (typically 4-14 C atoms) that is fully fluorinated and bonded to a
charged functional group [e.g., carboxylic acid, sulfonic acid, or phosphonic acid; Lau etal. (2007)].
More specifically, PFDA, PFNA, PFHxA, and PFBA are classified as perfluoroalkyl carboxylic acids
(PFCAs), and PFHxS is a perfluoroalkane sulfonic acid [PFSA; OECD (2015)]. PFCAs containing
seven or more perfluorinated carbon groups and PFSAs containing six or more perfluorinated
carbon units are considered long-chain PFAS fATSDR. 2018: OECD. 2015: Buck etal.. 20111. Thus,
PFDA, PFNA, and PFHxS are considered long-chain, and PFHxA and PFBA are short-chain. To
This document is a draft for review purposes only and does not constitute Agency policy.
2-2 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
simplify the terminology used throughout this protocol and the subsequent assessments, PFBA,
PFHxA, PFHxS, PFNA, and PFDA (and their salts) are referred to using the broad and more
recognizable term, PFAS, rather than using the more specific terms PFAAs, PFSAs, or PFCAs. The
chemical structures of PFDA, PFNA, PFHxA, PFHxS, and PFBA, and their related salts are shown in
Figure 2-1 (along with their CASRNs), and estimated or experimental values for their
physiochemical properties are provided in Table 2-1. Importantly, these values are intended for
general context and may no longer be accurate or current at the time of review and should not be
used for any purpose other than conveying generalities around physicochemical properties. For
example, even though the logP may be difficult to predict, the possibility that PFAS exist in the
ionized and nonionized fBeesoon et al.. 20121 form cannot be ignored and understanding the PFAS
dissociation and partitioning constants are important for understanding how PFAS interact with
the environment.
This document is a draft for review purposes only and does not constitute Agency policy.
2-3 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
PFDA
335-76-2
F'Y F
PFNA
375-95-1
PFHxA
307-24-4
PFHxS
355-46-4
O
OH
f-ff
PFBA
375-22-4
FFFFFFFFF
OH
NHf F
F F F
PFDA
sodium salt
3830-45-3
PFNA
sodium salt
21049-39-8
PFHxA
sodium salt
2923-26-4
PFBA
ammonium salt
10495-86-0
PFDA
ammonium salt
3108-42-7
0*^0
PFNA
ammonium salt
4149-60-4
NH< 0
PFHxA
ammonium salt
21615-47-4
F F F F F F P
F F F F F F 0
PFHxS
potassium salt
3871-99-6
Figure 2-1. Chemical structures of the per- and polyfluoroalkyl substances
(PFAS) being assessed.
Table 2-1. Predicted or experimental physiochemical property values for the
per- and polyfluoroalkyl substances (PFAS) being assessed (see
https: / / comptox.epa.gov/dashboard/)
Property
(unit)
PFDA + salts
PFNA + salts
PFHxA
PFHxS + salts
PFBA + salts
PFDA3
nh4+
saltb
Na
salt0
PFNAd
nh4+
salt6
Na
saltf
PFHxA®
PFHxSh
K
salt1
PFBAj
nh4+
saltk
Molecular wt.
(g/mol)
514
531
536
464
481
486
314
400
438
214
ND
Melting pt. (°C)
79.5
165*
71.4*
66.5
165*
ND
14.0
190
273
-17.5
ND
Boiling pt. (°C)
218
205*
205*
218
190*
196
157
345
236*
121
ND
This document is a draft for review purposes only and does not constitute Agency policy.
2-4 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Property
(unit)
PFDA + salts
PFNA + salts
PFHxA
PFHxS + salts
PFBA + salts
PFDA3
nh4+
salt"
Na
salt0
PFNAd
nh4+
salt6
Na
saltf
PFHxA®
PFHxSh
K
salt1
PFBAj
nh4+
saltk
Density (g/cm3)
1.79*
ND
ND
1.78*
ND
ND
1.69*
1.84*
1.84*
1.65
ND
Vapor pressure
(mm Hg)
3.89 x
10-2*
3.51 x
10-2*
3.51 x
10-2*
7.06 x
10-2*
7.38 x
10-2*
1.71 x
10"1
9.08 x
10"1
9.95 x
10-s*
9.95 x
10-s*
6.37
ND
Henry's law
constant
(atm-m3/mole)
3.55 x
10"10*
3.55 x
10"10*
3.55 x
10"10*
1.64 x
10"10*
1.64 x
10"10*
ND
2.51 x
10"10*
1.96 x
10"10*
1.96 x
10"10*
4.99 x
10-s*
ND
Water solubility
(mol/L)
8.3 x
10"6*
8.41 x
10-i*
8.41 x
10-i*
1.38 x
10-s*
8.45 x
10-i*
ND
9.39 x
10"5
6.08 x
10"4
2.96 x
10"4*
2.09 x
10"3
ND
pKa
-0.17*
ND
ND
-0.17*
ND
ND
-0.16
0.14*
ND
0.08*
ND
LogP
7.37*
4.36*
4.36*
6.59*
3.12*
ND
2.51
2.20
2.97*
1.43
ND
Soil adsorption
coefficient (L/kg)
397*
397*
397*
2,830*
2,830*
ND
1,070*
2,300*
2,300*
47.9*
ND
Bioconcentration
factor
789*
29.8*
29.8*
752*
4.95*
ND
41*
118*
5.94*
7.61*
ND
K = potassium; ND = no data; NH4+ = ammonium; Na = sodium; pt. = point; wt. = weight.
aCASRN 335-76-2. U.S. EPA (2018a) Chemistry Dashboard (https://comptox.epa.gov/dashboard/; search = PFDA)
for all values except pKa (ATSDR, 2018). Values are median experimental values (when available), or median or
average predicted values.
bCASRN 3108-42-7. Predicted average values from the U.S. EPA (2018a) Chemistry Dashboard (search = 3108-42-7)
CCASRN 3830-45-3. Predicted average values from the U.S. EPA (2018a) Chemistry Dashboard (search = 3830-45-3)
dCASRN 375-95-1. U.S. EPA (2018a) Chemistry Dashboard (search = PFNA) for all values except pKa (NLM, 2013).
Values are median experimental values (when available), or median or average predicted values.
eCASRN 4149-60-4. Predicted average values from the U.S. EPA (2018a) Chemistry Dashboard (search = 4149-60-4)
fCASRN 21049-39-8. ChemNet website:
http://www.chemnet.com/cas/en/21049-39-8/sodium-heptadecafluorononanoate.html for all values.
gCASRN: 307-24-4. U.S. EPA (2018a) Chemistry Dashboard (search = 307-24-4) for all values except pKa (NLM,
2016). Values are median or average experimental values (when available) or predicted values.
hCASRN 355-46-4. U.S. EPA (2018a) Chemistry Dashboard (search = 355-46-4) for all values except pKa (NLM,
2017). Values are median or average experimental values (when available) or predicted values.
'CASRN 3871-99-6. U.S. EPA (2018a) Chemistry Dashboard (search = 3871-99-6) for all values. Values are median
or average experimental values (when available) or predicted values.
JCASRN 375-22-4. U.S. EPA (2018a) Chemistry Dashboard (search = 375-22-4)
https://comptox.epa.gov/dashboard/dsstoxdb/results?utf8=%E2%9C%93&search=375-22-4 for all values except
pKa (ATSDR, 2018). Values are median or average experimental values (when available) or predicted values.
kCASRN 10495-86-0.
* Predicted value.
2.1.2. Sources, Production, and Use
1 PFAS are synthetic (man-made) compounds that have been used since the 1940s in
2 consumer products and industrial applications because of their resistance to heat, oil, stains,
3 grease, and water. They have been used in stain-resistant fabrics for clothing, carpets, and
This document is a draft for review purposes only and does not constitute Agency policy.
2-5 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
furniture; nonstick cookware; food packaging (e.g., popcorn bags, and fast-food containers); and
personal care products [e.g., dental floss, cosmetics, and sunscreen; ATSDR f20181]. Some PFAS
have also been used in firefighting foam and as industrial surfactants, emulsifiers, wetting agents,
additives, and coatings, and in the aerospace, automotive, building, and construction industries to
help reduce friction f ATSDR. 20181. Because of the widespread use of PFAS and their persistence in
the environment, most people in the United States have been exposed to them (see
https: //www.epa.gov/pfas/epas-pfas-action-plan for additional details). Although not exhaustive,
the bulleted list below provides some examples of how the five PFAS of interest have been used:
• PFDA has been used in stain and grease-proof coatings on food packaging, furniture,
upholstery, and carpet (Harbison et al.. 2015). and as a lubricant, wetting agent, plasticizer,
and corrosion inhibitor (Keml. 2015).
• PFNA has been used as a processing aid in the production of fluoropolymers, primarily
polyvinylidene fluoride (PVDF), which is a plastic designed to be temperature resistant and
chemically nonreactive fNTDWOI. 2017: Prevedouros etal.. 20061. It has also been used in
aqueous film-forming foam (AFFF) for fire suppression (Laitinen et al.. 2014).
• PFHxA is not currently a commercial product; it is a breakdown product of "stain- and
grease-proof coatings on food packaging and household products" (NTP. 2018b).
• PFHxS has been used as a surfactant to make fluoropolymers, and in water- and
stain-protective coatings for carpets, paper, and textiles (NTP. 2018a). It may also be
present in certain industrial and consumer products, such as "food-contact papers,
water-proofing agents, cleaning and polishing products either for intentional uses (as
surfactants or surface protection agents) or as unintentional impurities from industrial
production processes" (Norwegian Environment Agency. 2018). It has also been used in
AFFF for fire suppression (Laitinen et al.. 2014).
• PFBA is a breakdown product of other PFAS that are used in stain-resistant fabrics, paper
food packaging, and carpets; it was also used for manufacturing photographic film (MDH.
20091.
The U.S. Environmental Protection Agency (EPA) has been working with companies in the
fluorochemical industry since the early 2000s to phase outthe production and use of long-chain
PFAS (https://www.epa.gov/assessing-and-managing-
chemicals-under-tsca/risk-management-and-polyfluoroalkyl-substances-PFAS). Although
production of long-chain PFAS in Western Europe and Japan has declined (OECD. 2015). their
production in emerging economies in Asia (China and India) has increased (OECD. 2015). Given the
past production and use of these PFAS in some regions, and the increased production and use in
others, PFAS have been and are being released to the environment through various waste streams
(NLM. 2016. 2013). Also, because precursor products (e.g., fluorotelomer alcohols) or products
containing PFAS are still in use, they continue to be a source of environmental PFAS contamination
This document is a draft for review purposes only and does not constitute Agency policy.
2-6 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
through their disposal and subsequent breakdown into PFAS in the environment (Kim and Kannan.
20071.
Chemical reporting data (CRD) on production volumes are not available in EPA's ChemView
fU.S. EPA. 20191 for PFDA, PFNA, PFHxA, PFHxS, PFBA, or their salts. Also, because there are no
requirements to report releases to the environment from facilities manufacturing, processing, or
otherwise using PFAS, quantitative information is not available in EPA's Toxics Release Inventory
[TRI; ATSDRC20181: U.S. EPA C201911.
2.1.3. Environmental Fate and Transport
PFAS are very stable and persistent in the environment (ATSDR. 20181. and many are found
worldwide in the air, water, and soil and in the tissues of plants, animals, and humans
f https://www.epa. gov/assessi ng-and-
managing-chemicals-under-tsca/risk-management-and-polyfluoroalkyl-substances-PFAS). They
have been detected at a variety of sites, including private and federal facilities, and have been
associated with various sources, including AFFF, chrome-plating facilities, PFAS manufacturers, and
industries that use PFAS [e.g., textiles; ATSDR f20181]. The environmental fate and transport of
PFAS potentially includes releases to air to soil and surficial water bodies which can then lead to
migration to subsurface soils and ground water contamination [Guelfo etal. f20181:
https://www.atsdr.cdc.gov/pfas/index.html].
Some PFAS (PFNA, PFHxA, PFHxS) released to air are expected to exist solely in the vapor
phase given their vapor pressures (NLM. 2017. 2016. 2013: Kim and Kannan. 20071. although
particle-bound concentrations have also been measured for PFNA and PFDA (Kim and Kannan.
20071. Although vapor-phase PFAS are not susceptible to direct photolysis by sunlight fNLM. 2017.
2016. 20131 and are generally resistant to photo-oxidation (ATSDR. 20181. they can be degraded by
reaction with photochemically produced hydroxyl radicals fNLM. 2017. 2016. 20131. The
atmospheric half-life for these reactions is estimated to be 31 days for PFNA and PFHxA, and
115 days for PFHxS (NLM. 2017. 2016. 2013). Long-range atmospheric transport of PFAS is
possible, as indicated by the detection of PFHxS in remote arctic and marine air samples (NLM.
2017). Wet and dry deposition are potential removal processes for particle-bound PFAS in air
[e.g., to surface water or soil; ATSDR f20181]. Standardized analytical methods for measuring these
five PFAS in ambient air is an area of ongoing research.
In soil, the mobility of PFAS will vary depending on their soil adsorption coefficients (see
Table 2-1), with PFNA predicted to be the least mobile and PFBA the most mobile of the five PFAS
addressed here. Volatilization of PFNA, PFHxA, and PFHxS from moist soil is not expected to be an
important transport process (NLM. 2017. 2016. 2013). Uptake of soil PFAS to plants can occur
(ATSDR. 2018). Yoo etal. (2011) estimated grass-soil accumulation factors (grass concentration
divided by soil concentration) of 3.4, 0.12, and 0.10 for PFHxA, PFNA, and PFDA, respectively, based
on samples collected from a site with bio-solids-amended soil. Zhao etal. f 20161 observed that
This document is a draft for review purposes only and does not constitute Agency policy.
2-7 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
shorter chain PFAS like PFBA were transported more readily from the roots to the shoots of wheat
plants than longer chain PFAS.
PFNA, PFHxA, and PFHxS are expected to adsorb to suspended solids and sediments in
water fNLM. 2017. 2016. 20131. The potential for PFAS to bioaccumulate in aquatic organisms can
be generally assessed using the predicted bioconcentration factors, with the predicted potential for
PFDA and PFNA to bioaccumulate being high compared with PFHxA, PFHxS, and PFBA (see
Table 2-1). Note, however, that these predicted values may vary over a wide range depending on
the variables (i.e., species, habitat, etc.). As described in Section 2.2, standardized analytical
methods for measuring these five PFAS in drinking water exist (for four of the five PFAS to be
assessed) or are under development (i.e., for PFBA). Standardized nondrinking water methods are
currently under development
2.1.4. Environmental Concentrations
PFDA, PFNA, PFHxA, PFHxS, and PFBA have not been evaluated under the National Air
Toxics Assessment program (https: //www.epa.gov/national-air-toxics-assessment). However,
PFDA, PFNA, and PFHxS were measured at concentrations ranging from below the limit of detection
(L0D) to 1.56 pg/m3 in the vapor and particle phases of air samples collected from an urban area of
Albany, NY in 2006 fKim and Kannan. 20071. PFAS have also been measured in indoor air and dust,
and they may be associated with the indoor use of consumer products such as PFAS-treated carpets
or other textiles (ATSDR. 2018). For example, Kato etal. (2009) analyzed dust samples collected
from 39 homes in the United States, United Kingdom, Germany, and Australia for PFAS, including
PFDA, PFNA, PFHxA, and PFHxS. These PFAS were detected in 38.5, 25.6, 46.2, and 79.5% of the
samples, respectively. Likewise, Strvnar and Lindstrom f20081 analyzed dust samples from
110 homes and 10 day care centers in North Carolina and Ohio, and detected PFDA, PFNA, and
PFHxA in 30.4, 42.9, and 92.9% of the samples, respectively. Indoor air samples (n = 4) from a town
in Norway had mean concentrations of 3.4 pg/m3 for PFDA, 2.7 pg/m3 for PFNA, and <4.1 pg/m3 for
PFHxS (Barber et al.. 2007).
The levels of PFAS in soil and sediment surrounding perfluorochemical industrial facilities
have been measured at concentrations ranging from less than the LOD to 124 ng/g for PFBA and
less than the LOD to 3,470 ng/g for PFHxS fATSDR. 20181. PFDA, PFNA, PFHxA, PFHxS, and PFBA
were also detected at an Australian training ground where AFFFs had been used (Baduel etal..
20151. PFDA, PFNA, PFHxA, PFHxS, and PFBA were detected at 10 U.S. military sites in 67.0, 71.4,
70.3, 76.9, and 38.5% of the surface soil samples, respectively, and 48.5,12.1, 63.6, 72.7, and 24.2%
of the sediment samples, respectively (ATSDR. 2018). Table 2-2 shows the concentrations of these
PFAS in soil and sediment at these military sites.
EPA conducted monitoring for several PFAS in drinking water as part of the third
Unregulated Contaminant Monitoring Rule [UCMR; U.S. EPA f2016el]. Under the UCMR, all public
water systems (PWSs) serving more than 10,000 people and a representative sample of 800 PWSs
serving 10,000 or fewer people were monitored for 30 unregulated contaminants between
This document is a draft for review purposes only and does not constitute Agency policy.
2-8 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
January 2013 and December 2015. PFNA and PFHxS were among the 30 contaminants monitored.
PFNA was detected above the minimum reporting level (MRL) of 0.02 |ig/L in 14 of the 4,920 PWSs
tested and in 19 of the 36,972 samples collected. PFNA was also detected above the MRL
(0.096 |ig/L) in groundwater near an industrial site in New Jersey fPost etal.. 20131. PFHxS was
detected above the MRL of 0.03 |ig/L in 55 of the 4,920 PWSs tested and in 207 of the
36,971 samples collected. UCMR data were not available for PFDA, PFHxA, or PFBA. However,
samples from seven municipal wells in Oakdale, MN were analyzed for PFHxA and PFBA. The
concentrations ranged from <0.025 to 0.235 |ig/L and 0.0855 to 2.04 |ig/L, respectively (U.S. EPA.
2017b). Kim and Kannan f20071 analyzed lake water, rainwater, snow, and surface water from
Albany, NY, and reported concentrations of PFDA, PFNA, and PFHxS ranging from less than the LOD
to 0.0135 |ig/L. PFAS were detected at higher concentrations in groundwater samples from an
industrial site (3M Cottage Grove) in Minnesota. PFHxS and PFBA were detected in all seven wells
that were sampled at concentrations ranging from 6.47-40 |ig/L and 23.3-318 |ig/L, respectively
[WS (2007) as cited in ATSDR (2018)]. The concentrations of these five PFAS measured at National
Priorities List (NPL) sites are shown in Table 2-3 as reported in ATSDR (2018). and the
concentrations of PFAS measured in surface water and groundwater at 10 military installations are
given in Table 2-2 as reported in ATSDR f2018I
This document is a draft for review purposes only and does not constitute Agency policy.
2-9 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 2-2. Levels of the per- and polyfluoroalkyl substances (PFAS) being
assessed in environmental media at 10 military installations
Media
Measure
PFDA
PFNA
PFHxA
PFHxS
PFBA
Surface soil
Frequency of detection (%)
67.03
71.43
70.33
76.92
38.46
Reporting limit (ng/kg)
0.28
0.23
0.16
0.29
0.12
Median (ng/kg)
0.980
1.30
1.75
5.70
1.00
Maximum (ng/kg)
15.0
23.0
51.0
1,300
31.0
Subsurface soil
Frequency of detection (%)
12.50
14.42
65.38
59.62
29.81
Reporting limit (ng/kg)
0.30
0.24
0.16
0.31
0.13
Median (ng/kg)
1.40
1.50
1.04
4.40
0.960
Maximum (ng/kg)
9.40
6.49
140
520
14.0
Sediment
Frequency of detection (%)
48.48
12.12
63.64
72.73
24.24
Reporting limit (ng/kg)
0.46
0.38
0.26
0.48
0.21
Median (ng/kg)
1.90
1.10
1.70
9.10
1.70
Maximum (ng/kg)
59.0
59.0
710
2,700
140
Surface water
Frequency of detection (%)
52.00
36.00
96.00
88.00
84.00
Reporting limit (ng/L)
0.008
0.017
0.003
0.007
0.010
Median (ng/L)
0.067
0.096
0.320
0.710
0.076
Maximum (ng/L)
3.20
10.0
292
815
110
Groundwater
Frequency of detection (%)
34.78
46.38
94.20
94.93
85.51
Reporting limit (ng/L)
0.008
0.018
0.003
0.007
0.010
Median (ng/L)
0.023
0.105
0.820
0.870
0.180
Maximum (ng/L)
1.80
3.00
120
290
64.0
Source: ATS PR (2018).
Table 2-3. Levels of the per- and polyfluoroalkyl substances (PFAS) being
assessed in water, soil, and air at National Priorities List sites
Media
Measure
PFDA
PFNA
PFHxA
PFHxS
PFBA
Water
Median (ppb)
ND
ND
0.25
0.26
2.15
Geometric mean (ppb)
ND
ND
0.10
1.12
1.03
Soil
Median (ppb)
ND
27.2
1,175
5,585
1,600
Geometric mean (ppb)
ND
27.2
1,175
5,585
1,600
Air
Median (ppbv)
ND
ND
ND
ND
ND
Geometric mean (ppbv)
ND
ND
ND
ND
ND
ND = no data.
Source: ATS PR (2018).
1 Schecter etal. f20121 collected 10 samples of 31 food items from five grocery stores in
2 Texas and analyzed them for persistent organic pollutants, including PFDA, PFNA, PFHxA, and
3 PFHxS. PFDA, PFNA, and PFHxA were not detected in any of the foods targeted, and PFHxS was
4 detected in cod fish at a concentration of 0.07 ng/g wet weight. PFAS have been detected in fish
5 from U.S. lakes and rivers with concentrations ranging from less than the limit of quantification to
This document is a draft for review purposes only and does not constitute Agency policy.
2-10 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
15.0 ng/g for PFDA, and <1 to 0.47 ng/g for PFHxS (ATSDR. 2018). Stahl etal. (2014) characterized
PFAS in freshwater fish from 164 U.S. urban river sites and 157 Great Lakes sites. PFDA, PFNA,
PFHxA, PFHxS, and PFBA were detected in 92, 69,15, 45, and 16% of the samples, at maximum
concentrations of 13.0, 9.7, 0.8, 3.5, and 1.3 ng/g, respectively. Apart from fish, overall dietary data
for the United States are limited; however, Schaider etal. f20171 detected PFAS in food packaging
collected from U.S. fast food restaurants. Data from other countries (e.g., South Korea, Brazil, Saudi
Arabia) suggest that these PFAS can sometimes be detected in samples of food products, including
shellfish, dairy products, meats, vegetables, food packaging materials, and water [both tap and
bottled; Heo etal. f20141: Chen etal. f20181: Perez etal. f20141: Moreta and Tena f20141: Surma et
al. f20171]. The relevance of these detects (and the associated PFAS levels) to U.S. products is
unknown. Information on detection limits is available in the referenced studies.
2.1.5. Potential for Human Exposure
The general population may be exposed to PFAS through multiple routes, including
ingestion of drinking water and food, ingestion of dust, hand-to-mouth and dermal transfer in
products and materials containing these chemicals, and inhalation via indoor and outdoor air
(ATSDR. 2018: NLM. 2017. 2013). The oral route of exposure has been considered the most
important exposure route for PFAS in the general population fKlaunig etal.. 20151.
The presence of PFAS in human blood provides evidence of exposure among the general
population. PFAS have been monitored in the human population as part of the National Health and
Nutrition Examination Survey (NHANES). PFDA, PFNA, and PFHxS were measured in serum
samples collected in 2013-2014 from more than 2,000 survey participants. The results of these
analyses are presented in Table 2-4. PFDA and PFNA have also been observed in cord blood and
human milk (ATSDR. 2018). Pinnevetal. (2014) and Papadopoulou etal. (2016) observed
associations between breastfeeding and elevated levels of PFHxS in the blood of children.
Table 2-4. Serum concentrations of the per- and polyfluoroalkyl substances
(PFAS) being assessed based on National Health and Nutrition Examination
Survey (NHANES) 2013-2014 data (pg/L)
Population group
Measure
PFDA
PFNA
PFHxA
PFHxS
PFBA
Total population (n = 2,168)
Geometric mean
0.185
0.675
ND
1.35
ND
50th percentile
0.200
0.700
ND
1.40
ND
95th percentile
0.700
2.00
ND
5.60
ND
3 to 5 yr (n = 181)
Geometric mean
_a
0.764
ND
0.715
ND
50th percentile
0.100
0.620
ND
0.740
ND
95th percentile
0.370
3.49
ND
1.62
ND
6 to 11 yr (n = 458)
Geometric mean
_a
0.809
ND
0.913
ND
50th percentile
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
12 to 19 yr (n = 402)
Geometric mean
0.136
0.599
ND
1.27
ND
50th percentile
0.100
0.500
ND
1.10
ND
95th percentile
0.400
2.00
ND
6.30
ND
20 yr and older (n = 1,766)
Geometric mean
0.193
0.685
ND
1.36
ND
50th percentile
0.200
0.700
ND
1.40
ND
95th percentile
0.800
2.00
ND
5.50
ND
LOD = limit of detection; 0.1 (ng/L); ND = no data.
aNot calculated because the proportion of results below the LOD was considered too high to provide a valid result.
Source: CDC (2018). Fourth National Report on Human Exposure to Environmental Chemicals.
2.1.6. Populations and Lifestages with Potentially Greater Exposures
In addition to exposure scenarios that are expected to apply to the general population (see
Section 2.1.5), certain populations and lifestages may have greater exposures than the general
population. These groups include individuals in occupations that require frequent contact with
PFAS-containing products, such as firefighters or individuals who install and treat carpets (ATSDR.
20181. infants and young children (due to placental transfer, breastfeeding, or their increased hand-
to-mouth behaviors), and populations consuming contaminated drinking water. Rotander et al.
(2015) analyzed serum samples from 149 Australian firefighters at an AFFF training facility. Mean
and median PFHxS concentrations were 10 to 15 times higher than those in the general population
of Australia and Canada. Populations living near fluorochemical facilities where environmental
contamination has occurred may also be more highly exposed fATSDR. 20181. Also, because these
chemicals can be found in ski wax, individuals who engage in professional ski waxing may be more
highly exposed because PFAS such as PFHxA, PFNA, and PFDA in dust or fumes may be inhaled
during this process (Harbison et al.. 2015: Nilsson etal.. 2010a: Nilsson etal.. 2010b). Populations
living near military or airport fire training areas or industrial sites that use or manufacture PFAS
may be more likely to have high-level PFAS exposure through consumption of contaminated
drinking water (Hu etal.. 2016). Further, due to the high water solubility and mobility of PFAS in
groundwater (and lack of current remediation technology at many water treatment facilities) it is
possible for populations consuming drinking water from a contaminated watershed to receive
disproportionate PFAS exposure fSun etal.. 20161.
Populations that rely primarily on seafood for most of their diet, possibly including some
native American tribes (Byrne etal.. 2017). may also be disproportionately exposed. Christensen et
al. (2017) and Haug etal. (2010) used data on serum PFAS levels and 30-day, self-reported fish and
shellfish ingestion rates from NHANES 2007-2014 to explore potential relationships between PFAS
exposures and fish consumption. PFDA, PFNA, and PFHxS were among the PFAS detected in the
serum of at least 30% of the NHANES participants, and after adjusting for demographic
characteristics, total fish consumption was associated with elevated serum PFDE and PFNA.
Shellfish consumption was associated with elevated levels of all the PFAS examined.
This document is a draft for review purposes only and does not constitute Agency policy.
2-12 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
PFAS exposures to fetuses and infants are also important to consider as studies show the
potential for elevated exposures during these sensitive developmental periods. Animal testing
fBeesoon et al.. 2012: Hinderliter et al.. 20051. and human studies [e.g., Fei etal. f20071. Gao et al.
f20161. Mamsenetal. f20191. Mondal etal. f20141. Zhang etal. f2013al] suggest that PFAS cross
the blood-placental barrier with transfer efficiencies in humans that may depend on PFAS chain
length and binding affinity to serum and breastmilk-protein complexes. Studies also show that
breastmilk appears to be an important route of exposure to long-chain PFAS in breastfed infants,
although the extent of lactational transfer of the current long-chain PFAS—PFNA, PFDA, and
PFHxS—is less clear [e.g., Fromme etal. f20101. Haugetal. f20111. Mondal etal. f20141. Mogensen
etal. f20151. Karrmanetal. f20071].
2.1.7. Other Environmental Protection Agency (EPA) Assessments of Per- and
Polyfluoroalkyl Substances (PFAS)
EPA released two PFAS assessments for peer review in 2018. Specifically, the draft
assessments of (1) 2,3,3,3-tetrafluoro-2-(l,l,2,2,3,3,3-heptafluoropropoxy)propanoic acid (also
called hexafluoropropylene oxide [HFPO] dimer acid) (CASRN 13252-13-6) and its ammonium salt
2,3,3,3-tetrafluoro-2-(l,l,2,2,3,3,3-heptafluoropropoxy)propanoate (also called HFPO dimer acid)
(CASRN 62037-80-3), referred to as GenX chemicals, and (2) perfluorobutane sulfonic acid
(CASRN 375-73-5) and its salt potassium perfluorobutane sulfonate (CASRN 29420-49-3) referred
to as PFBS. These assessments summarized the available data on the potential human health
effects of lifetime exposure to these PFAS and included oral reference doses (RfDs), which estimate
(with uncertainty spanning perhaps an order of magnitude) a level of daily oral exposure to the
human population (including sensitive subgroups) that is likely to be without an appreciable risk of
deleterious noncancer health effects during a lifetime, and qualitative descriptions of the
carcinogenic potential of the chemicals. The PFBS assessment updates a Provisional Peer-Reviewed
Toxicity Value (PPRTV) assessment that was developed in support of the Superfund Program and
published in 2014 (PFBS PPRTV 20141. In addition, EPA released Drinking Water Health Advisories
for PFOA and PFOS in 2016, along with health effect support documents (Drinking Water Health
Advisories for PFOA and PFOS1. Health advisories are nonenforceable and nonregulatory
summaries of technical information on contaminants that can cause human health effects and are
known or anticipated to occur in drinking water.
2.1.8. Assessments and Toxicity Values from Other Sources
For the five PFAS addressed in this protocol, a summary of existing human health reference
values from national, international, and state agencies (currentas of March 2019), is provided in
Figure 2-2 (see Addendum A for a tabular summary, including derivation details of the displayed
values). The majority of current reference values are noncancer toxicity values based on oral
exposure studies in rodents, although a few inhalation toxicity values exist (see Table A-l in
Addendum A for more details).
This document is a draft for review purposes only and does not constitute Agency policy.
2-13 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
(a)
PFBA Oral Reference Values
Acute
&0
.§. 0.001
o
Q
Short Term
SMDH RfD
~ TCEQ RfD
Subchronic
Chronic
MDH RfD
jgdD TCEQ RfD
100 1,000
Duration (Days)
L
10,000
100,000
(b)
PFHxA Oral Reference Values
o.ooooi
o
Q
Acute
Short Term
Subchronic
Chronic
TCEQ RfD I
10 100 1,000
Duration (Days)
J-L-LJ-H L
10,000
100,000
This document is a draft for review purposes only and does not constitute Agency policy.
2-14 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
(c)
PFHxS Oral Reference Values
Acute
Short Term
BMDH RfD
~ NHDESRfD
~1TCEQ RfD
~ Australia TDI*
Subchronic
Chronic
Australia TDI* ~
NH DES RfD
®HD TCEQ RfD
10 100 1,000
Duration (Days)
* Applies to the sum of multiple PFAS, including PFHxS
(d)
PFNA Oral Reference Values
Ml
~ 0.00001
0)
O
Q
Acute
Short Term
¦ NHDESRfD
m TCEQ RfD
Subchronic
Chronic
TCEQ RfD M
NH DES RfD
¦
10 100 1,000
Duration (Days)
10,000
100,000
This document is a draft for review purposes only and does not constitute Agency policy.
2-15 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
(e)
PFDA Oral Reference Values
u
.§¦ 0.00001
0
O
o
Acute
Short Term
Subchronic
Chronic
TCEQRfD HD
100 1,000
Duration (Days)
LL-LLH L
10,000
100,000
Figure 2-2. Existing oral reference values for (a) perfluorobutanoic acid
(PFBA), (b) perfluorohexanoic acid (PFHxA), (c) perfluorohexanesulfonate
(PFHxS), (d) perfluorononanoic acid (PFNA), and (e) perfluorodecanoic acid
(PFDA). Abbreviations and additional details on the derivation of the values can be
found in Addendum A.
2.2. SCOPING SUMMARY
1 Given the numerous PFAS of potential interest to the Agency, an extensive scoping effort
2 was undertaken to prioritize PFAS for review. This effort was coordinated across EPA program and
3 regional offices, where staff discussed specific assessment needs as well as the timeliness of those
4 needs. While additional factors were considered during this scoping effort, Table 2-5 summarizes
5 the primary considerations for selecting the five PFAS described in this protocol, as well as two
6 other PFAS that were recently assessed by EPA: PFBS and GenX chemicals
7 (https://www.epa.gov/pfas/genx-and-pfbs-draft-toxicity-assessmentsl. In short, these PFAS:
8
9
were identified as a priority to inform decision making for EPA's Office of Water (0W),
Office of Land and Emergency Management (0LEM), Office of Chemical Safety and Pollution
This document is a draft for review purposes only and does not constitute Agency policy.
2-16 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
1 Prevention (OCSPP), Office of Children's Health Protection (OCHP), EPA's regional offices,
2 tribes, or state departments of environmental protection. Most of these PFAS were a
3 priority for multiple patrons;
4 • had been evaluated in in vivo studies of animals and thus might be used to derive toxicity
5 values; and
6 • had existing (or under development) standardized analytical methods to monitor
7 environmental levels to allow for site-specific application of toxicity values to regulatory
8 decision making.
Table 2-5. Environmental Protection Agency (EPA) considerations for the
selection of per- and polyfluoroalkyl substances (PFAS) for evaluation
PFAS
EPA interest
Animal
dose-response
data available3
Analytical detection
methods available13
Standards
Methods
PFDA
• OLEM priority0
Yes
Yes
Yes
PFNA
• OLEM priority0
• OW (UCMR) priority
• Found in industrial effluent and AFFF
Yes
Yes
Yes
PFHxA
• OCSPP priority01
• OLEM priority0
• Region 4 (Coosa and Tennessee
Rivers)
• Found in AFFF
Yes
Yes
Yes
PFHxS
• OCSPP priority
• OLEM priority0
• OW (UCMR) priority
• Region 4 (Tennessee River)
• Found in AFFF
Yes
Yes
Yes
PFBA
• OLEM priority0
• OCSPP priority01
• Found in AFFF
Yes
Yes
Under
development
This document is a draft for review purposes only and does not constitute Agency policy.
2-17 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
PFAS
EPA interest
Animal
dose-response
data available3
Analytical detection
methods available13
Standards
Methods
PFBS
• OLEM priority0
• OCSPP priorityd
• OW (UCMR) priority
• Found in AFFF
Yes
Yes
Yes
GenX
chemicals
• OCSPP priority6
• Region 3 priority
• Region 4 priority
Yes
Yes
Yes
GenX = perfluoro(2-methyl-3-oxahexanoic) acid (CASRN 13252-13-6); Unknown = status of validated standards
and methods was unknown at scoping.
aA survey of publicly available literature on PFAS other than PFOA and PFOS (i.e., a broad PubMed search and
review of recent assessments, including ATSDR (2018) was performed to identify in vivo animal studies that
tested multiple PFAS exposure levels and evaluated health endpoints. The quality of the studies was not
evaluated, and while multiple PFAS are evaluated in human studies, this was not a focus of the survey.
bAs of March 2019. The methods noted are for drinking water; nondrinking water methods are being
developed.
cFound at sites, including private and federal facilities and from various sources, including AFFF, chrome-plating
facilities, PFAS manufacturers, and industries that use PFAS (e.g., textiles and electronics). These PFAS have
also been detected in environmental media (e.g., surface water; biota).
dA significant number of new chemicals submitted to EPA are based on C6 and C4 chemistry. OCSPP often
evaluates risk for these compounds based on PFHxA and PFBS, which are the terminal degradation products of
certain C6 and C4 compounds.
Replacement for PFOA (e.g., for emulsifiers) and perfluoroethers. GenX chemicals are of concern based on
occurrence in NC and because EPA has received requests to review similar types of compounds (e.g., longer
chain ethers that might break down to GenX chemicals) as new chemicals.
1 As described in Section 2.1.5, exposure to these five PFAS can occur via the oral, inhalation,
2 and dermal routes, with oral (e.g., through diet and drinking water) being the predominant one
3 (Klaunigetal.. 20151. Given the potential regulatory applications of these PFAS assessments (see
4 Table 2-6), these assessments will consider PFAS exposures from all exposure routes. The
5 assessments will consider all potential health effects of exposure, both cancer and noncancer.
This document is a draft for review purposes only and does not constitute Agency policy.
2-18 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 2-6. Potential Environmental Protection Agency (EPA) needs and
applications for five per- and polyfluoroalkyl substances (PFAS)
EPA program
or regional
office
PFASa
Oral
Inhalation
Dermal
Potential regulatory application and explanation
(at the time scoping was conducted)
OLEM (in
coordination
with EPA
Regions 1-10)
PFDA
PFNA
PFHxA
PFHxS
PFBA
V
V
V
Comprehensive Environmental Response, Compensation,
and Liability Act (CERCLA). CERCLA authorizes EPA to
conduct short- or long-term cleanups at Superfund sites
and later recover cleanup costs from potentially
responsible parties under Section 107. PFAS toxicological
information may be used to make risk determinations for
response actions (e.g., short-term removals, long-term
remedial response actions). An evaluation of potential
actions at Superfund sites considers all routes of exposure.
Resource Conservation and Recovery Act (RCRA). RCRA
can be drawn upon to help address waste management
and cleanup needs, including accidental releases from
potentially hazardous waste management facilities.
OW
PFNA
PFHxS
V
Safe Drinking Water Act (SDWA) and Clean Water Act
(CWA). The SDWA requires EPA to periodically review the
National Primary Drinking Water Regulation (NPDWR) for
each contaminant and revise the regulation, if appropriate.
These potential applications focus on oral exposure.
OCSPP
PFHxA
PFHxS
PFBA
V
V
New chemical submissions to the Office of Pollution
Prevention and Toxics within OCSPP.
Region 4
PFHxA
PFHxS
V
Resource Conservation and Recovery Act (RCRA). RCRA
can be drawn upon to help address waste management
and cleanup needs, including accidental releases from
potentially hazardous waste management facilities. For
PFAS, the primary concern is potential oral exposure from
rivers in Region 4.
OCHP
PFDA
PFNA
PFHxA
PFHxS
PFBA
V
V
V
Executive Order 13045—Protection of Children from
Environmental Health Risks and Safety Risks: Policy on
Evaluating Health Risks to Children. In accordance with
EPA's 1995 policy and EO 13045, EPA instituted and
reaffirmed an Agency-wide commitment to "consider the
risks to infants and children consistently and explicitly as
part of risk assessments generated during its
decision-making process."
aPFAS to which this protocol applies (i.e., excluding PFBS and GenX chemicals).
This document is a draft for review purposes only and does not constitute Agency policy.
2-19 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
2.3. PROBLEM FORMULATION
2.3.1. Preliminary Literature Inventory for the Five Per- and Polyfluoroalkyl Substances
(PFAS) Being Assessed
1 As described in Section 2.1.1, several of these five PFAS have associated salts of potential
2 interest for human health assessment. Thus, the assessments will address each PFAS as follows:
3 • PFBA: PFBA (CASRN 375-22-4); PFBA ammonium salt (CASRN 10495-86-0)
4 • PFHxA: PFHxA (CASRN 307-24-4); PFHxA ammonium salt (CASRN 21615-47-4); PFHxA
5 sodium salt (CASRN 2923-26-4)
6 • PFHxS: PFHxS (CASRN 355-46-4); PFHxS potassium salt (CASRN 3871-99-6)
7 • PFNA: PFNA (CASRN 375-95-1); PFNA ammonium salt (CASRN 4149-60-4); PFNA sodium
8 salt (CASRN 21049-39-8)
9 • PFDA: PFDA (CASRN 335-76-2); PFDA ammonia salt (CASRN 3108-42-7); PFDA sodium salt
10 (CASRN 3830-45-3)
11 The results of a preliminary literature inventory of health effect-related studies on these
12 five PFAS and their associated salts are presented in Figure 2-3. The studies summarized in this
13 preliminary literature inventory reflect searches conducted in mid-2019 (and will be updated in the
14 context of the PFAS-specific assessments, but not this protocol) and are described on the project
15 pages for these five PFAS assessments in HERO fhttps://hero.epa.gov: see Section 1 for links to the
16 specific Health and Environmental Research Online [HERO] pages).
This document is a draft for review purposes only and does not constitute Agency policy.
2-20 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
-*-++ (-10+ studies)
- (-5 studies}
+ (-
1 2 studies) j - (Not Studied)
PFDA and salts
PFNA and salts
PFHxA and salts
PFHxS and salts
PFBA and salts
Oral:
Long1
Oral:
Short1
Oral:
Long1
Oral:
Short1
Oral:
Long1
Oral:
Short1
Oral:
Long1
Oral:
Short1
Oral:
Long1
Oral:
Short1
Cardiovascular
Developmental
Endocrine
(Thyroid)
Gastro-
intestinal
Hematologic
Hepatic
Musculo-
skeletal
Reproductive
Respiratory
General
Toxicity/ Other
Figure 2-3. Results of a preliminary literature inventory of five per- and polyfluoroalkyl substances (PFAS). Data
are approximated based on a cursory review of the literature search results for studies published through 2018 (see
Section 4 for details; this includes at-the-time-unpublished reports from NTP, see Section 4.1). Health effects are based on
groupings from the EPA's Integrated Risk Information System [IRIS] website
(https:JJ_cfpub.epa.gov/nceaJ_iris/search/index.cfml,a For this summary, metabolic effects are captured under "other" and
"hepatic" includes lipid and lipoprotein measures.
"'Oral; long" indicates subchronic or chronic oral exposure duration studies in animals and "Oral: short" reflects short-term and acute oral exposure studies in
animals, as well as reproductive and developmental studies.
This document is a draft for review purposes only and does not constitute Agency policy,
2-21 " DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Based on the results from the preliminary literature inventory in Figure 2-3, the following
health effects appear to be well studied for most PFAS of interest:
• Developmental effects
• Endocrine (primarily thyroid hormone) effects
• Hepatic effects, including lipid and lipoprotein measures
• Immune effects
• Reproductive effects in males or females
• Urinary effects
• General toxicity
As also shown in Figure 2-3, no studies of dermal exposure were identified. In addition,
data are sparse for assessing the potential health effects from chronic or subchronic oral exposure
or for inhalation exposure of any duration. Few studies have examined whether exposure to these
PFAS may result in carcinogenicity.
Given the potential future utility of comparing evidence across PFAS assessments (including
their respective data gaps), the five PFAS assessments will specifically address each of the potential
health effects enumerated above as "well studied." In addition, the potential for carcinogenicity will
be explicitly addressed in each assessment. Data on several other, variably studied endpoints
(i.e., cardiovascular effects, hematological effects, metabolic effects including diabetes, and nervous
system effects) will also be summarized when available. These summaries may be developed in
association with one of the health effects noted above, as a separate formal evaluation of hazard, or
as part of a qualitative summary on "other effects," depending on the assessment-specific data.
Information on other health effects, such as gastrointestinal effects; musculoskeletal effects; ocular
effects; and respiratory effects may be briefly summarized but will not be formally evaluated in any
of these assessments because of the paucity of available studies and the absence of exceptional
evidence in those that do exist New literature relating to these outcomes will be monitored during
literature search updates for potential inclusion.
2.4. KEY SCIENCE ISSUES
This section describes critical areas of scientific complexity that were identified based on
the preliminary literature inventory results summarized in the previous section. These scientific
issues are essential to consider during development of these assessments, and the specific methods
for doing so within these PFAS assessments are described in subsequent sections.
This document is a draft for review purposes only and does not constitute Agency policy.
2-22 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
2.4.1. Toxicokinetic Differences across Species and Sexes
The PFAS being evaluated are not metabolized, and reported half-lives in humans range
from several days (PFBA, PFBS) to multiple years (e.g., PFHxS). They are typically not stored in
body fat (see Section 2.1 for PFAS-specific chemical properties, including predicted LogP), but
accumulate in locations such as the blood, liver, and kidneys [and can be transferred to offspring
through placental transfer and breast milk; Postetal. (2012): ATSDR (2018): U.S. EPA (2016d): U.S.
EPA (2016c)]. However, as illustrated in Table 2-8, previous summaries of the existing literature
suggest there are pronounced half-life differences across species, sex, and type of PFAS. In general,
PFAS with longer chain lengths are reported to have a longer serum half-life. For the PFAS with
data available, serum half-life variation across species generally exhibits the following pattern:
rats
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
humans is the same as for animals, making the approach equally reliant on animal PK data. Also,
ti/2 measurements in animals are most reliable when ti/2 is in the range of hours to a few days. On
the other hand, measuring ti/2 is much more difficult when the value is very large, because it
requires long-term observation during which one must account for animal growth. But when ti/2 is
long, plasma levels will not vary rapidly from hour to hour, hence a measurement at the end of a
toxicity study should be a reliable measure of average internal dose. PFNA and PFDA have the
longest half-life in rats, averaging 1-2 months, but PFHxS also has ti/2 values reported up to 30 days
in male rats and in mice. Conveniently, matched blood and urine data were obtained for these three
in humans by Zhang etal. f 2 013b). allowing for a direct measure of CL in humans.
The apparent toxicokinetic differences may significantly affect the interpretation of toxic
effects across species or sexes. More directly, substantive toxicokinetic differences would be
expected to affect quantitative extrapolations of dose-response data from experimental animals to
humans. Thus, the half-life estimates for these five PFAS are likely to impact multiple assessment
decisions, and a critical review of the available ADME data for each PFAS will be important (see
discussion in Sections 5 and 9.2).
Although not identified during the preliminary literature inventory shown above,
physiologically based pharmacokinetic (PBPK) models for PFHxS fKim etal.. 20181 and PFDA and
PFNA fKim etal.. 20191 parameterized for adult male and female rats and humans have recently
been described. Fabregaetal. (2015) also described a PBPK model for multiple PFAS in humans,
including PFBS, PFHxS, PFHxA, PFNA, and PFDA. In addition, Verner etal. (2016) and Goeden et al.
(2019) described models for evaluating gestational and lactational transfer of PFAS from mothers
to their children, including PFHxS and PFOA. These models could prove useful for addressing
toxicokinetic questions in these assessments (see discussion in Sections 6.4 and 11.2).
This document is a draft for review purposes only and does not constitute Agency policy.
2-24 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 2-7. Preliminary serum half-life estimates of five per- and
polyfluoroalkyl substances (PFAS) across species and sexes
PFBA (C4)
PFHxA (C6)
PFHxS(C6)
PFNA (C9)
PFDA (C10)
Female
Male
Female
Male
Female
Male
Female
Male
Female
Male
Rat
1.0-1.8 h
6-9 h
0.4-0.6 h
1.0-1.6 h
1.8 d
6.8 d
1.4 d
30.6 d
58.6 d
39.9 d
Mouse
3 h
12 h
~1.2 h
~1.6 h
24-27 d
28-30 d
26-68 d
34-69 d
ND
Monkey
1.7 d
2.4 h
5.3 h
87 d
141 d
ND
ND
Human
3d
32 d
8.5 yr
4.3 yr
12 yr
"C" = carbon chain length; ND = no data.
Data are summarized in Lau (2015). Note that these values do not necessarily represent those that would be
used in qualitative or quantitative analyses for these PFAS assessments because the underlying data will be
reviewed and possibly supplemented with additional (e.g., newer) studies. Darker shading indicates longer
half-life (i.e., from hours to days to years).
2.4.2. Human Relevance of Effects in Animals that Involve Peroxisome
Proliferator-Activated Receptor Alpha (PPARa)
Activation of the peroxisome proliferator-activated receptor alpha (PPARa) by PFAS has
been reported, with in vitro evidence that the potency of human and mouse PPARa activation is
positively correlated with increasing PFCA chain length up to C9 (no human receptor activation was
noted for PFDA, although activation of the mouse receptor was only slightly less potent than PFNA)
and greater for carboxylates than sulfonates (Wolfetal.. 2014: Wolfetal.. 2008: Takacs and Abbott.
2007: Shipley etal.. 2004: Malonev and Waxman. 19991. It is not known whether PFAS distribute to
the nucleus and bind directly to PPARa in vivo, or whether these substances activate the receptor
indirectly. PPARa ligand binding causes a conformational change in the protein, release of
corepressors, heterodimerization with the retinoid X receptor (RXR), and binding to cognate
peroxisome proliferator response elements in the promoters of target genes (perhaps most notably,
those related to fatty acid (3-oxidation and energy homeostasis) to modulate gene transcription.
PPARa is a ligand-activated nuclear receptor expressed in many tissues and has been at the
forefront of a longstanding debate as to whether chemical-induced PPARa modulation in rodents,
particularly in the liver, is relevant to humans (Corton etal.. 2018: Filgo etal.. 2015: Guvton et al..
2009). PPARa is active in humans and responsive to the hypolipidemic effect of fibrate drugs that
lower serum lipid levels, but the human receptor is generally considered less sensitive than PPARa
in rodents (Corton etal.. 2014: Wolfetal.. 2014: Wolfetal.. 2008: Malonev and Waxman. 1999).
However, there are known human PPARa and other hepatic nuclear receptor polymorphisms
associated with increased susceptibility to liver disease fLi etal.. 2016: Li etal.. 20121. PPARa
activation has been extensively shown to induce peroxisome proliferation and result in
hepatocellular carcinoma, [reviewed in Liss and Finck (2017) and Corton et al. (2014)]. This
This document is a draft for review purposes only and does not constitute Agency policy.
2-25 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
phenomenon has not been observed in human models and is specific to rodents (Corton etal.. 2014:
Gonzalez and Shah. 2008: Holden and Tugwood. 19991. These effects are not observed in human
models fCorton et al.. 20141. Given the critical role PPARs' play as master regulators of lipid and
glucose metabolism in multiple cell types as described above, the role this family of nuclear
receptors plays in human metabolic diseases such as nonalcoholic liver disease (NAFLD) is an
active area of research (Liss and Finck. 2017).
It continues to be difficult to evaluate the relative sensitivity of humans and animal models
to PFAS-related PPARa inductions and to determine the extent to which differences relate to
differing toxicokinetics and/or intrinsic variations in biological sensitivities. For example, in some
contrast to observations in rodent models, longer duration administration of PFOA to nonhuman
primates (cynomolgus monkey) resulted in increasing absolute and relative liver-weight trends,
with statistically significant increases in relative liver weight at the higher dose, but in the absence
of histopathological changes (Bute nhoff etal.. 2002). These effects were concomitant with
significantly increased enzymatic markers of mitochondrial proliferation (not dose dependent) and
peroxisome proliferation at higher doses that further complicate interpretation. Evaluating the
human relevance of animal PPARa evidence is also complicated by a lack of comparable in vitro
model systems, including widely used primary cell lines that rapidly lose the capability to express
nuclear receptors such as PPARa fSoldatowetal.. 20131. and potential species-specific differences
in transcriptional coactivators and other pathway components. Finally, while toxicity studies
conducted with other more data-rich PFAS, such as PFOA and PFOS, may be informative to
characterizing data gaps and uncertainties, caution needs to be exercised when extrapolating
across PFAS, given that PFAS chain length, branching, and functional groups appear to be important
drivers influencing toxicokinetic and toxicological properties.
PPARa is also known to be important to other physiological processes in both rodents and
humans, including energy homeostasis, inflammation, reproduction, musculoskeletal function, and
development (NTDWOI. 2017: Corton etal.. 2014: Burri etal.. 2010: Abbott. 2009: Peraza etal..
2006: Corton etal.. 2000). Thus, although not extensively studied for PFAS, the modulation of
PPARa may be important to consider for developmental, metabolic, reproductive, and
immunological effects, as well as for hepatic effects.
There are additional complexities when considering the dependence on, and human
relevance of, PPARa activation by PFAS for certain health effects. The extent of PPARa activation is
likely to differ by PFAS type, making it harder to apply read-across (specifically, drawing
conclusions for one PFAS based on findings for another PFAS) or related approaches. In addition,
based on conclusions from other PFAS assessments and review articles, there is evidence to
indicate that many PFAS-mediated effects appear to include both PPARa-dependent and
PPARa-independent mechanisms, the latter of which include activation of PPARy,
phosphatidylinositol-3-kinase-serine/threonine kinase Akt (PI3K-Akt), constitutive androstane
receptor (CAR), mitochondrial damage, nuclear factor kappa B pathway (NF-kB), farnesoidX
This document is a draft for review purposes only and does not constitute Agency policy.
2-26 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
receptor, liver X receptor, and estrogen receptor a (Li etal.. 2017: NTDWOI. 2017: Rosen etal..
2017: FSANZ. 2016: U.S. EPA. 2014a. b; Foreman etal.. 20091.
Despite the complexities involved, it is important to evaluate the human relevance of some
PFAS exposure-mediated effects in animals (see discussion in Section 9.2.2).
2.4.3. Potential Confounding by Other Per- and Polyfluoroalkyl Substances (PFAS)
Exposures in Epidemiology Studies
Because different PFAS may be used in similar applications or result from similar sources,
potential confounding of associations by PFAS coexposures is an important area of uncertainty for
epidemiology studies. When associations are found for two or more moderately correlated PFAS in
a study, including those not the focus of these assessments (e.g., PFOS and PFOA), confounding is a
possible explanation. Based on a cursory review of studies identified during the preliminary
literature inventory, a complicating factor is that correlations between PFAS pairs vary
considerably across studies (see Section 6.2.1). When a study does notreportthe correlations in its
population, the interpretation of the risk of bias from confounding is particularly challenging. Even
when correlations are reported, there is no perfect method for eliminating confounding. Given this
variability, assessing the likelihood and impact of this source of potential confounding based on
reporting within individual studies is expected to be difficult (see discussion in Section 6.2).
2.4.4. Toxicological Relevance of Changes in Certain Urinary and Hepatic Endpoints in
Rodents
The scientific community has identified difficulties in interpreting the toxicological
relevance of changes in certain urinary and hepatic endpoints available in rodent studies (based on
the preliminary literature inventory) for some of the five PFAS assessments. The specific rodent
endpoints in question are chronic progressive nephropathy and related urinary histopathological
changes (including alpha 2u-globulin-mediated changes) and hepatic effects that may be
considered adaptive (e.g., increased liver weight; cellular hypertrophy; single cell
necrosis/apoptosis). For the former, some of these changes are not considered relevant to humans,
and methods exist for evaluating the dependency of observed changes on this rodent-specific
mechanism. For the latter, neither a clear scientific consensus nor specific EPA-wide guidance
defines exactly what level of change or constellation of effects is necessary to establish a cause for
concern. Thus, interpretations of the toxicological relevance of changes in these specific endpoints
are expected to require additional consideration (see discussion in Section 9.2).
2.4.5. Characterizing Uncertainty Due to Missing Chemical-Specific Information
Two PFAS, PFOA and PFOS (C8), have been studied more extensively than other PFAS.
Thus, this existing knowledge base may be useful in helping to characterize existing data gaps and
uncertainties in the current five PFAS assessments. Two recently developed EPA assessments
(PFBS and GenX chemicals) could also provide information during the development of these
This document is a draft for review purposes only and does not constitute Agency policy.
2-27 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
1 current assessments. For example, given knowledge regarding the health effects of PFOA and PFOS,
2 the potential lack of studies on immune effects for PFBA and developmental effects for PFHxA
3 (based on the preliminary literature inventory; see Section 2.3.2) appear to represent important
4 database uncertainties. In addition, given the potential for lifetime human exposure to PFAS by
5 multiple routes of exposure (see Section 2.1.5), the apparent scarcity of data on most of these five
6 PFAS other than short-term oral exposure studies in animals is expected to affect assessment
7 decisions and characterization of uncertainties (see discussion in Sections 10.2 and 11.2.3).
This document is a draft for review purposes only and does not constitute Agency policy.
2-28 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
3.0VERALL OBJECTIVES, SPECIFIC AIMS, AND
POPULATIONS, EXPOSURES, COMPARATORS,
AND OUTCOMES (PECO) CRITERIA
The overall objective of these five assessments is to identify adverse human health effects
and characterize exposure-response relationships for the effects of perfluorobutanoic acid (PFBA),
perfluorohexanoic acid (PFHxA), perfluorohexanesulfonate (PFHxS), perfluorononanoic acid
(PFNA), and perfluorodecanoic acid (PFDA) to support development of toxicity values. These
assessments will use systematic review methods to evaluate the epidemiological and toxicological
literature, including consideration of relevant mechanistic evidence (e.g., to inform key science
issues; see Section 2.4). The evaluations conducted in these assessments will be consistent with
relevant Environmental Protection Agency (EPA) guidance.3
The specific approach taken for these assessments of the potential health effects of PFBA,
PFHxA, PFHxS, PFNA, and PFDA (and their associated salts) was based on input received during
scoping, as well as a preliminary literature inventory of the health effects studied for these PFAS.
As outlined in Section 2.3.2, these assessments will evaluate the potential for PFAS exposure via the
oral or inhalation route to cause health effects in humans, specifically focusing on developmental
effects; endocrine (primarily thyroid hormone) effects; hepatic effects, including lipid and
lipoprotein measures; immune effects; reproductive effects in males or females; urinary effects;
general toxicity; and carcinogenicity (see Section 5 for preliminary decisions for grouping outcomes
and endpoints within each of these predetermined health effect categories). Data on cardiovascular
effects, hematological effects, metabolic effects including diabetes, and nervous system effects will
also be summarized when available. These summaries may be developed in association with one of
the health effects noted above either as a separate formal evaluation of hazard or as part of a
qualitative summary on "other effects," depending on the assessment-specific data. Given the
paucity of available studies and in the absence of exceptional evidence in any available studies,
information on other health effects (i.e., gastrointestinal effects; musculoskeletal effects; ocular
effects; and respiratory effects) will not be formally evaluated (these effects may be briefly
summarized) in any of these assessments; although new literature relating to these outcomes will
be monitored during literature search updates for potential inclusion. As outlined in the EPA PFAS
action plan,4 the characterization of the potential human health hazards from exposure to these
3EPA guidance documents: http://www.epa.gov/iris/basic-information-about-integrated-risk-information-
svstem#guidance/.
4EPA PFAS action plan: https://www.epa.gov/pfas/epas-pfas-action-plan.
This document is a draft for review purposes only and does not constitute Agency policy.
3-1 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
individual PFAS will be coupled with data generated from new advances in computational and
high-throughput toxicology to inform evaluations of other PFAS.
3.1. SPECIFIC AIMS
The aims of these assessments are to:
• Identify epidemiological (i.e., human) and toxicological (i.e., experimental animal) literature
reporting effects of exposure to PFBA, PFHxA, PFHxS, PFNA, and PFDA (and their associated
salts), as outlined in the PECO. These five systematic reviews will focus on identifying
studies following oral or inhalation exposure to PFAS.
• Evaluate mechanistic information (including toxicokinetic understanding) associated with
exposure to PFBA, PFHxA, PFHxS, PFNA, and PFDA, to inform the interpretation of findings
related to potential health effects in studies of humans and animals. The scope of these
analyses of mechanistic information will be determined by the complexity and confidence in
the phenotypic evidence in humans and animals, the likelihood of the analyses
(e.g., considering the mechanistic studies available based on the literature inventory; see
Section 4.2.2) to affect evidence synthesis conclusions for human health, and the directness
or relevance of the available model systems for understanding potential human health
hazards (see Section 9.2). The mechanistic evaluations will focus primarily on the key
science issues identified in Section 2.4.
• Conduct study evaluations for individual epidemiological and toxicological studies
(evaluating reporting quality, risk of bias, and sensitivity) and PBPK (scientific and technical
review). The evaluation of epidemiology studies will specifically consider, to the extent
possible, the likelihood and impact of potential confounding by other PFAS (see
Section 6.2.1).
• Extract data on relevant health outcomes from epidemiological and toxicological studies of
high, medium, and low confidence based on the study evaluations (full data extraction of low
confidence studies may not be performed for poorly studied health effects or for health
effects on which extensive medium and high confidence studies exist in the evidence base).
• Synthesize the evidence across studies, assessing similar health outcomes using a narrative
approach. To inform future comparisons across a range of PFAS structures and properties
(e.g., using high throughput screening, computational toxicology approaches, and chemical
informatics to fill in data gaps; see EPA PFAS action plan), each of the five PFAS assessments
will synthesize the available evidence (or lack thereof) for developmental effects; endocrine
(primarily thyroid hormone) effects; hepatic effects, including lipid and lipoprotein
measures (the latter of which are also applicable to interpreting the potential for
cardiovascular toxicity); immune effects; reproductive effects in males or females; urinary
effects; general toxicity; and carcinogenicity. Some assessments may include additional
evidence syntheses for other health effects. The toxicological relevance of changes in some
urinary and hepatic outcomes will be a point of focus in the evidence syntheses (see
Section 9.2.3).
This document is a draft for review purposes only and does not constitute Agency policy.
3-2 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
• For each health outcome (or grouping of outcomes), evaluate the strength of evidence
across studies (or subsets of studies) separately for studies of exposed humans and for
animal studies. Based on the focused mechanistic analyses specific to each PFAS
assessment (see Section 9.2), the mechanistic evidence will be used to inform evaluations of
the available health effects evidence (or lack thereof).
• For each health outcome (or grouping of outcomes), develop an integrated expert judgment
across evidence streams as to whether the evidence is sufficient (or insufficient) to indicate
that exposure to the PFAS has the potential to be hazardous to humans (in rare instances,
the evidence may be judged as sufficient to indicate that a hazard is unlikely). The judgment
will be directly informed by the evidence syntheses and based on structured review of an
adapted set of considerations for causality first introduced by Austin Bradford Hill [Hill
(1965): see Sections 9 and 10], including consideration (e.g., based on available mechanistic
information) and discussion of biological understanding. As part of the evidence
integration narrative, characterize the strength of evidence for the available database of
studies and its uncertainties, and identify and discuss issues concerning potentially
susceptible populations and lifestages.
• Derive toxicity values (e.g., oral reference doses [RfDs], inhalation reference concentrations
[RfCs], cancer risk estimates) as supported by the available data (see Section 10.2). Apply
toxicokinetic and dosimetry modeling (possibly including PBPK modeling) to account for
interspecies differences, as appropriate. Given the apparent species and sex differences in
the toxicokinetic profile of the different PFAS (see Section 2.4), methods to address these
potential differences will be a key consideration (see Section 9.2.1).
• Characterize uncertainties and identify key data gaps and research needs across each PFAS
database, such as limitations of the available evidence, limitations of the systematic review,
and consideration of dose relevance and toxicokinetic differences when extrapolating
findings from higher dose animal studies to lower levels of human exposure.
3.2. POPULATIONS, EXPOSURES, COMPARATORS, AND OUTCOMES
(PECO) CRITERIA
The PECO criteria are used to identify the evidence that addresses the specific aims of the
assessment and to focus the literature screening, including study inclusion/exclusion, in a
systematic review (see details on literature screening in Section 4.2). Given the expected lack of
studies on carcinogenicity for these PFAS based on the preliminary literature inventory,
genotoxicity studies were included in the PECO criteria (see Table 3-1).
In addition to those studies meeting the PECO criteria, studies containing supplemental
material that are potentially relevant to the specific aims of the assessment were tracked during the
literature screening process. Although these studies did not meet PECO criteria, they were not
excluded from further consideration. The categories used to track studies as "potentially relevant
supplemental material" are also described in Section 4.2.
This document is a draft for review purposes only and does not constitute Agency policy.
3-3 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 3-1. Populations, exposures, comparators, and outcomes (PECO)
criteria
PECO
element
Evidence
Populations
Human: Any population and lifestage (occupational or general population, including children and
other sensitive populations). The following study designs will be included: controlled exposure,
cohort, case-control, and cross-sectional. (Note: Case reports and case series will be tracked as
potential supplemental material.)
Animal: Nonhuman mammalian animal species (whole organism) of any lifestage (including
preconception, in utero, lactation, peripubertal, and adult stages).
Other: In vitro, in silico, or nonmammalian models of genotoxicity. (Note: Other in vitro, in silico,
or nonmammalian models will be tracked as potential supplemental material.)
Exposures
Human: Studies providing quantitative estimates of PFAS exposure based on administered dose
or concentration, biomonitoring data (e.g., urine, blood, or other specimens), environmental or
occupational-setting measures (e.g., water levels or air concentrations, residential location
and/or duration, job title, or work title). (Note: Studies that provide qualitative, but not
quantitative, estimates of exposure will be tracked as supplemental material.)
Animal: Oral or Inhalation studies including quantified exposure to a PFAS of interest based on
administered dose, dietary level, or concentration. (Note: Nonoral and noninhalation studies
will be tracked as potential supplemental material.) PFAS mixture studies are included if they
employ an experimental arm that involves exposure to a single PFAS of interest. (Note: Other
PFAS mixture studies are tracked as potential supplemental material.)
Studies must address exposure to one or more of the following: PFDA (CASRN 335-76-2), PFDA
ammonia salt (CASRN 3108-42-7), PFDA sodium salt (CASRN 3830-45-3), PFNA (CASRN 375-95-1),
PFNA ammonium salt (CASRN 4149-60-4), PFNA sodium salt (CASRN 21049-39-8), PFHxA
(CASRN 307-24-4), PFHxA sodium salt (CASRN 2923-26-4), PFHxA ammonium salt (CASRN 21615-
47-4), PFHxS (CASRN 355-46-4), PFHxS potassium salt (CASRN 3871-99-6), PFBA
(CASRN 375-22-4), or PFBA ammonium salt (CASRN 10495-86-0). [Note: although while these
PFAS are not metabolized or transformed in the body, there are precursor compounds known to
be biotransformed to a PFAS of interest; for example, 6:2 fluorotelomer alcohol is metabolized
to PFHxA and PFBA (Russell et al., 2015). Thus, studies of precursor PFAS that identify and
quantify a PFAS of interest will be tracked as potential supplemental material (e.g., for ADME
analyses or interpretations).]
Comparators
Human: A comparison or reference population exposed to lower levels (or no
exposure/exposure below detection levels) or for shorter periods of time.
Animal: Includes comparisons to historical controls or a concurrent control group that is
unexposed, exposed to vehicle-only or air-only exposures. (Note: Experiments including
exposure to PFAS across different durations or exposure levels without including one of these
control groups will be tracked as potential supplemental material [e.g., for evaluating key
science issues; Section 2.4].)
Outcomes
All cancer and noncancer health outcomes. (Note: Other than genotoxicity studies, studies
including only molecular endpoints [e.g., gene or protein changes; receptor binding or
activation] or other nonphenotypic endpoints addressing the potential biological or chemical
This document is a draft for review purposes only and does not constitute Agency policy.
3-4 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
PECO
element
Evidence
progression of events contributing towards toxic effects will be tracked as potential
supplemental material [e.g., for evaluating key science issues; Section 2.4].)
PBPK models
Studies describing physiologically based pharmacokinetic (PBPK) and other PK models for PFDA
(CASRN 335-76-2), PFDA ammonia salt (CASRN 3108-42-7), PFDA sodium salt (CASRN 3830-45-3),
PFNA (CASRN 375-95-1), PFNA ammonium salt (CASRN 4149-60-4), PFNA sodium salt
(CASRN 21049-39-8), PFHxA (CASRN 307-24-4), PFHxS (CASRN 355-46-4), PFHxS potassium salt
(CASRN 3871-99-6), PFBA (CASRN 375-22-4), or PFBA ammonium salt (CASRN 10495-86-0).
ADME = absorption, distribution, metabolism, and excretion; PK = pharmacokinetic.
This document is a draft for review purposes only and does not constitute Agency policy.
3-5 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
4.LITERATURE SEARCH AND SCREENING
STRATEGIES
The initial literature search was completed in July 2017 as part of a cross-Environmental
Protection Agency (EPA) workgroup that focused on a large set of PFAS, including the five PFAS
addressed in this protocol. Subsequent literature searches were refined and are being updated
regularly. In an effort to ensure that all pertinent studies were captured, the studies identified as
relevant to PFBA, PFHxA, PFHxS, PFNA, and PFDA are being shared simultaneously with release of
this protocol.5 These search efforts reflect studies published through February 2018, several
unpublished reports, and a few recent studies not yet identified through the formal literature
updating process; the literature is currently being updated and will be updated regularly in the
context of the five PFAS assessments (in their respective U.S. EPA Health and Environmental
Research Online [HERO] pages) until several months before public release of the draft
assessments.6 Accordingly, the methods for literature search and screening (as well as some of the
approaches to refining the evaluation plan based on the identified literature; see Section 5) are
described in the protocol using the past tense, whereas approaches for the other assessment
methods are outlined using the future tense.
4.1. LITERATURE SEARCH STRATEGIES
The initial literature search strategy performed in July 2017 was designed to identify a
broad range of topics relevant to PFAS, including studies on physicochemical properties,
environmental fate and occurrences, human exposures, and biological effects representative of all
types of evidence (i.e., human, animal, in vitro, in silico) and health outcomes. PFAS search terms
included PFAS names (including salt, cationic, and anionic forms), all known synonyms, and CAS
registry numbers. The literature search itself encompassed a non-date-limited query of the
following databases:
5PFBA: https://hero.epa.gov/hero/index.cfm/proiect/page/proiect id/2632
PFHxA: https://hero.epa.gov/hero/index.cfm/proiect/page/proiect id/2628
PFHxS: https://hero.epa.gov/hero/index.cfm/proiect/page/proiect id/2630
PFNA: https://hero.epa.gov/hero/index.cfm/proiect/page/proiect id/2633
PFDA: https://hero.epa.gov/hero/index.cfm/proiect/page/proiect id/2614.
6Although not identified as part of the formal literature searches through early 2019, several more recent PBPK
studies found through regular monitoring of new studies are included in this protocol (see Section 6.4) so that the
process for evaluating those data can be outlined.
This document is a draft for review purposes only and does not constitute Agency policy.
4-1 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
1 • PubMed fNational Library of Medicine 1
2 • Web of Science fThomson Reuters!
3 • Toxline (National Library of Medicine)
4 • TSCATS (Toxic Substances Control Act Test Submissions)
5 All literature identified in the initial search was loaded into the HERO database. In
6 February 2018, the literature search was updated for the PFAS in this assessment (i.e., PFBA,
7 PFHxA, PFHxS, PFNA, and PFDA). The updated literature query included all PFAS nomenclature
8 from the initial search as well as a broader non-date-limited search of several new PFAS synonyms
9 that were identified after the original search. This updated search was conducted by EPA's HERO
10 tool to search the same databases that were included in the initial literature query.
11 Because each database has its own search architecture, the resulting search strategy was
12 tailored to account for each database's unique search functionality. Full details of the July 2017 and
13 February 2018 search strategies are presented in Addendum B. No literature was restricted by
14 language.
15 Additional relevant literature not found through database searching was identified by:
16 • Review of studies cited in state, national (EPA, Food and Drug Administration [FDA], etc.),
17 and international (International Agency for Research on Cancer [IARC], World Health
18 Organization [WHO], European Chemicals Agency [ECHA], etc.) assessments on these five
19 PFAS, including parallel assessment efforts in progress (e.g., the draft Agency for Toxic
20 Substances and Disease Registry [ATSDR] assessment released publicly in 2018).
21 • Review of studies submitted to federal regulatory agencies and brought to the attention of
22 EPA. For example, studies submitted to EPA by the manufacturers of these five PFAS in
23 support of requirements under the Toxic Substances Control Act (TSCA). Such studies (or
24 data summaries) will only be tracked in the literature flow diagrams released with each of
25 the five assessments (and considered for inclusion in the assessment) when they can be
26 made publicly available. To facilitate the timely completion of these assessments, if
27 attempts to acquire a publicly accessible version of an identified study are unsuccessful
28 after 3 months of the initial request, these studies will be considered unobtainable and will
29 not be considered for inclusion in the assessment(s).
30 • Identification of studies during screening for other PFAS. For example, epidemiology
31 studies relevant to more than one of these five PFAS were sometimes identified by searches
32 focused on one PFAS, but not the others.
33 • Other gray literature (i.e., primary studies not indexed in typical databases, such as
34 technical reports from government agencies or scientific research groups; unpublished
35 laboratory studies conducted by industry; or working reports/white papers from research
36 groups or committees) brought to the attention of EPA during problem formulation,
This document is a draft for review purposes only and does not constitute Agency policy.
4-2 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
engagement with technical PFAS experts, and during future solicitation of Agency,
interagency, and public comment during the Integrated Risk Information System (IRIS)
assessment development and review process. For example, one such study was brought to
the attention of EPA on March 29, 2018 by the National Toxicology Program (NTP) when
NTP published study tables and individual animal data from a 28-day toxicity study of
multiple PFAS fNTP. 20111. A peer-reviewed NTP Technical Report was not yet available at
the time this protocol was developed,7 but these data have undergone standard NTP quality
assurance/control processing and are publicly available, and a protocol outlining the NTP
study methods is available in HERO
fhttps://hero.epa.gov/hero/index.cfm/reference/details/reference id/43097411
The number of studies on PFBA, PFHxA, PFHxS, PFNA, and PFDA returned from the
literature searches through February 2018 is documented in the literature flow diagrams in
Figure 4-1, which also reflect the literature screening decisions (see Section 4.2). Notably, the
identification and review of records submitted to EPA, which may include confidential business
information (CBI), is ongoing. This includes exploring the possibility of making the data within any
identified records publicly available. Any records included in the assessments will be reflected in
updates to the protocol and in the draft assessments. In addition, any identified companion
documents for the included studies, such as retractions, corrections and supplemental materials,
will also be included, and the assessments will incorporate the most recent publication materials
(note: these are tracked as separate, "included" records in the literature flow diagrams [see
Section 4.2.2] and HERO; companion documents in other screening categories such as "excluded,"
which are not relevant to the target PECO, are similarly tagged as separate records within that
screening category).
The literature searches will be updated throughout the assessments' development and
review process to identify newly published literature. The last full literature search update will be
conducted prior to (several months) the planned release of the draft document for public comment
The literature flow diagrams (see Section 4.2.2) presented in the assessment will be revised to
reflect these updates. Although uncommon, it is possible that additional literature searches may be
performed during assessment development and review (e.g., to supplement an analysis of a specific
mechanism or biological linkage). Any such ancillary searches will be documented in updates to the
protocol.
The IRIS Program takes extra steps to ensure identification of pertinent studies by
encouraging the scientific community and the public to identify additional studies and ongoing
research; by searching for publicly available data submitted under the TSCA and the Federal
Insecticide, Fungicide, and Rodenticide Act (FIFRA); and by considering late-breaking studies that
would affect the credibility of the conclusions, even during the review process. Studies identified
7The 28-day oral studies of PFDA, PFNA, PFHxA, and PFHxS in male and female rats are now final (updated after
public comments received). The published research reports will be cited in the assessments and are available on
HERO; however, the protocol discusses these reports in relation to their unpublished version.
This document is a draft for review purposes only and does not constitute Agency policy.
4-3 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
after peer review begins will only be considered for inclusion if they meet the PECO criteria and are
expected to fundamentally alter the assessment's conclusions. Release of the PECO-screened
literature in parallel with release of the protocol for public comment provides an opportunity for
stakeholders to identify any missing studies, which if identified, will be screened as outlined above
for adherence to the PECO criteria.
4.1.1. Non-Peer-Reviewed Data
IRIS assessments rely mainly on publicly accessible, peer-reviewed studies. However, it is
possible that gray literature (i.e., studies that are not reported in the peer-reviewed literature)
directly relevant to the PECO may be identified during assessment development (e.g., good
laboratory practice [GLP] studies submitted to EPA, dissertations, etc.). In this case, if the data
make a substantial impact on assessment decisions or conclusions (i.e., have potential to affect the
PECO statement, hazard conclusions, or dose-response analysis), EPA can obtain external peer
review if the owners of the data are willing to have the study details and results made publicly
accessible. This independent, contractor-driven peer review would include an evaluation of the
study, as is done for peer review of a journal publication. The contractor would identify and select
two to three scientists knowledgeable in scientific disciplines relevant to the topic as potential peer
reviewers. Persons invited to serve as peer reviewers would be screened for conflict of interest
before confirming their service. In most instances, the peer review would be conducted by letter
review. The study authors would be informed of the outcome of the peer review and given an
opportunity to clarify issues or provide missing details. The study and its related information, if
used in the IRIS assessment, would become publicly available. In the assessment, EPA would
acknowledge that the document underwent external peer review managed by EPA, and the names
of the peer reviewers would be identified. In certain cases, IRIS will conduct an assessment for
utility and data analysis based on having access to a description of study methods and raw data that
have undergone rigorous quality assurance/quality control review (e.g., ToxCast/Tox21 data;
results of National Toxicology Program [NTP] studies) but that have not yet undergone external
peer-review.
Unpublished (e.g., raw) data from personal author communication can supplement a
peer-reviewed study, if that information is made publicly available. If such ancillary information is
acquired, it will be documented in either the Health Assessment Workspace Collaborative (HAWC)
or HERO project page for the PFAS being assessed (depending on the nature of the information
received).
4.2. SCREENING PROCESS
As described below, PECO criteria or predefined inclusion and exclusion criteria (i.e., the
latter were used for the initial search) were used by two independent reviewers to screen and
inventory studies at the title and abstract level. For those studies considered relevant at the title
This document is a draft for review purposes only and does not constitute Agency policy.
4-4 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
1 and abstract level, these criteria were then used to determine inclusion or exclusion of a reference
2 based on the full text. In addition to the PECO criteria, the following exclusion criteria were
3 applied:
4 • Review, commentary, other agency assessment, letter, or other record that does not contain
5 original data (note that these records were tracked for potential use in identifying
6 study-specific, original data relevant to specific scientific questions during assessment
7 development, including scanning of reference lists for unidentified studies; any such studies
8 incorporated into the assessment will be tracked under "other" as the reference source in
9 updates to the protocol)
10 • Study available only as an abstract (e.g., conference abstract)
11 • Full text of the study is not available, and screening decisions could not be made at the
12 title/abstract level
13 In addition to including studies that meet PECO criteria, other studies containing material
14 that is potentially relevant to the assessments' objectives and specific aims were tracked during the
15 screening process as "potentially relevant supplemental material." Importantly, these studies were
16 not excluded, but they may not be incorporated into the assessments unless they are deemed to be
17 relevant to addressing the key science issues, specific aims (see Sections 2.4 and 3.1), or key
18 scientific uncertainties identified at later stages of assessment development (see Section 9). Studies
19 categorized as "potentially relevant supplemental material" include the following:
20 • In vivo mechanistic or mode-of-action studies, including non-PECO routes of exposure and
21 populations (e.g., nonmammalian models—generally, these are interpreted to be less
22 directly relevant to evaluating the potential for human disease, although exceptions do exist
23 for some endpoints)
24 • In vitro and in silico models
25 • AD ME and toxicokinetic studies (excluding models)8
26 • Exposure assessment or characterization (no health outcome) studies
27 • PFAS mixture studies (no individual PFAS comparisons)
28 • Human case reports or case-series studies
8Given the known importance of ADME data, this supplemental tagging was used as the starting point for a
separate screening and review of toxicokinetics data (see Section 9.2.1 for details on the PECO and screening
process for this separate literature identification effort).
This document is a draft for review purposes only and does not constitute Agency policy.
4-5 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
• Ecotoxicity studies
• Studies on PFAS manufacture/use
• Treatment/remediation studies
• Studies of PFAS analysis or other laboratory methods
• Environmental fate and transport studies
• Studies of other PFAS
Several of these categories of studies were further screened for consideration in addressing
the key science issues (described in Section 9.2).
Title and abstract screening. Following a pilot phase to calibrate screening guidance, two
screeners independently performed a title and abstract screen using a structured form in
DistillerSR (Evidence Partners; https://distillercer.com/products/distillersr-systematic-review-
software/). For citations with no abstract, the article was excluded if screening decisions could not
be made based on the title and other citation information (e.g., page length) and if additional
attempts to acquire the abstract or full text were unsuccessful. Screening conflicts were resolved
by discussion between the primary screeners, with consultation by a third reviewer or technical
advisor (if needed) to resolve any remaining disagreements. Eligibility status of non-English
studies was assessed using the same approach with online translation tools used as needed to
evaluate portions of the study text and assess eligibility at the title and abstract level.
Studies not meeting title/abstract criteria but identified as "potentially relevant
supplemental material" were categorized (i.e., tagged) during the title and abstract screening
process (further described in Section 4.3). Conflict resolution was not required during the
screening process to identify supplemental information (i.e., tagging by a single screener was
considered adequate to identify the study as potentially relevant supplemental material for possible
inclusion during draft development).
Before beginning the Integrated Risk Information System (IRIS) PFAS assessments project,
the EPA contractor that conducted the July 2017 literature search as part of an EPA-wide
workgroup had performed a title and abstract screen to bin studies into different categories
(e.g., human, in vivo animal, excluded). At the initiation of these PFAS assessments within IRIS, a
formalized effort was deployed with a new title and abstract screen of all studies identified in the
initial July 2017 search based on the PECO criteria in Table 3-1. For this initial literature screening,
specific inclusion/exclusion criteria were applied in the formalized title and abstract screen (see
Addendum B, Table B-6). Title and abstract screening of studies identified during literature search
updates will be conducted using the PECO criteria in Table 3-1 in DistillerSR using forms that
facilitate simultaneous initial tagging during screening (e.g., category of supplemental data;
This document is a draft for review purposes only and does not constitute Agency policy.
4-6 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
contains data on other PFAS of interest). An example of the questions and answers populating the
DistillerSR form for title/abstract and full-text (below) screening during literature search updates
is provided in Addendum B, Table B-7.
Full-text screening. Records that were not excluded based on the title and abstract were
advanced to full-text review. Full-text copies of these potentially relevant records were retrieved,
stored in the HERO database, and independently assessed by two screeners using a structured form
in DistillerSR to confirm eligibility. Screening conflicts were resolved by discussion among the
primary screeners with consultation by a third reviewer or technical advisor (as needed to resolve
any remaining disagreements). As with the title and abstract screening, some studies were also
identified as "potentially relevant supplemental material" based on full-text screening. Approaches
for language translation included engagement of a native speaker from within EPA or use of
fee-based translation services.
In addition to identifying studies as included, excluded, or potential supplemental material,
the reviewers used the DistillerSR screening forms to confirm the specific PFAS (or multiple PFAS)
evaluated and to document several important experimental features of the studies (see Section 4.3).
The results of this screening process are documented in the HERO database
fhttps://hero.epa.gov: see Section 4.1 for links to the specific HERO pages) and literature flow
diagrams (see Figure 4-1), with individual studies "tagged" in HERO according to their appropriate
category descriptors (e.g., reference source; screening decisions regarding inclusion, exclusion, or
identification as supplemental; type of study).
4.2.1. Multiple Publications of the Same Data
When there are multiple publications using the same or overlapping data, all publications
on the research were included, with one selected for use as the primary study; the others were
considered as secondary publications with annotation in HAWC indicating their relationship to the
primary record during data extraction. For epidemiology studies, the primary publication was
generally the one with the longest follow-up, the largest number of cases, or the most recent
publication date. For animal studies, the primary publication was typically the one with the most
recent publication date, longest duration of exposure, or the one that assessed the outcome (s) most
informative to the PECO. For both epidemiology and animal studies, the assessments will include
relevant data from all publications of the study, although if the same data are reported in more than
one study, the data will only be extracted once (see Section 8). For corrections, retractions, and
other companion documents to the included publications, a similar approach to annotation was
taken (see Section 4.1), and the most recently published data will be incorporated in the
assessments.
This document is a draft for review purposes only and does not constitute Agency policy.
4-7 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
4.2.2. Literature Flow Diagrams
1 Figure 4-1 presents the literature flow diagrams for PFBA (a), PFHxA (b), PFHxS (c), PFNA
2 (d), and PFDA (e).9 These figures reflect literature searches through 2018. A literature search
3 update has been conducted and the results will be reflected in the draft assessments (and the most
4 current results can be viewed at any time in the HERO project pages provided in Section 4.1). Note
5 that the potential for updates or revisions to these figures related to CBI data and other reference
6 decisions is discussed in the previous sections.
9Note that the literature searches included the associated salts for each of the five PFAS, as presented in Figure 2-1
(see Section 2.1.1). In addition, although not identified (yet) as part of the formal literature searches and not
included in these diagrams, several recent PBPK studies found through regular monitoring of new studies are
included in this protocol (see Section 6.4) so that the process for evaluating those studies can be outlined.
This document is a draft for review purposes only and does not constitute Agency policy.
4-8 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
(a)
PFBA
Literature Searches (through 2018)
r "\
PubMed WOS ToxLine TSCATS
(n = 461) (n = 456) (n = 28) (n = 0)
Other
ATSDR assessment (n= 1)
Submitted to EPA (n = 3)
V V
i
TITLE AND ABSTRACT SCREENING
This document is a draft for review purposes only and does not constitute Agency policy,
4-9 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
(b)
PFHxA
Literature Searches (through 2018)
PubMed
(n = 239)
WOS
(n = 245)
ToxLine TSCATS
(n = 17) (n = 0)
Other
ATSDR assessment (n = 17)
Submitted to EPA (n = 0)
NTP Report (n = 1)
I
TITLE AND ABSTRACT
Title & Abstract Screening
(285 records after duplicate removal)
FULL TEXT SCREENING
I
Studies Meeting PECO (n = 32)
• Human health effects studies (n = 17)
• Animal health effect studies (n = 10)
• Genotoxicity studies (n = 4)
• Susceptible population (n = 4)
Excluded (n= 205)
Not relevant to PECO (n = 205)
Excluded (n= 8)
• not relevant to PECO (n = 1), review,
commentary, or letter (n = 6), abstract-only
(n = 0), unable to obtain full text (n = 0),
other (n = 2)
Tagged as Supplemental (n= 45)
mechanistic or MOA (n = 8), ADME (n = 18),
qualitative exposure only (n = 12), mixture-
only (n = 3), non-PECO route of exposure (n
= 2), case report or case study (n = 2)
This document is a draft for review purposes only and does not constitute Agency policy,
4-10 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
PFHxS
Literature Searches (through 2018)
PubMed
(n = 476)
WOS
(n =517 )
ToxLine
(n = 266)
TSCATS
(n = 10)
Additional Strategies
(n=52)
1
TITLE AND ABSTRACT
Title & Abstract Screening
(507 records after duplicate removal)
FULL TEXT SCREENING
Full-Text Screening
(n = 234)
Excluded (n= 254)
• Not relevant to PECO (n = 254)
Excluded (n= 55)
• not relevant to PECO (n = 28), review or
regulatory document (n = 27), abstract-only
(n = 0)
Studies Meeting PECO (n = 114)
Human health effects studies (n = 98)
Animal health effect studies (n = 14)
Genotoxicity studies (n = 1)
PBPK models (n = 1)
Tagged as Supplemental (n= 70)
mechanistic or MOA (n = 26), ADME (n = 32),
qualitative exposure only (n = 12), mixture-only (n
= 0), non-PECO route of exposure (n = 0), case
report or case study (n = 0)
This document is a draft for review purposes only and does not constitute Agency policy,
4-11. DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
(d)
PFNA
Literature Searches (through 2018)
PubMed
(n = 1,111)
WOS
(n = 707)
ToxLine
(n = 862)
TSCATS
(n = 57)
Other
ATSDR assessment (n = 36)
Submitted to EPA (n = 0)
Non-English or non-peer
reviewed (n = 37)
NTP (2018) report (n = 1)
1
TITLE AND ABSTRACT SCREENING
This document is a draft for review purposes only and does not constitute Agency policy,
4-12' DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
(e)
PubMed
(n = 515)
PFDA
Literature Searches (through 2018)
wos
(n= 613)
ToxLine
(n= 120)
TSCATS
(n=l)
Other
From draft ATSDR
assessment (n = 21)
Submitted to EPA (n = 0)
NTP report (n=l)
1
TITLE AND ABSTRACT SCREENING
Figure 4-1. Literature flow diagrams for PFBA and its ammonium salt (a),
PFHxA and its ammonium and sodium salts (b), PFHxS and its potassium salt
(c), PFNA and its ammonium and sodium salts (d), and PFDA and its
ammonium and sodium salts (e).
4.3. SUMMARY-LEVEL LITERATURE INVENTORIES
1 As noted in Section 4.2, during title/abstract or full-text level screening, studies tagged
2 based on PECO eligibility were further categorized based on features such as evidence type (human,
3 animal, mechanistic, PBPK, etc.], health outcome(s], and/or endpointmeasure(s) included in the
4 study, and the specific PFAS (or multiple PFAS] addressed (see Addendum B, Table B-7 for
5 examples]. Literature inventories for PECO-relevant studies were created to develop
This document is a draft for review purposes only and does not constitute Agency policy,
4-13' DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
summary-level, sortable lists that include some basic study design information (e.g., study
population, exposure information such as doses administered or biomarkers analyzed,
age/lifestage10 of exposure, endpoints examined, etc.). These working literature inventories are for
internal use and facilitate subsequent review of individual studies or sets of studies by
topic-specific experts.
Inventories were also created for studies that were tagged as "potentially relevant
supplemental material" during screening, including in vitro or in silico models not addressing
genotoxicity, ADME studies, and studies on endpoints or routes of exposure that did not meet the
specific PECO criteria, but which may still be relevant to the research question(s). Here, the
objective was to create an inventory of studies that can be tracked and further summarized as
needed—for example, by model system, key characteristic [e.g., of carcinogens, Smith etal. f20161],
mechanistic endpoint, or key event—to support analyses of potentially critical mechanistic
questions that arise at various stages of the systematic review (see Section 9.2 for a description of
the process for determining the specific questions and pertinent mechanistic studies to be
analyzed). For example, ADME data and related information are important to the next steps of
evaluating the evidence from individual PECO-specific studies, and these data will be reviewed by
subject matter experts early in the assessment process. Thus, the comprehensive identification of
studies relevant to interpreting the ADME or toxicokinetic characteristics of these PFAS was
prioritized (see additional discussion in Section 5, and the specifics of the approach in Section 9.2).
10Age/lifestage was considered according to EPA's Guidance on Selecting Age Groups for Monitoring and Assessing
Childhood Exposures to Environmental Contaminants and EPA's A Framework for Assessing Health Risk of
Environmental Exposures to Children.
This document is a draft for review purposes only and does not constitute Agency policy.
4-14 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
5. REFINED EVALUATION PLAN
The primary purpose of this step is to outline any potential or expected refinements to the
set of populations, exposures, comparators, and outcomes (PECO)-relevant studies that would
narrow the scope of studies considered for use in evidence synthesis and beyond. This optional
step is typically applied to focus an assessment with a very large number of PECO-relevant studies
on review of the most informative evidence (e.g., when many studies examine the same health
outcome, focusing on toxicity studies including exposures below a specified range, those studies
examining more specific or objective measures of toxicity, or those that address lifestage- or
exposure duration-specific knowledge on how the health outcome develops). Given the relatively
small databases of animal toxicological studies for these five PFAS (see Section 2.3.2), this
narrowing is not considered applicable to these data. Thus, for these five PFAS assessments, all
relevant health outcomes in the animal toxicological studies meeting PECO criteria will be
considered.
In contrast to the animal studies, there are many epidemiology studies. To make the
systematic review of the epidemiology literature more pragmatic and efficient and focus the set of
studies undergoing study evaluation, one epidemiologist per outcome performed an initial review
of the available evidence examining at a high level the direction and consistency of observed
associations. Based on this initial review, outcomes were classified into one of three tiers:
(1) formal systematic review, (2) rapid review (reduced rigor; study evaluation with a single
reviewer), or (3) no further review (no study evaluation or synthesis of the evidence, although the
available database might be mentioned in the assessments to inform data gaps). Most outcomes
were classified into the first tier (formal systematic review). Outcomes with an a priori serious
concern for reverse causality (e.g., clear link to elimination of PFAS from the body, such as
outcomes related to menstruation or renal function) were classified into the second tier (rapid
review) because of the large amount of uncertainty in interpreting these results. These outcomes
included renal function (e.g., glomerular filtration rate), menstrual cycle characteristics,
endometriosis, polycystic ovary syndrome, and albumin. The third tier (no further review) was
used primarily for outcomes where the results for available studies were null and the study
sensitivity was poor, due to, for example, PFAS exposure levels being below or near the LOD. This
included penile width, sex ratio, hematologic effects, and mortality. For the second and third tiers,
it is possible that new data could change their classification. The outcomes included in the
assessment are summarized in Table 5-1, which also indicates those undergoing only rapid review.
This approach of tiered reviews is consistent with recommendations from the National
Academies of Science encouraging the U.S. Environmental Protection Agency (EPA) to explore ways
to make systematic review more feasible, including conducting a "rapid review in which
This document is a draft for review purposes only and does not constitute Agency policy.
5-1 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
Relevant human health
effect category3 '5
Examples of epidemiology outcomes included
Hepatic (toxicity)
• Serum liver enzymes (e.g., ALT, AST, total bilirubin from clinical chemistry)
• Liver disease
Cardiovascular (toxicity)
• Serum lipids (note: also, informative to hepatic)
• Blood pressure
• Atherosclerosis
• Cardiovascular disease
• Ventricular geometry
Immune (effects)
• Antibody response
• Hypersensitivity (asthma, allergy, atopic dermatitis)
• Infections
Urinary (toxicity)
• Renal function tests (e.g., glomerular filtration rate) (RR)
Endocrine (effects)
• Thyroid hormones
• Thyroid disease
Metabolic (effects)
• Diabetes
• Gestational diabetes
• Insulin resistance
• Serum glucose
• Adiposity (e.g., BMI, weight gain)
• Metabolic syndrome
Reproductive (toxicity)
Note: Evidence synthesis and
evidence integration
conclusions in assessments
are developed separately for
male and female reproductive
effects (toxicity)
• Reproductive hormones
• Fecundity
• Semen parameters
• Anogenital distance
This document is a draft for review purposes only and does not constitute Agency policy.
5-2 DRAFT-DO NOT CITE OR QUOTE
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
components of the systematic review process are simplified or omitted (e.g., the need for two
independent reviewers)" fNASEM. 20171.
Table 5-1. Epidemiology outcome grouping categories
-------
1
2
3
4
5
6
7
8
9
10
11
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Relevant human health
effect category3 '5
Examples of epidemiology outcomes included
• Female reproductive conditions (endometriosis, polycystic ovary syndrome)
(RR)
• Ovarian reserve
• Menstrual cycle characteristics (RR)
• Pubertal development
Developmental (effects)
Note: Evidence synthesis of
these endpoints in the
assessments is termed
"offspring growth and early
development," but evidence
integration conclusions will be
drawn on the broader
category of "developmental
effects" (which also considers
organ/system-specific effects
after exposure during
development)
• Birth size/fetal growth restriction
• Preterm birth/gestational duration
• Postnatal growth
• Spontaneous abortion
RR = rapid review; ALT = alanine aminotransferase; AST = aspartate aminotransferase; BMI = body mass index.
aThe primary focus of these assessments will be on developmental effects; endocrine effects; hepatic effects,
including lipid and lipoprotein measures; immune effects; reproductive effects in males or females; urinary
effects; general toxicity; and carcinogenicity. Data on cardiovascular effects, hematological effects, metabolic
effects including diabetes, and nervous system effects will be summarized when available. These summaries may
be developed in association with one of the health effects noted above, as a separate formal evaluation of
hazard, or as part of a qualitative summary on "other effects," depending on the assessment-specific data.
bSome outcomes are relevant to multiple health effects. These outcomes may be categorized under only a single
health effect in Table 5-2 for clarity. However, in the assessments, such outcome data would be discussed in the
first relevant health effect synthesis (syntheses will generally follow the pattern of most to least available
evidence) and then this synthesis will be cited in the syntheses of other relevant health effects. The evidence (for
or against an effect) will contribute to evidence integration decisions for all relevant health effects.
To promote consistency in evaluation and presentation across assessments, preliminary
decisions were made regarding the grouping of related endpoints for outcome-specific study
evaluations and discussion in the evidence synthesis. This helps implement the study evaluation
criteria (see Section 6) because those evaluations are outcome and analysis specific. Preliminary
decisions for grouping of endpoints from animal toxicological studies for discussion within each
assessed human health effect category are described in Table 5-2. Parallel groupings for outcomes
assessed in the available epidemiology studies are captured in Table 5-1. These groupings are
meant to serve as a starting place for consistency in presentation and analysis across studies and
assessments, although assessment-specific deviations are possible (e.g., depending on the
assessment-specific database of endpoints in the available studies or PFAS-specific understanding
of mechanistic relationships across outcomes).
This document is a draft for review purposes only and does not constitute Agency policy.
5-3 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 5-2. Animal endpoint grouping categories
Relevant human health
effect category3
Examples of animal endpoints included
Notes
General toxicity
• Body weight (not maternal or pup weights, or
weights after developmental-only exposure)
• Mortality, survival, or LD50S
• Growth curve
• Clinical observations (nonbehavioral)
• Clinical chemistry
endpoints are under
Hepatic or Hematologic
• Maternal or pup
body-weight endpoints
are under Developmental
• Pathology (including gross
lesions) is organ specific
Hepatic (toxicity)15
• Liver weight and histopathology
• Serum or tissue liver enzymes (e.g., ALT and
AST from clinical chemistry)
• Other liver tissue enzyme activity
(e.g., catalase) or protein/DNA content
• Other liver tissue biochemical markers (e.g.,
albumin; glycogen; glucose)
• Liver-specific serum biochemistry
(e.g., albumin; albumin/globulin)
• Liver tissue lipids: triglycerides, cholesterol
• Serum lipids (Note: also, informative to
cardiovascular0)
• Biochemical markers such
as albumin or glucose are
under Hematological
• Liver tissue cytokines are
under Immune
• Serum glucose is under
Metabolic
Cardiovascular (toxicity)b d
• Heart weight and histopathology
• Serum lipids (note: also, informative to
Hepatic)
• Blood pressure
• Blood measures of cardiovascular risk (e.g.,
C-reactive protein)
• Other blood measures are
under Hepatic, Immune,
or Hematologic
Hematologic (effects)b d
• Red blood cells
• Blood hematocrit or hemoglobin
• Corpuscular volume
• Blood platelets or reticulocytes
• Blood biochemical measures (e.g., sodium,
calcium, phosphorus)
• White blood cell count
and globulin are under
Immune
• Serum lipids are under
Cardiovascular
• Serum liver markers are
under Hepatic
This document is a draft for review purposes only and does not constitute Agency policy.
5-4 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Relevant human health
effect category3
Examples of animal endpoints included
Notes
Immune (effects)15
• Host resistance
• Allergic, autoimmune or infectious disease
• Hypersensitivity
• General immune assays (e.g., white blood cell
counts, immunological factors or cytokines in
blood, lymphocyte phenotyping or
proliferation)
• Any measure in lymphoid tissues (weight;
histopathology; cell counts; etc.)
• Immune cell counts or immune-specific
cytokines in nonlymphoid tissues
• Other immune functional assays
(e.g., antibody production, natural killer cell
function, DTH, MLR, CTL, phagocytosis or
bacterial killing by monocytes)
• Immune responses in the respiratory system
• Stress-related factors in blood
(e.g., glucocorticoids or other adrenal
markers)
• Red blood cells are under
Hematological
• Nonimmune measures of
pulmonary function are
under Respiratory
Urinary (toxicity)15
• Kidney weight and histopathology
• Urinary measures (e.g., protein, volume, pH,
specific gravity)
Nervous system (effects)b d
• Brain weight
• Behavioral measures (including FOB and
cage-side observations)
• Nervous system histopathology
Endocrine (effects)15
• Thyroid weight and histopathology
• Hormonal measures in any tissue or blood
(nonreproductive)
• Reproductive hormones
are under Reproductive
This document is a draft for review purposes only and does not constitute Agency policy.
5-5 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Relevant human health
effect category3
Examples of animal endpoints included
Notes
Metabolic (effects)bd
• Free fatty acids
• Serum glucose or insulin, or other measures
related to diabetes
• Pancreatic effects relevant to diabetes
• Induced-obesity or BMI
• Any of the above endpoints after
developmental exposure will be primarily
discussed in this health effect category, and
then referenced under developmental effects
Reproductive (toxicity)15
Note: Evidence synthesis
and evidence integration
conclusions in assessments
are developed separately
for male and female
reproductive effects
(toxicity)
• Reproductive organ weight and
histopathology
• Markers of sexual differentiation or
maturation (e.g., preputial separation in
males; vaginal opening or estrous cycling in
females)
• Mating parameters (e.g., success, mount
latency)
• Reproductive hormones
• Birth parameters
(e.g., litter size;
resorptions,
implantations, viability)
are under Developmental
• If data indicate altered
birth parameters are likely
attributable to female
fertility, these data may
be discussed under
Female Reproductive
Developmental (effects)15
Note: Evidence synthesis of
these endpoints in the
assessments is termed
"offspring growth and early
development," but
evidence integration
conclusions will be drawn
on the broader category of
"developmental effects"
(which also considers
organ/system-specific
effects after exposure
during development)
• Dam health (e.g., weight gain, food
consumption)
• Pup viability/survival or other birth
parameters (e.g., number of pups per litter)
• Pup weight or growth (includes measures
into adulthood after developmental-only
exposure)
• Developmental landmarks (eye opening, etc.,
but not including markers for other
organ/system-specific toxicities)
• Histopathology and
markers of development
specific to other systems
are organ/system-specific
(e.g., vaginal opening is
under Female
Reproductive; tests of
sensory maturation are
under Nervous System)
Carcinogenicity13
• Tumors
• Precancerous lesions (e.g., dysplasia)
ALT = alanine aminotransferase; AST = aspartate aminotransferase; BMI = body mass index; CTL = cytotoxic T
lymphocyte; DNA = deoxyribonucleic acid; DTH = delayed-type hypersensitivity; FOB = functional operational
battery; LD50 = median lethal dose; MLR = mixed leukocyte reaction.
This document is a draft for review purposes only and does not constitute Agency policy.
5-6 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
aGiven the paucity of available studies and the absence of exceptional new evidence, information on
gastrointestinal effects, musculoskeletal effects, ocular effects, and respiratory effects will not be formally
evaluated in these assessments, although short summaries of the evidence may be included for context, and new
literature relating to these outcomes will be monitored during literature search updates for potential inclusion.
bAny of the health effect-relevant endpoints observed after developmental exposure will be discussed primarily in
the health effect category indicated, and then referenced under developmental effects.
cSome outcomes are relevant to multiple health effects. These outcomes may be categorized under only a single
health effect in Table 5-2 for clarity. However, in the assessments, such outcome data would be discussed in the
first relevant health effect synthesis (syntheses will generally follow the pattern of most-to-least available
evidence) and then this synthesis will be cited in the syntheses of other relevant health effects. The evidence (for
or against an effect) will contribute to evidence integration decisions for all relevant health effects.
dThe primary focus of these assessments will be on developmental effects; endocrine effects; hepatic effects,
including lipid and lipoprotein measures; immune effects; reproductive effects in males or females; urinary
effects; general toxicity; and carcinogenicity. Data on cardiovascular effects, hematological effects, metabolic
effects including diabetes, and nervous system effects will be summarized when available. These summaries may
be developed in association with one of the health effects noted above, either as a separate formal evaluation of
hazard or as part of a qualitative summary on "other effects," depending on the assessment-specific data.
Assessment-specific refinements to the evaluation plan described in later sections of this
protocol may be justified after review of the key areas of scientific complexity outlined in
Section 2.4. Although not expected based on the relatively small database of studies for these PFAS,
one such refinement includes the potential prioritization of studies testing specific (lower)
exposure levels, exposure lifestages, or routes of exposure, as identified based on conclusions made
regarding the ADME properties of these PFAS. As noted in Section 2.4, consideration of the
available ADME data for these five PFAS will be prioritized (see additional discussion in
Section 9.2). This will serve multiple purposes, including updating the data in Table 2-7 on serum
half-lives across species and sexes. Notably, it is not expected that there will be enough data to
examine lifestage-specific differences in ADME (including metabolic pathways for toxification or
detoxification) that might inform evidence evaluation and synthesis decisions. (This is distinct
from lifestage-specific differences in exposure, for example, due to the higher intake of food per kg
body weight [BW] of young children, ingestion of dust, or maternal transfer via breastmilk.)
However, a few anticipatory refinements will be applied to study evaluations based on the
preliminary data presented in Table 2-7. Specifically, given the apparent sex-specific differences in
PFAS half-life in rats and mice (note: toxicological studies in nonhuman primates were not
identified for these PFAS), examining and reporting data for both sexes will be reviewed as a
potential source of study insensitivity during study evaluation (see Section 6.3), particularly for
PFAS that seem to vary largely for this parameter (e.g., PFHxS; PFNA). These half-life data will also
be considered when evaluating the experiment-specific sensitivity of the frequency of exposures
and the timing of endpoint testing after exposure in experimental animals (see Section 6.3), as well
as the potential for using exposure biomarkers in exposed animals and humans (see Sections 6.2
and 6.3). The apparent ADME differences across species will be a critical consideration in these
assessments. This consideration will be applied to evidence synthesis and integration decisions
(e.g., exploring ADME differences between rats and mice as a potential explanation if there are
This document is a draft for review purposes only and does not constitute Agency policy.
5-7 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
differences in sensitivity in outcome-specific responses; see Sections 9 and 10), as well as in
extrapolating dosimetry (i.e., exposure levels and duration) from experiments in animal models to
quantitative estimates relevant to humans, possibly including application within existing
pharmacokinetic (PK) and PBPK models (see additional discussion in Sections 6.4 and 11.2).
Lastly, based on the key areas of scientific complexity outlined in Section 2.4, some of the
analyses performed in support of these assessments may need to consider a broader array of
studies than those available for these five PFAS. One example includes the need to consider the
human relevance of certain outcomes observed in animals, including the role of receptors such as
PPARa (see additional discussion in Section 9.2). In addition, it is possible that additional literature
identification and evaluation strategies will be developed to address the other key areas of
scientific complexity outlined in Section 2.4, or other assessment-specific issues that arise during
review. Any such approaches will be documented in updates to the PFAS-specific assessment
protocol(s) (i.e., as assessment-specific updates to this document included as Addendum materials
for each of the five PFAS assessments).
This document is a draft for review purposes only and does not constitute Agency policy.
5-8 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
6.STUDY EVALUATION (REPORTING, RISK OF BIAS,
AND SENSITIVITY) STRATEGY
The general approach for evaluating PECO-relevant primary health effect studies is
described in Section 6.1 and is the same for epidemiology studies and animal toxicology
experiments, but the specifics of applying the approach differ; thus, they are described separately
for epidemiology and animal toxicological studies in Sections 6.2 and 6.3, respectively. No
controlled human exposure studies for these PFAS were identified (see Section 4). PBPK modeling
studies were recently identified for PFHxS fKim etal.. 20181 and for PFDA and PFNA fKim etal..
2019). although they were not formally identified by the systematic literature searches completed
prior to posting of this protocol. In addition, a two-compartment PK model for gestational and
lactational transfer of PFHxS in humans has been described by Verner etal. f20161. The specific
approach for reviewing these studies is described in Section 6.4. Different approaches are used to
evaluate mechanistic studies (see Sections 6.5 and 9.2).
6.1. STUDY EVALUATION OVERVIEW FOR HEALTH EFFECT STUDIES
Key concerns for the review of epidemiology and animal toxicological studies are potential
bias (factors that affect the magnitude or direction of an effect in either direction) and insensitivity
(factors that limit the ability of a study to detect a true effect; low sensitivity is a bias towards the
null when an effect exists). The study evaluations are aimed at discerning the expected magnitude
of any identified limitations (focusing on limitations that could substantively change a result),
considering also the expected direction of the bias. The study evaluation approach is designed to
address a range of study designs, health effects, and chemicals. The general approach for reaching
an overall judgment for the study (or a specific analysis within a study) regarding confidence in the
reliability of the results is illustrated in Figure 6-1.
This document is a draft for review purposes only and does not constitute Agency policy.
6-1 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
(a) Study evaluation process (b)
Individual evaluation domains
Develop assessment-
specific considerations
Pilot testing
(and possible refinement)
Independent evaluation by
two reviewers
Finalize domain judgments
and overall study rating
Animal
Epidemiology
Selection and performance
¦ Allocation
Participant selection
¦ Observational bias/blinding
Confounding/variable control
Confounding
Selective reporting and attrition
Selective reporting
Exposure methods sensitivity
¦ Chemical administration and
Exposure measurement
characterization
¦ Exposure timing, frequency, and duration
Outcome measures and results display
Outcome ascertainment
• Endpoint sensitivity and specificity
Analysis
• Results presentation
Reporting quality
Other sensitivity
Domain judgments
Judgment
0 Good
Adequate
I®
Deficient
Critically
Deficient
Interpretation
Appropriate study conduct relating to the domain and
minor deficiencies not expected to influence results
A study that may have some limitations relating to the
domain, but they are not likely to be severe or to
have a notable impact on results-
Identified biases or deficiencies interpreted as likely
to have had a notable impact on the results or
prevent reliable interpretation of study findings.
A serious flaw identified that makes the observed
effect(s) uninterpretable. Studies with a critical
deficiency will almost always be considered
*un informative" overall.
Overall study rating for an outcome
Rating
Interpretation
High
No notable deficiencies or concerns identified; potential
for bias unlikely or minimal; sensitive methodology.
Medium
Possible deficiencies or concerns noted, but resulting
bias or lack of sensitivity is unlikely to be of a notable
degree.
Low
Deficiencies or concerns were noted, and the potential
for substantive bias or inadequate sensitivity could have
a significant impact on the study results or their
interpretation.
Uninformative
Serious flaw(s) makes study results unusable for hazard
identification or dose response.
Figure 6-1. Overview of Integrated Risk Information System (IRIS) study
evaluation process, [a] An overview of the general evaluation process (note: see
Section 5 for deviations from independent evaluation by two reviewers for some
health outcomes in epidemiology studies), (b) The evaluation domains and
definitions for ratings (i.e., domain and overall judgments, performed on an
outcome-specific basis].
This document is a draft for review purposes only and does not constitute Agency policy,
6-2 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
With the exceptions noted in the refined evaluation plan for select outcomes reported in
epidemiology studies (see Section 5), at least two reviewers will independently evaluate health
effect studies to identify characteristics that bear on the informativeness (i.e., validity and
sensitivity) of the results. The independent reviewers will use the structured platform for study
evaluation housed within the Environmental Protection Agency's (EPA's) version of the Health
Assessment Workplace Collaboration (HAWC)11 to record separate judgements for each domain
and the overall study for each outcome, to reach consensus between reviewers, and when
necessary, resolve differences by discussion between the reviewers or consultation with additional
independent reviewers. For some domains, additional chemical- or outcome-specific knowledge
will be applied to evaluating the experimental design and methodology, as described below.
In general, considerations for reviewing a study with regard to its conduct for specific
health outcomes is based on the use of existing guidance documents when available, including EPA
guidance for carcinogenicity, neurotoxicity, reproductive toxicity, and developmental toxicity (U.S.
EPA. 2005a. 1998.1996b. 1991a). For some aspects of the study evaluations (e.g., review of
exposure assessment in epidemiology studies), additional considerations are developed in
consultation with topic-specific technical experts. To calibrate the assessment-specific
considerations, the study evaluations will include a pilot phase to assess and refine the evaluation
process. Additionally, as reviewers examine a group of studies, additional chemical-specific
knowledge or methodologic concerns may emerge and a second pass of all pertinent studies may
become necessary. Refinements to the study evaluation process made during the pilot phase and
subsequent implementation across all relevant studies will be acknowledged as updates to the
protocol.
Authors may be queried to obtain critical information, particularly that involving missing
reporting quality information or other data (e.g., information on variability) or additional analyses
that could address potential study limitations. The decision on whether to seek missing
information includes consideration of what additional information would be useful, specifically
with respect to any information that could result in a reevaluation of the overall study confidence
for an outcome. Outreach to study authors will be documented and considered unsuccessful if
researchers do not respond to an email or phone request within one month of the attempt to
contact. Only information or data that can be made publicly available (e.g., within HAWC or Health
and Environmental Research Online [HERO]) will be considered.
nHAWC is a free and open source software application that provides a modular, web-based interface to help
develop human health assessments of chemicals: https://hawcproiect.org/portal/. Standard operating
procedures provided to the reviewers to facilitate consistent and relevant documentation of their judgments
using the HAWC software can be found as attachments embedded within the online tool
(https://hawcprd.epa.gov/assessment/100000Q39/).
This document is a draft for review purposes only and does not constitute Agency policy.
6-3 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
1 When evaluating studies12 that examine more than one outcome, the evaluation process will
2 be performed separately for each outcome, because the utility of a study can vary for different
3 outcomes. If a study examines multiple endpoints for the same outcome,13 evaluations may be
4 performed at a more granular level if appropriate, but these measures may still be grouped for
5 evidence synthesis.
6 During review, the reviewers will reach a consensus judgment of good, adequate, deficient,
7 not reported, or critically deficient for each evaluation domain. If a consensus is not reached, a third
8 reviewer will perform conflict resolution. It is important to stress that these evaluations are
9 performed in the context of the study's utility for identifying individual hazards. While limitations
10 specific to the usability of the study for dose-response analysis are useful to note for informing
11 those later decisions, they do not contribute to the study confidence classifications.
12 These four categories are applied to each evaluation domain for each study as follows:
13 • Good represents a judgment that the study was conducted appropriately in relation to the
14 evaluation domain and that any minor deficiencies noted would not be expected to
15 influence the study results.
16 • Adequate indicates a judgment that there may be methodological limitations relating to the
17 evaluation domain, but they are not likely to be severe or to have a notable impact on the
18 results.
19 • Deficient denotes identified biases or deficiencies that are interpreted as likely to have had a
20 notable impact on the results or that prevent interpretation of the study findings.
21 • Not reported indicates that the information necessary to evaluate the domain question was
22 not available in the study. Generally, this term carries the same functional interpretation as
23 deficient for the purposes of the study confidence classification (described below).
24 Depending on the number of unreported items and severity of other limitations identified in
25 the study, it may or may not be worth reaching out to the study authors for this information
26 (see discussion above).
27 • Critically deficient reflects a judgment that the study conduct relating to the evaluation
28 domain question introduced a serious flaw that is interpreted to be the primary driver of
29 any observed effect(s) or makes the study uninterpretable. Studies with a determination of
30 critically deficient in an evaluation domain will not be used for hazard identification or
31 dose-response analysis, but they may be used to highlight possible research gaps. Given
12"study" is used instead of a more accurate term (e.g., "experiment") throughout these sections owing to an
established familiarity within the field for discussing a study's risk of bias or sensitivity, etc. However, all
evaluations discussed herein are explicitly conducted at the level of an individual outcome within a population or
cohort of humans or animals exposed in a similar manner (e.g., unexposed or exposed at comparable lifestages
and for the same duration of exposure), or to a sample of the study population within a study.
13Note: "outcome" will be used throughout these methods; the same methods also apply to an endpoint assessed
separately within a larger outcome. "Endpoint" refers to a more granular measurement (e.g., for the outcome of
liver histopathology, different endpoints might include necrosis and cellular hypertrophy).
This document is a draft for review purposes only and does not constitute Agency policy.
6-4 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
this potential for exclusion, this classification is used infrequently and with extreme care;
methodological limitations warranting this classification are defined a priori on an
exposure- and outcome-specific basis and are inherently severe enough to warrant
exclusion based on a single critical deficiency. Serious flaws that do not warrant study
exclusion will be classified as deficient.
Once the evaluation domains have been rated, the identified strengths and limitations will
be considered as a whole to reach a study confidence classification of high, medium, or low
confidence, or uninformative for a specific health outcome. This classification is based on the
reviewer judgments across the evaluation domains and considers the likely impact that inadequate
reporting or the noted deficiencies in bias and sensitivity have on the outcome-specific results.
There are no predefined weights for the domains, and the reviewers are responsible for applying
expert judgment to make this determination. The classifications, which reflect a consensus
judgment between reviewers, are defined as follows:
• High confidence: No notable deficiencies or concerns were identified; the potential for bias
is unlikely or minimal, and the study used sensitive methodology. High confidence studies
generally reflect judgments of good across all or most evaluation domains.
• Medium confidence: Possible deficiencies or concerns were noted, but the limitations are
unlikely to be of a notable degree. Generally, medium confidence studies include adequate
or good judgments across most domains, with the impact of any identified limitation not
being judged as severe.
• Low confidence: Deficiencies or concerns are noted, and the potential for bias or inadequate
sensitivity could have a significant impact on the study results or their interpretation.
Typically, low confidence studies have a deficient evaluation for one or more domains,
although some medium confidence studies may have a deficient rating in domain(s)
considered to have less influence on the magnitude or direction of the outcome-specific
results). Low confidence results are given less weight than high or medium confidence
results during evidence synthesis and integration (see Section 10.1, Table 10-3 and
Table 11-1), and are generally notused for hazard identification or dose-response analyses
unless they are the only studies available or they inform data gaps unexamined in the high
or medium confidence studies. Studies rated as medium or low confidence only because of
sensitivity concerns about bias towards the null will be asterisked or otherwise noted
because they may require additional consideration during evidence synthesis. Effects
observed in studies biased toward the null may increase confidence in the results, assuming
the study is otherwise well conducted (see Section 9).
• Uninformative: Serious flaw(s) make the study results unusable for hazard identification.
Studies with critically deficient judgements in any evaluation domain are almost always
classified as uninformative (see explanation above). Studies with multiple deficient
judgments across domains may also be considered uninformative. Uninformative studies
will not be considered further in the synthesis and integration of evidence, except perhaps
to highlight possible research gaps.
This document is a draft for review purposes only and does not constitute Agency policy.
6-5 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
As previously noted, study evaluation determinations reached by each reviewer and the
consensus judgment between reviewers will be recorded in HAWC. Final study evaluations housed
in HAWC, including for each domain and overall study confidence, will be made available when the
draft is publicly released. These classifications and their rationales will be carried forward and
considered as part of evidence synthesis (see Section 9) to help interpret the results across studies.
6.2. EPIDEMIOLOGY STUDY EVALUATION
Evaluation of epidemiology studies of health effects to assess risk of bias and study
sensitivity will be conducted for the following domains: exposure measurement, outcome
ascertainment, participant selection, potential confounding, analysis, study sensitivity, and selective
reporting. Bias can result in false positives and negatives (i.e., Types I and II errors), while study
sensitivity is typically concerned with identifying the latter.
The principles and framework used for evaluating epidemiology studies are based on the
Cochrane Risk of Bias in Nonrandomized Studies of Interventions [ROBINS-I; Sterne etal. f20161]
but modified to address environmental and occupational exposures. The underlying philosophy of
ROBINS-I is to describe attributes of an "ideal" study with respect to each of the evaluation domains
(e.g., exposure measurement, outcome classification, etc.). Core and prompting questions are used
to collect information to guide evaluation of each domain.
Core and prompting questions for each domain are presented in Table 6-1. Core questions
represent key concepts, while the prompting questions help the reviewer focus on relevant details
under each key domain. Table 6-1 also includes criteria that apply to all exposures and outcomes.
PFAS-specific criteria are described in Section 6.2.1. As mentioned in Section 6.1, any additions to
or refinements of the criteria will be documented as updates to the protocol.
This document is a draft for review purposes only and does not constitute Agency policy.
6-6 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 6-1. Questions and criteria for evaluating each domain in epidemiology studies
Domain and
core question
Prompting questions
Follow-up questions
Criteria that apply to most exposures and outcomes
Exposure
measurement
Does the
exposure
measure reliably
distinguish
between levels
of exposure in a
time window
considered most
relevant for a
causal effect
with respect to
the development
of the outcome?
For all:
• Does the exposure measure
capture the variability in exposure
among the participants,
considering intensity, frequency,
and duration of exposure?
• Does the exposure measure reflect
a relevant time window? If not,
can the relationship between
measures in this time and the
relevant time window be
estimated reliably?
• Was the exposure measurement
likely to be affected by a
knowledge of the outcome?
• Was the exposure measurement
likely to be affected by the
presence of the outcome (i.e.,
reverse causality)?
For case-control studies of occupational
exposures:
• Is exposure based on a
comprehensive job history
describing tasks, setting, time-
period, and use of specific
materials?
Is the degree of
exposure
misclassification
likely to vary by
exposure level?
If the correlation
between exposure
measurements is
moderate, is there an
adequate statistical
approach to
ameliorate variability
in measurements?
If there is a concern
about the potential
for bias, what is the
predicted direction
or distortion of the
bias on the effect
estimate (if there is
enough
information)?
Good
• Valid exposure assessment methods used, which represent
the etiologically relevant time-period of interest.
• Exposure misclassification is expected to be minimal.
Adequate
• Valid exposure assessment methods used, which represent
the etiologically relevant time-period of interest.
• Exposure misclassification may exist but is not expected to
greatly change the effect estimate.
Deficient
• Valid exposure assessment methods used, which represent
the etiologically relevant time-period of interest. Specific
knowledge about the exposure and outcome raise concerns
about reverse causality, but there is uncertainty whether it is
influencing the effect estimate.
• Exposed groups are expected to contain a notable proportion
of unexposed or minimally exposed individuals, the method
did not capture important temporal or spatial variation, or
there is other evidence of exposure misclassification that
would be expected to notably change the effect estimate.
This document is a draft for review purposes only and does not constitute Agency policy.
6-7 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Domain and
core question
Prompting questions
Follow-up questions
Criteria that apply to most exposures and outcomes
For biomarkers of exposure, general
population:
• Is a standard assay used? What
are the intra and interassay
coefficients of variation? Is the
assay likely to be affected by
contamination? Are values less
than the limit of detection dealt
with adequately?
• What exposure time-period is
reflected by the biomarker? If the
half-life is short, what is the
correlation between serial
measurements of exposure?
Critically deficient
• Exposure measurement does not characterize the
etiologically relevant time-period of exposure or is not valid.
• There is evidence that reverse causality is very likely to
account for the observed association.
• Exposure measurement was not independent of outcome
status.
Outcome
ascertainment
Does the
outcome
measure reliably
distinguish the
presence or
absence (or
degree of
severity) of the
outcome?
For all:
• Is outcome ascertainment likely to
be affected by knowledge of, or
presence of, exposure
(e.g., consider access to health
care, if based on self-reported
history of diagnosis)?
For case-control studies:
• Is the comparison group without
the outcome (e.g., controls in a
case-control study) based on
objective criteria with little or no
likelihood of inclusion of people
with the disease?
Is there a concern
that any outcome
misclassification is
nondifferential,
differential, or both?
What is the predicted
direction or
distortion of the bias
on the effect
estimate (if there is
enough
information)?
Good
• High certainty in the outcome definition (i.e., specificity and
sensitivity), minimal concerns with respect to
misclassification.
• Assessment instrument was validated in a population
comparable to the one from which the study group was
selected.
Adequate
• Moderate confidence that outcome definition was specific
and sensitive, some uncertainty with respect to
misclassification but not expected to greatly change the
effect estimate.
• Assessment instrument was validated but not necessarily in a
population comparable to the study group.
This document is a draft for review purposes only and does not constitute Agency policy.
6-8 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Domain and
core question
Prompting questions
Follow-up questions
Criteria that apply to most exposures and outcomes
For mortality measures:
• How well does cause-of-death
data reflect occurrence of the
disease in an individual? How well
do mortality data reflect incidence
of the disease?
For diagnosis of disease measures:
• Is the diagnosis based on standard
clinical criteria? If it is based on
self-report of the diagnosis, what
is the validity of this measure?
For laboratory-based measures
(e.g., hormone levels):
• Is a standard assay used? Does
the assay have an acceptable level
of interassay variability? Is the
sensitivity of the assay appropriate
for the outcome measure in this
study population?
Deficient
• Outcome definition was not specific or sensitive.
• Uncertainty regarding validity of assessment instrument.
Critically deficient
• Invalid/insensitive marker of outcome.
• Outcome ascertainment is very likely to be affected by
knowledge of, or presence of, exposure.
Note: Lack of blinding should not be automatically construed to be
critically deficient.
Participant
selection
Is there evidence
that selection
into or out of the
study (or analysis
sample) was
jointly related to
exposure and to
outcome?
For longitudinal cohort:
• Did participants volunteer for the
cohort based on knowledge of
exposure and/or preclinical
disease symptoms? Was entry
into the cohort or continuation in
the cohort related to exposure and
outcome?
Were differences in
participant
enrollment and
follow-up evaluated
to assess bias?
If there is a concern
about the potential
for bias, what is the
predicted direction
or distortion of the
bias on the effect
Good
• Minimal concern for selection bias based on description of
recruitment process (e.g., selection of comparison
population, population-based random sample selection,
recruitment from sampling frame including current and
previous employees).
• Exclusion and inclusion criteria specified and would not
induce bias.
This document is a draft for review purposes only and does not constitute Agency policy.
6-9 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Domain and
core question
Prompting questions
Follow-up questions
Criteria that apply to most exposures and outcomes
For occupational cohort:
• Did entry into the cohort begin
with the start of the exposure?
• Was follow-up or outcome
assessment incomplete, and if so,
was follow-up related to both
exposure and outcome status?
• Could exposure produce
symptoms that would result in a
change in work assignment/work
status ("healthy worker survivor
effect")?
For case-control study:
• Were controls representative of
population and time periods from
which cases were drawn?
• Are hospital controls selected from
a group whose reason for
admission is independent of
exposure?
• Could recruitment strategies,
eligibility criteria, or participation
rates result in differential
participation relating to both
disease and exposure?
For population-based survey:
• Was recruitment based on
advertisement to people with
knowledge of exposure, outcome,
and hypothesis?
estimate (if there is
enough
information)?
Were appropriate
analyses performed
to address changing
exposures over time
in relation to
symptoms?
Is there a comparison
of participants and
nonparticipants to
address whether
differential selection
is likely?
• Participation rate is reported at all steps of study (e.g., initial
enrollment, follow-up, selection into analysis sample). If rate
is not high, there is appropriate rationale for why it is unlikely
to be related to exposure (e.g., comparison between
participants and nonparticipants or other available
information indicates differential selection is not likely).
Adequate
• Enough of a description of the recruitment process to be
comfortable that there is no serious risk of bias.
• Inclusion and exclusion criteria specified and would not
induce bias.
• Participation rate is incompletely reported but available
information indicates participation is unlikely to be related to
exposure.
Deficient
• Little information on recruitment process, selection strategy,
sampling framework, and/or participation. Or aspects of
these processes raise the potential for bias (e.g., healthy
worker effect, survivor bias).
Critically deficient
• Aspects of the processes for recruitment, selection strategy,
sampling framework, or participation result in concern that
selection bias is likely to have had a large impact on effect
estimates (e.g., convenience sample with no information
about recruitment and selection, cases and controls are
recruited from different sources with different likelihood of
exposure, recruitment materials stated outcome of interest
and potential participants are aware of or are concerned
about specific exposures).
This document is a draft for review purposes only and does not constitute Agency policy.
6-10 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Domain and
core question
Prompting questions
Follow-up questions
Criteria that apply to most exposures and outcomes
Confounding
Is confounding of
the effect of the
exposure likely?
Is confounding adequately addressed by
considerations in:
• Participant selection (matching or
restriction)?
• Accurate information on potential
confounders and statistical
adjustment procedures?
• Lack of association between
confounder and outcome, or
confounder and exposure in the
study?
• Information from other sources?
Is the assessment of confounders based on
a thoughtful review of published literature,
potential relationships (e.g., as can be
gained through directed acyclic graphing),
and minimizing potential overcontrol
(e.g., inclusion of a variable on the pathway
between exposure and outcome)?
If there is a concern
about the potential
for bias, what is the
predicted direction
or distortion of the
bias on the effect
estimate (if there is
enough
information)?
Good
• Conveys strategy for identifying key confounders. This may
include a priori biological considerations, published
literature, causal diagrams, or statistical analyses, with the
recognition that not all "risk factors" are confounders.
• Inclusion of potential confounders in statistical models not
based solely on statistical significance criteria (e.g., p < 0.05
from stepwise regression).
• Does not include variables in the models that are likely to be
influential colliders or intermediates on the causal pathway.
• Key confounders are evaluated appropriately and considered
to be unlikely sources of substantial confounding. This often
will include:
o Presenting the distribution of potential confounders by
levels of the exposure of interest and/or the outcomes of
interest (with amount of missing data noted);
o Consideration that potential confounders were rare
among the study population, or were expected to be
poorly correlated with exposure of interest;
o Consideration of the most relevant functional forms of
potential confounders;
o Examination of the potential impact of measurement
error or missing data on confounder adjustment; or
o Presenting a progression of model results with
adjustments for different potential confounders, if
warranted.
This document is a draft for review purposes only and does not constitute Agency policy.
6-11 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Domain and
core question
Prompting questions
Follow-up questions
Criteria that apply to most exposures and outcomes
Adequate
• Similar to good but may not have included all key
confounders, or less detail may be available on the
evaluation of confounders (e.g., sub-bullets in good). It is
possible that residual confounding could explain part of the
observed effect, but concern is minimal.
Deficient
• Does not include variables in the models that have been
shown to be influential colliders or intermediates on the
causal pathway.
• And any of the following:
o The potential for bias to explain some of the results is
high based on an inability to rule out residual
confounding, such as a lack of demonstration that key
confounders of the exposure-outcome relationships
were considered;
o Descriptive information on key confounders (e.g., their
relationship relative to the outcomes and exposure
levels) are not presented; or
o Strategy of evaluating confounding is unclear or is not
recommended (e.g., only based on statistical
significance criteria or stepwise regression [forward or
backward elimination]).
Critically deficient
• Includes variables in the models that are colliders and/or
intermediates in the causal pathway, indicating that
substantial bias is likely from this adjustment; or
• Confounding is likely present and not accounted for,
indicating that all the results were most likely due to bias.
This document is a draft for review purposes only and does not constitute Agency policy.
6-12 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Domain and
core question
Prompting questions
Follow-up questions
Criteria that apply to most exposures and outcomes
Analysis
Does the analysis
strategy and
presentation
convey the
necessary
familiarity with
the data and
assumptions?
• Are missing outcome, exposure,
and covariate data recognized, and
if necessary, accounted for in the
analysis?
• Does the analysis appropriately
consider variable distributions and
modeling assumptions?
• Does the analysis appropriately
consider subgroups or lifestages of
interest (e.g., based on variability
in exposure level or duration or
susceptibility)?
• Is an appropriate analysis used for
the study design?
• Is effect modification considered,
based on considerations
developed a priori?
• Does the study include additional
analyses addressing potential
biases or limitations
(i.e., sensitivity analyses)?
If there is a concern
about the potential
for bias, what is the
predicted direction
or distortion of the
bias on the effect
estimate (if there is
enough
information)?
Good
• Use of an optimal characterization of the outcome variable,
including presentation of subgroup- or lifestage-specific
comparisons (as appropriate for the outcome).
• Quantitative results presented (effect estimates and
confidence limits or variability in estimates) (i.e., not
presented only as a p-value or "significant"/"not significant").
• Descriptive information about outcome and exposure
provided (where applicable).
• Amount of missing data noted and addressed appropriately
(discussion of selection issues—missing at random vs.
differential).
• Where applicable, for exposure, includes LOD (and
percentage below the LOD), and decision to use log
transformation.
• Includes analyses that address robustness of findings, for
example, examination of exposure-response (explicit
consideration of nonlinear possibilities, quadratic, spline, or
threshold/ceiling effects included, when feasible); relevant
sensitivity analyses; effect modification examined based only
on a priori rationale with sufficient numbers.
• No deficiencies in analysis evident. Discussion of some
details may be absent (e.g., examination of outliers).
Adequate
• Same as good, except:
• Descriptive information about exposure provided (where
applicable) but may be incomplete; might not have discussed
missing data, cutpoints, or shape of distribution(s).
This document is a draft for review purposes only and does not constitute Agency policy.
6-13 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Domain and
core question
Prompting questions
Follow-up questions
Criteria that apply to most exposures and outcomes
• Includes analyses that address robustness of findings
(examples in good), but some important analyses are not
performed.
Deficient
• Does not conduct analysis using optimal characterization of
the outcome variable.
• Descriptive information about exposure levels not provided
(where applicable).
• Effect estimate and p-value presented, without standard
error or confidence interval.
• Results presented as statistically "significant"/"not
significant."
Critically deficient
• Results of analyses of effect modification examined without
clear a priori rationale and without providing main/principal
effects (e.g., presentation only of statistically significant
interactions that were not hypothesis driven).
• Analysis methods are not appropriate for design or data of
the study.
Selective
reporting
Is there reason
to be concerned
about selective
reporting?
• Were results provided for all the
primary analyses described in the
methods section?
• Is there appropriate justification
for restricting the amount and
type of results that are shown?
• Are only statistically significant
results presented?
If there is a concern
about the potential
for bias, what is the
predicted direction
or distortion of the
bias on the effect
estimate (if there is
enough
information)?
Good
• The results reported by study authors are consistent with the
primary and secondary analyses described in a registered
protocol or methods paper.
Adequate
• The authors described their primary (and secondary)
analyses in the methods section, and results were reported
for all primary analyses.
This document is a draft for review purposes only and does not constitute Agency policy.
6-14 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Domain and
core question
Prompting questions
Follow-up questions
Criteria that apply to most exposures and outcomes
Deficient
• Concerns were raised based on previous publications, a
methods paper, or a registered protocol indicating that
analyses were planned or conducted that were not reported,
or that hypotheses originally considered to be secondary
were represented as primary in the reviewed paper.
• Only subgroup analyses were reported, suggesting that
results for the entire group were omitted.
• Only statistically significant results were reported.
Sensitivity
Is there a
concern that
sensitivity of the
study is not
adequate to
detect an effect?
• Is the exposure range adequate to
detect associations and
exposure-response relationships?
• Was the appropriate population or
lifestage included?
• Was the length of follow-up
adequate? Is the time/age of
outcome ascertainment optimal
given the interval of exposure and
the health outcome?
• Are there other aspects related to
risk of bias or otherwise that raise
concerns about sensitivity?
Adequate
• The range of exposure levels provides adequate variability to
evaluate the associations of relevance.
• The population was exposed to levels expected to have an
impact on response.
• The study population was sensitive to the development of
the outcomes of interest (e.g., ages, lifestage, sex).
• The timing of outcome ascertainment was appropriate given
expected latency for outcome development (i.e., adequate
follow-up interval).
• The study was adequately powered to observe an effect.
• No other concerns raised regarding study sensitivity.
Deficient
• Concerns were raised about the issues described for
adequate that are expected to notably decrease the
sensitivity of the study to detect associations for the
outcome.
1
This document is a draft for review purposes only and does not constitute Agency policy.
6-15 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
6.2.1. Epidemiology Study Evaluation Criteria Specific to These Five Per- and
Polyfluoroalkyl Substances (PFAS)
The exposure criteria described in Table 6-2 below are modified from the criteria
developed by NTP 0HAT14for their assessment of the association between PFOA and immune
effects.
The estimated serum half-lives of PFAS in humans were presented in Table 2-7 (see
Section 2.4.1). In considering temporality concerns, some PFAS (PFHxS, PFDA, and PFNA) are
persistent compounds with longer (multiple year) half-lives in humans, so current exposure levels
may be indicative of critical exposure windows that were narrow or past exposures that extended
beyond the anticipated half-lives. In contrast, other PFAS appear to have half-lives of 1 month or
less (PFBA, PFHxA), and current exposure levels may not be indicative of past exposures that
extend beyond the anticipated half-lives. Some evidence suggests that the half-lives vary based on
sex, parity, interval between pregnancy, reproductive hormones, and gynecological disorders (Lau
etal.. 2007): therefore, these factors will be considered depending on the population(s), critical
windows, and outcomes being examined.
Standard analytical methods of individual PFAS in serum or whole-blood using quantitative
techniques such as liquid chromatography-triple quadrupole mass spectrometry are considered to
be well-established methods (CDC. 2019a. b; ATSDR. 2018: CDC. 2015: U.S. EPA. 2014a. b; CDC.
20091.
"National Toxicology Program (NTP) Report:
https://ntp.niehs.nih.gov/ntp/ohat/pfoa pfos/pfoa pfosmonograph 508.pdf.
NTP protocol: https://ntp.niehs.nih.gov/ntp/ohat/pfoa pfos/protocol 201506 508.pdf.
This document is a draft for review purposes only and does not constitute Agency policy.
6-16 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 6-2. Criteria for evaluating exposure measurement in epidemiology
studies of per- and polyfluoroalkyl substances (PFAS) and health effects
Rating
Criteria
Good
• Evidence that exposure was consistently assessed using well-established analytical
methods that directly measure exposure (e.g., measurement of PFAS in blood, serum, or
plasma).
OR
• Exposure was assessed using less established methods (e.g., measurement of PFAS in
breast milk) or methods that indirectly measure exposure (e.g., drinking water
concentrations and residential location/history, questionnaire or occupational exposure
assessment by a certified industrial hygienist) that are validated against well-established
direct methods (i.e., intermethods validation: one method vs. another) in the target
population of interest.
And all the following:
• Exposure was assessed in a relevant time window (i.e., temporality is established, and
sufficient latency occurred before disease onset) for development of the outcome based
on current biological understanding.
• There is evidence that sufficient exposure data measurements are above the limit of
quantification for the assay.
• The laboratory analysis included data on standard quality control measures with
demonstrated precision and accuracy.
Adequate
• Exposure was assessed using less established methods or indirect measures that are
validated but not in the target population of interest.
OR
• Evidence that exposure was consistently assessed using methods described in good, but
there were some concerns about quality control measures or other potential for
nondifferential misclassification.
And all the following:
• Exposure was assessed in a relevant time window for development of the outcome
• There is evidence that sufficient exposure data measurements are above the limit of
quantification for the assay.
• The laboratory analysis included some data on standard quality control measures with
demonstrated precision and accuracy.
Deficient
Any of the following:
• Some concern, but no direct evidence, that the exposure was assessed using methods
that have not been validated or empirically shown to be consistent with methods that
directly measure exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
6-17 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Rating
Criteria
• Exposure was assessed in a relevant time window(s) for development of the outcome,
but there could be some concern about the potential for bias due to reverse causality3
between exposure and outcome, yet no direct evidence that it is present.
Critically
deficient
Any of the following:
• Exposure was assessed in a time window that is unknown or not relevant for
development of the outcome. This could be due to clear evidence of bias from reverse
causality between exposure and outcome, or other concerns such as the lack of temporal
ordering of exposure and disease onset, insufficient latency, or having exposure
measurements that are not reliable measures of exposure during the etiologic
window(s).
• Direct evidence that bias was likely because the exposure was assessed using methods
with poor validity.
• Evidence of differential exposure misclassification (e.g., differential recall of self-reported
exposure).
• There is evidence that an insufficient number of the exposure data measurements were
above the limit of quantification for the assay.
aReverse causality refers to a situation in which an observed association between exposure and outcome is not due
to causality from exposure to outcome, but rather due to the outcome of interest causing a change in the
measured exposure.
1 In addition, there are PFAS-specific considerations for the evaluation of confounding. As
2 discussed in Section 2.4.3, confounding across PFAS is an important area of uncertainty when
3 interpreting the results of epidemiology studies for individual PFAS (i.e., quantifying the effected of
4 an individual PFAS can potentially be confounded by other PFAS). Based on preliminary analyses,
5 correlations differ across the PFAS (see Figure 6-2). While some pairs have correlation coefficients
6 consistently above 0.6 (e.g., PFNA and PFDA), the correlations for mostvary from 0.1 to 0.6
7 depending on the study, and little data is available on correlations with less commonly occurring or
8 detected PFAS like PFBA and PFHxA.
This document is a draft for review purposes only and does not constitute Agency policy.
6-18 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
PFBA
PFDA
PFHxA
PFHxS
PFNA
PFOA
PFOS
PFBA*
1.00
0.01
0.45
0.14
0.15
0.03
0.07
PFDA
1.00
-0.03
0.28
0.73
0.42
0.48
PFHxA*
1.00
0.08
-0.07
0.19
-0.04
PFHxS
1.00
0.35
0.43
0.50
PFNA
1.00
0.54
0.51
Figure 6-2. Preliminary mean correlation coefficients across per- and
polyfluoroalkyl substances (PFAS) among studies in the inventory, for all
media types.
*PFBA and PFHxA correlations were based on three studies for PFOS, PFOA, and PFHxS, two studies for each other,
and one study for PFNA and PFDA, so these estimates are less stable than the other PFAS, which were all based
on >10 studies.
PFOS = perfluorooctane sulfonate.
Rather than rating each study with lower confidence because of this issue, potential
confounding by other PFAS will be considered during the evidence synthesis phase, primarily when
there is some support for an association with adverse health effects in the epidemiology evidence.
This may include looking across studies in populations with different exposure profiles
(e.g., observing an association in a population with much higher exposure to one PFAS due to
proximity to an industrial plant would increase confidence for that PFAS). In situations where
there is considerable uncertainty regarding the impact of residual confounding across PFAS, this
will be captured as a factor that decreases evidence strength (see Section 10).
6.3. EXPERIMENTAL ANIMAL STUDY EVALUATION
Using the principles described in Section 6.1, the evaluation of animal studies of health
effects to assess risk of bias and sensitivity will be conducted for the following domains: reporting
quality, risk of bias (selection or performance bias, confounding/variable control, and reporting or
attrition bias), and study sensitivity (exposure methods sensitivity, and outcome measures and
results display) (see Table 6-3). Several additional considerations specific to assessing these five
PFAS are outlined in Section 6.3.1.
The rationale for judgments will be documented clearly and consistently at the outcome
level. In addition, for domains other than reporting quality, the evaluation documentation in HAWC
will include the identified limitations and consider their impact on the overall confidence level, a
procedure similar to the evaluation of epidemiology studies. This will, to the extent possible, reflect
an interpretation of the potential influence on the outcome-specific results (including the direction
and/or magnitude of influence).
This document is a draft for review purposes only and does not constitute Agency policy.
6-19 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 6-3. Considerations to evaluate domains from animal toxicological
studies
Evaluation
concern
Domain—core question
Prompting questions
General considerations
cr
no
c
t
o
Q.
a>
cc
Reporting quality
Does the study report
information for evaluating
the design and conduct of
the study for the
endpoints/outcomes of
interest?
Note:
This domain is limited to
reporting. Other aspects of
the exposure methods,
experimental design, and
endpoint evaluation
methods are evaluated
using the domains related
to risk of bias and study
sensitivity.
Does the study report the
following?
Critical information necessary
to perform study evaluation:
• Species, test article name,
levels and duration of
exposure, route (e.g., oral;
inhalation), qualitative or
quantitative results for at
least one endpoint of
interest
Important information for
evaluating the study methods:
• Test animal: strain, sex,
source, and general
husbandry procedures
• Exposure methods: source,
purity, method of
administration
• Experimental design:
frequency of exposure,
animal age, and lifestage
during exposure and at
endpoint/outcome
evaluation
• Endpoint evaluation
methods: assays or
procedures used to measure
the endpoints/outcomes of
interest
A judgment and rationale for this domain
will generally be given for the study. In the
rationale, reviewers will also indicate when
a study adhered to GLP, or to OECD (or
similar) testing guidelines.
• Good: All critical and important
information is reported or inferable for
the endpoints/outcomes of interest.
• Adequate: All critical information is
reported, but some important
information is missing. However, the
missing information is not expected to
significantly impact the study
evaluation.
• Deficient: All critical information is
reported, but important information is
missing that is expected to significantly
reduce the ability to evaluate the study.
• Critically deficient: Study report is
missing any pieces of critical
information. Studies that are critically
deficient for reporting are
uninformative for the overall rating and
not considered further.
This document is a draft for review purposes only and does not constitute Agency policy.
6-20 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Evaluation
concern
Domain—core question
Prompting questions
General considerations
Allocation
For each study:
A judgment and rationale for this domain
Were animals assigned to
experimental groups using a
method that minimizes
selection bias?
• Did each animal or litter
have an equal chance of
will be given for each cohort or experiment
in the study.
being assigned to any
experimental group
(i.e., random allocation3)?
• Good: Experimental groups were
randomized, and any specific
randomization procedure was
• Is the allocation method
described?
described or inferable
(e.g., computer-generated scheme).
(Note that normalization is not the
• Aside from randomization,
same as randomization [see response
.5
_Q
were any steps taken to
for adequate].)
ai
£
the specific procedure used
4—
o
*_
ai
(e.g., "animals were randomized").
yi
Q.
-a
Alternatively, authors used a
CC
c
ro
c
o
'+¦»
u
ai
ai
l/l
nonrandom method to control for
important modifying factors (i.e., with
respect to the outcome of interest)
across experimental groups
(e.g., body-weight normalization).
• Not reported (interpreted as deficient):
No indication of randomization of
groups or other methods
(e.g., normalization) to control for
important modifying factors across
experimental groups.
• Critically deficient: Bias in the animal
allocations was reported or inferable.
This document is a draft for review purposes only and does not constitute Agency policy.
6-21 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Evaluation
concern
Domain—core question
Prompting questions
General considerations
Observational bias/blinding
For each endpoint/outcome or
A judgment and rationale for this domain
Did the study implement
grouping of outcomes in a study:
will be given for each endpoint/outcome or
measures to reduce
observational bias?
• Does the study report
blinding or other
methods/procedures for
reducing observational bias,
as appropriate for the assays
of interest?
group of outcomes investigated in the study.
• Good: Measures to reduce
observational bias were described
(e.g., blinding to conceal treatment
groups during endpoint evaluation;
consensus-based evaluations of
• If not, did the study use a
histopathology lesions3).
"D
0)
design or approach for
• Adequate: Methods for reducing
3
C
which such procedures can
observational bias (e.g., blinding) can
c
o
be inferred?
be inferred or were reported but
"D
0)
u
V)
.w
• What is the expected impact
described incompletely.
C
of failure to implement (or
• Not reported: Measures to reduce
c
o
ai
LO
automated/computer-driven
systems, standard laboratory kits,
relatively simple objective
measures (e.g., body or tissue
weight), or screening-level
evaluations of histopathology.
o (Interpreted as deficient): The
potential impact on the results is
large (e.g., outcome measures are
highly subjective).
• Critically deficient: Strong evidence for
observational bias that impacted the
results.
This document is a draft for review purposes only and does not constitute Agency policy.
6-22 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Evaluation
concern
Domain—core question
Prompting questions
General considerations
¦a
a>
o
u
.5
o
u
_ai
-Q
.5
ro
>
o
u
Confounding
Are variables with the
potential to confound or
modify results controlled
for and consistent across all
experimental groups?
For each study:
• Are there differences across
the treatment groups
(e.g., coexposures, vehicle,
diet, palatability, husbandry,
health status, surgery) that
could bias the results?
• If differences are identified,
to what extent are they
expected to impact the
results?
A judgment and rationale for this domain
will be given for each cohort or experiment
in the study, noting when the potential for
confounding is restricted to specific
endpoints/outcomes.
• Good: Outside of the exposure of
interest, variables that are likely to
confound or modify results appear to
be controlled for and consistent across
experimental groups.
• Adequate: Some concern that variables
likely to confound or modify the results
were uncontrolled or inconsistent
across groups, but these are expected
to have a minimal impact on the
results.
• Deficient: Notable concern that
potentially confounding variables were
uncontrolled or inconsistent across
groups and that they are expected to
substantially impact the results.
• Critically deficient: Confounding
variables were presumed to be
uncontrolled or inconsistent across
groups, and they are expected to be a
primary driver of the results.
This document is a draft for review purposes only and does not constitute Agency policy.
6-23 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Evaluation
concern
Domain—core question
Prompting questions
General considerations
Risk of bias (continued)
Selective reporting and attrition bias
Selective reporting and
attrition
Did the study report results
for all prespecified
outcomes and tested
animals?
Note:
This domain does not
consider the
appropriateness of the
comparisons/results
presentation. This aspect of
study quality is evaluated in
another domain.
For each study:
Selective reporting bias:
• Are all results presented for
endpoints/outcomes
described in the methods
(see note)?
Attrition bias:
• Do the results account for all
animals?
• If there are discrepancies, do
the authors provide an
explanation (e.g., death or
unscheduled sacrifice during
the study)?
• If unexplained results
omissions and/or attrition
are identified, what is the
expected impact on the
interpretation of the results?
A judgment and rationale for this domain
will be given for each cohort or experiment
in the study.
• Good: Quantitative or qualitative
results were reported for all
prespecified outcomes (explicitly stated
or inferred), exposure groups, and
evaluation time points. Data not
reported in the primary article are
available from supplemental material.
If results omissions or animal attrition
are identified, the authors provide an
appropriate explanation, and the
omissions or attrition are not expected
to impact the interpretation of the
results.
• Adequate: Quantitative or qualitative
results are reported for most
prespecified outcomes (explicitly stated
or inferred), exposure groups, and
evaluation time points. Omissions
and/or attrition are not explained, but
they are not expected to significantly
impact the interpretation of the results.
• Deficient: Quantitative or qualitative
results are missing for many
prespecified outcomes (explicitly stated
or inferred), exposure groups and
evaluation time points, or there is high
animal attrition; omissions and/or
attrition are not explained and are
expected to significantly impact the
interpretation of the results.
• Critically deficient: Extensive results
omission and/or animal attrition are
identified and prevent comparisons of
results across treatment groups.
This document is a draft for review purposes only and does not constitute Agency policy.
6-24 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Evaluation
concern
Domain—core question
Prompting questions
General considerations
>
">
a>
to
>
">
-a
o
.c
a!
E
ai
o
a.
x
Chemical administration
and characterization
Did the study adequately
characterize exposure to
the chemical of interest and
the exposure administration
methods?
Note:
These considerations are
limited to oral exposure, as
only a single inhalation
study focusing on acute
toxicity (i.e., after PFNA
exposure) was identified
(see Section 2.3.2).
For each study:
• Does the study report the
source and purity and/or
composition (e.g., identity
and percent distribution of
different isomers) of the
chemical? If not, can the
purity and/or composition
be obtained from the
supplier (e.g., as reported on
the website)?
• Was independent analytical
verification of the test article
purity and composition
performed?
• Are there concerns about
the methods used to
administer the chemical
(e.g., gavage volume)?
• If necessary, based on
consideration of
chemical-specific knowledge
(e.g., instability in solution;
volatility) and/or exposure
design (e.g., the frequency
and duration of exposure),
were the chemical
concentrations in the dosing
solutions or diet analytically
confirmed?
A judgment and rationale for this domain
will be given for each cohort or experiment
in the study.
• Good: Chemical administration and
characterization is complete
(i.e., source, purity, and analytical
verification of the test article are
provided). There are no concerns
about the composition, stability, or
purity of the administered chemical, or
the specific methods of administration.
• Adequate: Some uncertainties in the
chemical administration and
characterization are identified, but
these are expected to have minimal
impact on interpreting the results
(e.g., source and vendor-reported
purity are presented, but not
independently verified; purity of the
test article is suboptimal but not
concerning).
• Deficient: Uncertainties in the exposure
characterization are identified and
expected to substantially impact the
results (e.g., source of the test article is
not reported; levels of impurities are
substantial or concerning; deficient
administration methods, such as use of
a gavage volume considered too large
for the species and/or lifestage at
exposure).
• Critically deficient: Uncertainties in the
exposure characterization are
identified, and there is reasonable
certainty that the results are largely
attributable to factors other than
exposure to the chemical of interest
(e.g., identified impurities are expected
to be a primary driver of the results).
This document is a draft for review purposes only and does not constitute Agency policy.
6-25 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Evaluation
concern
Domain—core question
Prompting questions
General considerations
Exposure timing,
For each endpoint/outcome or
A judgment and rationale for this domain
frequency, and duration
grouping of outcomes in a study:
will be given for each endpoint/outcome or
Was the timing, frequency,
and duration of exposure
sensitive for the
endpoint(s)/outcome(s) of
interest?
• Does the exposure period
include the full critical
group of outcomes investigated in the study.
• Good: The duration and frequency of
¦C
a>
window of sensitivity, based
on current biological
the exposure was sensitive, and the
exposure included the critical window
c
'+¦»
understanding?
of sensitivity (if known).
"D
c
o
u
• Was the duration and
• Adequate: The duration and frequency
0)
3
>
+¦»
frequency of exposure
of the exposure was sensitive, and the
c
'+¦»
sensitive for detecting the
exposure covered most of the critical
c
o
u
l/l
c
0)
endpoint of interest?
window of sensitivity (if known).
>
U)
U)
• Deficient: The duration and/or
>
o
.c
frequency of the exposure is not
c
ai
l/l
+¦»
ai
sensitive and did not include most of
E
a>
3
U)
O
a.
X
LU
the critical window of sensitivity (if
known). These limitations are expected
to bias the results towards the null.
• Critically deficient: The exposure design
was not sensitive and is expected to
strongly bias the results towards the
null. The rationale should indicate the
specific concern(s).
This document is a draft for review purposes only and does not constitute Agency policy.
6-26 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Evaluation
concern
Domain—core question
Prompting questions
General considerations
Endpoint sensitivity and
For each endpoint/outcome or
A judgment and rationale for this domain
specificity
grouping of outcomes in a study:
will be given for each endpoint/outcome or
Are the procedures
• Are there concerns
regarding the sensitivity,
specificity, and/or validity of
the outcome measurement
group of outcomes investigated in the study.
sensitive and specific for
evaluating the
endpoint(s)/outcome(s) of
interest?
Examples of potential concerns include:
• Selection of protocols that are
insensitive or nonspecific for the
>
protocols?
endpoint of interest.
Q.
Note:
• Are there serious concerns
• Evaluations did not include all
"D
0)
¦5
U)
+J
3
U)
• Sample size alone is not
a reason to conclude
regarding the sample size?
• Are there concerns
treatment groups (e.g., only control and
high dose).
C
01
an individual study is
regarding the timing of the
• Use of unreliable or invalid methods to
c
o
u
¦a
c
ro
critically deficient.
endpoint assessment?
assess the outcome.
>
+¦»
U)
01
• Considerations related
• Assessment of endpoints at
3
U)
ro
to adjustments/
inappropriate or insensitive ages, or
l/l
c
01
£
corrections to endpoint
without addressing known endpoint
01
l/l
01
measurements
variation (e.g., due to circadian
E
o
(e.g., organ weight
rhythms, estrous cyclicity).
+¦»
3
o
corrected for body
weight) are addressed
under results
presentation.
• Decreased specificity or sensitivity of
the response due to the timing of
endpoint evaluation, as compared with
exposure (e.g., immediate endpoint
assessment after exposure to chemicals
with short-acting depressant or irritant
effects; insensitivity due to prolonged
period of nonexposure before testing).
This document is a draft for review purposes only and does not constitute Agency policy.
6-27 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Evaluation
concern
Domain—core question
Prompting questions
General considerations
-a
ai
>
">
ai
to
¦a
ai
o
a.
ai
*_
T3
TO
ai
E
ai
E
o
Results presentation
Are the results presented in
a way that makes the data
usable and transparent?
Note:
Potential issues associated
with statistical analyses will
be flagged for review by
EPA statisticians and
possible reanalysis (if
information is available to
do so, any reanalysis will be
transparently presented).
Any remaining limitations
will be discussed during
evidence synthesis or
dose-response analyses
(depending on the identified
issue).
For each endpoint/outcome or
grouping of outcomes in a study:
• Does the level of detail allow
for an informed
interpretation of the results?
• Are the data analyzed,
compared, or presented in a
way that is inappropriate or
misleading?
A judgment and rationale for this domain
will be given for each endpoint/outcome or
group of outcomes investigated in the study.
Examples of potential concerns include:
• Nonpreferred presentation
(e.g., developmental toxicity data
averaged across pups in a treatment
group, when litter responses are more
appropriate; presentation of absolute
organ-weight data when relative
weights are more appropriate).
• Failing to present quantitative results
either in tables or figures.
• Pooling data when responses are
known or expected to differ
substantially (e.g., across sexes or
lifestages).
• Failing to report on or address overt
toxicity when exposure levels are
known or expected to be highly toxic.
• Lack of full presentation of the data
(e.g., presentation of mean without
variance data; concurrent control data
are not presented).
a>
¦c
o
u
a>
>
O
Overall confidence
Considering the identified
strengths and limitations,
what is the overall
confidence rating for the
endpoint(s)/outcome(s) of
interest?
Note:
Reviewers will mark studies
that are rated lower than
high confidence due only to
low sensitivity (i.e., bias
towards the null) for
additional consideration
during evidence synthesis.
If the study is otherwise well
conducted and an effect is
observed, the confidence
may be increased.
For each endpoint/outcome or
grouping of outcomes in a study:
• Were concerns
(i.e., limitations or
uncertainties) related to the
reporting quality, risk of
bias, or sensitivity
identified?
• If yes, what is their expected
impact on the overall
interpretation of the
reliability and validity of the
study results, including
(when possible)
interpretations of impacts
on the magnitude or
direction of the reported
effects?
The overall confidence rating considers the
likely impact of the noted concerns
(i.e., limitations or uncertainties) in
reporting, bias and sensitivity on the results.
A confidence rating and rationale will be
given for each endpoint/outcome or group
of outcomes investigated in the study.
Confidence rating definitions are described
above (see Section 6.1).
This document is a draft for review purposes only and does not constitute Agency policy.
6-28 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
aSeveral studies have characterized the relevance of randomization, allocation concealment, and blind outcome
assessment in experimental studies (Hirst et al., 2014; Krauth et al., 2013; Macleod, 2013; Higgins and Green,
2011).
GLP = good laboratory practice; OECD = Organisation for Economic Co-operation and Development.
6.3.1. Animal Toxicological Study Evaluation Considerations Specific to These Five Per- and
Polyfluoroalkyl Substances (PFAS)
A key uncertainty in these assessments involves the toxicokinetics of the five PFAS. The
apparent differences in toxicokinetics across animal species will not be addressed at the individual
study level but will be considered during evidence integration (see Section 10) and is expected to
be most influential when developing toxicity values for potential human health hazards (see
Section 11). However, based on Table 2-7 (see Section 2.4.1), the clearance of some of these PFAS,
and the sex-specific differences in serum half-lives, represent important considerations for
potential sources of insensitivity during study evaluation. Specifically, studies may be judged as
insensitive if they fail to account for the short serum half-lives of PFBA in female rats and mice
(half-lives of ~l-3 hours; half-lives in males are close to half a day and of less concern) and PFHxA
in rats and mice of both sexes (half-lives of ~0.5-1.5 hours) by including, for example, multiple
daily exposures. Half-lives in rodents for PFDA, PFNA, and PFHxS are on the order of days or
longer, so insensitivity due to the short half-life in rodents does not represent a concern for these
PFAS (note: no nonhuman primate health effect studies were identified). Similarly, given the
profound apparent differences in clearance between male and female rats for PFNA (i.e., females
appear to clear PFNA 25 x faster), studies that examined both sexes are preferred, and any study
that tested female rats only may be judged as insensitive. This consideration may also be applied,
but to a lesser extent, for studies of PFHxS in rats and PFBA in mice or rats (females appear to clear
these PFAS ~4-6x faster than males).
These five PFAS are considered stable and nonreactive, and the presence of potentially toxic
impurities within these readily available chemicals has not been identified as an issue in the
literature. Thus, failure to describe preparation and storage of dosing solutions will not be
considered an issue of concern, and a lack of information on chemical purity will not be considered
a significant limitation. This interpretation is consistent with quality control and solubilization
information on these PFAS performed by the EPA as part of the ongoing ToxCast testing
fhttps;//co mptox.epa.gov/dashboard/chemical lists/EPAPFASINVl. None of these PFAS were
flagged as problematic (e.g., based on volatility and solubility, or degradation-type issues), or raised
concerns during analytical testing, although CIO (PFDA) can become less soluble in water at very
high concentrations. Given the more relevant possibility of contamination of these five PFAS with
other PFAS, a lack of analytical verification of the test article will be flagged as a limitation, although
this alone will not significantly affect overall study confidence ratings.
A wide variety of outcomes have been assessed in the available animal studies for these five
PFAS. Considerations specific to each outcome are not included in this protocol (outcome-specific
This document is a draft for review purposes only and does not constitute Agency policy.
6-29 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
concerns will be available in HAWC when the assessments are released). As examples, a few
specific considerations that will be applied include better domain ratings for studies that address
potential differences in time of day for evaluations of hormone levels (due to fluctuations
throughout the day), and for studies that address fasting status for metabolic-related
measurements.
6.4. PHARMACOKINETIC MODEL EVALUATION
A similar approach for evaluation will be applied to the full PBPK models for PFHxS (Kim et
al.. 20181 and for PFDA and PFNA fKim etal.. 20191. as well as to the two-compartment PK model
for gestational and lactational transfer of PFHxS in humans described bvVerner etal. f20161.
Models will be preferred for use in these assessments when an applicable one exists and no equal
or better alternative for dosimetric extrapolation is available. Given these preferences, sound
justification will be provided for not using a PBPK (or classical PK) model when an applicable one
exists and no equal or better alternative for dosimetric extrapolation is available. Note, however,
that these preferences only apply to models that faithfully represent current scientific knowledge
and accurately translate the science into computational code in a reproducible, transparent
manner. In practice, many models have errors that affect their predictions to varying degrees;
hence, an evaluation of a model is required before it can be used in an assessment Thus, the
currently available models and any other models identified at later stages of developing these
assessments will be evaluated as described below.
Considerations for judging the suitability of a model are separated into two categories:
scientific and technical. The scientific criteria focus on whether the biology, chemistry, and other
information available for chemical mode(s) of action (M0A[s]) are appropriately represented by
the model structure and equations. Scientific criteria are easier to evaluate in judging a model's
suitability because they can be judged by reading the publication or report that describes the
model, without requiring an evaluation of the computer code. Preliminary technical criteria include
the availability of the computer code and apparent completeness of parameter listing and
documentation. The in-depth technical and scientific criteria focus on the accurate implementation
of the conceptual model in the computational code, use of correct or biologically consistent
parameters in the model, and reproducibility of model results reported in journal publications and
other documents. Specific details for this evaluation are provided in the Quality Assurance Project
Plan for PBPK models flJ.S. FPA. 2018bl
6.5. MECHANISTIC STUDY EVALUATION
Sections 9 and 10 outline an approach for focused consideration of information from
mechanistic studies (including in vitro, in vivo, ex vivo, and in silico studies) where the specific
analytical approach is targeted to the assessment needs, depending in part on the extent and nature
of the phenotypic human and animal evidence. In this way, the mechanistic synthesis for a given
This document is a draft for review purposes only and does not constitute Agency policy.
6-30 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
health effect might range from a high-level summary (or detailed analysis) of potential mechanisms
of action to specific, focused questions needed to address important and impactful assessment
uncertainties unaddressed by the available phenotypic studies (e.g., expected shape of the
dose-response curve in the low-dose region, applicability of the animal evidence to humans,
addressing susceptible populations). Individual study-level evaluation of mechanistic endpoints
will not typically be pursued. However, it may be necessary to identify assay-specific
considerations for study endpoint evaluations on a case-by-case basis to provide a more detailed
summary and evaluation for the most relevant individual mechanistic studies addressing a key
assessment uncertainty. This may be done, for example, when the scientific understanding of a
critical mechanistic event or MOA lacks scientific consensus, when the reported findings on a
critical mechanistic endpoint are conflicting, when the available mechanistic evidence addresses a
complex and influential aspect of the assessment, or when in vitro or in silico data make up the bulk
of the evidence base and there is little or no evidence from epidemiological studies or animal
bioassays. As noted in Section 3 and Section 4, genotoxicity studies were identified as meeting
PECO criteria; these data will be summarized in each PFAS assessment to describe evidence
relevant to carcinogenicity even in the absence of more phenotypic data. Based on the
considerations above, if these studies are interpreted as adequate to draw a hazard conclusion
(i.e., other than "insufficient"), individual study-level evaluations of some or all the genotoxicity
studies will be informative to this decision. If necessary, based on the assessment-specific issues
identified during study evaluation and evidence synthesis (see Section 9.2), the specific approach to
evaluating individual studies other than those addressed in Sections 6.2-6.4 will be outlined in the
assessments and included as an update to the protocol.
This document is a draft for review purposes only and does not constitute Agency policy.
6-31 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
7.ORGANIZING THE HAZARD REVIEW
The organization and scope of the hazard evaluation is determined by the available
evidence for each PFAS regarding routes of exposure, metabolism and distribution, outcomes
evaluated, and number of studies pertaining to each outcome, as well as the results of the
evaluation of sources of bias and sensitivity. The hazard evaluations will be organized around
organ systems (e.g., nervous system) informed by one or multiple related outcomes, as described in
Section 5, and a decision will be made as to what level (e.g., organ system or subsets of outcomes
within an organ system) to organize the synthesis.
Table 7-1 lists some questions that may be asked of the evidence to assist with this decision.
These questions extend from considerations and decisions made during development of the refined
evaluation plan to include review of the concerns raised during individual study evaluations as well
as the direction and magnitude of the study-specific results. Resolution of these questions will then
inform critical decisions about the organization of the hazard evaluation and help determine what
studies may be useful in dose-response analyses.
This document is a draft for review purposes only and does not constitute Agency policy.
7-1 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 7-1. Querying the evidence to organize syntheses for human and animal
evidence
Evidence
Questions
Follow-up questions
ADME
Given the known ADME issues for these PFAS, do the
data appear to differ by route of exposure studied,
lifestage when exposure occurred, sex, species, or
dosing regimens used?
Will separate analyses be needed by factors
such as sex, route of exposure, or by methods
of dosing within a route of exposure (e.g., are
large differences expected between gavage
and dietary exposures)?
Are data available to inform which lifestages
and what dosing regimens are more relevant
to human exposure scenarios?
Is there toxicity information for metabolites that also
should be evaluated for hazard?
What exposures will be included in the
evaluation?
Outcomes
What outcomes are reported in studies? Are the
data reported in a comparable manner across studies
(similar output metrics at similar levels of specificity,
such as adenomas and carcinomas quantified
separately)?
At what level (hazard, grouped outcomes, or
individual outcomes) will the synthesis be
conducted?
What commonalities will the outcomes be
grouped by:
• health effect,
• exposure levels,
• functional or population-level
consequences (e.g., endpoints all
ultimately leading to decreased
fertility or impaired cognitive
function),
• involvement of related biological
pathways
How well do the assessed human and animal
outcomes relate within a level of grouping?
Are there interrelated outcomes? If so, consider
whether some outcomes are more useful and/or of
greater concern than others.
Does the evidence indicate greater sensitivity to
effects (at lower exposure levels or severity) in
certain subgroups (by age, sex, ethnicity, lifestage)?
Should the hazard evaluation include a subgroup
analysis?
Does incidence or severity of an outcome increase
with duration of exposure or a particular window of
exposure. What exposure time frames are relevant
to development or progression of the outcome?
Is there mechanistic evidence that informs how
outcomes might be grouped together?
How robust is the evidence for specific outcomes?
• What outcomes are reported by both human
and animal studies and by one or the other?
Were different animal species and sexes (or
other important population-level
differences) tested?
• In general, what are the study confidence
conclusions of the studies (high, medium,
low, uninformative) for the different
outcomes? Is there enough evidence from
high and medium confidence studies to draw
conclusions about causality?
What outcomes should be highlighted?
Should the others be synthesized at all?
Would comparisons by specific limitations be
informative?
This document is a draft for review purposes only and does not constitute Agency policy.
7-2 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Evidence
Questions
Follow-up questions
Dose-
response
Did some outcomes include better coverage of
exposure ranges that may be most relevant to human
exposure than others?
What outcomes and studies are informative
for developing toxicity values?
For which outcomes are there sufficient data
available to draw conclusions about dose-response?
Are there outcomes with study results of sufficient
similarity (e.g., an established linkage in a biological
pathway) to allow examination or calculation of
common measures of effect across studies? Do the
mechanistic data identify surrogate or precursor
outcomes that are adequate for dose-response
analysis?
Are there subgroups that exhibit responses at lower
exposure levels than others?
Are there findings from ADME studies that could
inform data-derived extrapolation factors, or link
toxicity observed via different routes of exposure, or
link effects between humans and experimental
animals?
What studies might be used to develop
nondefault UFs? Is there a common internal
dose metric that can be used to compare
species or routes of exposure?
ADME = absorption, distribution, metabolism, and excretion; UF = uncertainty factor.
This document is a draft for review purposes only and does not constitute Agency policy.
7-3 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
8. DATA EXTRACTION OF STUDY METHODS AND
RESULTS
Data extraction and content management will be carried out using the Health Assessment
Workplace Collaborative (HAWC; web links will be shared in the individual assessments). A
consistent approach to data extraction will be applied across these PFAS assessments to facilitate
their anticipated future use in addressing poorly studied PFAS (e.g., through coupling with
computational toxicology data generated as described in the Environmental Protection Agency
[EPA] PFAS action plan). Data extraction elements that may be collected from epidemiological,
controlled human exposure, animal toxicological, and in vitro studies are described in HAWC
(https://hawcprd.epa.gov/about/). Not all studies that meet the PECO criteria go through data
extraction. Studies evaluated as being Uninformative are not considered further and therefore
would not undergo data extraction. In addition, outcomes determined to be less relevant during
PECO refinement (see Section 5) may not go through data extraction or may have only minimal data
extraction. The same may be true for low confidence studies if enough medium and high confidence
studies (e.g., on an outcome) are available. All findings are considered for extraction, regardless of
statistical significance. The level of extraction for specific outcomes within a study may differ
(i.e., ranging from a narrative to full extraction of dose-response effect size information). In part,
this extraction level is determined based on the level of detail to be discussed in the evidence
synthesis for that health effect (e.g., a detailed extraction will not be necessary for health effects
with very few available studies; these will only be briefly summarized in a short narrative).
Similarly, decisions about data extraction for low confidence studies are typically made while
implementing the protocol and are based on the quality and extent of the available evidence. If
necessary, the version of the protocol released with the draft assessment will outline how low
confidence studies were treated for extraction and evidence synthesis.
The data extraction results for included studies will be presented in the assessment (and
made available for download from EPA HAWC in Excel format) when the draft is publicly released.
(Note: The following browsers are supported for accessing HAWC: Google Chrome [preferred],
Mozilla Firefox, and Apple Safari. There are errors in functionality when viewed with Internet
Explorer.) For quality control, data extraction will be performed by one member of the evaluation
team and independently verified by at least one other member. Discrepancies in data extraction
will be resolved by discussion or consultation with a third member of the evaluation team. Digital
rulers, such as WebPlotDigitizer (http: //arohatgi.info/WebPlotDigitizer/). will be used to extract
numerical information from figures, and their use will be documented during extraction.
This document is a draft for review purposes only and does not constitute Agency policy.
8-1 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
1 As previously described, routine attempts will be made to obtain missing information from
2 human and animal health effect studies, if it is considered influential during study evaluations (see
3 Section 6) or when it can provide information important for dose-response analysis or
4 interpretations of significance (e.g., missing group size or variance descriptors such as standard
5 deviation or confidence interval). Missing data from individual mechanistic (e.g., in vitro) studies
6 generally will not be sought Outreach to study authors or designated contact persons will be
7 documented and considered unsuccessful if they do not respond to email or phone requests within
8 1 month of initial attempt(s) to contact
8.1. STANDARDIZING REPORTING OF EFFECT SIZES
9 In addition to providing quantitative outcomes in their original units for all study groups,
10 results from outcome measures will be transformed, when possible, to a common metric to help
11 compare distinct but related outcomes that are measured with different scales. These standardized
12 effect size estimates facilitate systematic evaluation and evidence integration for hazard
13 identification (see Section 9.1). The following summary of effect size metrics by data type outlines
14 issues in selecting the most appropriate common metric for a collection of related endpoints
15 fVesterinen et al.. 20141.
16 Common metrics for continuous outcomes include:
17 • Absolute difference in means. This metric is the difference between the means in the control
18 and treatment groups, expressed in the units in which the outcome is measured. When the
19 outcome measure and its scale are the same across all studies, this approach is the simplest
20 to implement.
21 • Percent control response (or normalized mean difference [NMD]). Percent control group
22 calculations are based on means. Standard deviation (or standard error) values presented
23 in the studies for these normalized effect sizes can also be estimated if sufficient
24 information has been provided. Note that some outcomes reported as percentages, such as
25 mean percentage of affected offspring per litter, can lead to distorted effect sizes when
26 further characterized as percentage change from control. Such measures are better
27 expressed as absolute difference in means, or rather preferably transformed to incidences
28 using approaches for event or incidence data (see below).
29 • Standardized mean difference. The NMD approach above is relevant to ratio scales, but
30 sometimes it is not possible to infer what a "normal" animal would score, such as when data
31 for animals without lesions are not available. In these circumstances, standardized mean
32 differences can be used. The difference in group means is divided by a measure of the
33 pooled variance to convert all outcome measures to a standardized scale with units of
34 standard deviations. This approach can also be applied to data for which different
35 measurement scales are reported for the same outcome measure (e.g., different measures of
36 lesion size such as infarct volume and infarct area).
This document is a draft for review purposes only and does not constitute Agency policy.
8-2 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Common metrics for event or incidence data include:
• Percent change from control. This metric is analogous to the NMD approach described for
continuous data above.
• For binary outcomes such as the number of individuals that developed a disease or died,
and with only one treatment evaluated, data can be represented in a 2 x 2 table. Note that
when the value in any cell is 0, 0.5 is added to each cell to avoid problems with the
computation of the standard error. For each comparison, the odds ratio (OR) and its
standard error can be calculated. Odds ratios are normally combined on a logarithmic scale.
An additional approach for epidemiology studies is to extract adjusted statistical estimates
when possible rather than unadjusted or raw estimates.
It is important to consider the variability associated with effect size estimates, with better
powered studies generally showing more precise estimates. Effect size estimation can be affected,
however, by such factors as variances that differ substantially across treatment groups, or by lack of
information to characterize variance, especially for animal studies in biomedical research
(Vesterinen et al.. 2014). The assessments will consider the nature of any variance issues and
ensure that the associated uncertainties are clarified and accounted for during the evidence
synthesis process (see Section 9).
8.2. STANDARDIZING ADMINISTERED DOSE LEVELS/CONCENTRATIONS
Exposures will be standardized to common units. Exposure levels in oral studies will be
expressed in units of mg PFAS/kg-day. When the study authors provide exposure levels in
concentrations in the diet or drinking water, dose conversions will be made using study-specific
food or water consumption rates and body weights when available. Otherwise, EPA defaults will be
used fU.S. EPA. 19881. addressing age and study duration as relevant for the species/strain and sex
of the animal of interest. Exposure levels in inhalation studies will be expressed in units of mg/m3.
Assumptions used in performing dose conversions will be documented in HAWC or the specific
assessments.
This document is a draft for review purposes only and does not constitute Agency policy.
8-3 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
9.SYNTHESIS OF EVIDENCE
For the purposes of this assessment, evidence synthesis and integration are considered
distinct, but related processes. As described below, for each assessed health effect the evidence
syntheses provide a summary discussion of each body of evidence considered in the review
(i.e., human, animal, and mechanistic evidence). These separate summaries directly inform
interpretations regarding the support for causation provided by each body of evidence and the
evidence as a whole. In other words, the syntheses of separate bodies of evidence described in this
section will directly inform the integration across evidence to draw an overall judgment for each of
the assessed human health effects (as described in Section 10). The phrase "evidence integration"
used here is analogous to the phrase "weight of evidence" used in some other assessment processes
fEFSA. 2017: U.S. EPA. 2017a: NRC. 2014: U.S. EPA. 2005a1"
For each potential human health effect (or smaller subset of related outcomes), the U.S.
Environmental Protection Agency (EPA) will separately synthesize the available phenotypic human
and animal evidence pertaining to that potential health effect. Mechanistic evidence will also be
considered in targeted analyses conducted before, during, and after developing syntheses of the
phenotypic human and animal evidence. The results of the analyses of mechanistic evidence will be
used to help resolve key uncertainties; as a result, the scope of the mechanistic analyses will
generally depend on the extent and nature of the human and animal evidence (see Sections 9.2 and
10). Thus, apart from the predefined mechanistic analyses (see Sections 9.2.1-9.2.3), the human
and animal evidence syntheses (or the lack of phenotypic data in humans and animals) help
determine the approach to be taken in synthesizing the available mechanistic evidence (see
Section 9.2.4). In this way, a mechanistic evidence synthesis might range from a high-level
summary of potential toxicity mechanisms discussed in the published literature to a detailed
analysis of multiple potential modes of action, or it might evaluate specific, focused questions that
inform key uncertainties unaddressed by the phenotypic human and animal evidence (e.g., shape of
the dose-response curve at low doses, applicability of the animal evidence to humans, addressing
susceptible populations). Each synthesis will provide a summary discussion of the available
evidence that addresses considerations regarding causation (see Table 9-1). These considerations
15This revision has been adopted primarily based on the 2014 NAS review of IRIS (NRC, 2014): 'The present
committee found that the phrase weight of evidence has become far too vague as used in practice today and thus
is of little scientific use. In some accounts, it is characterized as an oversimplified balance scale on which evidence
supporting hazard is placed on one side and evidence refuting hazard on the other... The present committee
found the phrase evidence integration to be more useful and more descriptive of what is done at this point in an
IRIS assessment—that is, IRIS assessments must come to a judgment about whether a chemical is hazardous to
human health and must do so by integrating a variety of evidence."
This document is a draft for review purposes only and does not constitute Agency policy.
9-1 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
1 are adapted from considerations for causality introduced by Austin Bradford Hill (Hill.
2 19651: consistency, dose-response relationship, strength of the association, temporal relationship,
3 biological plausibility, coherence, and "natural experiments" in humans [see additional discussion
4 in U.S. EPA f2005al and U.S. EPA T19941]. Importantly, the evidence synthesis process explicitly
5 considers and incorporates the conclusions from the individual study evaluations (see Section 6).
This document is a draft for review purposes only and does not constitute Agency policy.
9-2 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 9-1. Information most relevant to describing primary considerations
for assessing causality during evidence syntheses
Consideration
Description of the consideration and its application in IRIS syntheses
Study confidence
Description: Incorporates decisions about study confidence within each of the
considerations.
Application: In evaluating the evidence for each of the causalitv considerations described
in the following rows, the syntheses will consider study confidence decisions. High
confidence studies carry the most weight. The syntheses will consider the specific
limitations and strengths identified during study evaluation and describe how these
informed each consideration.
Consistency
Description: Examines the similarity of results (e.g., direction; magnitude) across studies.
Application: Syntheses will evaluate the homogeneity of findings on a given outcome or
endpoint across studies. When inconsistencies exist, the syntheses consider whether
results were "conflicting" (i.e., unexplained positive and negative results in similarly
exposed human populations or in similar animal models) or "differing" (i.e., mixed results
explainable by, for example, differences between human populations, animal models,
exposure conditions, or studv methods) (U.S. EPA, 2005a). These considerations are
based on analyses of potentially important explanatory factors such as:
• Confidence in studies' results, including study sensitivity (e.g., some study
results that appear to be inconsistent may be explained by potential biases or
other attributes that affect sensitivity).
• Exposure, including route (if applicable) and administration methods, levels,
duration, timing with respect to outcome development (e.g., critical windows),
and exposure assessment methods (i.e., in epidemiology studies), including
analytical units and specific groups being compared.
• Specificity and sensitivity of the endpoint for evaluating the health effect in
question (e.g., functional measures can be more sensitive than organ weights).
• Populations or species, including consideration of potential susceptible groups
or differences across lifestage at exposure or endpoint assessment.
• Toxicokinetic information explaining observed differences in responses across
route of exposure, other aspects of exposure, species, sexes, or lifestages.
The interpretation of consistency will emphasize biological significance, to the extent
that it is understood, over statistical significance. Statistical significance from suitably
applied tests (this may involve consultation with an EPA statistician) adds weight when
biological significance is not well understood. Consistency in the direction of results
increases confidence in that association even in the absence of statistical significance. In
some cases, it may be helpful to consider the potential for publication bias to provide
context to interpretations of consistency.3
This document is a draft for review purposes only and does not constitute Agency policy.
9-3 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Consideration
Description of the consideration and its application in IRIS syntheses
Strength (effect
magnitude) and
precision
Description: Examines the effect magnitude or relative risk, based on what is known
about the assessed endpoint(s), and considers the precision of the reported results
based on analyses of variability (e.g., confidence intervals; standard error). This may
include consideration of the rarity or severity of the outcomes.
Application: Syntheses will analvze results both within and across studies and mav
consider the utility of combined analyses (e.g., meta-analysis). While larger effect
magnitudes and precision (e.g., p < 0.05) help reduce concerns about chance, bias, or
other factors as explanatory, syntheses should also consider the biological or
population-level significance of small effect sizes.
Biological gradient/
dose-response
Description: Examines whether the results (e.g., response magnitude; incidence;
severity) change in a manner consistent with changes in exposure (e.g., level; duration),
including consideration of changes in response after cessation of exposure.
Application: Syntheses will consider relationships both within and across studies,
acknowledging that the dose-response relationship (e.g., shape) can vary depending on
other aspects of the experiment, including the biology underlying the outcome and the
toxicokinetics of the chemical. Thus, when dose-dependence is lacking or unclear, the
synthesis will also consider the potential influence of such factors on the response
pattern.
Coherence
Description: Examines the extent to which findings are cohesive across different
endpoints that are related to, or dependent on, one another (e.g., based on known
biology of the organ system or disease, or mechanistic understanding such as
toxicokinetic/dynamic understanding of the chemical or related chemicals). In some
instances, additional analyses of mechanistic evidence from research on the chemical
under review or related chemicals that evaluate linkages between endpoints or
organ-specific effects may be needed to interpret the evidence. These analyses may
require additional literature search strategies.
Application: Syntheses will consider potentially related findings, both within and across
studies, particularly when relationships are observed within a cohort or within a
narrowly defined category (e.g., occupation; strain or sex; lifestage of exposure).
Syntheses will emphasize evidence indicative of a progression of effects, such as
temporal- or dose-dependent increases in the severity of the type of endpoint observed.
If an expected coherence between findings is not observed, possible explanations should
be explored, including those related to the biology of the effects as well as the sensitivity
and specificity of the measures used.
This document is a draft for review purposes only and does not constitute Agency policy.
9-4 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Consideration
Description of the consideration and its application in IRIS syntheses
Mechanistic evidence
related to biological
plausibility
Description: There are multiple uses for mechanistic information, and this consideration
overlaps with "coherence." This consideration examines the biological support (or lack
thereof) for findings from the human and animal health effect studies and becomes
more influential on the hazard conclusions when notable uncertainties in the strength of
those sets of studies exist. These analyses can also improve understanding of dose- or
duration-related development of the health effect. In the absence of human or animal
evidence of apical health endpoints, the synthesis of mechanistic information may drive
evidence integration conclusions (when such information is available).
Application: Syntheses can evaluate evidence on precursors, biomarkers, or other
molecular or cellular changes related to the health effect(s) of interest to describe the
likelihood that the observed effects result from exposure. This evaluation will entail an
analysis of existing evidence, and not simply speculate whether a theoretical pathway
can be postulated. This analysis may not be limited to evidence relevant to the PECO but
may also include evaluations of biological pathways (e.g., for the health effect;
established for other, possibly related, chemicals). Any such synthesis of mechanistic
evidence will consider the sensitivity of the mechanistic changes and the potential
contribution of alternative or previously unidentified mechanisms of toxicity.
Natural experiments
Description: Specific to epidemiology studies and rarelv available, this consideration
examines effects in populations that have experienced well-described, pronounced
changes in chemical exposure (e.g., lead exposures before and after banning lead in
gasoline).
Application: Compared with other observational designs, natural experiments have the
benefit of dividing people into exposed and unexposed groups without them influencing
their own exposure status. During synthesis, associations in medium and high
confidence natural experiments can substantially reduce concerns about residual
confounding.
Publication bias involves the influence of the direction, magnitude, or statistical significance of the results on the
likelihood of a paper being published; it can result from decisions made, consciously or unconsciously, by study
authors, journal reviewers, and journal editors (Dickersin, 1990). When evidence of publication bias is present for
a set of studies, less weight may be placed on the consistency of the findings for or against an effect during
evidence synthesis and integration.
PECO = populations, exposures, comparators, and outcomes.
1
2
3
4
16Various terms have been used to characterize populations that may be at increased risk of developing health
effects from exposure to environmental chemicals, including "susceptible," "vulnerable," and "sensitive."
Furthermore, these terms have been inconsistently defined across the scientific literature. The term susceptibility
is used in this protocol to describe populations or lifestages at increased risk, focusing on intrinsic biological
factors that can modify the effect of a specific exposure, but also considering social determinants or behaviors
that may increase susceptibility. However, factors resulting in higher exposures to specific groups (e.g., proximity,
housing, occupation) will typically not be analyzed to describe increased risk among specific populations or
subgroups.
This document is a draft for review purposes only and does not constitute Agency policy.
9-5 DRAFT-DO NOT CITE OR QUOTE
Data permitting, the syntheses will also discuss analyses relating to potential susceptible
populations.16 These analyses will be based on knowledge about the health outcome or organ
system affected, demographics, genetic variability, lifestage, health status, behaviors or practices,
and social determinants (see Table 9-2). This information will be used to draw conclusions
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
regarding potential susceptibility among specific populations or subgroups in a separate section
(see Section 10.3). This summary will describe concerns across the available evidence for all
potential human health effects and will be used for both hazard identification and dose-response
analyses.
Table 9-2. Individual and social factors that may increase susceptibility to
exposure-related health effects
Factor
Examples
Demographic
Sex, age, race/ethnicity, education, income, occupation, geography
Genetic variability
Polymorphisms in genes regulating cell cycle, DNA repair, cell division,
cell signaling, cell structure, gene expression, apoptosis, and metabolism
Lifestage
In utero, childhood, puberty, pregnancy, women of childbearing age, old
age
Health status
Pre-existing conditions or disease such as psychosocial stress, elevated
body mass index, frailty, nutritional status, chronic disease
Behaviors or practices
Diet, mouthing, smoking, alcohol consumption, pica, subsistence or
recreational hunting and fishing
Social determinants
Income, socioeconomic status, neighborhood factors, health care
access, and social, economic, and political inequality
EPA ExpoBox Exposure Assessment Tools, based on EPA's Guidelines for Exposure Assessment (U.S. EPA, 1992).
DNA = deoxyribonucleic acid.
9.1. HUMAN AND ANIMAL HEALTH EFFECTS EVIDENCE
The syntheses of the human and animal health effects evidence will focus on describing
aspects of the evidence that best inform causal interpretations, including the exposure context
examined in the sets of studies. Each evidence synthesis will be based primarily on studies of high
and medium confidence. Low confidence studies may be used if few or no studies with higher
confidence are available to help evaluate consistency, or if the study designs of the low confidence
studies address notable uncertainties in the set of high or medium confidence studies on a given
health effect If low confidence studies are used, then a careful examination of risk bias and
sensitivity with potential impacts on the evidence synthesis conclusions will be included in the
narrative.
As previously described, these syntheses will articulate the strengths and the weaknesses of
the available evidence organized around the considerations described in Table 9-1, as well as issues
that stem from the evaluation of individual studies (e.g., concerns about bias or sensitivity). If
possible, results across studies will be compared using graphs and charts or other data
visualization strategies. The analysis will typically include examination of results stratified by any
This document is a draft for review purposes only and does not constitute Agency policy.
9-6 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
or all of the following: study confidence classification (or specific issues within confidence
evaluation domains), population or species, exposures (e.g., level, patterns [intermittent or
continuous], duration, intensity), sensitivity (e.g., low vs. high), and other factors that may have
been identified during study evaluation or analyses of key science issues (see Section 2.4). The
number of studies and the differences encompassed by the studies will determine the extent to
which specific factors can be examined for use in stratifying study results. Additional analyses
across studies (e.g., meta-analyses) may also be conducted for both the human and animal evidence
syntheses, if supported by available data.
9.2. MECHANISTIC INFORMATION
The synthesis of mechanistic information informs the integration of health effects evidence
for both hazard identification (i.e., biological plausibility or coherence of the available human or
animal evidence; inferences regarding human relevance, or the identification of susceptible
populations and lifestages across the human and animal evidence) and dose-response evaluation.
As introduced in previous sections, several key science issues that are essential to consider in these
five assessments will involve a focused analysis and synthesis of mechanistic information (see
Sections 9.2.1-9.2.3). Other potential assessment-specific uncertainties for which mechanistic
analyses might be conducted, and the considerations for including those analyses in an assessment,
are outlined in Section 9.2.4. Deviations from the approaches described in Sections 9.2.1-9.2.3, as
well as the specific methods for any analyses conducted based on the considerations described in
Section 9.2.4, will be tracked as updates to the protocol.
Mechanistic evidence includes any experimental measurement related to a health outcome
that provides information about the biological or chemical events associated with phenotypic
effects; these measurements can improve understanding of the mechanisms involved in the toxic
effects following exposure to a chemical but are not generally considered adverse outcomes.
Mechanistic data are reported in a diverse array of observational and experimental studies across
species, model systems, and exposure paradigms, including in vitro, in vivo (by various routes of
exposure), ex vivo, and in silico studies, and across a wide spectrum of diverse endpoints.
Evaluations of mechanistic information typically differ from evaluations of phenotypic
evidence (e.g., from routine toxicological studies). This is primarily because mechanistic data
evaluations consider the support for and involvement of specific events or sets of events within the
context of a broader research question (e.g., support for a hypothesized mechanism; consistency
with known biological processes), rather than evaluations of individual apical endpoints considered
in relative isolation. Such analyses are complicated because a chemical may operate through
multiple mechanistic pathways, even if one hypothesis dominates the literature (U.S. EPA. 2005a).
Similarly, multiple mechanistic pathways might interact to cause an adverse effect. Thus, pragmatic
and stepwise approaches to considering and reviewing this evidence for these PFAS assessments
are outlined below. The format of these syntheses is expected to vary from a short narrative
This document is a draft for review purposes only and does not constitute Agency policy.
9-7 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
summary of existing knowledge to an in-depth analysis and weighing of the evidence underlying
multiple mechanistic events, depending on data availability and the criticality of the
assessment-specific uncertainties.
9.2.1. Toxicokinetic Information and Pharmacokinetic (PK)/Physiologically Based
Pharmacokinetic (PBPK) Models
One key mechanistic issue involves the toxicokinetics of these five PFAS, particularly their
serum half-life values because these values are useful for extrapolating doses from exposed animals
to humans. Toxicokinetic studies were extracted for consideration (from the broad PFAS literature
searches) by subject matter experts using two different methods: (1) tagging of studies during
literature screening (see Sections 4.2-4.3), noting that this tagging was not conducted by ADME
subject matter experts, and (2) use of SWIFT Review software
[https: //www.sciome.com/swift-review/: Howard etal. (2016)] to categorize the literature via
health outcome tags for ADME from the title and abstract For identification of ADME-related
studies to be reviewed using SWIFT Active Screener
f https: //www.sciome.com /swift-activescreener/]. the results using the health outcome tags for
ADME embedded within SWIFT Review were confirmed using a search string developed by experts
in toxicokinetics within the IRIS Program.17 This tagging resulted in 813 potentially relevant
studies that were imported into SWIFT Active Screener for review by two independent reviewers
with demonstrated expertise in ADME (conflicts were resolved through discussion). A basic set of
PECO criteria were used for this review:
• Population: in vivo studies in humans, nonhuman primates, rats, or mice. (Note: in vitro
studies in these species were tagged as potentially supportive; see explanation below.)
• Exposure: any route of administration of a single chemical compound that is expected to
occur for human exposure for PFBA, PFHxS, PFHxA, PFDA, or PFNA. Exposure to metabolic
precursors of these chemicals was also included. (Note: intraperitoneal [i.p.] injection
studies and in vitro studies were tagged as potentially supportive; see explanation below.)
17tiab: (adme OR admet OR bile OR biliary OR bioavail* OR biodistribut* OR biologic-avail* OR biological-avail* OR
biologically-avail* OR biotrans* OR clearance OR detox* OR distribut* OR dosim* OR eliminat* OR endocytosis
OR enterohepatic OR "entero hepatic" OR excret* OR exhalation OR hepatobiliary OR inhalation OR metaboli* OR
"partition coefficient" OR permeability OR persistence OR phagocytosis OR pharmacokinetic* OR
physiologic-avail* OR physiological-avail* OR physiologically-avail* OR pinocytosis OR protein-bind* OR
reabsorption OR retention OR secretion* OR toxicokinetic* OR transport OR uptake OR urination OR ((absorb OR
absorbs OR absorbed OR absorption* OR deposition) NOT (atomic OR optical OR spectra* OR spectros* OR
spectrum* OR infrared)) OR title : ("gas exchange" AND (alveolar OR lung OR lungs OR pulmonary OR respirat*))
OR mesh_mh: ("biological transport" OR "enterohepatic circulation" OR pharmacokinetics) OR mesh_sh:
(pharmacokinetics) OR mesh_mh: (toxicokinetics)).
This string identified two fewer potentially relevant studies than the SWIFT review (including all studies identified
using the non-SWIFT string). So, the studies identified by SWIFT Review were screened in SWIFT Active Screener.
This document is a draft for review purposes only and does not constitute Agency policy.
9-8 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
• Comparator: vehicle control or reference population.
• Outcome: data to quantify ADME processes, steady state analysis, empirical
pharmacokinetic (PK), full PBPK.
This screening (i.e., to 96% predicted completion based on the machine-learning software)
resulted in the identification of 99 studies relevant to toxicokinetics across the five PFAS
assessments. These data will be considered for use in the assessments as described below.
All PK and PBPK models will be formally evaluated for use in the assessments, as described
in Section 6.4. The specific approaches for determining the most appropriate method for
dosimetric extrapolation, if necessary for these assessments (note: this is likely to be necessary,
based on the preliminary literature inventory), as well as other potential quantitative approaches
for using the PK/PBPK models and ADME data, are outlined in Section 11.2.
To draw conclusions regarding the most appropriate serum half-life measures, the ADME
studies identified by the screening methods described above will be considered as outlined in U.S.
EPA (2018b). Briefly, the studies relevant to updating the data presented in Table 2-7 in
Section 2.4.1, including the studies underlying the current data in the table, will be reviewed, and
data that are highly unreliable will be excluded (e.g., data points below the limit of detection [LOD];
values based on uncertain exposure estimates, or other unvalidated assumptions). Study
characteristics that will be reviewed by subject matter experts to determine whether studies are
informative to the PFAS-specific half-life values include appropriateness of the analytic method, the
number of exposure levels tested, the human relevance of the exposure range, and the number of
time points and tissues sampled. Although ADME data from in vitro studies and i.p. injection
studies were tracked as potentially relevant during screening, additional considerations will apply
to the potential incorporation of these data into the assessments, given their inherent uncertainties
(e.g., difficulties interpreting the relevance of bioavailability or peak concentration data from i.p.
injection studies). Specifically, regarding in vitro studies, it is expected that there may be no in vivo
toxicokinetic data on the rate of conversion of precursor compounds to the PFAS of interest, in
which case conversion rates measured in vitro can be extrapolated to in vivo as the next best means
of predicting this mechanism of exposure. Even if such extrapolation is determined to be
quantitatively uncertain, these data might still provide useful qualitative information.
While data and careful PBPK modeling of PFOA and PFOS have revealed nonlinear kinetics
attributed to a mechanism of saturable renal resorption [e.g., Loccisano etal. (2011)]. initial
evaluation of PFAS data for the compounds addressed in this protocol do not show such a clear
pattern; that is, studies evaluating PK parameters at high and low doses do not show a significant
dose-dependence in clearance. Such dose-dependence is taken to be distinct from time-dependent
biphasic distribution patterns, whereby an initial, relatively rapid distribution phase is followed by
a slower (terminal) elimination phase. The distribution phase is more rapid because the decline in
blood or plasma concentration reflects both elimination and distribution to peripheral tissues, but
This document is a draft for review purposes only and does not constitute Agency policy.
9-9 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
the corresponding half-lives may be independent of dose. EPA's analysis of the PK data will seek to
identify a common elimination-phase half-life (or clearance) among all doses and studies for a given
PFAS in each animal species and sex, or among humans, separated into men and women given
sufficient data. Variation in the rate of absorption (for oral dosing) and distribution phases is
expected to occur between and within studies from random factors that are not dose dependent,
and between tissues within a study due to differing distribution characteristics. Various features of
study design will be considered in evaluating apparent variation (e.g., if the duration of a study is
too short or the sensitivity of the analytic method too low to observe the terminal elimination
phase, such that the apparent clearance is likely to be due to distribution within the animal or
subject). If this analysis reveals a clear dose dependence (nonlinearity) in the elimination phase,
separate from these other sources of variation, the analysis will then focus on identifying and using
the half-life at low doses, considered most relevant to animal-human extrapolation.
Because significant differences in the half-life between males and females of a given species
have been observed for some PFAS, these sex differences will be assumed to be real in general
across species. Specifically, when feasible, the data for males and females of each species, for each
PFAS, will be analyzed separately, even if the difference is not statistically significant. If the values
for the elimination-phase half-life differ significantly across studies for the same species/strain/sex,
a more detailed review of the study methods indicated above will be conducted to determine
whether one study is more likely to provide accurate information than another.
Given that PFAS inhalation exposure is expected to be via adsorption to particulates, if
sufficient data are available for any of the assessed PFAS, inhalation exposure rates for
PFAS-containing particles will be predicted using the multiple-path particle dosimetry (MPPD)
model fhttps://www.ara.com/products/multiple-path-particle-dosimetrv-model-mppd-v-3 041
This model predicts inhaled particle deposition in laboratory animals and humans as a function of
particle size. Particle sizes used in controlled animal studies or measured in ambient
environmental or workplace exposure studies in humans will be used as inputs. If PK data are
identified that allow the bioavailability of inhaled particulate PFAS to be estimated, the mass
deposition predicted by the MPPD model will be adjusted accordingly. Otherwise 100% absorption
of PFAS from inhalation deposition will be assumed. Note that while some inhaled particles are
later moved by the mucociliary apparatus to the larynx and swallowed, the PK bioavailability for
oral ingestion can then be applied to that fraction. Any predictions will be considered for use in
comparing findings across oral and inhalation routes of exposure during evidence integration (see
Section 10). In addition, see Section 11.2 for the application of these predictions to developing
quantitative estimates. If necessary, inhalation of PFAS in free ionic form can be estimated based
on the inhalation uptake of other chemicals with high liquid: air partition coefficients (i.e., assuming
nearly complete absorption of any free ions that contact the airway lining).
This document is a draft for review purposes only and does not constitute Agency policy.
9-10 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
9.2.2. Peroxisome Proliferator-Activated Receptor Alpha (PPARa) Dependence for Health
Effect(s) Observed in Animals
A second area of focused mechanistic analysis is evaluating the human relevance of effects
in animals that appear to involve (at least in part) a PPARa-mediated MOA. The approach outlined
below focuses on hepatic effects, which are expected to be the primary health effect area in these
assessments for which this analysis is useful, and for which there are likely to be data for analysis.
The specifics of applying this approach may vary across the five PFAS assessments, depending on
the availability of data to address this question and the strength of the evidence indicating PPARa
involvement. During assessment development, for other health effects with evidence that a
PPARa-mediated MOA might be operant, the mechanistic syntheses will include consideration of
this issue. These analyses will depend on the amount of information available and the strength of
the evidence indicating PPARa involvement. Thus, the analyses might range from a short summary
of the available evidence when data are sparse to an evaluation approximating the one described
below when extensive data are available.
To identify the literature most relevant to addressing the question of the
PPARa-dependence of hepatic effects observed in experimental animals, a PFAS assessment with
extensive evidence of liver effects and potential PPARa involvement will screen18 the "potentially
relevant supplemental material" studies on a given PFAS at the full-text level as follows:
• Population: in vivo animal studies in mammalian models; in vitro and human experiments
using primary or immortalized liver cell lines
• Exposure: PFAS of interest (parent compound only)
• Comparator: vehicle control
• Outcome: mechanistic outcomes relevant to the hepatobiliary system (e.g., in liver tissues or
cells)
Any additional assessment-specific strategies for identifying other information of potential
relevance on molecular mechanistic data for these five PFAS, or from the more extensive literature
on perfluorooctanoic acid (PFOA) and PFAS (e.g., as points of comparison), will be described as
updates to this protocol.
The pool of studies identified based on the strategies outlined above will be inventoried into
a database to allow for the organization and evaluation of these data. Specifically, the following
information will be extracted for each reference: a reference identifier; test compound; exposure
18Although the specifics of this screening process may vary across PFAS, this protocol describes that screening will
occur by at least two reviewers and use of DistillerSR to track decisions and resolve differences. Any deviations
from this will be tracked on an assessment-specific basis as updates to the protocol.
This document is a draft for review purposes only and does not constitute Agency policy.
9-11 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
route and duration; the sex, species, and strain of the organism; age at exposure; and endpoint
evaluation of the test organism or test system. Additionally, the inventory(ies) will capture a
succinct description of the assessed endpoints and the potential mechanistic event(s) informed by
those endpoints in each study. The mechanistic events in the proposed mechanisms pathway for
which there are data will then be organized according to the following levels of biological
organization: molecular target(s), cellular response(s), tissue/organ response(s), and organism
response(s), in accordance with the levels of biological organization used to develop adverse
outcome pathways (AOPs).19
Although refinements based on the assessment-specific evidence are anticipated, these
assessments will first consider the use of the preliminary pathway outlined in Figure 9-1 as an
organizing AOP for these data. The preliminary, proposed AOP displayed in Figure 9-1 is based on
molecular initiating events, key events, and adverse outcomes identified in previous evaluations on
PFOS and PFOA and proposed AOPs for chemical-induced noncancer liver toxicity [see Li etal..
2017. Mellor et al. ("20161. Wang et al. ("20141. U.S. EPA C2016dl. U.S. EPA C2016dl. AT SDR ("20181.
andNTDWOI (20171], Prior evaluations of PFOS and PFOA have discussed studies using wild-type,
PPARa knockout and humanized PPARa (hPPARa) mice showing that exposure leads to fatty acid
and triglyceride accumulation in the liver and steatosis via both PPARa-dependent
and -independent pathways (ATSDR. 2018: Li etal.. 2017: Viberg and Eriksson. 20171. In addition
to PPARa, these reviews have implicated other nuclear receptor (NR) and cell signaling pathways
with PFOA- and PFOS-induced noncancer liver effects, including PPARp/5, PPARy, constitutive
androstane receptor (CAR) and pregnane X receptor (PXR), the farnesoid X receptor (FXR), the
phosphatidylinositol 3-kinase-serine/threonine kinase Akt (PI3K-Akt) signal transduction pathway,
and the nuclear factor kappa B pathway (NF-kB) (Li etal.. 2017: Viberg and Eriksson. 2017).
Activation of these pathways can be associated with alterations in lipid and glucose metabolism,
increased cellular stress, and inflammation fMackowiak etal.. 2018: Li etal.. 2017: Mellor etal..
2016: Wang etal.. 20141. Thus, the potential involvement and contribution of these different
signaling responses to hepatic effects after exposure to the five PFAS will also be considered.
19Although the World Health Organization (WHO)-lnternational Programme on Chemical Safety (IPCS)-MOA and
the Organisation for Economic Co-operation and Development (OECD)-AOP frameworks are similar in the
identification and analysis of key events following modified Bradford Hill criteria (Meek et al., 2014), AOPs are
chemical agnostic whereas MOA analyses are intended to inform health assessments of individual (or groups of)
chemical(s) (Edwards et al., 2016).
This document is a draft for review purposes only and does not constitute Agency policy.
9-12 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Figure 9-1. Preliminary proposed mechanistic pathway for per- and
polyfluoroalkyl substances (PFAS)-induced noncancer liver effects. Based on
previous reviews of perfluorooctane sulfonate (PFOS)- and perfluorooctanoic acid
(PFOA)-induced noncancer liver effects in animals fATSDR. 2018: Li etal.. 2017:
Viberg and Eriksson. 2017: U.S. EPA. 2016c. d), and proposed adverse outcome
pathways for hepatic steatosis fMellor et al.. 20161.
NAFLD = nonalcoholic fatty liver disease; ROS = reactive oxygen species; TNFa = tumor necrosis factor alpha;
XME = xenobiotic metabolizing enzymes.
The analysis of the involvement of PPARa and these other signaling cascades in hepatic
toxicity after exposure to these five PFAS will focus on the concordance of changes in the specific
mechanistic events or separate pathways to effects (i.e., in Figure 9-1, and as otherwise identified
during assessment-specific evaluations) across species to ascertain the relevance of animal studies
to human health. The analyses of evidence for each mechanistic event and potential pathway will
be qualitatively analyzed for various aspects of the Hill considerations outlined in the EPA Cancer
Guidelines framework for MOA analysis (U.S. EPA. 2005a). Given the focus of these analyses, the
review will stress the aspects of consistency, coherence, and biological plausibility to ascertain the
level of support (or lack thereof), depending on the availability of data. To facilitate this analysis,
the following prompting questions and clarifying considerations will be used, depending on the
assessment-specific data:
• What is the level of evidentiary support (or lack thereof) for the mechanistic events or
signaling pathways, based on the assessment-specific PFAS data? In parallel, are
assessment-specific data available to inform the strength of the linkages between events in
the pathway or across pathways? In general, well-conducted, independent studies using
different experimental models and reporting consistent or coherent findings would provide
This document is a draft for review purposes only and does not constitute Agency policy.
9-13 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
1 strong supportive evidence for a mechanistic (potentially key) event or pathway (or
2 linkages between events in a pathway), with a lesser degree of support provided by
3 individual experimental observations or sets of studies reporting some consistent or
4 coherent findings as well as some equivocal results or findings that vary from one model to
5 another without explanation.
6 • Are sufficient assessment-specific data available to inform exposure duration- or
7 level-dependencies for any of the evaluated mechanistic events or pathways?
8 • Is the assessment-specific evidence (on specific events or pathways in general) consistent
9 with the general biology of the human liver or mechanisms known to be associated with
10 noncancer liver effects in humans? To consider this question, assessments will compare the
11 endpoint-level results across studies on a particular PFAS against the mechanistic
12 understanding/underlying biology for similar effects in the human liver. (Note: this analysis
13 might be informed by studies or reviews on the more robust PFOA/perfluorooctane
14 sulfonate [PFOS] evidence bases.)
15 • Are responses across studies for these five PFAS assessments indicative of activation of
16 specific mechanisms or signaling pathways conserved across experimental models and
17 designs? To consider this question, assessments will include an evaluation of consistency
18 and coherence across different species and strains of animals, human and animal cell
19 culture models, and in vivo humanized animal models, depending on data availability.
20 • Does the assessment-specific mechanistic information indicate the likelihood of populations
21 or lifestages that may be more susceptible to PFAS-induced liver effects?
22 The assessment-specific conclusions (and attendant uncertainties) regarding these
23 questions will be used to draw judgments regarding the human relevance of these animal effects,
24 and the rationale for these judgments will be documented transparently within each assessment.
25 As described in EPA guidance (U.S. EPA. 2005a). human relevance is the default, and mechanistic
26 evidence will need to be compelling and strong to conclude otherwise (i.e., to conclude that findings
27 in animals are not relevant to humans).
9.2.3. Toxicological Relevance of Select Outcomes Observed in Animals
28 The preliminary literature inventory identified studies on several health outcomes relating
29 to potential urinary and hepatic effects (see below) for which it is expected to be difficult to
30 determine whether any observed changes (or a lack of changes) are toxicologically relevant It is
31 expected that in some instances, the synthesis will need to address this issue to inform whether the
32 effects in animals are relevant to interpreting the potential for PFAS exposure to cause a human
33 health effect, and in other instances addressing this issue may be necessary to identify a level of
34 change for use in determining the potential for adversity or in dose-response analysis. It is possible
35 that additional outcomes with similar questions of health relevance might be identified during the
36 development of these assessments. If so, the specifics of the approach selected to address those
This document is a draft for review purposes only and does not constitute Agency policy.
9-14 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
outcomes will be documented in the assessment(s) and as an update to this protocol. For the
aforementioned outcomes, different approaches will be taken, specifically:
1) Kidney changes in rats, including chronic progressive nephropathy (CPN) and effects that
appear to be mediated by an alpha 2u-gIobuIin MOA. Because the rodent (i.e., male
rat)-specific alpha 2u-globulin MOA is not considered relevant to humans, assessments with
evidence indicating its involvement will include an evaluation and judgment of the evidence
supporting (or not supporting) dependence on this MOA. Specifically, these data will be
evaluated against the predefined criteria established by the U.S. EPA f!991al and/or more
recently established criteria, such as those published by Swenberg and Lehman-McKeeman
(19991. Relatedly (and possibly overlapping the evaluation of alpha 2u-globulin, because
this MOA may exacerbate CPN), there is no human disease analog to the constellation of
changes observed in rodent CPN. CPN represents a complex disease process in rats, and its
etiology is unknown. Thus, these evaluations will include judgments as to whether all or a
subset of the observed changes have adequate evidence to identify dependence on
rodent-specific processes, including whether it can be concluded (i.e., based on biological
understanding) that the observed kidney endpoints are associated with CPN and the
potential for exacerbation of human-relevant disease processes can be ruled out. Data
permitting, the assessments will consider whether these conclusions vary by exposure level.
2) Hepatic changes. Some individual liver endpoints (and even some constellations of
endpoints) might be considered adaptive in nature, possibly leading to the interpretation
that some statistically significant changes are not indicative of adverse effects. These
endpoints include increased liver weight, cellular hypertrophy, and single cell
necrosis/apoptosis. To draw inferences regarding the adversity of these types of liver
effects, these assessments will consider the panel recommendations outlined by Hall et al.
f20121 to draw assessment-specific judgments regarding adversity. Briefly, these include
evaluation of the available histological data and results suggesting structural degeneration
or cellular demise (e.g., apoptosis, oncosis, and/or necrosis), and clinical evidence of
hepatocyte damage. As the recommendations were developed in the context liver tumor
formation, consultation of additional relevant information will be considered to interpret
the adversity of noncancer liver effects over a lifetime exposure, taking into account that
effects perceived as adaptive can progress into more severe responses and lead to cell
injury (Hall etal.. 2012). These considerations include the EPA 2002 Guidance Document on
Hepatocellular Hypertrophy fU.S. EPA. 2002al. reference materials on clinical and
histopathology data fThoolen et al.. 2010: EMEA. 2008: Boone etal.. 20051 and publications
describing potential mechanisms of chemical-induced liver disease such as fatty liver
disease/steatohepatitis (Wahlangetal.. 2019: Toshi-Barve etal.. 2015: Wahlangetal.. 2013).
Each assessment will include an explanatory rationale documenting the application of the
Hall etal. f20121 recommendations (and any other considerations) to the available
evidence.
This document is a draft for review purposes only and does not constitute Agency policy.
9-15 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
9.2.4. Other Focused Mechanistic Analyses
Other analyses within the syntheses of mechanistic information will focus on the evidence
most useful for informing key uncertainties in the human or animal health effect evidence, both
qualitative and quantitative.
This means that, for example, if extensive and consistent high confidence human or animal
evidence is available, the need to synthesize all relevant mechanistic evidence will likely be
diminished. In these cases, the analyses will focus on reviewing and interpreting smaller sets of
mechanistic studies that specifically address controversial or outstanding issues that are expected
to have a substantial impact on the assessment conclusions. Generally, key uncertainties will be
addressed in the mechanistic evidence syntheses by considering the biological understanding of
how the effect(s) in question develop or are related. In this way, the analyses can provide
information on, for example, (1) potential precursor events when the apical data are uncertain (or
unusable for dose-response analyses), (2) animal results for which the human relevance is unclear
or controversial and the human evidence is weak, (3) the shape of the dose-response curve at low
exposure levels when this understanding is highly uncertain and data informing this uncertainty
are known to exist, or (4) the identification of likely susceptible populations and lifestages. Thus,
consideration of biological understanding represents an important component of the evidence
analysis. However, mechanistic understanding is not a prerequisite for drawing a conclusion that a
chemical causes a given health effect (NTP. 2015: NRC. 2014).
To identify the focused set(s) of studies to use in analyzing critical mechanistic questions
other than those outlined in Sections 9.2.1-9.2.3, a stepwise approach will be applied to
progressively define the scope of the mechanistic information to be considered throughout
assessment development This stepwise scoping begins during the literature search and screening
steps and depends primarily on the potential health hazard signals that arise from the individual
human and/or animal health effect studies, or from mechanistic studies that signal potential health
hazards not examined in studies of phenotypic, potentially adverse effects. Examples of the focused
questions or scenarios triggering these mechanistic evaluations, as well as when during the
systematic review they are likely to apply, are listed in Table 9-3. While the specific methods for
evaluating the evidence most relevant to each question will vary, some general considerations for
judging the evidence strength in these syntheses are provided below, and if necessary,
assessment-specific refinements will be included as updates to the protocol.
This document is a draft for review purposes only and does not constitute Agency policy.
9-16 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 9-3. Examples of questions and considerations that can trigger focused
analysis and synthesis of mechanistic information
Key
assessment-
specific
uncertainties
Examples of questions and PFAS-specific considerations for identifying the
uncertainties and key evidence to analyze
Addressing
database
completeness
based on
literature
inventories of
human,
animal, and
mechanistic
information
• Are there mechanistic studies on an organ system or potential health hazard that were
not examined by human or animal studies meeting the PECO criteria?
o Depending on the extent of the available data, consider the utility of developing a
separate synthesis of evidence versus the utility of a concise, narrative summary (or
evidence mapping) to describe these knowledge gaps. Consider whether the
mechanistic evidence might be sufficient to substantiate a conclusion on its own (if
so, a separate synthesis will be developed).
Addressing
questions of
inconsistency
within the
human and
animal
evidence
• For the health effects of potential concern, is a mechanistic evaluation(s) warranted to
inform questions regarding the consistency of the available human or animal studies?
Typically, this consideration would focus on health effects that show some indication of
an association in epidemiological studies or causality in experimental studies during
evidence synthesis. Based on the literature inventory, consider whether mechanistic
data are available to inform the specific, key uncertainties in question. Examples of
specific scenarios for evaluation include:
o If cancer has been observed and tumor types appear to differ across populations
(e.g., species or sex), review the literature inventory for mechanistic data that might
be relevant to interpreting such differences, and conduct analyses as warranted
based on that review. Approaches outlined in the EPA Cancer Guidelines (U.S. EPA,
2005a) that mav be relevant to these analyses will be applied, as appropriate.
o If pronounced and unexplained differences in health effect(s)-specific responses are
observed across lifestages or populations (e.g., animal strain; human demographic),
first consider toxicokinetic differences for the specific PFAS, and then the
mechanistic evidence relevant to assessing the potential for health effect-specific
biological differences in response (toxicodynamics). Further, inconsistent evidence
(i.e., heterogeneous results) across different animal species or human populations
might be clarified by a review of the evidence relevant to whether different
mechanisms may be operant in the different populations (e.g., evidence
demonstrating that certain species are more or less sensitive to a certain biological
perturbation; evidence that gene polymorphisms are related to variability in
response).
This document is a draft for review purposes only and does not constitute Agency policy.
9-17 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Key
assessment-
specific
uncertainties
Examples of questions and PFAS-specific considerations for identifying the
uncertainties and key evidence to analyze
Addressing
questions of
biological
plausibility3
and coherence
within the
human and
animal
evidence, and
coherence
across bodies
of evidence
• For the health effects of potential concern, would a mechanistic evaluation(s) of
biological plausibility (usually for an individual outcome) or coherence (usually across
outcomes) provide meaningful information for interpreting the evidence strength?
Typically, this consideration would focus on effects for which the evidence strength for
an individual outcome (either for or against an effect) is questionable (e.g., primarily
studies of low confidence), when a substantial outstanding methodological concern(s)
across the relevant studies exists, or when evidence exists for multiple, potentially
related (e.g., biologically) outcomes. Based on the literature inventory, consider
whether there are mechanistic data available to inform the specific, key uncertainties in
question. Examples of specific scenarios for evaluation may include:
o If the evidence for a given outcome is weak or uncertain, or when unaddressed
methodological concerns identify critical uncertainties in the human or animal
findings for a health effect, identify data on mechanistic changes in exposed
humans or animals that are likely to be linked to the development or occurrence of
the health outcome in question. If enough suitable studies are available, analyze
data on changes expected to be related to the phenotypic finding(s) of interest,
which can either increase or decrease the evidence strength that the finding(s) is
real. It is important to note that the absence of a mechanistic explanation for an
association (e.g., the MOA is not understood) will not be used to reduce
confidence in observations from human or animal studies. However, the
plausibility of an association observed in human or animal studies may be
diminished if expected mechanistic findings (e.g., based a known biological
dependence) are tested and not apparent. The mechanistic evidence on possible
precursors or effects that are known to co-occur with the health outcome of
interest are particularly impactful when the changes are observed in the same
exposed population presenting the outcome of interest. An understanding of
mechanistic pathways (e.g., by identifying and analyzing mechanistic precursor
events linked qualitatively or quantitatively to apical health effect[s]; see Section
9.2.2 for additional context) can inform the strength of the evidence integration
judgments (see Section 10).
o If evidence on multiple health outcomes within an organ system, or possibly across
organ systems (e.g., thyroid and nervous system), is available and the strength of
the evidence for any single outcome is uncertain, identify biological data that can
inform understanding of the relatedness of outcomes within and across systems.
Biological understanding or strong mechanistic support (e.g., a shared mechanistic
event) for linkages across outcomes can increase the strength of the evidence
when changes are related. However, evidence strength may be diminished if an
expected pattern among biologically linked outcomes is not observed.
Interpretation of the pattern of changes across the outcomes will consider the
underlying biological understanding (e.g., one outcome may be expected to
precede the other, or be more sensitive). These same considerations inform
analyses of the coherence of observed effects across evidence streams during
evidence integration (see Section 10.2).
This document is a draft for review purposes only and does not constitute Agency policy.
9-18 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Key
assessment-
specific
uncertainties
Examples of questions and PFAS-specific considerations for identifying the
uncertainties and key evidence to analyze
Addressing
questions on
the human
relevance of
findings in
animals
• For the health effects of potential concern, does the available evidence raise questions
of human relevance? Typically, this consideration applies when human evidence is
lacking or has results that differ from animal studies, given that responses can differ
between humans and animals [e.g., for cancer, site concordance is not a requirement
for determining the relevance of animal data for humans (U.S. EPA, 2005a); for
noncancer nervous system effects, behavioral changes can manifest differently
between animals and humans]. The identification of potential differences will also
consider ADME information across species, primarily relating to distribution (e.g., to the
likely target tissue) and PFAS half-life. Examples of information to identify from the
literature inventory, as well as specific scenarios and considerations for these analyses
may include:
o If there is no evidence indicating that the animal results are irrelevant to humans,
summarize existing knowledge on the development of the health effect in each
species, including potential differences in PFAS toxicokinetics, and assess the
relatedness across species. Note that in the absence of sufficient evidence to the
contrarv, effects in animal models are assumed to be relevant to humans (ATSDR,
2018; NTP, 2015; U.S. EPA, 2005a).b
o If there is evidence indicating that the mechanisms underlying the effects in animals
may not operate in humans, or that the available animal model(s) may not be
suitable for the human health outcome(s) of interest, present and analyze the
strength of the evidence for and against the human relevance of the observed
findings. In addition to considerations specific to the outcome of interest, the
analysis will evaluate observations of mechanistic changes in exposed humans for
similarities or biological coherence with mechanistic or toxicological changes in
experimental animals interpreted to be associated with the health outcome under
evaluation. It may also include an evaluation of findings across species known or
presumed to be more or less relevant for interpreting potential human toxicity for
the health effect(s) in question. In rare instances or for controversial decisions that
are likely to drive key assessment conclusions, the analysis may extend to a detailed
analysis of a plausible mechanistic pathway(s) or MOA(s) within which each key
event and key event relationship is evaluated regarding the likelihood of similarities
(e.g., in presence or function) across species. These analyses, regardless of their
rigor, will lead to a definitive judgment about whether the animal response is
relevant to humans during evidence integration (see Section 10).
This document is a draft for review purposes only and does not constitute Agency policy.
9-19 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Key
assessment-
specific
uncertainties
Examples of questions and PFAS-specific considerations for identifying the
uncertainties and key evidence to analyze
Addressing
questions on
potential
susceptibility
for hazard
identification
and
dose-response
analysis
• For the health effects of potential concern, do the results from the human and animal
health effect studies appear to differ by categories that indicate the apparent presence
of susceptible populations (e.g., across demographics, species, strains, sexes, or
lifestages)? Separately, are there human or animal study data that could identify or
clarify population differences in response (e.g., experiments testing sensitivity of
responses across lifestages or across genetic variations; observed differences
attributable to genetic polymorphisms)? Are there mechanistic data (i.e., based on the
literature inventory) that address potential susceptibility factors?0 If evidence exists for
any of these scenarios, information on susceptibility will be reviewed and, if impactful
to assessment conclusions, analyzed in detail. Examples of when these analyses are
important include:
o If the analysis of evidence indicates the likely presence of a sensitive population or
lifestage in humans, the groups likely to be at greatest risk will be captured in the
evidence integration narrative (see Section 10). In addition, this narrative will
discuss whether the appropriate analogous exposures and populations or lifestages
were adequately represented or tested in the available human or animal studies,
and if not, will identify studies on the most susceptible populations or lifestages as
key research needs (see Section 10).
o If the analysis of evidence indicates the likely presence of a sensitive population or
lifestage in humans, this information will be used to select studies for quantitative
analysis (e.g., prioritizing those studies that include such populations [see
Section 11]). If specific studies addressing these susceptibilities are unusable for
quantitative analysis, susceptibility data may be used to support refined human
variability uncertainty factors or probabilistic uncertainty analyses (see Section 11).
This document is a draft for review purposes only and does not constitute Agency policy.
9-20 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Key
assessment-
specific
uncertainties
Examples of questions and PFAS-specific considerations for identifying the
uncertainties and key evidence to analyze
Addressing
questions on
biological
understanding
to optimize
dose-response
analysis
• If the human and/or animal health effect data amenable to dose-response analysis are
weakd or only at high exposure levels, or if the selection of critical parameters for
modeling is uncertain, the following analyses will be considered:
o When the apical health effect data are highly uncertain or cannot be used with
confidence for the purpose of deriving quantitative estimates, mechanistic
precursor events linked qualitatively or quantitatively to the phenotypic effect can
be evaluated for use as surrogate markers (e.g., based on the strength and
completeness of the linkage between mechanistic and phenotypic effects) for
deriving quantitative estimates.
o When understanding of the appropriate exposure metric, biomarker, or modeling
parameter for developing quantitative estimates is notably lacking, then
toxicokinetic and mechanistic understanding of the development of the health
effect can inform the most biologically appropriate measure.
o When there are dose-response modeling decisions or uncertainties that would be
substantially improved by biological or toxicokinetic understanding, mechanistic
analyses can improve selection of particular models (e.g., a linear, nonlinear, or
threshold model) and help evaluate the appropriateness of integrating/combining
data across related outcomes (e.g., based on biological coherence or a conserved
MOA). For cancer toxicity values, existing guidance will be consulted (U.S. EPA,
2005a).
aAs applied herein, biological plausibility describes mechanistic information that either strengthens or weakens an
interpretation of the likelihood of an association between exposure and the health effect. The interpretation of
biological plausibility considers the existing biological understanding of how the health effect develops and can
involve analyses of information at different levels of biological complexity (e.g., molecular, cellular, tissue).
bAs described in the EPA RfD/RfC Technical Report (2002), "one of the major default assumptions in EPA's risk
assessment guidelines is that animal data are relevant for humans [e.g., U.S. EPA (1998), U.S. EPA (1991a), and
U.S. EPA (1996a)1. Such defaults are intended to be used in the absence of experimental data that can provide
direct information on the relevance of animal data" (U.S. EPA, 2002b).
Susceptibility factors include lifestage, demographics and social determinants, behavioral factors, health status,
and genetic variability. Although not considered in these analyses, factors that can increase vulnerability, such as
other pollutant exposures or differential proximity to exposure sources, are typically considered during a full risk
assessment.
dNote that "weak" here refers to the study's usability for dose-response analysis specifically. Such studies may be
judged to be of medium or high confidence for the purposes of identifying potential hazards but possess
limitations preventing their use for deriving reliable quantitative estimates.
PECO = populations, exposures, comparators, and outcomes; RfC = inhalation reference concentration; RfD = oral
reference dose.
If focused areas for additional mechanistic evaluations are identified that can help address
key assessment-specific uncertainties (e.g., by applying Table 6-3), the assessments will identify the
most influential studies for evaluation. This could represent only a subset of the potentially
relevant studies, particularly if there are many mechanistic studies relevant to the specific
This document is a draft for review purposes only and does not constitute Agency policy.
9-21 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
question(s). Because the potential influence of the information provided by the available studies
can vary depending on the question(s) or the associated mechanistic events or pathways, the rigor
of the analyses will likewise vary from cursory insights drawn from sets of unanalyzed results to
detailed evaluations of a subset of the most relevant, individual mechanistic studies. Although the
specifics that might be applicable across potential mechanistic topic areas cannot be predefined, the
analyses will first consider the studies based on their toxicological relevance to answering the
specific question (e.g., model system; specificity of the assay for the effect of interest), potentially
refining the focus to a subset of the most relevant studies. This will be particularly important when
the set(s) of studies are inconsistent and potentially conflicting. If available, emphasis will
generally be placed on more informative studies that challenge the necessity of proposed
mechanistic relationships between exposure and an apical effect (e.g., altering a receptor-mediated
pathway through chemical intervention or using knockout animals). The analysis may also
consider whether particular study design aspects in some or all of the relevant studies are likely to
have significant flaws or important uncertainties (e.g., for certain questions, a preliminary review of
the exposure methods across the relevant mechanistic studies can flag serious deficiencies). In
general, across these assessments, relevant mechanistic information from in vivo studies will be
prioritized, with preference given to PFAS- and endpoint-relevant exposure routes and exposure
designs. Analysis of ex vivo and in vitro studies will then be considered, prioritizing those most
informative for evaluating the mechanistic events indicated by the in vivo data, including studies
conducted under conditions most relevant to human exposures and in model systems best
replicating in vivo human biology.
In some instances, additional literature searches may be warranted, targeting mechanistic
events or biological pathways that are not specific to a particular PFAS or group of PFAS. When
more rigorous mechanistic analyses are deemed necessary, the review will be aided by using
pathway-based organizational methods and, if available, established evidence evaluation
frameworks. These approaches provide transparency and objectivity to integrate and interpret the
mechanistic events and pathways anchored to the specific questions that have been identified
(e.g., anchored to a specific health effect) across diverse sets of relevant data (e.g., human, animal,
and in vitro studies). The approaches may be facilitated by using organizational tools or
frameworks, such as AOPs (see example in Section 9.2.2). As noted above, any additional
assessment-specific literature searches and evaluation methods will be described in updates to the
protocol.
Based on the analyses and considerations outlined in Sections 9.2.1-9.2.4, the results of the
health effect- and assessment-specific mechanistic evidence syntheses will inform both evidence
integration and dose-response analyses (see Sections 10 and 11).
This document is a draft for review purposes only and does not constitute Agency policy.
9-22 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
10. EVIDENCE INTEGRATION
For analyzing human health outcomes that might result from chemical exposure, these PFAS
assessments will draw integrated judgments across human, animal, and mechanistic evidence for
each assessed health effect The evidence integration judgments include interpretations drawn
regarding the support provided by the individual bodies of evidence (i.e., human, animal, and
mechanistic evidence) based on the structured application of an adapted set of considerations first
introduced by Austin Bradford Hill fHill. 19651. which are directly informed by the summary
discussions of each body of evidence during evidence synthesis (see Section 9). As previously
discussed in Section 9.2, the approach to evaluating the mechanistic evidence relevant to each
assessed health effect will follow a stepwise approach and is expected to vary depending on the
nature and impact of the uncertainties identified within each evidence base, as well as the specific
mechanistic information available to address those uncertainties. This includes evaluations of
mechanistic evidence relevant to the identified key science issues (see Section 2.4) prior to or in
parallel with evaluations of the phenotypic data in human and animal studies, as well as other
focused mechanistic analyses identified during draft development to address key assessment
uncertainties (see Section 9.2.4 for a discussion of these scenarios). During evidence integration, a
structured and documented process will be used, as follows (and depicted in Figure 10-1):
• Building from the separate syntheses of the human and animal evidence (see Section 9.1),
the strength of the evidence from the available human and animal health effect studies will
be summarized in parallel, but separately, using a structured evaluation of an adapted set of
considerations first introduced by Austin Bradford Hill fHill. 19651. Table 10-2 describes
these structured evaluations and the explicit consideration of study confidence within each
evaluation domain. Based on the approaches and considerations described in Section 9.2,
these summaries will incorporate the relevant mechanistic evidence (or MOA
understanding) that informs the biological plausibility and coherence within the available
human or animal health effect studies.
• The strength of the animal and human evidence will be considered together in light of
inferences across evidence streams. Specifically, the inferences considered during this
integration include the human relevance of the animal and mechanistic evidence, coherence
across the separate bodies of evidence, and other important information (e.g., judgments
regarding susceptibility). Note that without evidence to the contrary, the human relevance
of animal findings is assumed.
This document is a draft for review purposes only and does not constitute Agency policy.
10-1 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
• A summary judgment is drawn as to whether the available evidence base for each potential
human health effect as a whole is sufficient (or insufficient) to indicate that PFAS exposure
has the potential to be hazardous to humans.20
Evidence Stream Evaluation
Based on the structured review of adapted Hill
considerations (including biological understanding),
as part of the evidence integration narrative:
• Qualitatively summarize the strength of the
evidence from studies in humans.
• Qualitatively summarize the strength of the
evidence from animal studies.
Figure 10-1. Process for evidence integration. Note that "sufficient evidence"
could indicate a judgment of "sufficient evidence for hazard" or "sufficient evidence
to judge that a hazard is unlikely," depending on the nature and extent of the
available evidence (see Table 10-3],
The decision points within the structured evidence integration process will be summarized
in an evidence profile table for each health effect category (see Table 10-1 for a preliminary
template version) in support of the evidence integration narrative. The specific decision
frameworks for the structured evaluation of the strength of the human and animal evidence
streams and for drawing the overall evidence integration judgment are described in Section 10.1.
This process is similar to that used by the Grading of Recommendations Assessment, Development,
and Evaluation [GRADE; Morgan etal. (20161: Guvattetal. (20111: Schunemann etal. (2011)].
which arrives at an overall integration conclusion based on consideration of each body of evidence.
As described in Section 9, the human, animal, and mechanistic evidence syntheses serve as inputs
providing a foundation for the evidence integration decisions; thus, the major conclusions from
these syntheses will be summarized in the evidence profile table (see Table 10-1 for a preliminary
template version) supporting the evidence integration narrative. The evidence profile tables on
20Due to the expected rarity of scenarios where there is "sufficient evidence to judge that a hazard is unlikely" (see
description in Table 10-3 and Section 10.2) and to improve readability, this judgment is not specified in some
instances.
Inference Across Evidence Streams
•Information on the human relevance of
the animal and mechanistic evidence
•Coherence across bodies of evidence or
with related effects
•Other (e.g., read-across; susceptibility)
Evidence Integration
Summary Judgment
Overall judgments across
evidence for each
potential human health
effect, including evidence
basis rationale
This document is a draft for review purposes only and does not constitute Agency policy,
10-2' DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
1 each potential human health effect evaluated will summarize the judgments and evidence basis for
2 each step of the structured evidence integration process. Separate sections are included for
3 summarizing the human and animal evidence, for the inference drawn across evidence streams, and
4 for the overall evidence integration judgment The table presents the key information from the
5 different bodies of evidence that informs each decision.
This document is a draft for review purposes only and does not constitute Agency policy.
10-3 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 10-1. Evidence profile table template
Evidence integration summary judgment (for each health effect or outcome grouping)
Describe judgment regarding whether there is sufficient (or insufficient) evidence to identify a potential human health hazard, providing the primary
interpretations from each evidence stream as well as a summary of the models and range of dose levels in the studies upon which the judgment is primarily
reliant.
Summary of human, animal, and mechanistic evidence
Studies, outcomes,
and confidence
Factors that increase
strength or certainty
Factors that
decrease strength
or certainty
Key findings and
interpretation
Evidence stream summary
Inferences across
evidence streams
Evidence from studies of exposed humans (may be separated by exposure route or other study design characteristic3)
Mav be separate
rows bv outcome
• References
• Study confidence
• Study design
description (if
informative)
• Consistency
• Dose-response gradient
• Coherence of observed
effects
• Effect size
• Mechanistic evidence
providing plausibility
• Medium or high
confidence studies'5
• Unexplained
inconsistency
• Imprecision
• Low confidence
studies'5
• Evidence
demonstrating
implausibility
• Description of the
primary results
and interpretation
• If sensitivity issues
were identified,
describe the
impact on
reliability of the
reported findings
Summary of the strength of the evidence
from human studies based on the factors
at left, including the primary evidence
basis and considering:
• Results across human epidemiological
and controlled exposure studies
• Interpretations regarding any human
mechanistic evidence informing
biological plausibility (e.g., precursor
events linked to adverse outcomes)
• Human
relevance of
findings in
animals
• Cross-stream
coherence
• Summary of
potential
susceptible
populations or
lifestages
• Other
inferences:
o MOA analysis
inferences
o Relevant
information
from other
sources
Evidence from in vivo animal studies (may be separated by exposure route or other study design characteristic3)
Mav be separate
rows by outcome
• References
• Study confidence
• Consistency and/or
replication
• Dose-response gradient
• Coherence of observed
effects
• Effect size
• Unexplained
inconsistency
• Imprecision
• Low confidence
studies'5
• Description of the
primary results
and interpretation
• If sensitivity issues
were identified,
describe the
impact on
Summary of the strength of the evidence
for an effect in animals based on the
factors at left, including the primary
evidence basis and considering:
• Results across animal toxicological
studies
This document is a draft for review purposes only and does not constitute Agency policy.
10-4 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
• Study design
description (if
informative)
• Mechanistic evidence
providing plausibility
• Medium or high
confidence studies'5
• Evidence
demonstrating
implausibility
reliability of the
reported findings
• Interpretations regarding any animal
mechanistic evidence informing
biological plausibility (e.g., precursor
events linked to adverse outcomes)
(e.g., read-
across)
Biological events or
pathways (or other
information)
Species or model systems
Key findings, limitations, and
interpretation (for each row below)
Evidence stream summary
Mechanistic evidence or supplemental information (may be separated by, for example, exposure route3 or focused topic of analysis)
Separate rows bv
biological events or
other feature of the
approach used for
analvsis
• Generally, studies
are not listed, but
will cite the
synthesis section
• Does not have to
be chemical-
specific (e.g., read-
across)
Mav be multiple rows
(e.g., bv measurement
method)
Typically includes:
• Evidence base(s)
(e.g., number of studies;
suite of HTS assays; new
approach methods
[NAMs])
• Species (may include
lifestage- or sex-specific
description, if important
to interpretation)
• System (in vivo; in vitro; in
silico)
• Range of exposure levels
and durations tested
Interpretation: Summary of expert
interpretation for the body of evidence
and supporting rationale
Key findings: Summary of findings across
the body of evidence (may focus on or
emphasize highly informative designs or
findings)
Limitations: summary of key sources of
uncertainty or limitations of the study
designs tested (e.g., regarding the
biological event or pathway being
examined)
Overall summary of expert interpretation
across the assessed set of biological
events, potential mechanisms of toxicity,
or other analysis approach (e.g., AOP).
• Includes the primary evidence
supporting the interpretation(s)
• Describes and substantiates the extent
to which the evidence influences
inferences across evidence streams
• Characterizes the limitations of the
evaluation and highlights existing data
gaps
• May have overlap with factors
summarized for other streams
aln addition to exposure route, the summaries of each evidence stream may include multiple rows (e.g., by study confidence, population, or species, if they informed the analysis
of results heterogeneity or other features of the evidence). When data within an evidence stream are lacking or otherwise not informative to the evidence integration
decisions, the summary subrows for that evidence stream may be abbreviated to more easily present this information.
bStudy confidence, based on evaluation of risk of bias and study sensitivity (see Section 6), and information on susceptibility will be considered when evaluating each of the
other factors that increase or decrease strength (e.g., consistency). Notably, lack of findings in studies deemed insensitive neither increases nor decreases strength.
This document is a draft for review purposes only and does not constitute Agency policy.
10-5 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
10.1. EVALUATING THE STRENGTH OF THE HUMAN AND ANIMAL
EVIDENCE STREAMS
Before drawing overall evidence integration judgments about whether exposure to one of
these five PFAS has the potential to cause certain health effect(s) in humans given relevant
exposure circumstances, the strength of evidence for the available human and animal evidence will
be evaluated and summarized. For each assessed health effect or health effect grouping (see
Section 5 for examples of the endpoints that will be considered within each health effect category),
the relevant mechanistic evidence in exposed humans and animals (or in their cells, relevant new
approach methods [NAMs] or in silico models), which will be synthesized based on the approaches
and considerations in Section 9.2, will be integrated with the evidence from the available studies of
phenotypic effects in humans and animals. The considerations previously outlined in Table 9-1 (the
different features of the evidence considered and summarized during evidence synthesis) will be
evaluated by the specific PFAS assessment teams within the context of how they affect judgments of
the strength of evidence (see Table 10-2), which will directly inform the overall evidence
integration judgment (see Section 10.2). The evaluation of the strength of the human or animal
health effects evidence will preferably occur at the most specific health outcome level possible
(e.g., an analysis at the level of decreased pulmonary function is generally preferable to an analysis
of respiratory system effects), if there is an adequate set of studies for analyses at this level and
considering the interrelatedness of the available outcomes. If studies on a target system are sparse
or varied, or if the evidence strength relies largely on the interpretation of coherence across related
outcomes, then the analyses may need to be conducted at a broader health effect level. The factors
judged to increase or decrease the strength of the evidence will be summarized in tabular format
using the evidence profile table template in Table 10-1 to transparently convey expert judgments
made throughout the evidence synthesis and integration processes. The evidence profile table
allows for consistent documentation of the supporting rationale for each decision.
This document is a draft for review purposes only and does not constitute Agency policy.
10-6 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 10-2. Considerations that inform evaluations of the strength of the human and animal evidence
Consideration
Increased evidence strength
(of the human or animal evidence)
Decreased evidence strength
(of the human or animal evidence)
The structured categories and criteria in Table 10-3 (see Section 10.2) will guide the application of strength-of-evidence judgments for an outcome or health
effect. Evidence synthesis scenarios that do not warrant an increase or decrease in evidence strength for a given consideration will be considered "neutral"
and are not described in this table (and, in general, will not be captured in the assessment-specific evidence profile tables).
Risk of bias; sensitivity
(across studies)
• An evidence base of high or medium confidence
studies increases strength.
• An evidence base of mostly low confidence studies decreases
strength. An exception to this is an evidence base of studies in
which the primary issues resulting in low confidence are related to
insensitivity. This may increase evidence strength in cases where
an association is identified because the expected impact of study
insensitivity is towards the null.
• Decisions to increase strength for other considerations in this table
should generally not be made if there are serious concerns for risk
of bias.
Consistency
• Similarity of findings for a given outcome
(e.g., of a similar magnitude, direction) across
independent studies or experiments increases
strength, particularly when consistency is
observed across populations
(e.g., geographical location) or exposure
scenarios in human studies, and across
laboratories, populations (e.g., species), or
exposure scenarios (e.g., duration; route;
timing) in animal studies.
• Unexplained inconsistency N.e., conflicting evidence; see U.S. EPA
(2005a)l decreases strength. Generally, strength should not be
decreased if discrepant findings can be reasonably explained by
study confidence conclusions; variation in population or species,
sex, or lifestage; exposure patterns (e.g., intermittent or
continuous); exposure levels (low or high); or exposure duration.
This document is a draft for review purposes only and does not constitute Agency policy.
10-7 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Consideration
Increased evidence strength
(of the human or animal evidence)
Decreased evidence strength
(of the human or animal evidence)
Strength (effect magnitude)
and precision
• Evidence of a large magnitude effect
(considered either within or across studies) can
increase strength. Effects of a concerning rarity
or severity can also increase strength, even if
they are of a small magnitude.
• Precise results from individual studies or across
the set of studies increases strength, noting that
biological significance is prioritized over
statistical significance.
• Strength may be decreased if effect sizes that are small in
magnitude are concluded not to be biologically significant, or if
there are only a few studies with imprecise results.
Biological
gradient/dose-response
• Evidence of dose-response increases strength.
Dose-response may be demonstrated across
studies or within studies and it can be dose- or
duration-dependent. It also may not be a
monotonic dose-response (monotonicity should
not necessarily be expected, e.g., different
outcomes may be expected at low vs. high
doses because of activation of different
mechanistic pathways or induction of systemic
toxicity at very high doses).
• Decreases in a response after cessation of
exposure (e.g., symptoms of current asthma)
also may increase strength by increasing
certainty in a relationship between exposure
and outcome (this is most applicable to
epidemiology studies because of their
observational nature).
• A lack of dose-response when expected based on biological
understanding and having a wide range of doses/exposures
evaluated in the evidence base can decrease strength.
• In experimental studies, strength may be decreased when effects
resolve under certain experimental conditions (e.g., rapid
reversibility after removal of exposure). However, many reversible
effects are of high concern. Deciding between these situations is
informed by factors such as the toxicokinetics of the chemical and
the conditions of exposure [see U.S. EPA (1998)1, endpoint
severity, judgments regarding the potential for delayed or
secondary effects, as well as the exposure context focus of the
assessment (e.g., addressing intermittent or short-term
exposures).
• In rare cases, and typically only in toxicological studies, the
magnitude of effects at a given exposure level might decrease with
longer exposures (e.g., due to tolerance or acclimation). Like the
discussion of reversibility above, a decision about whether this
decreases evidence strength depends on the exposure context
focus of the assessment and other factors.
• If the data are not adequate to evaluate a dose-response pattern,
then strength is neither increased nor decreased.
This document is a draft for review purposes only and does not constitute Agency policy.
10-8 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Consideration
Increased evidence strength
(of the human or animal evidence)
Decreased evidence strength
(of the human or animal evidence)
Coherence
• Biologically related findings within an organ
system, or across populations (e.g., sex)
increase strength, particularly when a
temporal- or dose-dependent progression of
related effects is observed within or across
studies, or when related findings of increasing
severity are observed with increasing exposure.
• An observed lack of expected coherent changes
(e.g., well-established biological relationships) will typically
decrease evidence strength. However, the biological relationships
between the endpoints being compared and the sensitivity and
specificity of the measures used need to be carefully examined.
The decision to decrease evidence strength depends on the
availability of evidence across multiple related endpoints for which
changes would be anticipated, and it considers factors (e.g., dose
and duration of exposure, strength of expected relationship) across
the studies of related changes.
Mechanistic evidence
related to biological
plausibility
• Mechanistic evidence of precursors or health
effect biomarkers in well-conducted studies of
exposed humans or animals, in appropriately
exposed human or animal cells, or other
relevant human, animal, or in silico models
(including new approach methods [NAMs])
increases strength, particularly when this
evidence is observed in the same
cohort/population exhibiting the phenotypic
health outcome.
• Evidence of changes in biological pathways or
support for a proposed MOA in appropriate
models also increases strength, particularly
when support is provided for rate-limiting or
key events or across multiple components of
the pathway or MOA.
• Mechanistic understanding is not a prerequisite for drawing a
conclusion that a chemical causes a given health effect; thus, an
absence of knowledge should not be used a basis for decreasing
strength (NTP, 2015; NRC, 2014).
• Mechanistic evidence in well-conducted studies (see examples of
evidence types at left) that demonstrates that the health effect(s)
are unlikely to occur, or only likely to occur under certain scenarios
(e.g., above certain exposure levels), can decrease evidence
strength. A decision to decrease strength depends on an
evaluation of the strength of the mechanistic evidence for and
against biological plausibility, as well as the strength of the health
effect-specific findings (e.g., stronger health effect data require
more certainty in mechanistic evidence opposing plausibility).
This document is a draft for review purposes only and does not constitute Agency policy.
10-9 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
1 For human and animal evidence, the analyses of each consideration in Table 10-2 will be
2 used to qualitatively summarize the strength-of-evidence for the separate evidence streams in the
3 evidence integration narrative. Table 10-3 provides the criteria that will guide how to draw the
4 judgment for each health effect, and the terms that will be used to summarize those evidence
5 integration judgments.
10.2. OVERALL EVIDENCE INTEGRATION JUDGMENTS
6 Evidence integration combines decisions regarding the strength of the animal and human
7 evidence with considerations regarding mechanistic information on the human relevance of the
8 animal evidence, relevance of the mechanistic evidence to humans (especially in cases where
9 animal evidence is lacking), coherence across bodies of evidence, and information on susceptible
10 populations and lifestages, based on the considerations and analyses outlined in Section 9.2. This
11 evidence integration decision process will culminate in an evidence integration narrative that
12 summarizes the judgments regarding the evidence for each potential health effect (i.e., each
13 noncancer health effect and specific type of cancer, or broader grouping of related outcomes). For
14 each health effect, this narrative will include:
15 • A descriptive summary of the primary judgments about the evidence informing the
16 potential for health effects in exposed humans, based on the following analyses:
17 o evaluations of the strength of the available human and animal evidence (see
18 Section 10.1);
19 o consideration of the coherence of findings (i.e., the extent to which the evidence for
20 health effects and relevant mechanistic changes are similar) across human and animal
21 studies;
22 o other information on the human relevance of findings in animals (see Section 9.2); and
23 o conclusions drawn based on the predefined mechanistic analyses (see
24 Sections 9.2.1-9.2.3), as well as those based on analyses identified during stepwise
25 consideration of the health effect-specific evidence during draft development (see
26 Section 9.2.4).
27 • A summary of key evidence supporting these judgments, highlighting the evidence that was
28 the primary driver of these judgments and any notable issues (e.g., data quality; coherence
29 of the results), and a narrative expression of confidence (a summary of strengths and
30 remaining uncertainties) for these judgments.
31 • Information on the general conditions of expression of these health effects (e.g., exposure
32 routes and levels in the studies that were the primary drivers of these judgments), noting
33 that these conditions will be clarified during dose-response analysis (see Section 11).
This document is a draft for review purposes only and does not constitute Agency policy.
10-10 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
1
2
3
• Indications of potentially susceptible populations or lifestages (i.e., an integrated summary
of the available evidence on potential susceptible populations and lifestages drawn across
the syntheses of the human, animal, and mechanistic evidence).21
4
5
• A summary of key assumptions used in the analysis, which are generally based on EPA
guidelines and which are largely captured in this protocol.
6
7
8
9
10
11
12
• Strengths and limitations of the evidence integration judgments, including key uncertainties
and data gaps, as well as the limitations of the systematic review. As noted in Section 4.2.2,
for one or more of these five PFAS assessments, characterization of the uncertainties in the
animal evidence is expected to include a discussion of the reliance on short-term oral
exposure studies in rats. Similarly, the characterization of uncertainty in the human
evidence is expected to include a discussion of potential confounding by PFAS other than
the PFAS of interest.
13
In short, the evidence integration narrative will present a qualitative summary of the
14 strength of each evidence stream and an overall judgment across all relevant evidence, with
15 exposure context provided. For each health effect or specific cancer type of potential concern, the
16 first sentence of the evidence integration narrative will include the summary judgment [see
17 description below for how these judgments help inform selection of a descriptor for carcinogenicity
18 fU.S. EPA. 2005al], Assessments will also include an evidence profile table (see Table 10-1) to
19 support the evidence integration narrative by providing the major decisions and supporting
20 rationale. Table 10-3 describes the categories of evidence integration judgments that will be used
21 in these PFAS assessments and provides examples of database scenarios that fit each category of
22 evidence. These summary judgments provide a succinct and clear representation of the decisions
23 from the more detailed analyses of whether (or not) the evidence strength indicates that PFAS
24 exposure has the potential to cause the human health effect(s) under the necessary conditions of
25 exposure. Consistent with EPA noncancer and cancer guidelines, a judgment that the evidence
26 supports an apparent lack of an effect of PFAS exposure on the health effect(s) will only be used
27 when the available data are considered robust for deciding that there is no basis for human hazard
28 concern; lesser levels of evidence suggesting a lack of an effect will be characterized as
29 "insufficient"
21One or more of these five PFAS assessments may include consideration of information outside of their
PFAS-specific database to address this aspect of the evidence integration narrative. These PFAS-specific data gaps
and uncertainties appear to extend beyond poorly studied health effects, and the discussion of missing
information on potential populations, sexes, or lifestages that are likely to be more susceptible to developing a
specific health effect may consider information from reviews of other PFAS.
This document is a draft for review purposes only and does not constitute Agency policy.
10-11 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 10-3. Evidence integration judgments for characterizing potential human health hazards in the evidence
integration narrative
Evidence
integration
judgment3
Evidence in studies of humans
Evidence in animal studies
Inferences across evidence
streams
Sufficient
evidence for
hazard
A judgment of sufficient evidence for hazard
animal studies, incorporating the consideratior
span a broad range of overall evidence strengt
• Strong mechanistic evidence in well-
conducted studies of exposed humans
(medium or high confidence) or human
cells (including NAMs), in the absence of
other substantive data, in which an
informed evaluation has determined that
the data are reliable for assessing toxicity
relevant to humans and the mechanistic
events have been causally linked to the
development of the health effect of
interest.0
• A single high or medium confidence study
demonstrating an effect with one or more
factors that increase evidence strength,
such as: a large magnitude or severity of
the effect, a dose-response gradient,
unique exposure or outcome scenarios
(e.g., a natural experiment), or supporting
coherent evidence, including mechanistic
evidence from exposed humans. There are
no comparable studies of similar
confidence and sensitivity providing
conflicting evidence, or if there are, the
differences can be reasonably explained
squires tha
is outlined
i and exam
l/>
o
*i_
LU
i_
(D
DO
C
o
1_
+-»
LO
V
a scenario below is met for either the eviden
under inferences across evidence streams. Th
pies are provided below, starting with the we
• Strong mechanistic evidence in well-
conducted studies of animals or animal
cells (including NAMs), in the absence of
other substantive data, in which an
informed evaluation has determined the
assays are reliable for assessing toxicity
relevant to humans and the mechanistic
events have been causally linked to the
development of the health effect.0
• A single high or medium confidence
experiment in the absence of comparable
experiment(s) of similar confidence and
sensitivity providing conflicting evidence01
[evidence that cannot be reasonably
explained, e.g., by respective study
designs or differences in animal model;
(U.S. EPA, 2005a)l.
• At least one high or medium confidence
study with supporting information
increasing the strength of the evidence.
Although the results are largely
consistent, notable uncertainties remain.
However, in scenarios when inconsistent
evidence or evidence indicating
ce in studies of humans or evidence in
e scenarios justifying this judgment
akest evidence.15
•Supplemental evidence
(e.g., structure-activity data;
chemical class information; other
NAMs) is judged to increase the
strength of limited or near-
equivocal, chemical-specific human
or animal evidence to sufficient
evidence for hazard.
•Coherent or biologically consistent
findings across evidence streams
increases the strength of limited or
near-equivocal human or animal
evidence (e.g., single or few high or
medium confidence studies with
some conflicting evidence) to
sufficient evidence for hazard.
•The strength of the evidence is
decreased because mechanistic
information (even if it does not
provide MOA understanding) raises
uncertainties regarding the human
and/or animal evidence, but overall
the evidence is still considered
strong enough to result in a
This document is a draft for review purposes only and does not constitute Agency policy.
10-12 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
[e.g., by the populations or exposure levels
studied (U.S. EPA, 2005a)l.
• Multiple studies showing generally
consistent findings, including at least one
high or medium confidence study and
supporting evidence, but with some
residual uncertainty due to potential
chance, bias, or confounding (e.g., effect
estimates of low magnitude or small effect
sizes given what is known about the
endpoint; uninterpretable patterns with
respect to exposure levels). Associations
with related endpoints, including
mechanistic evidence from exposed
humans, can address uncertainties relating
to exposure response, temporality,
coherence, and biological plausibility, and
any conflicting evidence is not from a
comparable body of higher confidence,
sensitive studies.d
• A set of high or medium confidence
independent studies reporting an
association between the exposure and the
health outcome, with reasonable
confidence that alternative explanations,
including chance, bias, and confounding,
can be ruled out across studies. The set of
studies is primarily consistent, with
reasonable explanations when results
differ; and an exposure response gradient
is demonstrated. Supporting evidence,
such as associations with biologically
related endpoints in human studies
(coherence) or large estimates of risk or
severity of the response, may help rule out
alternative explanations. Similarly,
l/>
o
*i_
(D
DO
C
o
nonspecific effects exist, it is not judged
to reduce or discount the level of concern
regarding the positive findings, or it is not
from a comparable body of higher
confidence, sensitive studies 4. The
additional support provided includes
either consistent effects across
laboratories or species; coherent effects
across multiple related endpoints; an
unusual magnitude of effect, rarity, age
at onset, or severity; a strong
dose-response relationship; or consistent
observations across exposure scenarios
(e.g., route, timing, duration), sexes, or
animal strains. Mechanistic evidence in
animals may serve to provide this
support or otherwise address residual
uncertainties.
A set of high or medium confidence
experiments with consistent findings of
adverse or toxicologically significant
effects across multiple laboratories,
exposure routes, experimental designs
(e.g., a subchronic study and a
two-generation study), or species; and
the experiments reasonably rule out the
potential for nonspecific effects to have
caused the effects of interest. Any
inconsistent evidence (evidence that
cannot be reasonably explained based on
study design or differences in animal
model) is from a set of experiments of
lower confidence or sensitivity. To
reasonably rule out alternative
explanations, multiple additional factors
in the set of experiments exist, such as:
coherent effects across biologically
judgment of sufficient evidence for
hazard.
•The strength of the evidence is
decreased because findings across
evidence streams are conflicting
(U.S. EPA, 2005a) or biologically
inconsistent, but a judgment of
sufficient evidence for hazard is
supported by review of the
adversity and human relevance
(prioritizing findings relevant to
human toxicity) of the effects.
•The strength of the evidence is
neither increased or decreased due
to a lack of experimental
information on the human
relevance of the animal evidence
or mechanistic understanding
(mechanistic evidence may exist,
but it is inconclusive); in these
cases, the animal data are judged
not to conflict with current
biological understanding and thus
are assumed to be relevant, while
findings in humans and animals are
presumed to be real unless proven
otherwise.
• For the strongest animal evidence,
there is mechanistic understanding
that the findings are expected to
occur and progress in humans.
Most notably, an MOA interpreted
with reasonable certainty would
rule out alternative explanations.
This document is a draft for revi
^evhs/piir
purposes only and does not constitute Agency policy.
10-13 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
mechanistic evidence from exposed
humans may serve to address uncertainties
related to exposure-response, temporality,
coherence, and biological plausibility
(i.e., providing evidence consistent with an
explanation for how exposure could cause
the health effect based on current
biological knowledge).
related endpoints; an unusual magnitude
of effect, rarity, age at onset, or severity;
a strong dose-response relationship; or
consistent observations across animal
lifestages, sexes, or strains. Similarly,
mechanistic evidence (e.g., precursor
events linked to adverse outcomes) in
animal models may exist to address
uncertainties in the evidence base.
• For the strongest evidence, there is
adequate testing of potentially
susceptible lifestages and
populations, based on the effect(s)
of interest and chemical knowledge
(e.g., toxicokinetics).
Insufficient
evidence
A judgment of insufficient evidence requires that a scenario below is met for both the evidence in studies of humans and evidence in animal
studies, incorporating the considerations outlined under inferences across evidence streams.
• A body of evidence, including scenarios
with one or more high or medium
confidence studies reporting an association
between exposure and the health
outcome, where either (1) conflicting
evidence exists in studies of similar
confidence and sensitivityd e, (2) a single
study without a factor that increases
evidence strength as described in sufficient
evidence for hazard), or (3) considerable
methodological uncertainties remain
across the body of evidence (typically
related to exposure or outcome
ascertainment, including temporality), and
there is no supporting coherent evidence
that increases the overall evidence
strength.
• A set of only low confidence studies.
• No studies of exposed humans or well-
conducted studies of human cells.
• A body of evidence, including scenarios with one or
more high or medium confidence experiments
reporting effects but without supporting coherent
evidence that increases the overall evidence
strength, where conflicting evidence exists from a
set of sensitive experiments of similar or higher
confidence (can include mechanistic evidence).d e
• A set of only low confidence experiments.
• No animal studies or well-conducted studies of
animal cells.
• The available endpoints are not informative to the
hazard question under evaluation.
• A set of largely null studies that does not meet the
criteria for sufficient evidence to judge that a
hazard is unlikely.
•The evidence in animal studies
meets a scenario for sufficient
evidence for hazard, but strong
experimental evidence (e.g., an
MOA interpreted with reasonable
certainty) indicates the findings in
animals are unlikely to be relevant
to humans.
•The evidence meets a scenario for
sufficient evidence to judge that a
hazard is unlikely, but there is
inadequate testing of susceptible
populations and lifestages or of
data conflict across evidence
streams.
•The evidence in animal studies
meets a scenario for sufficient
evidence to judge that a hazard is
unlikely, but the database lacks
experimental support that the
models are relevant to humans for
the effect of interest.
This document is a draft for review purposes only and does not constitute Agency policy.
10-14 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
• A set of largely null studies that does not
meet a scenario for sufficient evidence to
judge that a hazard is unlikely.
Sufficient
evidence to
judge that a
hazard is
unlikely1
A iudgment of sufficient evidence to iudae that a hazard is unlikelv reauires that a scenario below is met for either the evidence in studies of
humans or evidence in animal studies, incorporating the considerations outlined under inferences across evidence streams.
• Several high confidence studies showing
null results (for example, an odds ratio of
1.0), ruling out alternative explanations
including chance, bias, and confounding
with reasonable confidence. Each of the
studies should have used an optimal
outcome and exposure assessment and
adequate sample size (specifically for
higher exposure groups and for susceptible
populations). The overall set should
include the full range of levels of exposures
that human beings are known to
encounter, and an evaluation of an
exposure-response gradient.
• A set of high confidence experiments examining a
reasonable spectrum of endpoints relevant to a type
of toxicity that demonstrate a lack of biologically
significant effects across multiple species, both
sexes (if applicable), and a broad range of exposure
levels. The data are compelling in that the
experiments have examined the range of scenarios
across which health effects in animals could be
observed, and an alternative explanation
(e.g., inadequately controlled features of the
studies' experimental designs; inadequate sample
sizes) for the observed lack of effects is not
available. The experiments were designed to
specifically test for the effects of interest, including
suitable exposure timing and duration, postexposure
latency, and endpoint evaluation procedures.
•There is adequate testing of
susceptible populations and
lifestages.
•When the evidence in animal
studies meets a scenario for this
judgment, there is experimental
support that the models are
relevant to humans for the effect
of interest and no conflicting
human evidence exists.
•When the evidence in studies of
humans meets a scenario for this
judgment and conflicting animal
data exist, mechanistic information
indicates that the animal data are
unlikely to be relevant to humans.
•When multiple high confidence
animal experiments and studies in
humans indicate lack of an effect,
but the evidence does not meet a
scenario for sufficient evidence to
judge that a hazard is unlikely,
strong mechanistic evidence in
models relevant to humans
supports lack of an effect such that
the totality of evidence supports
this judgment.
This document is a draft for review purposes only and does not constitute Agency policy.
10-15 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
aThese categories are based on those indicated for use in hazard characterization from the existing EPA guidelines for noncancer health effects [i.e., U.S. EPA
(1988); U.S. EPA (1991b); U.S. EPA (1996b)1. As described in those guidance documents, the judgments depend heavily on expert judgment (note: as applied
herein, the process of "evidence integration" is synonymous with "weight of evidence"). The evidence integration judgment for each assessed health effect
will be included as part of an evidence integration narrative, the specific documentation of the various expert decisions and evidence-based (or default)
rationales are summarized in an evidence profile table, and the judgement will be contextualized based on the primary supporting evidence (experimental
model or observed population, and exposure levels tested or estimated). Importantly, as discussed in Section 10.1, these judgments may be based on
analyses of grouped outcomes at different levels of granularity (e.g., motor activity vs. neurobehavioral effects vs. nervous system effects) depending on the
specifics of the health effect evidence base. Health effects characterized as having sufficient evidence for hazard will be evaluated for use in dose-response
assessment.
bQualitative descriptions of differences in the strength of the evidence across different health effects judged as having sufficient evidence for hazard are useful
for other assessment decisions, including prioritizing outcomes in quantitative analyses and characterizing assessment uncertainties. Thus, for all evidence
scenarios, but particularly for those in the lower end of this range, it is important to characterize the uncertainties in the evidence base within the evidence
integration narrative and convey the evidence strength to subsequent steps, including toxicity values developed based on those effects. Existing guidance
defines the minimum evidence necessary to judge that a health hazard could exist as one adverse endpoint from one well-conducted study (U.S. EPA, 1998);
this has been expanded in this table to better incorporate mechanistic evidence, including new approach methods (NAMs) data.
Scientific understanding of toxicity mechanisms and of the human implications of new toxicity testing methods (e.g., from high-throughput screening, from
short-term in vivo testing of alternative species, or from new in vitro and in silico testing and other NAMs) will continue to increase. Thus, the sufficiency of
mechanistic evidence alone for identifying potential human health hazards is expected to increase as the science evolves. The decision to identify a potential
human hazard based on these data is an expert judgment dependent on the state of the science at the time of review.
Scenarios with unexplained heterogeneity across sets of studies with similar confidence and sensitivity can be considered either sufficient evidence for hazard
or insufficient evidence, depending on the expert judgment of the overall weight of evidence. Specifically, this judgment considers the level of support (or
lack thereof) provided by evaluations of the magnitude or severity of the effects, coherence of related findings (including mechanistic evidence), dose-
response, and biological plausibility, as well as the comparability of the supporting and conflicting evidence (e.g., the specific endpoints tested, or the
methods used to test them; the specific sources of bias or insensitivity in the respective sets of studies). The evidence-specific factors supporting either
evidence integration judgment will be clearly articulated in the evidence integration narrative.
eWhen the database includes at least one well-conducted study and a hazard characterization judgment of insufficient evidence is drawn, quantitative analyses
may still be useful for some purposes (e.g., providing a sense of the magnitude and uncertainty of estimates for health effects of potential concern, ranking
potential hazards, or setting research priorities), but not for others [see related discussions in U.S. EPA (2005a)l. It is critical to transparently convey the
extreme uncertainty in any such estimates.
the criteria for this category are intentionally more stringent than those justifying a conclusion of sufficient evidence for hazard, consistent with the "difficulty
of proving a negative" [as discussed in U.S. EPA (1988); U.S. EPA (1991b); U.S. EPA (1996b)1.
This document is a draft for review purposes only and does not constitute Agency policy.
10-16 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Evaluations of carcinogenicity will be consistent with EPA's Cancer Guidelines (U.S. EPA.
2005a)- One of EPA's standardized cancer descriptors will be used as a shorthand characterization
of the evidence integration narrative, describing the overall potential for human carcinogenicity
across all potential cancer types. These are (1) carcinogenic to humans, (2) likely to be carcinogenic
to humans, (3) suggestive evidence of carcinogenic potential, (4) inadequate information to assess
carcinogenic potential, or (5) not likely to be carcinogenic to humans. More than one descriptor can
be used when a chemical's effects differ by exposure level or route (U.S. EPA. 2005a): if the
database supports such an analysis, these decisions will be clarified based on a more thorough
review of the mechanistic evidence or more detailed dose-response analysis (see Section 11). In
some cases, mutagenicity will also be evaluated (e.g., when there is evidence of carcinogenicity),
because it influences the approach to dose-response assessment and subsequent application of
adjustment factors for exposures early in life fU.S. EPA. 2005a. b).
An appropriate cancer descriptor will be selected as described in EPA Cancer Guidelines
(U.S. EPA. 2005a). For each cancer subtype, an evidence integration narrative and summary
judgment will be provided, as described above. The cancer descriptor will consider the
interrelatedness of cancer types potentially due to PFAS exposure, consistency across the human
and animal evidence for any cancer type [noting that site concordance is not required fU.S. EPA.
2005a)], and the uncertainties associated with each assessment-specific conclusion. In general,
however, if a systematic review of more than one cancer type was conducted, then the overall
judgment and discussion of evidence strength in the evidence integration narrative for the cancer
type(s) with the strongest evidence for hazard will be used to inform selection of the cancer
descriptor, with each assessment providing a transparent description of the decision rationale. The
cancer descriptor and evidence integration narrative for potential carcinogenicity, including
application of the MOA framework, will consider the conditions of carcinogenicity, including
exposure (e.g., route; level) and susceptibility (e.g., genetics; lifestage), as the data allow fFarland.
2005: U.S. EPA. 2005a. hi.
10.3. HAZARD CONSIDERATIONS FOR DOSE-RESPONSE
This section outlines how these assessments will consider and describe the transition from
hazard identification to dose-response analysis, highlighting (1) information that will inform the
selection of outcomes or broader health effect categories for which toxicity values will be derived,
(2) whether toxicity values can be derived to protect specific populations or lifestages, (3) how
dose-response modeling will be informed by toxicokinetic information, and (4) information aiding
the identification of biologically based benchmark response (BMR) levels. The pool of outcomes
and study-specific endpoints will be discussed to identify which categories of effects and study
designs are considered the strongest and most appropriate for quantitative assessment of a given
health effect Health effects that were analyzed in human studies in relation to exposure levels
within or closer to the range of exposures encountered in the environment will be considered
This document is a draft for review purposes only and does not constitute Agency policy.
10-17 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
particularly informative, as are animal studies testing a broad range of exposure levels and
including levels in the lower dose region. When there are multiple endpoints for an organ/system,
considerations for characterizing the overall impact on this organ/system will be discussed,
including the severity and longevity of the effects. For example, if there are multiple
histopathological alterations relevant to liver function changes, liver necrosis may be selected as
the most representative endpointto consider for dose-response analysis. This section may review
or clarify which endpoints or combination of endpoints in each organ/system characterize the
overall effect for dose-response analysis. For cancer types, consideration will be given to the
overall risk of multiple types of tumors. Multiple tumor types (if applicable) will be discussed and a
rationale given for any grouping.
Biological considerations that are important for dose-response analysis (e.g., that could help
with selection of a BMR) will be discussed. The impact of route of exposure on toxicity to different
organs/systems will be examined, if appropriate. The existence and validity of PBPK models or
toxicokinetic information that may allow the estimation of internal dose for route-to-route
extrapolation will be presented (see additional discussion and decision points in Section 11.2). In
addition, mechanistic evidence analyses that will influence the dose-response analyses will be
highlighted (see Section 9.2 for specific considerations), for example, evidence related to
susceptibility or potential shape of the dose-response curve.
This section will also describe the evidence regarding populations and lifestages that
appear to be susceptible to the health hazards identified and factors that are likely to increase the
risk of developing (or exacerbating) these health effects, depending on the available evidence. This
section will include this discussion even if there are no specific data on the effects of exposure to
the PFAS of interest in the potentially susceptible population. Table 9-2 in Section 9 outlines some
of the specific factors that will be considered for discussion and summaries of the evidence with
respect to patterns across studies pertinent to consistency, coherence, and the magnitude and
direction of effect measures. At a minimum, consideration will be given to discussion of
information relevant to infants and children, pregnant women, and women of childbearing age.
The section will consider options for using susceptible population data in the dose-response
analysis. In particular, an attempt will be made to highlight where it might be possible to develop
separate risk estimates for a specific population or lifestage or to determine whether evidence is
available to select a data-derived uncertainty factor.
This document is a draft for review purposes only and does not constitute Agency policy.
10-18 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
11. DOSE-RESPONSE ASSESSMENT: SELECTING
STUDIES AND QUANTITATIVE ANALYSIS
The previous sections of this protocol describe how systematic review principles will be
applied to evaluate studies (for potential bias and sensitivity) and reach evidence integration
conclusions on potential human health effects associated with exposure to the PFAS of interest
Selection of specific data sets for dose-response assessment and performance of the dose-response
assessment will be conducted after hazard identification is complete and involves database- and
chemical-specific biological judgments that build from decisions made at earlier stages of
assessment development Several Environmental Protection Agency (EPA) guidance and support
documents describe data requirements and other considerations for dose-response modeling,
especially EPA's Benchmark Dose Technical Guidance (U.S. EPA. 20121. EPA's Review of the Reference
Dose and Reference Concentration Processes (U.S. EPA. 2002b). Guidelines for Carcinogen Risk
Assessment fU.S. EPA. 2005al. and Supplemental Guidance for Assessing Susceptibility from Early-Life
Exposure to Carcinogens (U.S. EPA. 2005b). This section of the protocol provides an overview of
considerations for conducting the dose-response assessment, particularly statistical considerations
specific to dose-response analysis that support quantitative risk assessment Importantly, these
considerations do not supersede existing EPA guidance.
Dose-response assessments will be performed for both noncancer and cancer health
hazards, and for both oral and inhalation routes of exposure following exposure22 to the chemical of
interest, if supported by existing data. For noncancer hazards, an oral reference dose (RfD) and/or
an inhalation reference concentration (RfC) will be derived when possible. An RfD or an RfC is an
estimate, with uncertainty spanning perhaps an order of magnitude, of an exposure to the human
population (including susceptible subgroups) that is likely to be without an appreciable risk of
deleterious health effects over a lifetime (U.S. EPA. 2002b). In addition to an RfD and/or RfC, when
feasible and if the available data are appropriate for doing so, the assessments will derive a less-
than-lifetime toxicity value (a "subchronic" reference value) for noncancer hazards. Likewise, part
of the process for deriving an oral or inhalation reference value will include developing separate
values specific to each hazard ("organ- or system-specific" reference values). Both less-than-
lifetime and hazard-specific values may be useful to EPA risk assessors within specific decision
22For most health outcomes (e.g., this would not apply to outcomes related to developmental toxicity),
dose-response assessments will be preferably based on studies of chronic exposure. However, analyses will also
be conducted for shorter durations, particularly when the evidence base for a PFAS indicates potential risks
associated with shorter exposures to the chemical (U.S. EPA, 2002b).
This document is a draft for review purposes only and does not constitute Agency policy.
11-1 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
contexts. Reference values are not predictive risk values; that is, they provide no information about
risks at higher or lower exposure levels.
Reference values may also be derived for cancer effects [e.g., in a case where a nonlinear
MOA is concluded that indicates a key precursor event necessary for carcinogenicity does not occur
below a specific exposure level fU.S. EPA. 2005al: see Section 11.2.3], When low-dose linear
extrapolation for cancer effects is supported, particularly for chemicals with direct mutagenic
activity or those for which the data indicate a linear component below the point of departure (POD),
an oral slope factor (OSF) and/or an inhalation unit risk (IUR) will be used to estimate human
cancer risks. In general, this will also be the case when no data are available to inform the
evaluation of linearity. An OSF is a plausible upper bound lifetime cancer risk from chronic
ingestion of a chemical per unit of mass consumed per unit body weight per day (mg/kg-day). An
IUR is a plausible upper bound lifetime cancer risk from chronic inhalation of a chemical per unit of
air concentration (expressed as ppm or [ig/m3). In contrast with reference values (RfVs), an OSF or
IUR can be used in conjunction with exposure information to predict cancer risk at a given dose.
As discussed in Section 2 "Scoping and Problem Formulation Summary" for these PFAS
assessments, the Integrated Risk Information System (IRIS) Program will conduct the assessments
with a goal of developing any toxicity values that are reasonably supported by the available data,
based on judgments of the evidence drawn during hazard identification and the suitability of
studies for dose-response analysis.
The derivation of reference values and cancer risk estimates will depend on the nature of
the health hazard conclusions drawn during evidence integration (see Section 10.2). Specifically,
EPA generally conducts dose-response assessments and derives cancer values for chemicals that
are classified as carcinogenic or likely to be carcinogenic to humans. When there is suggestive
evidence of carcinogenicity to humans, EPA generally would not conduct a dose-response
assessment or derive a cancer value except when the evidence includes a well-conducted study and
quantitative analyses may be useful for some purposes, for example, providing a sense of the
magnitude and uncertainty of potential risks, ranking potential hazards, or setting research
priorities (U.S. EPA. 2005a). A parallel approach will be taken for potential noncancer health effects
in these assessments. Specifically, for noncancer outcomes these assessments will generally
include dose-response assessments when the evidence integration judgments indicate there is
sufficient evidence for hazard, with preference given to health effects with stronger evidence
scenarios within that category (see Section 10.2), and quantitative analyses generally will not be
attempted for insufficient evidence.
11.1. SELECTING STUDIES FOR DOSE-RESPONSE ASSESSMENT
The dose-response assessment will begin with a review of the important health effects
highlighted during hazard identification, particularly among the studies of highest quality and that
exemplify the study attributes summarized in Table 11-1. This review will also consider whether
This document is a draft for review purposes only and does not constitute Agency policy.
11-2 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
there are opportunities for quantitative evidence integration, although it is considered unlikely that
the data available to do so will be available for these assessments based on the preliminary
literature inventory. Examples of quantitative integration, from simplest to more complex, include
(1) the combination of results for an outcome across sex (within a study); (2) characterizing overall
toxicity, as in combining effects that constitute a syndrome, or occur on a continuum
(e.g., precursors and overt toxicity, benign tumors that progress to malignant tumors); and
(3) meta-analysis or metaregression of all studies addressing a category of important health effects.
Some studies that were used qualitatively for hazard identification may or may not be
considered useful quantitatively for dose-response analysis in these five assessments because of
factors like the lack of quantitative measures of exposure or of variability measures for response
data. If the needed information cannot be located (e.g., by contacting study authors and making any
information publicly available), a semiquantitative analysis (e.g., via no-observed-adverse-effect
level [NOAEL]/lowest-observed-adverse-effect level [LOAEL]) will be considered. Studies of low
sensitivity may be considered less useful if they failed to detect an effect or reported points of
departure with wide confidence limits, but such studies will still be considered for inclusion in a
meta-analysis.
Among the studies that support the evidence integration conclusions, those that are most
useful for dose-response analysis will generally have at least one exposure level in the region of the
dose-response curve near the benchmark response (the response level to be used for deriving
toxicity values) to minimize low-dose extrapolation. Such studies will also have more exposure
levels and larger sample sizes overall (U.S. EPA. 2012). These attributes support a more complete
characterization of the shape of the exposure-response curve and decrease the uncertainty in the
associated exposure-response metric (e.g., IUR or RfC) by reducing statistical uncertainty in the
POD and minimizing the need for low-dose extrapolation. In addition to these more general
considerations, specific issues that may be considered for their potential to affect the feasibility of
dose-response modeling for individual data sets are described in more detail in the Benchmark Dose
Technical Guidance (U.S. EPA. 2012).
This document is a draft for review purposes only and does not constitute Agency policy.
11-3 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table 11-1. Attributes used to evaluate studies for deriving toxicity values
Study attributes
Considerations
Human studies
Animal studies
Rationale for choice of
species
Human data are preferred over animal data to eliminate
interspecies extrapolation uncertainties (e.g., in
toxicodynamics, relevance of specific health outcomes
to humans, and in toxicokinetics, especially given
minimal human TK data).
Animal studies provide supporting evidence when adequate human
studies are available and are considered principal studies when
adequate human studies are not available. For some hazards, studies
of animal species known to respond similarly to humans would be
preferred over studies of other species.
Relevance of
exposure
paradigm
Exposure
route
Studies involving human environmental exposures (oral,
inhalation).
Studies by a route of administration relevant to human
environmental exposure are preferred. A validated toxicokinetic
model can also be used to extrapolate across exposure routes.
Exposure
durations
When developing a chronic toxicity value, chronic- or subchronic-duration studies are preferred over studies of acute exposure.
Exceptions exist, such as when a susceptible population or lifestage is more sensitive in a certain time window
(e.g., developmental exposure).
Exposure
levels
Exposures near the range of typical environmental human exposures are preferred. Studies with a broad exposure range and
multiple exposure levels are preferred to the extent that they can provide information about the shape of the
exoosure-resDonse relationship [see the EPA Benchmark Dose Technical Guidance (U.S. EPA, 2012)1 and facilitate extrapolation
to more relevant (generally lower) exposures.
Subject selection
Studies that provide risk estimates in the most susceptible groups are preferred.
Controls for possible
confounding3
Studies with a design (e.g., matching procedures, blocking) or analysis (e.g., covariates or other procedures for statistical
adjustment) that adequately address the relevant sources of potential critical confounding for a given outcome are preferred.
Measurement of exposure
Studies that can reliably distinguish between levels of
exposure in a time window considered most relevant
for a causal effect with respect to the development of
the outcome are preferred. Exposure assessment
methods that reduce measurement error and methods
that provide measurement of exposure at the level of
the individual are preferred. Measurements of
exposure should not be influenced by knowledge of
health outcome status.
Studies providing actual measurements of exposure (e.g., analytical
inhalation concentrations vs. target concentrations) are preferred.
Relevant internal dose measures may facilitate extrapolation to
humans, as would availability of a suitable animal PBPK model in
conjunction with an animal study reported in terms of administered
exposure.
This document is a draft for review purposes only and does not constitute Agency policy.
11-4 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Study attributes
Considerations
Human studies
Animal studies
Measurement of health
outcome(s)
Studies that can reliably distinguish the presence or absence (or degree of severity) of the outcome are preferred. Outcome
ascertainment methods using generally accepted, standardized approaches) are preferred.
Studies with individual data are preferred in general. Examples include characterizing experimental variability more realistically
and characterizing overall incidence of individuals affected by related outcomes (e.g., phthalate syndrome).
Study size and design
Preference is given to studies using designs reasonably expected to have power to detect responses of suitable magnitude.15
This does not mean that studies with substantial responses but low power would be ignored, but that they should be
interpreted in the context of a confidence interval or variance for the response. Studies that address changes in the number at
risk (through decreased survival, loss to follow-up) are preferred.
aAn exposure or other variable that is associated with both exposure and outcome but is not an intermediary between the two.
bPower is an attribute of the design and population parameters, based on a concept of repeatedly sampling a population; it cannot be inferred post hoc using
data from one experiment (Hoenig and Heisev, 2001).
This document is a draft for review purposes only and does not constitute Agency policy.
11-5 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
11.2. CONDUCTING DOSE-RESPONSE ASSESSMENTS
Consistent with EPA practice, these PFAS assessments will apply a two-step approach for
dose-response assessment that distinguishes analysis of the dose-response data in the range of
observation from any inferences about responses at lower environmentally relevant exposure
levels fIJ.S. EPA. 2012. 2005a1:
1) Within the observed dose range, the preferred approach will be to use dose-response
modeling to incorporate as much of the data set as possible into the analysis. This modeling
to derive a POD should include an exposure level ideally near the lower end of the range of
observation, without significant extrapolation to lower exposure levels (see Section 11.2.1
for more details).
2) As derivation of cancer risk estimates and reference values nearly always involves
extrapolation to exposures lower than the POD; the approaches to be applied in these
assessments are described in more detail in Section 11.2.2 and Section 11.2.3, respectively.
When sufficient and appropriate human and laboratory animal data are available for the
same outcome, human data will be generally preferred for the dose-response assessment because
its use eliminates the need to perform interspecies extrapolations.
For reference values, these assessments will typically derive a candidate value from each
suitable data set, whether in humans or animals (see Section 11.1). Evaluation of these candidate
values grouped within a given organ/system will yield a single organ/system-specific value for each
organ/system under consideration. Next, evaluation of these organ/system-specific values will
result in the selection of a single overall reference value to cover all health outcomes across all
organs/systems. While this overall reference value represents the focus of these dose-response
assessments, the organ/system-specific values can be useful for subsequent cumulative risk
assessments that consider the combined effect of multiple PFAS (or other agents) acting at a
common organ/system.
For cancer, if there are multiple tumor sites that can be quantified individually, the final
cancer risk estimate(s) will typically address overall cancer risk, to the extent the data allow.
For both cancer and noncancer toxicity values, uncertainties in these estimates will be
transparently characterized and discussed.
11.2.1. Dose-Response Analysis in the Range of Observation
Toxicodynamic ("biologically based") modeling is generally preferred when there are
sufficient, reliable data to ascertain the MOA and quantitatively support model parameters that
represent rates and other quantities associated with the key precursor events of the MOA. Such
data, however, do not appear to be available for these five PFAS.
This document is a draft for review purposes only and does not constitute Agency policy.
11-6 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Because a toxicodynamic model will not be available for dose-response assessment,
empirical modeling will be used to fit the data (on the apical outcome or a key precursor event) in
the range of observation. For this purpose, EPA has developed a standard set of models
fhttp: //www.epa.gov/ncea/bmds] that can be applied to typical data sets, including those that are
nonlinear. In situations where there are alternative models with significant biological support
(e.g., when the available evidence provides strong support for a threshold MOA), the decision
maker will be informed by the presentation of these alternatives in the assessment(s) along with
the models' strengths and uncertainties. EPA has developed guidance on modeling dose-response
data, assessing model fit, selecting suitable models, and reporting modeling results [see the EPA
Benchmark Dose Technical Guidance fU.S. EPA. 20121], Additional judgment or alternative analyses
will be used if the procedure fails to yield reliable results; for example, if the fit is poor, modeling
may be restricted to the lower doses, especially if there is competing toxicity at higher doses.
For each modeled response, a POD from the observed data will be estimated to mark the
beginning of extrapolation to lower doses. The POD is an estimated dose (expressed in
human-equivalent terms) near the lower end of the observed range without significant
extrapolation to lower doses. The POD will be used as the starting point for subsequent
extrapolations and analyses. For linear extrapolation of cancer risk, the POD will be used to
calculate an OSF or IUR, and for nonlinear extrapolation, the POD will be used in the calculation of
an RfD or RfC.
The response level at which the POD is calculated will be guided by the severity of the
endpoint. If linear extrapolation is used, standard values near the low end of the observable range
will generally be used (for example, 10% extra risk for cancer bioassay data, 1% for epidemiologic
data, lower for rare cancers). For nonlinear approaches, both statistical and biological significance
will be considered. For dichotomous data, a response level of 10% extra risk will generally be used
for minimally adverse effects, 5% or lower for more severe effects. For continuous data, a response
level ideally will be based on an established definition of biologic significance. In the absence of
such definition, one control standard deviation from the control mean will generally be used for
minimally adverse effects, and one-half standard deviation for more severe effects. The point of
departure will be the 95% lower bound on the dose associated with the selected response level.
EPA has developed standard approaches to determine the relevant dose for use in
dose-response modeling in the absence of appropriate toxicokinetic modeling. These standard
approaches can also aide comparison across exposure patterns and species in the absence of a
validated pharmacokinetic (PK) model (see below). The general approaches and considerations to
be used to extrapolate PFAS dosimetry from (1) shorter to longer durations within studies, (2) from
animals to humans, and (3) across routes of exposure are outlined below:
• Intermittent study exposures will be standardized to a daily average over the duration of
exposure. For chronic effects, daily exposures will be averaged over the life span.
This document is a draft for review purposes only and does not constitute Agency policy.
11-7 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Exposures during a critical period, however, will not be averaged over a longer duration
(U.S. EPA. 2005a. 1991a). Note that this will typically be done after modeling because the
conversion is linear.
• The preferred approach for dosimetry extrapolation from animals to humans will be
through PBPK or PK modeling. This approach will be considered first for PFAS and
lifestages with existing PBPK models or where an existing model structure can be readily
adapted (see Section 6.4 on PBPK modeling).
• Because there are PK data for the PFAS being evaluated in at least one relevant animal
species (rats or monkeys) and in humans (see Section 2.4.1), a data-informed extrapolation
approach will also be considered for any PFAS that either lacks a PK model or has a model
determined to be of inadequate quality. Briefly, the ratio of the elimination half-life in
animal to that in the human, ti/2A- ti/2H, or the ratio of the clearance in the human to the
animal, CLh:CLa, will be considered for use in converting an oral dose-rate in animals
(mg/kg-day) to a human equivalent dose rate (i.e., the human exposure that should yield
the same blood concentration as the animal exposure from which it is being extrapolated).
Note that clearance and half-life are inversely related. The assessments will consider these
metrics as follows:
° Of these two metrics, ti/2 and CL, the half-life is a less complete measure of elimination
but one that can be evaluated from more minimal PK data. A half-life can be estimated
by observing the decline in an individual's blood concentration of a compound after an
exposure has ended. In this way, the total exposure or body burden of the chemical
does not have to be known. However, PFAS elimination may go through several phases
during which distinct half-lives apply, and the blood concentration that occurs during
ongoing exposure may effectively reflect an average among these. The specific
approaches and considerations for estimating PFAS half-life are outlined in
Section 9.2.1.
° The clearance, on the other hand, is a measure of average elimination but requires more
data to estimate. One must also quantify a companion variable, the volume of
distribution (Vd), which in turn requires a measure of total exposure or dose in
well-conducted studies. Although more rigorous assessment-specific evaluations will
be performed, based on a preliminary review of studies in the literature inventory, the
data necessary for the reliable quantification of Vd in humans are expected to be lacking.
Specifically, accurate estimates of dose do not seem to be available in human exposure
studies, and the identified animal studies demonstrate considerable interstudy
variability in Vd estimates.
° Using an estimate of human CL based on Vd measured in rats, for example, yields the
same rat-human conversion factor as using the ratio of half-lives, but in a less
transparent way: that is, because the underlying assumption that human Vd equals rat
Vd is not clear, a reviewer might assume that use of CL indicates a more complete
evaluation of human PK. However, it is expected that Vd in humans will be similar to
that in non-human primates based on the more similar physiology and biochemical
parameters, so it is reasonable to use a primate Vd to estimate human CL. Hence human
CL values will only be used if they are based on an independent direct measurement of
CL in humans (e.g., using subject-paired measurements of a PFAS in human serum and
This document is a draft for review purposes only and does not constitute Agency policy.
11-8 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
urine), direct estimation of Vd in humans using controlled exposures to PFAS, or using
T0.5 determined from human data together with a Vd from non-human primates.
° As indicated in Section 9.2.1, if the PK data clearly indicate a dose-dependent half-life,
the ti/2 at lower doses, most relevant to human health extrapolation, will be used.
• Based on the selection of half-life as the preferred metric and a POD identified from a
health-effects study in animals, the human equivalent dose (HED) will be calculated as:
° Here, the [s] in the subscript indicates that the value may be sex specific. When there
are sex-specific values (significant differences between males and females) in both
animals and humans, the CL or ti/2 values for females would be used to extrapolate
health effects in female animals to women, the CL or ti/2 values for males used to
extrapolate male animal health effects to men. If human data are available to estimate
separate half-lives for women and men, the CL or ti/2 for women will likewise be used to
estimate HED values in women and the CL or ti/2 in men used to estimate HEDs in men.
If human data are not sufficient to provide distinct values for men and women, a
common ti/2 for humans will be used.
• In the absence of PK data/half-lives, oral doses will be scaled allometrically using BW3/4 as
the equivalent dose metric across species. Allometric scaling pertains to equivalence across
species, not across lifestages, and will not be used to scale doses from adult humans or
mature animals to infants or children (U.S. EPA. 2011a. 2005a. 1994). Using this approach,
the HED will be calculated as:
° If half-life data are available in humans and rats but not mice, for example, then
allometric scaling may be used to estimate the mouse half-life from the rat value
(i.e., using two species closer in BW). This extrapolated mouse half-life can then be used
with the measured human half-life to estimate an HED as described above, making the
greatest possible use of available TK data.
• Inhalation exposures will be scaled using dosimetry models that apply species-specific
physiologic and anatomic factors and consider whether the effect occurs at the site of first
contact or after systemic circulation (U.S. EPA. 2012.1994).
• It can be informative to convert doses across exposure routes. If this is done, the
assessment will describe the underlying data, algorithms, and assumptions (U.S. EPA.
HED = (CLh[s]/CLA[s]) x POD or
HED = (tl/2A[s]/ tl/2H[s]) x POD
(11-1)
HED = (BWh/BWa)0-25 x POD (mg/kg-day)
(11-2)
This document is a draft for review purposes only and does not constitute Agency policy.
11-9 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
1 2005a). Depending on the availability of sufficient data (see Section 9.2) and/or suitable
2 models (see Section 6.4), route-to-route extrapolations in these assessments will be
3 accomplished by using the inhalation exposure rates for PFAS-containing particles
4 predicted using the MPPD model (see Section 9.2) as an ingestion rate in the PK analysis
5 (PBPK/PK model or ADME adjustment), under the assumption that once absorbed into
6 general circulation, the toxic effect is only a function of the body burden or blood
7 concentration.
8 • In the absence of study-specific data on, for example, intake rates or body weight, the EPA
9 has developed recommended values for use in dose-response analysis fU.S. EPA. 19881.
11.2.2. Extrapolation: Slope Factors and Unit Risk
10 An OSF or IUR will be used to estimate human cancer risks when low-dose linear
11 extrapolation for cancer effects is supported by the PFAS-specific evidence, particularly for PFAS
12 with direct mutagenic activity or those for which the data indicate a linear component below the
13 POD. Low-dose linear extrapolation will also be used as a default when the data are insufficient to
14 establish the MOA (U.S. EPA. 2005a). If the PFAS-specific data are sufficient to ascertain that one or
15 more modes of action are consistent with low-dose nonlinearity, or to support their biological
16 plausibility, low-dose extrapolation will use the reference-value approach when suitable data are
17 available fU.S. EPA. 2005al: see Section 11.2.3 below.
18 Differences in susceptibility will be considered for use in deriving multiple slope factors or
19 unit risks, with separate estimates for susceptible populations and lifestages fU.S. EPA. 2005al If
20 appropriate chemical-specific data on susceptibility from early life exposures are available, then
21 these data will be used to develop cancer slope factors or unit risks that specifically address any
22 potential for differential potency in early lifestages (Farland. 2005: U.S. EPA. 2005a). If such data
23 are not available, the evidence integration analyses supports a mutagenic MOA for carcinogenicity,
24 and the extrapolation approach is linear, the dose-response assessment will indicate to decision
25 makers that in the development of risk estimates, the default age-dependent adjustment factors
26 should be used with the cancer slope factor or unit risk and age-specific estimates of exposure fU.S.
27 EPA. 2005a. b). In this scenario, the final cancer risk value presented in the assessment(s) will
28 reflect this adjustment, with the requisite calculations provided.
29 The derivation of an OSF and IUR for any of these five PFAS conducted as part of the current
30 assessments will be performed in a manner consistent with EPA guidance.
11.2.3. Extrapolation: Reference Values
31 Reference value derivation is EPA's most frequently used type of nonlinear extrapolation
32 method, and it will be used in these PFAS assessments for noncancer effects. This approach will
33 also be used for cancer effects if the available data are sufficient to ascertain the MOA and conclude
34 that it is not linear at low doses (see Section 11.2.2). In this case, reference values for each relevant
This document is a draft for review purposes only and does not constitute Agency policy.
11-10 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
route of exposure will be developed following EPA's established practices (U.S. EPA. 2005a): in
general, the reference value will be based not on tumor incidence, but on a key precursor event in
the MOA that is necessary for tumor formation. The derivation of an RfD or RfC (if feasible)
conducted as part of the assessments for perfluorobutanoic acid (PFBA), PFHxA,
perfluorohexanesulfonate (PFHxS), perfluorononanoic acid (PFNA), and perfluorodecanoic acid
(PFDA) will be performed in a manner consistent with EPA guidance.
For each data set selected, reference values will be estimated by applying relevant
adjustments (i.e., uncertainty factors [UFs]) to the PODs to account for the conditions of the
reference value definition. These factors account for human variation, extrapolation from animals
to humans, extrapolation to chronic exposure duration, extrapolation to a minimal level of risk (if
not observed in the data set), and database deficiencies, as outlined below. Increasingly, data-based
adjustments fU.S. EPA. 2014c! probabilistic approaches fChiu etal.. 2018: Chiu and Slob. 20151.
and Bayesian methods for characterizing population variability (NAS. 2014) are becoming feasible
and may be distinguished from the UF considerations outlined below, if such data exist for these
five PFAS. These assessments will discuss the scientific bases (or lack thereof) for each selected UF,
including any data-based adjustments based on the following considerations:
• Animal-to-human extrapolation: If animal results are used to make inferences about
humans, the reference value derivation will incorporate the potential for cross-species
differences, which may arise from differences in toxicokinetics or toxicodynamics. The POD
will be standardized to equivalent human terms or be based on toxicokinetic or dosimetry
modeling that may range from detailed chemical-specific to default approaches (U.S. EPA.
2014c. 2011a). and a factor of 10°5 (rounded to 3) will be applied to account for the
remaining uncertainty involving toxicokinetic and toxicodynamic differences. Data-derived
adjustments for toxicodynamic differences across species may include qualitative decisions
regarding key science issues (e.g., if, during evaluation of PPARa-dependency, it is
concluded that humans are not more sensitive than rodents).
• Human variation: The assessments will account for variation in susceptibility across the
human population and the possibility that the available data may not represent individuals
who are most susceptible to the effect. If appropriate data or models for the effect or for
characterizing the internal dose are available, the potential for data-based adjustments for
toxicodynamics or toxicokinetics will also be considered (U.S. EPA. 2014c. 2002b).23'24
When sufficient data are available, an intraspecies UF either less than or greater than
10-fold may be justified fU.S. EPA. 2002bl. A reduction in this UF will be considered if the
23Examples of adjusting the toxicokinetic portion of interhuman variability include the Integrated Risk Information
System (IRIS) boron assessment's use of non-chemical-specific kinetic data [e.g., glomerular filtration rate in
pregnant humans as a surrogate for boron clearance (U.S. EPA, 2004)1 and the IRIS trichloroethylene assessment's
use of population variability in trichloroethylene metabolism, via a PBPK model, to estimate the lower 1st
percentile of the dose metric distribution for each POD (U.S. EPA, 2011b).
24Note that when a PBPK model is available for relating human internal dose to environmental exposure, relevant
portions of this UF may be more usefully applied prior to animal-to-human extrapolation, depending on the
correspondence of any nonlinearities (e.g., saturation levels) between species.
This document is a draft for review purposes only and does not constitute Agency policy.
11-11 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
POD is derived from or adjusted specifically for susceptible individuals, but not for a general
population that includes both susceptible and nonsusceptible individuals (U.S. EPA. 2002b.
1998.1996b. 1994.1991al. In general, when the use of such data or modeling is not
supported, a UF with a default value of 10 will be used.
• LOAEL to NOAEL: When a POD is based on a LOAEL, the assessment will include an
adjustment to an exposure level where such effects are not expected. This can be a matter
of great uncertainty if no evidence is available at lower exposures. A factor of 3 or 10 will
generally be applied to extrapolate to a lower exposure expected to be without appreciable
effects. A factor other than 10 may also be considered, depending on the magnitude and
nature of the response and the shape of the dose-response curve (U.S. EPA. 2002b. 1998.
1996b. 1994.1991 at
• Subchronic-to-chronic exposure: When using studies of less-than-chronic exposure to make
inferences about chronic/lifetime exposure, the assessment will consider whether lifetime
exposure could reasonably be interpreted to result in effects at lower levels of exposure,
including consideration of the specific health outcome(s) in question. A factor of up to 10
will be considered, depending on the duration of the studies and the nature of the response
riJ.S. EPA. 2002b. 1998.19941.
• Database deficiencies: In addition to the adjustments above, if database deficiencies raise
concern that further studies might identify a more sensitive effect, organ system, or
lifestage, the assessment will apply a database UF fU.S. EPA. 2002b. 1998.1996b. 1994.
1991a). The size of the factor will depend on the nature of the database deficiency. For
example, EPA typically follows the recommendation that a factor of 10 be applied if both a
prenatal toxicity study and a two-generation reproduction study are missing and a factor of
10°5 (i.e., 3) if either one or the other is missing fU.S. EPA. 2002bl. As noted in Section 2.4.5,
the evaluation of database completeness for these five PFAS will also consider existing
knowledge gained through reviewing other, potentially similar, PFAS to identify data gaps.
For example, there is the potential for exposure to PFAS to cause developmental effects
(based on reviews of perfluorooctanoic acid [PFOA] and perfluorooctane sulfonate [PFOS])
and there appears to be a lack of such studies for PFHxA. Thus, consideration of the
potential for PFHxA exposure to cause developmental effects might review knowledge
gained through the assessment of the other C6 PFAS, PFHxS, or the other short-chain
perfluoroalkyl carboxylic acid, perfluorobutanoic acid (PFBA). In such cases, an
interpretation of the relatedness between the PFAS of interest and the PFAS used for
comparison will inform selection of the uncertainty factor.
The POD for a particular RfV will be divided by the product of these factors. As discussed in
the technical document reviewing the RfD/RfC process (U.S. EPA. 2002b). any composite factor that
exceeds 3,000 represents excessive uncertainty; thus, values with >3,000 UFc will not be used to
derive RfVs. An RfD/RfC may be based on the POD for a single endpoint within a study, or on a
collection of related PODs within or across studies, if such biological relationships are substantiated
by the evidence. Confidence in any derived toxicity value(s) will be described based on three
factors: confidence in the study(ies) used in the derivation of the toxicity value; confidence in the
evidence base for the hazard(s) underlying the toxicity value, and confidence in the quantitative
This document is a draft for review purposes only and does not constitute Agency policy.
11-12 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
derivation of the toxicity value. The confidence description(s) will be separate from consideration
of the composite uncertainty factor applied to derive the toxicity value.
This document is a draft for review purposes only and does not constitute Agency policy.
11-13 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
12. PROTOCOL HISTORY
1 March 2020: Comments on this protocol were provided in the public docket (see Docket
2 ID: EPA-HO-QRD-2019-0275 for detailed comments) during a 45-day public comment period from
3 November 8th, 2019 to December 23rd, 2019. Approximately 107 individual comments were
4 provided across a range of stakeholder groups. We thank the public commenters for their
5 constructive and informative reviews. The comments have been addressed in this update to the
6 protocol and were considered during development of the five draft IRIS PFAS assessments. A
7 summary of the comment topics and updates to the protocol is provided in Table 12-1 and below.
Table 12-1. Topic areas of public comments on the protocol and how
comments were addressed in this update (generally ordered based on
descending number of comments on the topic areas)
Topic areas raised by commenter(s)
Protocol updates and responses
Toxicokinetics
(see Protocol Sections 2.4, 6.4, 9.2, and 11)
Summary of comments on the use of ADME data
in study selection and parameter choice for
dose-response analysis: the protocol should add
specificity on the approaches.
Added clarifying text to Section 11.1 on considering
uncertainty in toxicokinetics across species when selecting
studies for dose-response analysis, as well as how ADME
information can influence selection of Cmax versus AUC as
measure of risk. Also, as described below, the literature
screening process for ADME data has been emphasized.
Summary of comments on the use of clearance
versus half-life data: the protocol should add
specificity on the approaches.
Added clarifying text to Sections 9.2.1 and 11.2.1 to explain
that while clearance is preferred, these data will only be used
if it is not based on assuming the same volume of distribution
(Vd) in the human as in a rodent. Otherwise this hides the
assumption and gives the same result as the half-life ratio.
Notably, the protocol maintains that analysis of the PK data
will assume a single half-life estimate (half-life is assumed not
to vary with dose); the text clarifies that use of a comparison
across data sets will allow for an evaluation of this
assumption. Specifically, if no evidence of nonlinearity is
demonstrated, then it is assumed to be irrelevant at
experimental levels. If evidence of nonlinearity is
demonstrated, the analyses will focus on lower dose PK data.
Summary of comments on the use of allometric
scaling versus data-specific adjustments:
allometric scaling should be used for short-chain
PFAS.
Added text to Section 11.2.1 that if half-life data are available
in humans and rats but not mice, for example, then allometric
scaling may be used to estimate the mouse half-life from the
rat value (i.e., using two species closer in BW). This
extrapolated mouse half-life can then be used with the
This document is a draft for review purposes only and does not constitute Agency policy.
12-14 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
measured human half-life to estimate an HED as described in
the protocol (prior to this update). EPA guidance indicates a
preference for data-specific adjustments over BW3/4 and says
the latter should only be used in the absence of data. Hence
the protocol was not revised in this regard, and the protocol
still indicates use of a data-specific adjustment (e.g., a half-life
ratio) when adequate data for doing so are available,
irrespective of the PFAS chain length or magnitude of the half-
life.
Human relevance and adversity of rodent responses (hepatic effects)
(see Protocol Sections 2.4.2 and 9.2.2)
Summary of comments on interpreting PPARa
responses in hepatic (and possibly other) health
outcomes: comments varied, including both
support for and against the human relevance of
hepatic effects.
Expanded the discussion (and references) in Sections 2.4.2 and
9.2.2 on current information and uncertainties regarding the
relative sensitivity of humans and animal models to PFAS-
related PPARa inductions, incorporating information on the
extent to which differences may relate to differing
toxicokinetics as well as intrinsic variations in biological
sensitivities. Examples and discussion have also been added
regarding how prior evaluations of the role of PPARa will be
considered. The updated protocol notes difficulties in
applying read-across approaches related to the human
relevance of rodent responses due to PFAS structural
differences.
Summary of comments on interpreting rodent
noncancer liver endpoint adversity: caution
should be exercised in applying criteria
developed in the context of cancer—such as the
Hall et al. criteria, to noncancer endpoints.
Expanded the discussion on the applicability of the Hall
criteria to noncancer hepatic endpoints and the development
of lifetime toxicity values has been added to Section 9.2.3.
This new text highlights the need to consider additional
factors when applying the Hall et al. criteria during assessment
development.
Study Evaluation
(see Protocol Sections 5 and 6).
Summary of comments on overall study
confidence: comments varied, including that the
approach should not use the confidence ratings
in a manner similar to "scoring" and that the
approach should be more quantitative in the
method for arriving at confidence.
Revised the protocol (see Section 6) to clarify that domain
judgments and the specific limitations identified in the study
are made available with the assessment and are carried
forward to inform the synthesis; however, the overall
approach was not changed. The overall study confidence
ratings are not used as "scores" and are not provided without
context. Text was also added to clarify that the overall study
confidence is reached using expert judgment on the impact of
the identified deficiencies for each specific study, and that
there are no predefined weights for combining the domains.
Summary of comments on the use of critically
deficient domain ratings: the approach should
not exclude a study due to one critical deficiency.
Revised text in Section 6 to clarify deficient and critically
deficient domain ratings; however, the overall approach was
not changed. Specifically, the critically deficient category is
used rarely and only in situations where the limitation is
severe enough to warrant excluding a study as uninformative
for the purposes of the assessment. Serious flaws that do not
warrant study exclusion will be classified as deficient.
Typically, domains rated as deficient are judged to reduce the
This document is a draft for review purposes only and does not constitute Agency policy.
12-15 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
reliability of the reported results. Studies with numerous
deficient ratings may be excluded as uninformative.
Summary of comments on prioritizing
epidemiology outcomes: the approach should
not use one evaluator—as noted in the protocol
for prioritization of some outcomes, or it should
be clarified that the review of those outcomes
performed with one evaluator were not
systematic reviews.
Amended the approach in Section 5 to include two
independent study evaluations for most outcomes. There is
still a tiering system in place to prioritize outcomes, but the
protocol now more clearly distinguishes the methodological
rigor of "rapid reviews" from systematic reviews. Classification
into the rapid review tier is now based on serious concern for
reverse causality, determined a priori.
Summary of comments on study evaluation
criteria: the approach should consider conflict of
interest.
The approach in Section 6 was not modified to consider
conflict of interest. The evaluations of risk of bias and
sensitivity by subject matter experts are designed to
encompass the primary aspects of methodological design that
could engender concern, irrespective of the sponsoring entity.
Summary of comments on applying the exposure
domain for study evaluation: caution should be
exercised when assessing PFAS, in general, due
to potential issues relating to analytical
chemistry or physiochemical properties.
This was determined to not be a significant issue of concern
for the specific PFAS being assessed. Additional support
relating to this decision has been added to the protocol (see
Section 6.3) based on review of data in the EPA Chemistry
Dashboard.
Use of Mechanistic Information
(see Protocol Sections 9.2,11.2.1, and 11.2.2).
Summary of comments on the use of mechanistic
evidence and mode-of-action (MOA)
understanding to inform dose-response
(provided in the context of cancer): expand the
discussion; note: as very few studies relevant to
evaluating cancer are available for these five
PFAS, these comments were viewed and
addressed as more broadly applicable to any
outcome.
Added clarifying text to Section 9.2.4 to emphasize that such
data are considered for potential use quantitatively as well as
qualitatively, and to Section 11.2.1 to indicate their potential
use when a nonlinear (threshold) dose-response relationship
is supported by the evidence.
Summary of comments on the use of mechanistic
evidence and mode-of-action understanding to
inform evaluation of key science topics: expand
the discussion.
Expanded the discussions on the explicit consideration of
mechanistic evidence for critical scientific topics, such as the
human relevance and adversity of hepatic changes, as well as
toxicokinetic interpretations, in Sections 2.4 and 9.2.
Literature Identification
(see Protocol Sections 3.2 and 4)
Summary of comments on PECO "outcome"
criteria: ADME studies should not be
"supplemental" for these PFAS.
Added clarifying text to Section 4.2 directing the reader to the
separate literature identification and review process of ADME
studies in Section 9.2.1.
Summary of comments on PECO "population"
criteria: nonmammalian models are not included
in the PECO, but they can be important.
Added a caveat to Section 4.2 clarifying the situationally
increased utility of some nonmammalian models for certain
outcomes.
Other Comments
Summary comments on the use of the 10 key
characteristics of carcinogens: the key
characteristics should not be used to conduct the
The protocol (see Section 9.2) presents the key characteristics
as an example approach to organize mechanistic evidence in a
literature inventory, and not as a means to conduct the
This document is a draft for review purposes only and does not constitute Agency policy.
12-16 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
analysis of mechanistic evidence relevant to
cancer.
evaluation or develop MOA judgments regarding those data;
thus, this text was not revised.
Summary of comments on addressing data gaps
and uncertainties using information from other,
better-studied PFAS: comments varied, ranging
from requests for an increased emphasis on use
of information from more well-studied PFAS, up
to including formal [reassessments of PFOA and
PFOS, to requests not to use data from other
PFAS to influence interpretations regarding the
five PFAS being assessed.
The scope of the assessments was not expanded to include
PFOA and PFOS (see additional discussion in bullets below).
However, as described in the protocol (see Sections 2.4.5, 9.2
and 10), the breadth of information on PFOA and PFOS (and
other well-studied PFAS) is still considered potentially
informative to these assessments. Examples of how these
data are expected to be used include helping to identify key
areas of potential concern (e.g., health outcomes associated
with PFOA or PFOS exposure) that have not been examined
for the PFAS of interest, and information (e.g., MOA
information; studies on health outcomes of interest) on other
PFAS pertinent to interpreting the reliability, adversity, and/or
human relevance of effects observed in studies of the PFAS of
interest. In addition, text has been added to Section 2.4 to
emphasize that caution will be taken in drawing judgments for
a PFAS of interest based on evidence on other PFAS due to
cross-PFAS differences in toxicokinetics and toxicodynamics.
Summary of comments on evidence integration:
the "weight of evidence" approach should not
only be applied when integrating across evidence
streams, but also in the analyses of individual
streams.
The commenters misinterpreted the approach to evidence
integration, as it lays out an evaluation of evidence strength
within each stream of evidence as well as across evidence
streams. For example: "Building from the separate syntheses
of the human and animal evidence (see Section 9.1), the
strength of the evidence from the available human and animal
health effect studies will be summarized in parallel, but
separately, using a structured evaluation of an adapted set of
considerations first introduced by Sir Bradford Hill."
Additional text emphasizing this point has been added to the
introductory materials in both Sections 9 and 10.
Summary of corrections and editorial
suggestions: numerous comments, on a variety
of topic areas.
Made edits to improve the accuracy of the protocol text;
however, not all editorial suggestions were incorporated.
Most notably, a number of suggestions related to Section 2.1
(Background) were not addressed. Because this section is
meant to provide brief, contextual summaries and not
comprehensive systematic reviews of the current information
(as suggested by some of the comments), a clarifying
introduction has been added regarding the purpose of this
section. However, emphasis was added regarding certain
aspects of exposure (e.g., the discussion of drinking water
exposure) and potential susceptible population and lifestages
(e.g., breast-fed infants) to improve context (see
Sections 2.1.5 and 2.1.6). The section was also retitled to,
"Summary of Background Information," and the summaries of
existing toxicity values for these PFAS were moved to Section
2.1.8 from Section 2.2.1) due to similar observations regarding
comprehension and updating.
1 In addition to the changes in Table 12-1 made in direct response to public comments,
2 several other edits to the protocol were made during this update, specifically:
This document is a draft for review purposes only and does not constitute Agency policy.
12-17 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
• A technical edit was conducted. Grammatical errors were corrected, and some editorial
changes were made that did not affecting the scientific approaches.
• Reference arrays providing a snapshot of existing toxicity values for these PFAS were
moved from Protocol Section 2.3 (Problem Formulation) to Protocol Section 2.1
(Background) and edited to remove values other than RfDs or RfD-like values (e.g., drinking
water standards); these were also removed from Addendum A. These values were removed
to reduce the potential for inappropriate comparisons.25
• Additional clarifications on considering CBI data for inclusion (specific to the timeliness of
their availability in a publicly available form) are provided in Protocol Section 4.1.
• Text in Protocol Section 9.2.2 revised to reflect internal EPA discussions to minimize the use
of single words or phrases to summarize weight-of-evidence-related judgments within the
narratives for individual evidence streams (e.g., mechanistic evidence).
• Text in Protocol Sections 9.2 and 11.2.1 related to incorporating the available PBPK models
(or their data) and toxicokinetics information was updated based on preliminary
conclusions from a more robust evaluation of the available PBPK models and toxicokinetics
data. In particular, the identified PBPK models do not appear adequate for direct
application in these assessments, so alternatives are now emphasized.
• An updated version of the evidence profile table template was inserted in Section 10. The
basic information included is unchanged; however, the presentation has been altered to
increase transparency. In addition, text indicating that sufficient evidence for hazard can
be judged based on a single epidemiology study without other supporting information has
been edited to indicate that such evidence scenarios will be judged as insufficient evidence.
• Text in Protocol Section 11 was updated to indicate that, when adequate data are available
and it is appropriate to do so, less-than-lifetime ("subchronic") and hazard-specific ("organ-
or system-specific") toxicity values will be derived in addition to an RfD and/or RfC. The
derivation of these values is methodologically consistent with approaches already described
in the protocol. During the protocol's public comment period, EPA partners indicated that
such less-than-lifetime values were potentially useful for certain decision contexts.
• Clarification was added to Section 11.2.3 that any toxicity values derived will be
accompanied by a description of confidence.
• Protocol "Appendix" materials were renamed "Addendum" materials for clarity, as this
protocol will be included as an Appendix to each of the five IRIS PFAS assessments.
25 IRIS does not derive drinking water standards or advisories. These values, which include information on human
exposure and other considerations, are the purview of EPA programs (e.g., OW) and regional risk assessors.
Although such standards or advisory levels may consider in their derivation reference values developed by IRIS,
drinking water standards or advisories are not comparable to reference values, and thus they were removed so as
not to convey an inappropriate comparison.
This document is a draft for review purposes only and does not constitute Agency policy.
12-18 DRAFT-DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
In addition to the comments outlined in Table 12-1, other topics were raised which were
outside of the scope of the protocol and thus did not warrant changes. The topic areas for these
comments and the rationale for not updating the protocol in response to these comments are
described below:
• Comments providing general support or criticism, or directing IRIS staff to other resources
(e.g., other conclusions on these PFAS, or opinions on those other conclusions), did not
result in changes to this protocol. Similarly, public comments recommending toxicity values
that should be adopted by the EPA are not addressed by this protocol. As a reminder, IRIS
assessment conclusions rely on independent evaluations of primary research studies,
unless otherwise indicated as part of EPA scoping and problem formulation decisions
(e.g., adopting a well-established conclusion).
• Comments relating to sites with potential PFAS contamination, recommendations for
conducting PFAS exposure assessments, and requests for these assessments to include
instruction on addressing coexposure to multiple PFAS are not addressed by this protocol,
because such issues are the purview of other EPA programs, EPA regions, tribes, and states.
• A number of comments were related to the scope covered by these five assessments. This
included recommendations that EPA should evaluate and regulate PFAS either as a class or
individually (this opinion varied across commenters), or that other PFAS (e.g., PFOA; PFOS)
should be (re)assessed simultaneously. In addition, some commenters wanted additional
details regarding next steps for the EPA Action Plan; plans for addressing inconsistency in
values developed by different federal agencies and states; or future EPA plans for
monitoring, regulation, and enforcement None of these comments are addressed by this
protocol. Based on EPA program and regional needs, specific chemicals or substances
(e.g., diesel exhaust) are nominated to the IRIS Program for independent, scientific
assessment of potential human health hazards and dose-response analyses. The decisions
on the PFAS for which an IRIS assessment would be useful, as well as the broader EPA plan
for addressing PFAS, are not the purview of the IRIS Program.
• Several commenters requested details on the operating procedures used within the IRIS
Program, specifically referencing the "IRIS Handbook." These comments are not addressed
by this protocol. The "IRIS Handbook" is not a public document and its development for
public release are separate from the development of these assessments.
• A few commenters were interested in PFAS assessment-specific decisions rather than the
methods and approaches for assessment development As a reminder, the PFAS-specific
literature screening decisions and updates will be available in HERO (www.hero.epa.gov):
individual study evaluation decisions will be available in HAWC (www.hawcproiect.org):
and decisions regarding studies and values used in support of assessment conclusions
(e.g., studies and values selected to represent individual PFAS half-lives in different species;
decisions regarding the human relevance of particular findings in animal studies) will be
summarized and discussed in the specific PFAS assessments. As such, these comments are
not further addressed in this protocol.
This document is a draft for review purposes only and does not constitute Agency policy.
12-19 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
REFERENCES
3M (3M Company). (2003). Oral (gavage) combined repeated dose toxicity study of T-7706 with the
reproduction/developmental toxicity screening test [TSCA Submission], (89-000811850U;
8EHQ-0903-00373). St. Paul, MN.
Abbott. BP. (2009). Review of the expression of peroxisome proliferator-activated receptors alpha
(PPAR alpha), beta (PPAR beta), and gamma (PPAR gamma) in rodent and human
development [Review], Reprod Toxicol 27: 246-257.
http://dx.doi.Org/10.1016/i.reprotox.2008.10.001
ATSDR (Agency for Toxic Substances and Disease Registry). (2018). Toxicological profile for
perfluoroalkyls. Draft for public comment Atlanta, GA: U.S. Department of Health and
Human Services, Centers for Disease Control and Prevention.
https://www.atsdr.cdc.gOv/toxprofiles/tp200.pdf
Baduel. C: Paxman. CI: Mueller. IF. (2015). Perfluoroalkyl substances in a firefighting training
ground (FTG), distribution and potential future release. J Hazard Mater 296: 46-53.
http://dx.doi.Org/10.1016/i.ihazmat2015.03.007
Barber. TL: Berger. U: Chaemfa. C: Huber. S: Tahnke. A: Temme. C: Tones. KC. (2007). Analysis of per-
and polyfluorinated alkyl substances in air samples from Northwest Europe. J Environ
Monit9: 530-541. http ://dx. doi.org/10.1039/b701417a
Beesoon. S: Genuis. ST: Benskin. TP: Martin. TW. (2012). Exceptionally high serum concentrations of
perfluorohexanesulfonate in a Canadian family are linked to home carpet treatment
applications. Environ Sci Technol 46: 12960-12967. http://dx.doi.org/10.1021/es3034654
Boone. L: Meyer. D: Cusick. P: Ennulat. D: Bolliger. AP: Everds. N: Meador. V: Elliott. G: Honor. D:
Bounous. D: Tordan. H. (2005). Selection and interpretation of clinical pathology indicators
of hepatic injury in preclinical studies [Review], Vet Clin Pathol 34: 182-188.
http://dx.doi.Org/10.llll/i.1939-165X.2005.th00041.x
Buck. RC: Franklin. 1: Berger. U: Conder. TM: Cousins. IT: de Voogt. P: Tensen. AA: Kannan. K: Maburv.
SA: van Leeuwen. SP. (2011). Perfluoroalkyl and polyfluoroalkyl substances in the
environment: terminology, classification, and origins [Review], Integr Environ Assess
Manag 7: 513-541. http://dx.doi.org/10.1002/ieam.258
Burri. L: Thoresen. GH: Berge. RK. (2010). The Role of PPARa Activation in Liver and Muscle. PPAR
Research 2010. http://dx.doi.org/10.1155/201Q/542359
Butenhoff. 1: Costa. G: Elcombe. C: Farrar. D: Hansen. K: Iwai. H: Tung. R: Kennedy. G: Lieder. P:
Olsen. G: Thomford. P. (2002). Toxicity of ammonium perfluorooctanoate in male
cynomolgus monkeys after oral dosing for 6 months. Toxicol Sci 69: 244-257.
http://dx.doi.Org/10.1093/toxsci/69.l.244
Byrne. S: Seguinot-Medina. S: Miller. P: Waghivi. V: von Hippel. FA: Buck. CL: Carpenter. DO. (2017).
Exposure to polybrominated diphenyl ethers and perfluoroalkyl substances in a remote
population of Alaska Natives. Environ Pollut 231: 387-395.
http://dx.doi.Org/10.1016/i.envpol.2017.08.020
CDC (Centers for Disease Control and Prevention). (2009). Fourth national report on human
exposure to environmental chemicals. Atlanta, GA: U.S. Department of Health and Human
Services, Centers for Disease Control and Prevention, http: //www.cdc.gov/exposurereport/
This document is a draft for review purposes only and does not constitute Agency policy.
R-l DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
CDC (Centers for Disease Control and Prevention). (2015). Fourth national report on human
exposure to environmental chemicals, updated tables, February 2015. Atlanta, GA.
https://www.cdc.gov/biomonitoring/pdf/FourthReport UpdatedTables Feb2015.pdf
CDC (Centers for Disease Control and Prevention). (2018). Fourth national report on human
exposure to environmental chemicals, updated tables, March 2018, volume one. Atlanta, GA:
U.S. Department of Health and Human Services.
https://www.cdc.gov/exposurereport/pdf/FourthReport UpdatedTables Volumel Mar20
18.pdf
CDC (Centers for Disease Control and Prevention). (2019a). Fourth National Report on Human
Exposure to Environmental Chemicals, Updated Tables, January 2019, Volume 1. Centers for
Disease Control and Prevention, National Health and Nutrition Examination Survey.
https://www.cdc.gov/exposurereport/pdf/FourthReport UpdatedTables Volumel Tan201
9-508.pdf
CDC (Centers for Disease Control and Prevention). (2019b). Fourth National Report on Human
Exposure to Environmental Chemicals, Updated Tables, January 2019, Volume 2. Atlanta,
GA: U.S. Department of Health and Human Services.
https://www.cdc.gov/exposurereport/pdf/FourthReport UpdatedTables Volume2 Tan201
9-508.pdf
Chang. S: Butenhoff. TL: Parker. GA: Coder. PS: Zitzow. TP: Krisko. RM: Bjork. TA: Wallace. KB: Seed.
TG. (2018). Reproductive and developmental toxicity of potassium
perfluorohexanesulfonate in CD-I mice. Reprod Toxicol 78: 150-168.
http://dx.doi.Org/10.1016/i.reprotox.2018.04.007
Chen. WL: Bai. FY: Chang. YC: Chen. PC: Chen. CY. (2018). Concentrations of perfluoroalkyl
substances in foods and the dietary exposure among Taiwan general population and
pregnant women. J Food Drug Anal 26: 994-1004.
http://dx.doi.Org/10.1016/j.jfda.2017.12.011
Chiu. WA: Guvton. KZ: Martin. MT: Reif. DM: Rusvn. I. (2018). Use of high-throughput in vitro
toxicity screening data in cancer hazard evaluations by IARC Monograph Working Groups.
ALTEX 35: 51-64. http://dx.doi.org/10.14573/altex.1703231
Chiu. WA: Slob. W. (2015). A unified probabilistic framework for dose-response assessment of
human health effects [Review], Environ Health Perspect 123: 1241-1254.
http://dx.doi.org/10.1289/ehp.14Q9385
Christensen. KY: Raymond. M: Blackowicz. M: Liu. Y: Thompson. BA: Anderson. HA: Turvk. M.
(2017). Perfluoroalkyl substances and fish consumption. Environ Res 154: 145-151.
http://dx.doi.Org/10.1016/i.envres.2016.12.032
Corton. TC: Anderson. SP: Stauber. A. (2000). Central role of peroxisome proliferator-activated
receptors in the actions of peroxisome proliferators [Review], Annu Rev Pharmacol Toxicol
40: 491-518. http://dx.doi.Org/10.1146/annurev.pharmtox.40.l.491
Corton. TC: Cunningham. ML: Hummer. BT: Lau. C: Meek. B: Peters. TM: Popp. TA: Rhomberg. L: Seed.
T: Klaunig. TE. (2014). Mode of action framework analysis for receptor-mediated toxicity:
The peroxisome proliferator-activated receptor alpha (PPARa) as a case study [Review],
Crit Rev Toxicol 44: 1-49. http: //dx.doi.org/10.3109/10408444.2Q13.835784
Corton. TC: Peters. TM: Klaunig. TE. (2018). The PPARa-dependent rodent liver tumor response is not
relevant to humans: addressing misconceptions [Review], Arch Toxicol 92: 83-119.
http://dx.doi.Org/10.1007/s00204-017-2094-7
Danish EPA (Danish Environmental Protection Agency). (2015). Perfluoroalkylated substances:
PFOA, PFOS and PFOSA. Evaluation of health hazards and proposal of a health based quality
criterion for drinking water, soil and ground water. (Environmental project No. 1665).
https://www2.mstdk/Udgiv/publications/2 015/04/978-87-93 283-01-5.pdf
This document is a draft for review purposes only and does not constitute Agency policy.
R-2 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Das. KP: Grey. BE: Rosen. MB: Wood. CR: Tatum-Gibbs. KR: Zehr. RD: Strvnar. MT: Lindstrom. AB:
Lau. C. (2015). Developmental toxicity of perfluorononanoic acid in mice. Reprod Toxicol
51: 133-144. http://dx.doi.Org/10.1016/i.reprotox.2014.12.012
Dickersin. K. (1990). The existence of publication bias and risk factors for its occurrence. JAMA 263:
1385-1389.
DOE (U.S. Department of Energy). (2016). Temporary emergency exposure limits for chemicals:
methods and practice. (DOE-HDBK-1046-2016).
https://www.standards.doe.gov/standards-documents/1000/1046-Bhdbk-2Q16/
DOE (U.S. Department of Energy). (2018). Table 2: Protective Action Criteria (PAC) Rev. 29a based
on applicable 60-minute AEGLs, ERPGs, or TEELs.
https://edms.energy.gov/pac/docs/Revision 29A Table2.pdf
Edwards. SW: Tan. YM: Villeneuve. PL: Meek. ME: Mcqueen. CA. (2016). Adverse outcome
pathways-organizing toxicological information to improve decision making [Review], J
Pharmacol Exp Ther 356: 170-181. http://dx.doi.org/10.1124/ipetll5.228239
EFSA (European Food Safety Authority). (2017). Guidance on the use of the weight of evidence
approach in scientific assessments. EFSA J15: 1-69.
http://dx.doi.Org/10.2903/i.efsa.2017.4971
EMEA (European Medicines Agency). (2008). Non-clinical guideline on drug induced hepatotoxicity.
(Doc. Ref. EMEA/CHMP/SWP/150115/2006). London, UK.
http://www.ema.europa.eu/docs/en GB/document library/Scientific guideline/2009/09/
WC500003355.pdf
Fabrega. F: Kumar. V: Benfenati. E: Schuhmacher. M: Domingo. TL: Nadal. M. (2015). Physiologically
based pharmacokinetic modeling of perfluoroalkyl substances in the human body. Toxicol
Environ Chem 97: 814-827. http://dx.doi.org/10.1080/02772248.2015.1060976
Fang. X: Feng. Y: Wang. 1: Dai. 1. (2010). Perfluorononanoic acid-induced apoptosis in rat spleen
involves oxidative stress and the activation of caspase-independent death pathway.
Toxicology 267: 54-59. http://dx.doi.Org/10.1016/i.tox.2009.10.020
Farland. WH. (2005). [Memo to Science Policy council regarding implementation of the cancer
guidelines and accompanying supplemental guidance - Science Policy Council Cancer
Guidelines. Implementation Workgroup communication I: Application of the mode of action
framework in mutagenicity determinations for carcinogenicity]. Available online at
https://www.epa.gov/sites/production/files/2015-
01/documents/cgiwgcommuniation i.pdf
Fei. C: Mclaughlin. IK: Tarone. RE: Olsen. 1. (2007). Perfluorinated chemicals and fetal growth: A
study within the Danish National Birth Cohort Environ Health Perspect 115: 1677-1682.
http://dx.doi.org/10.1289/ehp.10506
Filgo. AT: Ouist. EM: Hoenerhoff. MT: Brix. AE: Kissling. GE: Fenton. SE. (2015). Perfluorooctanoic
acid (PFOA)-induced liver lesions in two strains of mice following developmental
exposures: PPARalpha is not required. Toxicol Pathol 43: 558-568.
http://dx.doi. org/10.1177/0192623314558463
Foreman. IE: Chang. SC: Ehresman. DT: Butenhoff. TL: Anderson. CR: Palkar. PS: Kang. BH: Gonzalez.
FT: Peters. TM. (2009). Differential Hepatic Effects of Perfluorobutyrate Mediated by Mouse
and Human PPAR-alpha. Toxicol Sci 110: 204-211.
http ://dx. doi. or g/10.109 3 /toxsci /kfp 077
Fromme. H: Mosch. C: Morovitz. M: Alba-Alejandre. I: Boehmer. S: Kiranoglu. M: Faber. F: Hannibal.
I: Genzel-Boroviczenv. 0: Koletzko. B: Volkel. W. (2010). Pre- and postnatal exposure to
perfluorinated compounds (PFCs). Environ Sci Technol 44: 7123-7129.
http://dx.doi.org/10.1021/eslQ1184f
FSANZ (Food Standards Australia New Zealand). (2016). Hazard assessment report:
Perfluorooctane sulfonate (PF0S), perfluorooctanoic acid (PF0A), perfluorohexane
This document is a draft for review purposes only and does not constitute Agency policy.
R-3 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
sulfonate (PFHxS).
http://www.health.gov.au/internet/main/publishing.nsf/Content/2200FE086D480353CA
2580C900817CDC/$File/6.sdl-Hazard-assessment-report.pdf
Gao. K: Gao. Y: Li. Y: Fu. 1: Zhang. A. (2016). A rapid and fully automatic method for the accurate
determination of a wide carbon-chain range of per- and polyfluoroalkyl substances (C4-
C18) inhuman serum. J Chromatogr A1471: 1-10.
http://dx.doi.Org/10.1016/i.chroma.2016.09.050
Goeden. HM: Greene. CW: Tacobus. TA. (2019). A transgenerational toxicokinetic model and its use in
derivation of Minnesota PFOA water guidance. J Expo Sci Environ Epidemiol 29: 183-195.
http://dx.doi. org/10.1038/s41370-018-0110-5
Gonzalez. FT: Shah. YM. (2008). PPARalpha: mechanism of species differences and
hepatocarcinogenesis of peroxisome proliferators [Review], Toxicology 246: 2-8.
http://dx.doi.Org/10.1016/i.tox.2007.09.030
Guelfo. TL: Marlow. T: Klein. DM: Savitz. DA: Frickel. S: Crimi. M: Suuberg. EM. (2018). Evaluation
and Management Strategies for Per- and Polyfluoroalkyl Substances (PFASs) in Drinking
Water Aquifers: Perspectives from Impacted U.S. Northeast Communities. Environ Health
Perspect 126: 065001. http://dx.doi.org/l0.1289/EHP2727
Guvatt. G: Oxman. AD: Akl. EA: Kunz. R: Vist. G: Brozek. 1: Norris. S: Falck-Ytter. Y: Glasziou. P:
DeBeer. H: Taeschke. R: Rind. D: Meerpohl. 1: Dahm. P: Schiinemann. HI. (2011). GRADE
guidelines: 1. Introduction-GRADE evidence profiles and summary of findings tables. J Clin
Epidemiol 64: 383-394. http://dx.doi.Org/10.1016/i.iclinepi.2010.04.026
Guvton. KZ: Chiu. WA: Bateson. TF: Tinot. I: Scott. CS: Brown. RC: Caldwell. TC. (2009). A
reexamination of the PPAR-alpha activation mode of action as a basis for assessing human
cancer risks of environmental contaminants [Review], Environ Health Perspect 117: 1664-
1672. http://dx.doi.org/10.1289/ehp.0900758
Hall. AP: Elcombe. CR: Foster. TR: Harada. T: Kaufmann. W: Knippel. A: Kiittler. K: Malarkev. DE:
Maronpot. RR: Nishikawa. A: Nolte. T: Schulte. A: Strauss. V: York. Ml. (2012). Liver
hypertrophy: a review of adaptive (adverse and non-adverse) changes-conclusions from
the 3rd International ESTP Expert Workshop [Review], Toxicol Pathol 40: 971-994.
http: //dx.doi.org/10.1177/0192623312448935
Harbison. RD: Bourgeois. MM: Tohnson. GT. (2015). Hamilton & Hardys industrial toxicology (6th
ed.). Hoboken, NJ: John Wiley & Sons, Inc.
http://www.wiley.eom/WileyCDA/WileyTitle/productCd-0470929731.subiectCd-
CH50.html
Haug. LS: Huber. S: Becher. G: Thomsen. C. (2011). Characterisation of human exposure pathways to
perfluorinated compounds-comparing exposure estimates with biomarkers of exposure.
Environ Int 37: 687-693. http://dx.doi.Org/10.1016/i.envint.2011.01.011
Haug. LS: Thomsen. C: Brantsaeter. AL: Kvalem. HE: Haugen. M: Becher. G: Alexander. 1: Meltzer.
HM: Knutsen. HK. (2010). Diet and particularly seafood are major sources of perfluorinated
compounds in humans. Environ Int 36: 772-778.
http://dx.doi.Org/10.1016/j.envint2010.05.016
Heo. II: Lee. TW: Kim. SK: Oh. IE. (2014). Foodstuff analyses show that seafood and water are major
perfluoroalkyl acids (PFAAs) sources to humans in Korea. J Hazard Mater 279: 402-409.
http://dx.doi.Org/10.1016/i.ihazmat2014.07.004
Higgins. TPT: Green. S. (2011). Cochrane handbook for systematic reviews of interventions. Version
5.1.0 (Updated March 2011): The Cochrane Collaboration. Retrieved from
http://handbook.cochrane.org/
Hill. AB. (1965). The environment and disease: Association or causation? Proc R Soc Med 58: 295-
300.
This document is a draft for review purposes only and does not constitute Agency policy.
R-4 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Hinderliter. PM: Mvlchreest. E: Gannon. SA: Butenhoff. TL: Kennedy. GL. (2005). Perfluorooctanoate:
Placental and lactational transport pharmacokinetics in rats. Toxicology 211: 139-148.
http://dx.doi.Org/10.1016/i.tox.2005.03.010
Hirst. TA: Howick. 1: Aronson. IK: Roberts. N: Perera. R: Koshiaris. C: Heneghan. C. (2014). The need
for randomization in animal trials: an overview of systematic reviews [Review], PLoS ONE
9: e98856. http://dx.doi.org/10.1371/iournal.pone.0098856
Hoenig. TM: Heisev. DM. (2001). The abuse of power: The pervasive fallacy of power calculations for
data analysis. Am Stat 55: 19-24.
Holden. PR: Tugwood. ID. (1999). Peroxisome proliferator-activated receptor alpha: role in rodent
liver cancer and species differences [Review], J Mol Endocrinol 22: 1-8.
Howard. BE: Phillips. 1: Miller. K: Tandon. A: Mav. D: Shah. MR: Holmgren. S: Pelch. KE: Walker. V:
Roonev. AA: Macleod. M: Shah. RR: Thayer. K. (2016). SWIFT-Review: a text-mining
workbench for systematic review. SystRev 5: 87. http: //dx.doi.org/10.1186/sl3643-Q16-
0263-z
Hu. XC: Andrews. DO: Lindstrom. AB: Bruton. TA: Schaider. LA: Grandjean. P: Lohmann. R: Carignan.
CC: Blum. A: Balan. SA: Higgins. CP: Sunderland. EM. (2016). Detection of poly- and
perfluoroalkyl substances (PFASs) in U.S. drinking water linked to industrial sites, military
fire training areas, and wastewater treatment plants. Environ Sci Technol Lett 3: 344-350.
http://dx.doi.Org/10.1021/acs.estlett.6b00260
Toshi-Barve. S: Kirpich. I: Cave. MC: Marsano. LS: McClain. CI. (2015). Alcoholic, nonalcoholic, and
toxicant-associated steatohepatitis: Mechanistic similarities and differences [Review],
CMGH 1: 356-367. http://dx.doi.org/10.1016/i.icmgh.2015.05.006
Karrman. A: Ericson. I: van Bavel. B: Darnerud. PO: Aune. M: Glynn. A: Lignell. S: Lindstrom. G.
(2007). Exposure of perfluorinated chemicals through lactation: Levels of matched human
milk and serum and a temporal trend, 1996-2004, in Sweden. Environ Health Perspect 115:
226-230. http://dx.doi.org/10.1289/ehp.9491
Kato. K: Calafat. AM: Needham. LL. (2009). Polyfluoroalkyl chemicals in house dust. Environ Res
109: 518-523. http://dx.doi.Org/10.1016/i.envres.2009.01.005
Kawashima. Y: Kobavashi. H: Miura. H: Kozuka. H. (1995). Characterization of hepatic responses of
rat to administration of perfluorooctanoic and perfluorodecanoic acids at low levels.
Toxicology 99: 169-178. http://dx.doi.org/10.1016/0300-483X(95)03027-D
KemI (Swedish Chemicals Agency). (2015). International chemical identification:
Nonadecaflourodecanoic acid (PFDA) and its ammonium and sodium salts.
https://echa.europa.eu/documents/10162/0ddl63d4-56e5-3f3b-8a91-745ec9af2a9f
Kim. ST: Choi. El: Choi. GW: Lee. YB: Cho. HY. (2019). Exploring sex differences in human health risk
assessment for PFNA and PFDA using a PBPK model. Arch Toxicol 93: 311-330.
http: //dx.doi.org/10.1007/s00204-018-2365-v
Kim. ST: Shin. H: Lee. YB: Cho. HY. (2018). Sex-specific risk assessment of PFHxS using a
physiologically based pharmacokinetic model. Arch Toxicol 92: 1113-1131.
http: //dx.doi.org/10.1007/s00204-017-2116-5
Kim. SK: Kannan. K. (2007). Perfluorinated acids in air, rain, snow, surface runoff, and lakes:
relative importance of pathways to contamination of urban lakes. Environ Sci Technol 41:
8328-8334. http://dx.doi.org/10.1021 /es072107t
Kinney. LA: Chromev. NC: Kennedy. GL. (1989). Acute inhalation toxicity of ammonium
perfluorononanoate. Food Chem Toxicol 27: 465-468. http://dx.doi.org/10.1016/0278-
6915C89190033-1
Klaunig. IE: Shinohara. M: Iwai. H: Chengelis. CP: Kirkpatrick. IB: Wang. Z: Bruner. RH. (2015).
Evaluation of the chronic toxicity and carcinogenicity of perfluorohexanoic acid (PFHxA) in
Sprague-Dawley rats. Toxicol Pathol 43: 209-220.
http://dx.doi.org/10.1177/0192623314530532
This document is a draft for review purposes only and does not constitute Agency policy.
R-5 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Krauth. D: Woodruff. TT: Bero. L. (2013). Instruments for assessing risk of bias and other
methodological criteria of published animal studies: a systematic review [Review], Environ
Health Perspect 121: 985-992. http://dx.doi.org/10.1289/ehp.12Q6389
Laitinen. TA: Koponen. 1: Koikkalainen. 1: Kiviranta. H. (2014). Firefighters' exposure to
perfluoroalkyl acids and 2-butoxyethanol present in firefighting foams. Toxicol Lett 231:
227-232. http://dx.doi.org/10.1016/i.toxlet.2014.09.007
Lau. C. (2015). Perfluorinated compounds: An overview. In JC DeWitt (Ed.), Toxicological effects of
perfluoroalkyl and polyfluoroalkyl substances (pp. 1-21). New York: Springer.
http://dx.doi. org/10.1007/978-3-319-15518-0 1
Lau. C: Anitole. K: Hodes. C: Lai. D: Pfahles-Hutchens. A: Seed. I. (2007). Perfluoroalkyl acids: a
review of monitoring and toxicological findings [Review], Toxicol Sci 99: 366-394.
http: / /dx. doi. or g/10.109 3 /toxsci/kfml 2 8
Lau. C: Thibodeaux. TR: Hanson. RG: Narotskv. MG: Rogers. TM: Lindstrom. AB: Strvnar. Ml. (2006).
Effects of perfluorooctanoic acid exposure during pregnancy in the mouse. Toxicol Sci 90:
510-518. http: //dx. doi. or g/10.109 3 /toxsci /kfj 105
Li. K: Gao. P: Xiang. P: Zhang. X: Cui. X: Ma. LP. (2017). Molecular mechanisms of PFOA-induced
toxicity in animals and humans: Implications for health risks [Review], Environ Int 99: 43-
54. http://dx.doi.Org/10.1016/i.envint2016.ll.014
Li. T: Yu. RT: Atkins. AR: Downes. M: Tukev. RH: Evans. RM. (2012). Targeting the pregnane X
receptor in liver injury [Review], Expert OpinTher Targets 16: 1075-1083.
http: //dx.doi.org/10.1517/14728222.2012.715634
Li. XL: Sui. TO: Lu. LL: Zhang. NN: Xu. X: Dong. OY: Xin. YN: Xuan. SY. (2016). Gene polymorphisms
associated with non-alcoholic fatty liver disease and coronary artery disease: a concise
review [Review], Lipids Health Dis 15: 53. http://dx.doi.org/10.1186/sl2944-016-0221-8
Liss. KH: Finck. BN. (2017). PPARs and nonalcoholic fatty liver disease [Review], Biochimie 136: 65-
74. http://dx.doi.Org/10.1016/j.biochi.2016.ll.009
Loccisano. AE: Campbell. TL. Ir: Andersen. ME: Clewell. HI. III. (2011). Evaluation and prediction of
pharmacokinetics of PFOA and PFOS in the monkey and human using a PBPK model. Regul
Toxicol Pharmacol 59: 157-175. http://dx.doi.Org/10.1016/i.yrtph.2010.12.004
Luebker. DT: Case. MT: York. RG: Moore. TA: Hansen. KT: Butenhoff. TL. (2005). Two-generation
reproduction and cross-foster studies of perfluorooctanesulfonate (PFOS) in rats.
Toxicology 215: 126-148. http://dx.doi.Org/10.1016/i.tox.2005.07.018
Mackowiak. B: Hodge. 1: Stern. S: Wang. H. (2018). The roles of xenobiotic receptors: Beyond
chemical disposition [Review], Drug Metab Dispos 46: 1361-1371.
http://dx.doi.org/10.1124/dmd.118.081042
Macleod. MR. (2013). Systematic reviews of experimental animal studies. Presentation presented at
Workshop on weight of evidence; US National Research Council Committee to review the
Integrated Risk Information System (IRIS) process, March 27-28, 2013, Washington, DC.
Malonev. EK: Waxman. DT. (1999). trans-Activation of PPARalpha and PPARgamma by structurally
diverse environmental chemicals. Toxicol Appl Pharmacol 161: 209-218.
http://dx.doi.org/10.1006/taap.1999.8809
Mamsen. LS: Biorvang. RD: Mucs. D: Vinnars. MT: Papadogiannakis. N: Lindh. CH: Andersen. CY:
Damdimopoulou. P. (2019). Concentrations of perfluoroalkyl substances (PFASs) in human
embryonic and fetal organs from first, second, and third trimester pregnancies. Environ Int
124: 482-492. http://dx.doi.Org/l 0.1016/i.envint.2019.01.010
MDH (Minnesota Department of Health). (2009). East Metro Perfluorochemical Biomonitoring Pilot
Project. (8EHQ-80-373). St Paul, MN: 3M Company.
MDH (Minnesota Department of Health). (2018). Health based guidance for water: Toxicological
summary for: Perfluorobutanoate. Health Based Guidance for Water Health Risk
Assessment Unit, Environmental Health Division.
This document is a draft for review purposes only and does not constitute Agency policy.
R-6 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
https://www.health.state.mn.us/communities/environment/risk/docs/guidance/gw/pfba
2summ.pdf
MDH (Minnesota Department of Health). (2019). Health Based Guidance for Water: Toxicological
Summary for: Perfluorohexane sulfonate (PFHxS).
https://www.health.state.mn.us/communities/environment/risk/docs/guidance/gw/pfhx
s.pdf
Meek. ME: Boobis. A: Cote. I: Dellarco. V: Fotakis. G: Munn. S: Seed. 1: Vickers. C. (2014). New
developments in the evolution and application of the WHO/IPCS framework on mode of
action/species concordance analysis [Review], J Appl Toxicol 34: 1-18.
http ://dx. doi.org/10.10Q2/iat.2949
Mellor. CL: Steinmetz. FP: Cronin. MT. (2016). The identification of nuclear receptors associated
with hepatic steatosis to develop and extend adverse outcome pathways [Review], Crit Rev
Toxicol 46: 138-152. http://dx.doi.org/10.3109/10408444.2015.1089471
Mogensen. UB: Grandiean. P: Nielsen. F: Weihe. P: Budtz-largensen. E. (2015). Breastfeeding as an
Exposure Pathway for Perfluorinated Alkylates. Environ Sci Technol 49: 10466-10473.
http://dx.doi.org/10.1021/acs.est5b02237
Mondal. D: Weldon. RH: Armstrong. BG: Gibson. LI: Lopez-Espinosa. MT: Shin. HM: Fletcher. T.
(2014). Breastfeeding: a potential excretion route for mothers and implications for infant
exposure to perfluoroalkyl acids. Environ Health Perspect 122: 187-192.
http://dx.doi.org/10.1289/ehp.1306613
Moreta. C: Tena. MT. (2014). Determination of perfluorinated alkyl acids in corn, popcorn and
popcorn bags before and after cooking by focused ultrasound solid-liquid extraction, liquid
chromatography and quadrupole-time of flight mass spectrometry. J Chromatogr A 1355:
211-218. http://dx.doi.Org/10.1016/i.chroma.2014.06.018
Morgan. RL: Thayer. KA: Bero. L: Bruce. N: Falck-Ytter. Y: Ghersi. D: Guvatt. G: Hooiimans. C:
Langendam. M: Mandrioli. D: Mustafa. RA: Rehfuess. EA: Roonev. AA: Shea. B: Silbergeld. EK:
Sutton. P: Wolfe. MS: Woodruff. TT: Verbeek. TH: Hollowav. AC: Santesso. N: Schiinemann. HI.
(2016). GRADE: Assessing the quality of evidence in environmental and occupational health.
Environ Int 92-93: 611-616. http://dx.doi.Org/10.1016/i.envint2016.01.004
NAS (National Academy of Sciences). (2014). Review of EPA's Integrated Risk Information System
(IRIS) process. Washington, DC: National Academies Press (US).
http://dx.doi.org/10.17226/18764
NASEM (National Academies of Sciences, Engineering, and Medicine). (2017). Application of
systematic review methods in an overall strategy for evaluating low-dose toxicity from
endocrine active chemicals. Washington, D.C.: The National Academies Press.
http://dx.doi.org/10.17226/24758
New Hampshire PES (New Hampshire Department of Environmental Services). (2019). Technical
background report for the june 2019 proposed maximum contaminant levels (MCLs) and
ambient groundwater quality standards (AGQSs) for perfluorooctane sulfonic acid (PF0S),
perfluorooctanoic acid (PF0A), perfluorononanoic acid (PFNA), and perfluorohexane
sulfonic acid (PFHXs).
Nilsson. H: Karrman. A: Rotander. A: van Bavel. B: Lindstrom. G: Westberg. H. (2010a). Inhalation
Exposure to Fluorotelomer Alcohols Yield Perfluorocarboxylates in Human Blood? Environ
Sci Technol 44: 7717-7722. http: //dx.doi.org/10.1021 /esl 019511
Nilsson. H: Karrman. A: Westberg. H: Rotander. A: van Bavel. B: Lindstrom. G. (2010b). A time trend
study of significantly elevated perfluorocarboxylate levels in humans after using fluorinated
ski wax. Environ Sci Technol 44: 2150-2155. http://dx.doi.org/10.1021/es9034733
NTDWOI (New Jersey Drinking Water Quality Institute). (2017). Public review draft: Health-based
maximum contaminant level support document: Perfluorooctane sulfonate (PF0S) (CAS #:
1763-23-1; Chemical Formula: C8HF1703S). NJDWQI Health Effects Subcommittee.
This document is a draft for review purposes only and does not constitute Agency policy.
R-7 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
https://cswab.org/wp-content/uploads/2018/09/New-Tersey-Scientific-Documentation-
Supporting-PF0S-13-ppt.pdf
NLM (National Institutes of Health, National Library of Medicine). (2013). HSDB: Perfluoro-n-
nonanoic acid. Available online at https: //toxnet.nlm.nih.gov /cgi-
bin/sis/search/a?dbs+hsdb:@term+@DQCNO+8040
NLM (National Institutes of Health, National Library of Medicine). (2016). HSDB: Perfluorohexanoic
acid. Available online at https://toxnetnlm.nih.gov/cgi-
bin/sis/search/a?dbs+hsdb:@term+@DQCNO+8299
NLM (National Institutes of Health, National Library of Medicine). (2017). HSDB:
Perfluorohexanesulfonic acid. Available online at https://toxnet.nlm. nih. gov/cgi-
bin/sis /search/a?dbs+hsdb: @term+@DQCN0+82 74
Norwegian Environment Agency. (2018). Investigation of sources to PFHxS in the environment.
Munich, Germany: BiPRO GmbH.
https://www.miliodirektoratet.no/globalassets/publikasioner/M961/M961.pdf
NRC (National Research Council). (2014). Review of EPA's Integrated Risk Information System
(IRIS) process. Washington, DC: The National Academies Press.
http://www.nap.edu/catalog.php7record id=18764
NTP (National Toxicology Program). (2011). NTP Protocol Outline of the Class Study of
Perfluorinated Chemicals [perfluorobutane sulfonate (PFBS), perfluorohexane sulfonate
potassium salt (PFHSKslt), perfluorooctane sulfonate (PFOS), perfluorohexanoic acid
(PFHxA), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluorodecanoic
acid (PFDA) and WY 14643 (WY)] in Harlan Sprague Dawley rats administered daily by
gavage for 28-Days [NTP],
NTP (National Toxicology Program). (2015). Handbook for conducting a literature-based health
assessment using OHAT approach for systematic review and evidence integration. U.S. Dept.
of Health and Human Services, National Toxicology Program.
https://ntp.niehs.nih.gov/ntp/ohat/pubs/handbookian2015 508.pdf
NTP (National Toxicology Program). (2018a). TOX-96: 1-Perfluorobutanesulfonic acid (375-73-5),
potassium perfluorohexanesulfonate (3871-99-6), perfluorooctane sulfonate (1763-23-1),
WY-14643 (50892-23-4). Chemical Effects in Biological Systems (CEBS). Available online at
https://manticore.niehs.nih.gov/cebssearch/publication/T0X-96 (accessed August 6,
2018).
NTP (National Toxicology Program). (2018b). TOX-97: Perfluorohexanoic acid (307-24-4),
perfluorooctanoic acid (335-67-1), perfluorononanoic acid (375-95-1), perfluorodecanoic
acid (335-76-2), WY-14643 (50892-23-4). Chemical Effects in Biological Systems (CEBS).
Available online at https://manticore.niehs.nih.gov/cebssearch/publication/T0X-97
(accessed April 24, 2018).
NTP (National Toxicology Program). (2019). TOX-96: Toxicity ReportTables and Curves for Short-
term Studies: Perfluorinated Compounds: Sulfonates. Available online at
https://tools.niehs.nih.gov/cebs3/views/?action=main.dataReview&bin id=3874
OECD (Organisation for Economic Co-operation and Development). (2015). Working towards a
global emission inventory of PFASS: focus on PFCAS - status quo and the way forward. Paris,
France.
http://www.oecd.org/chemicalsafetv/Working%20Towards%20a%20Global%20Emission
%20Inventory%20of%2 OPFASS.pdf
Papadopoulou. E: Sabaredzovic. A: Namork. E: Nvgaard. UC: Granum. B: Haug. LS. (2016). Exposure
of Norwegian toddlers to perfluoroalkyl substances (PFAS): The association with
breastfeeding and maternal PFAS concentrations. Environ Int 94: 687-694.
http://dx.doi.Org/10.1016/j.envint2016.07.006
This document is a draft for review purposes only and does not constitute Agency policy.
R-8 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Peraza. MA: Burdick. AD: Marin. HE: Gonzalez. FT: Peters. TM. (2006). The toxicology of ligands for
peroxisome proliferator-activated receptors (PPAR) [Review], Toxicol Sci 90: 269-295.
http ://dx. doi.org/10.1093 /toxsci /kfi 062
Perez. F: Llorca. M: Kock-Schulmever. M: Skrbic. B: Oliveira. LS: da Boit Martinello. K: Al-Dhabi. NA:
Antic. I: Farre. M: Barcelo. D. (2014). Assessment of perfluoroalkyl substances in food items
at global scale. Environ Res 135: 181-189. http://dx.doi.Org/10.1016/i.envres.2014.08.004
Pinnev. SM: Biro. FM: Windham. GC: Herrick. RL: Yaghivan. L: Calafat. AM: Succop. P: Sucharew. H:
Ball. KM: Kato. K: Kushi. LH: Bornschein. R. (2014). Serum biomarkers of polyfluoroalkyl
compound exposure in young girls in Greater Cincinnati and the San Francisco Bay Area,
USA. Environ Pollut 184: 327-334. http://dx.doi.Org/10.1016/i.envpol.2013.09.008
Post. GB: Cohn. PD: Cooper. KR. (2012). Perfluorooctanoic acid (PFOA), an emerging drinking water
contaminant: a critical review of recent literature [Review], Environ Res 116: 93-117.
http://dx.doi.Org/10.1016/i.envres.2012.03.007
Post. GB: Louis. IB: Lippincott. RL: Procopio. NA. (2013). Occurrence of perfluorinated compounds
in raw water from New Jersey public drinking water systems. Environ Sci Technol 47:
13266-13275. http://dx.doi.org/10.1021/es402884x
Prevedouros. K: Cousins. IT: Buck. RC: Korzeniowski. SH. (2006). Sources, fate and transport of
perfluorocarboxylates [Review], Environ Sci Technol 40: 32-44.
http://dx.doi.org/10.1021/es0512475
Rosen. MB: Das. KP: Roonev. 1: Abbott. B: Lau. C: Corton. TC. (2017). PPARa-independent
transcriptional targets of perfluoroalkyl acids revealed by transcript profiling. Toxicology
387: 95-107. http://dx.doi.org/l 0.1016/i.tox.2017.05.013
Rotander. A: Karrman. A: Toms. LM: Kay. M: Mueller. IF: Gomez Ramos. Ml. (2015). Novel
fluorinated surfactants tentatively identified in firefighters using liquid chromatography
quadrupole time-of-flight tandem mass spectrometry and a case-control approach. Environ
Sci Technol 49: 2434-2442. http://dx.doi.org/10.1021/es503653n
Russell. MH: Himmelstein. MW: Buck. RC. (2015). Inhalation and oral toxicokinetics of 6:2 FTOH
and its metabolites in mammals. Chemosphere 120:328-335.
http://dx.doi.Org/10.1016/i.chemosphere.2014.07.092
Schaider. LA: Balan. SA: Blum. A: Andrews. DO: Strvnar. Ml: Dickinson. ME: Lunderberg. DM: Lang.
TR: Peaslee. GF. (2017). Fluorinated compounds in US fast food packaging. Environ Sci
Technol Lett4: 105-111. http://dx.doi.org/10.1021/acs.estlett6b00435
Schecter. A: Malik-Bass. N: Calafat. AM: Kato. K: Colacino. TA: Gent. TL: Hvnan. LS: Harris. TR: Malla.
S: Birnbaum. L. (2012). Polyfluoroalkyl compounds in Texas children from birth through 12
years of age. Environ Health Perspect 120: 590-594.
http://dx.doi.org/10.1289/ehp.1104325
Schiinemann. H: Hill. S: Guvatt. G: Akl. EA: Ahmed. F. (2011). The GRADE approach and Bradford
Hill's criteria for causation. J Epidemiol Community Health 65: 392-395.
http://dx.doi.org/10.1136/iech.201Q.119933
Shipley. TM: Hurst. CH: Tanaka. SS: Deroos. FL: Butenhoff. TL: Seacat. AM: Waxman. DT. (2004).
Trans-activation of PPARalpha and induction of PPARalpha target genes by
perfluorooctane-based chemicals. Toxicol Sci 80: 151-160.
http ://dx. doi. or g/10.109 3 /toxsci/kfh 130
Smith. MT: Guvton. KZ: Gibbons. CF: Fritz. TM: Portier. CI: Rusvn. I: DeMarini. DM: Caldwell. TC:
Kavlock. RT: Lambert. PF: Hecht. SS: Bucher. TR: Stewart. BW: Baan. RA: Cogliano. VI: Straif.
K. (2016). Key characteristics of carcinogens as a basis for organizing data on mechanisms
of carcinogenesis [Review], Environ Health Perspect 124: 713-721.
http://dx.doi.org/10.1289/ehp.15Q9912
Soldatow. VY: Lecluvse. EL: Griffith. LG: Rusvn. I. (2013). In vitro models for liver toxicity testing.
Toxicology Research 2: 23-39. http://dx.doi.org/10.1039/c2tx20051a
This document is a draft for review purposes only and does not constitute Agency policy.
R-9 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Stahl. LL: Snyder. BP: Olsen. AR: Kincaid. TM: Wathen. IB: McCarty. HB. (2014). Perfluorinated
compounds in fish from U.S. urban rivers and the Great Lakes. Sci Total Environ 499: 185-
195. http: / /dx. doi. o r g/10.1016 /i. scitotenv. 2 014.0 7.12 6
Sterne. TAC: Hernan. MA: Reeves. BC: Savovic. I: Berkman. ND: Viswanathan. M: Henry. D: Altman.
DG: Ansari. MT: Boutron. I: Carpenter. TR: Chan. AW: Churchill. R: Peeks. TT: Hrobiartsson. A:
Kirkham. 1: Tiini. P: Loke. YK: Pigott. TP: Ramsay. CR: Regidor. D: Rothstein. HR: Sandhu. L:
Santaguida. PL: Schiinemann. HI: Shea. B: Shrier. I: Tugwell. P: Turner. L: Valentine. 1C:
Waddington. H: Waters. E: Wells. GA: Whiting. PF: Higgins. TPT. (2016). ROBINS-I: A tool for
assessing risk of bias in non-randomised studies of interventions. Br Med J 355: i4919.
Strvnar. MT: Lindstrom. AB. (2008). Perfluorinated compounds in house dust from Ohio and North
Carolina, USA. Environ Sci Technol 42: 3751-3756. http://dx.doi.org/10.1021/es7032058
Sun. M: Arevalo. E: Strvnar. M: Lindstrom. A: Richardson. M: Kearns. B: Pickett. A: Smith. C: Knappe.
DRU. (2016). Legacy and emerging perfluoroalkyl substances are important drinking water
contaminants in the Cape Fear River Watershed of North Carolina. Environ Sci Technol Lett
3: 415-419. http: //dx. doi. o r g/10.10 21 /acs. e stlett 6b0 0 3 9 8
Surma. M: Piskula. M: Wiczkowski. W: Zieliriski. H. (2017). The perfluoroalkyl carboxylic acids
(PFCAs) and perfluoroalkane sulfonates (PFSAs) contamination level in spices. European
Food Research and Technology 243: 297-307. http://dx.doi.org/10.1007/s00217-016-
2744-7
Swenberg. TA: Lehman-McKeeman. LP. (1999). Alpha 2-urinary globulin-associated nephropathy as
a mechanism of renal tubule cell carcinogenesis in male rats. In CC Capen; E Pybing; JM
Rice; JP Wilbourn (Eds.), Species Pifferences in Thyroid, Kidney, and Urinary Bladder
Carcinogenesis (pp. 95-118). Lyon, France: International Agency for Research on Cancer.
http://apps.who.int/bookorders/anglais/detartl.isp?sesslan=l&codlan=l&codcol=73&cod
cch=147
Takacs. ML: Abbott. BP. (2007). Activation of mouse and human peroxisome proliferator-activated
receptors (alpha, beta/delta, gamma) by perfluorooctanoic acid and perfluorooctane
sulfonate. Toxicol Sci 95: 108-117. http: / / dx. do i. o r g /10.109 3 /toxs ci /kfl 13 5
TCEO (Texas Commission on Environmental Quality). (2016). Perfluoro compounds (PFCs).
https://www.tceq.texas.gov/assets/public/implementation/tox/evaluations/pfcs.pdf
Thomford. PI. (2002). 104-week dietary chronic toxicity and carcinogenicity study with
perfluorooctane sulfonic acid potassium salt (PFOS; T-6295) in rats. Study No. 6329-183.
Madison, WI: Convance Laboratories.
Thoolen. B: Maronpot. RR: Harada. T: Nvska. A: Rousseaux. C: Nolte. T: Malarkev. PE: Kaufmann. W:
Kiittler. K: Peschl. U: Nakae. P: Gregson. R: Vinlove. MP: Brix. AE: Singh. B: Belpoggi. F:
Ward. TM. (2010). Proliferative and nonproliferative lesions of the rat and mouse
hepatobiliary system [Review], Toxicol Pathol 38: 5S-81S.
http://dx.doi.org/10.1177/0192623310386499
U.S. EPA (U.S. Environmental Protection Agency). (1988). Recommendations for and documentation
of biological values for use in risk assessment [EPA Report] (pp. 1-395). (EPA/600/6-
87/008). Cincinnati, OH: U.S. Environmental Protection Agency, Office of Research and
Pevelopment, Office of Health and Environmental Assessment.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=34855
U.S. EPA (U.S. Environmental Protection Agency). (1991a). Guidelines for developmental toxicity
risk assessment (pp. 1-71). (EPA/600/FR-91/001). Washington, PC: U.S. Environmental
Protection Agency, Risk Assessment Forum.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=23162
U.S. EPA (U.S. Environmental Protection Agency). (1991b). Guidelines for developmental toxicity
risk assessment. Fed Reg 56: 63798-63826.
This document is a draft for review purposes only and does not constitute Agency policy.
R-10 PRAFT-P0 NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
U.S. EPA (U.S. Environmental Protection Agency). (1992). Guidelines for exposure assessment.
Federal Register 57(104):22888-22938 [EPAReport], In Risk Assessment Forum.
(EPA/600/Z-92/001). Washington, DC.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=15263
U.S. EPA (U.S. Environmental Protection Agency). (1994). Methods for derivation of inhalation
reference concentrations and application of inhalation dosimetry [EPA Report],
(EPA/600/8-90/066F). Research Triangle Park, NC: U.S. Environmental Protection Agency,
Office of Research and Development, Office of Health and Environmental Assessment,
Environmental Criteria and Assessment Office.
https://cfpub.epa. gov/ncea/risk/recordisplay.cfm?deid=71993&.CFID=51174829&.CFTOKE
N=25006317
U.S. EPA (U.S. Environmental Protection Agency). (1996a). Guidelines for reproductive toxicity risk
assessment. Fed Reg 61: 56274-56322.
U.S. EPA (U.S. Environmental Protection Agency). (1996b). Guidelines for reproductive toxicity risk
assessment (pp. 1-143). (EPA/630/R-96/009). Washington, DC: U.S. Environmental
Protection Agency, Risk Assessment Forum.
https://www.epa.gov/sites/production/files/2Q14-
11 /documents/guidelines repro toxicity.pdf
U.S. EPA (U.S. Environmental Protection Agency). (1998). Guidelines for neurotoxicity risk
assessment [EPAReport] (pp. 1-89). (ISSN 0097-6326
EISSN 2167-2520
EPA/630/R-95/001F). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment
Forum, http://www.epa.gov/risk/guidelines-neurotoxicity-risk-assessment
U.S. EPA (U.S. Environmental Protection Agency). (2002a). Hepatocellular hypertrophy. HED
guidance document#G2002.01 [EPAReport], Washington, DC.
U.S. EPA (U.S. Environmental Protection Agency). (2002b). A review of the reference dose and
reference concentration processes. (EPA/630/P-02/002F). Washington, DC: U.S.
Environmental Protection Agency, Risk Assessment Forum.
https://www.epa.gov/sites/production/files/2014-12/documents/rfd-final.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2004). Toxicological review of boron and
compounds. In support of summary information on the Integrated Risk Information System
(IRIS) [EPAReport], (EPA/635/04/052). Washington, DC: U.S. Environmental Protection
Agency, IRIS. http://nepis.epa.gov/exe/ZyPURL.cgi?Dockey=P1006CK9.txt
U.S. EPA (U.S. Environmental Protection Agency). (2005a). Guidelines for carcinogen risk
assessment [EPAReport], (EPA/630/P-03/001B). Washington, DC: U.S. Environmental
Protection Agency, Risk Assessment Forum.
https://www.epa.gov/sites/production/files/2013-
09/documents/cancer guidelines final 3-25-05.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2005b). Supplemental guidance for assessing
susceptibility from early-life exposure to carcinogens [EPA Report], (EPA/630/R-03/003F).
Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum.
https://www3.epa.gov/airtoxics/childrens supplement final.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2011a). Recommended use of body weight 3/4
as the default method in derivation of the oral reference dose (pp. 1-50). (EPA/100/R-
11/0001). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum,
Office of the Science Advisor, https://www.epa.gov/sites/production/files/2013-
09/documents/recommended-use-of-bw34.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2011b). Toxicological review of
trichloroethylene (CASRN 79-01-6) in support of summary information on the Integrated
Risk Information System (IRIS) [EPA Report], (EPA/635/R-09/011F). Washington, DC.
This document is a draft for review purposes only and does not constitute Agency policy.
R-ll DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
https://cfpub.epa.gov/ncea/iris/iris documents/documents/toxreviews/0199tr/0199tr.p
df
U.S. EPA (U.S. Environmental Protection Agency). (2012). Benchmark dose technical guidance.
(EPA/100/R-12/001). Washington, DC: U.S. Environmental Protection Agency, Risk
Assessment Forum, https: //www.epa.gov/risk/benchmark-dose-technical-guidance
U.S. EPA (U.S. Environmental Protection Agency). (2014a). Draft: Health effects document for
perfluorooctane sulfonate (PFOS) [EPA Report], (EPA/822/R-14/002). Washington, DC: U.S.
Environmental Protection Agency, Office of Water, Health and Ecological Criteria Division.
https://peerreview.versar.com/epa/pfoa/pdf/Health-Effects-Document-for-
Perfluorooctane-Sulfonate-rPFOSl.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2014b). Draft: Health effects document for
Perfluorooctanoic Acid (PFOA) [EPA Report], (EPA Document Number: 822R14001). Office
of Water, Health and Ecological Criteria Division.
https://peerreview.versar.com/epa/pfoa/pdf/Health-Effects-Document-for-
Perfluorooctanoic-Acid-(PFOA).pdf
U.S. EPA (U.S. Environmental Protection Agency). (2014c). Guidance for applying quantitative data
to develop data-derived extrapolation factors for interspecies and intraspecies
extrapolation [EPA Report], (EPA/100/R-14/002F). Washington, DC: Risk Assessment
Forum, Office of the Science Advisor, https://www.epa.gov/sites/production/files/2015-
01/documents/ddef-final.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2016a). Drinking water health advisory for
perfluorooctane sulfonate (PFOS) [EPA Report], (EPA 822-R-16-004). Washington, DC: U.S.
Environmental Protection Agency, Office of Water.
https://www.regulations.gov/document?D=EPA-H0-OW-2014-0138-0038
U.S. EPA (U.S. Environmental Protection Agency). (2016b). Drinking water health advisory for
perfluorooctanoic acid (PFOA). (EPA 822-R-16-005). Washington, DC: U.S. Environmental
Protection Agency, Office of Water, https: / /www.regulations. gov /document?D=EPA-H0-
QW-2014-0138-0041
U.S. EPA (U.S. Environmental Protection Agency). (2016c). Health effects support document for
perfluorooctane sulfonate (PFOS) [EPA Report], (EPA 822-R-16-002). Washington, DC: U.S.
Environmental Protection Agency, Office of Water, Health and Ecological Criteria Division.
https://www.epa.gOv/sites/production/files/2016-05/documents/pfos hesd final 508.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2016d). Health effects support document for
perfluorooctanoic acid (PFOA) [EPA Report], (EPA 822-R-16-003). Washington, DC: U.S.
Environmental Protection Agency, Office of Water, Health and Ecological Criteria Division.
https://www.epa.gOv/sites/production/files/2016-05/documents/pfoa hesd final-
plain.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2016e). The Third Unregulated Contaminant
Monitoring Rule. Available online at https: //www.epa.gov/dwucmr/third-unregulated-
contaminant-monitoring-rule
U.S. EPA (U.S. Environmental Protection Agency). (2017a). Guidance to assist interested persons in
developing and submitting draft risk evaluations under the Toxic Substances Control Act.
(EPA/740/R17/001). Washington, DC: U.S Environmental Protection Agency, Office of
Chemical Safety and Pollution Prevention.
https://www.epa.gov/sites/production/files/2017-
06/documents/tsca ra guidance final.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2017b). The Third Unregulated Contaminant
Monitoring Rule (UCMR 3): Data summary, January 2017. (EPA 815-S-17-001). Washington,
DC: U.S Environmental Protection Agency, Office of Water.
This document is a draft for review purposes only and does not constitute Agency policy.
R-12 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
https://www.epa.gOv/sites/production/files/2017-02/documents/ucmr3-data-summary-
ianuarv-2017.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2018a). Chemistry Dashboard. Washington, DC.
Retrieved from https: //comptox.epa.gov/dashboard
U.S. EPA (U.S. Environmental Protection Agency). (2018b). An umbrella Quality Assurance Project
Plan (QAPP) for PBPK models [EPA Report], (ORD QAPP ID No: B-0030740-QP-1-1).
Research Triangle Park, NC.
U.S. EPA (U.S. Environmental Protection Agency). (2019). ChemView [Database], Retrieved from
https://chemview.epa.gov/chemview
van Otterdiik. FM. (2007a). Repeated dose 28-day oral toxicity study with MTDID-8391 by daily
gavage in the rat, followed by a 21-day recovery period. (Study Number 06-226).
Maple wood, MN: 3M.
van Otterdiik. FM. (2007b). Repeated dose 90-day oral toxicity study with MTDID 8391 by daily
gavage in the rat followed by a 3-week recovery period. (Study Number 06-398).
Maple wood, MN: 3M.
Verner. MA: Ngueta. G: Tensen. ET: Fromme. H: Voelkel. W: Nvgaard. UC: Granum. B: Longnecker.
MP. (2016). A simple pharmacokinetic model of prenatal and postnatal exposure to
perfluoroalkyl substances (PFASs). Environ Sci Technol 50: 978-986.
http://dx.doi.org/10.1021/acs.est5b04399
Vesterinen. HM: Sena. ES: Egan. KT: Hirst. TC: Churolov. L: Currie. GL: Antonic. A: Howells. DW:
Macleod. MR. (2014). Meta-analysis of data from animal studies: a practical guide. J
Neurosci Methods 221: 92-102. http://dx.doi.org/10.1016/i.ineumeth.2013.09.010
Viberg. H: Eriksson. P. (2017). Chapter 43 - Perfluorooctane sulfonate and perfluorooctanoic acid.
In RC Gupta (Ed.), Reproductive and Developmental Toxicology (2nd ed., pp. 811-827). San
Diego, CA: Academic Press. http://dx.doi.org/10.1016/B978-0-12-804239-7.00043-3
Wahlang. B: Beier. II: Clair. HB: Bellis-Tones. HI: Falkner. K: Mcclain. CI: Cave. MC. (2013). Toxicant-
associated steatohepatitis [Review], Toxicol Pathol 41: 343-360.
http: //dx.doi.org/10.1177/0192623312468517
Wahlang. B: Tin. I: Beier. II: Hardestv. IE: Daly. EF: Schnegelberger. RD: Falkner. KC: Prough. RA:
Kirpich. IA: Cave. MC. (2019). Mechanisms of environmental contributions to fatty liver
disease [Review], Curr Environ Health Rep 6: 80-94. http://dx.doi.org/10.1007/s40572-
019-00232-w
Wambaugh. IF: Setzer. RW: Pitruzzello. AM: Liu. I: Reif. DM: Kleinstreuer. NC: Wang. NC: Sipes. N:
Martin. M: Das. K: Dewitt. TC: Strvnar. M: Tudson. R: Houck. KA: Lau. C. (2013). Dosimetric
anchoring of in vivo and in vitro studies for perfluorooctanoate and
perfluorooctanesulfonate. Toxicol Sci 136: 308-327.
http: / /dx. doi. or g/10.109 3 /toxsci /kft2 04
Wang. Y: Wang. L: Li. 1: Liang. Y: Ti. H: Zhang. 1: Zhou. 0: Tiang. G. (2014). The mechanism of
immunosuppression by perfluorooctanoic acid in BALB/c mice. Toxicology Research 3: 205-
213. http://dx.doi.org/10.1039/c3tx50096a
Wolf. CI: Rider. CV: Lau. C: Abbott. BP. (2014). Evaluating the additivity of perfluoroalkyl acids in
binary combinations on peroxisome proliferator-activated receptor-a activation. Toxicology
316: 43-54. http://dx.doi.org/10.1016/i.tox.2013.12.002
Wolf. CI: Takacs. ML: Schmid. IE: Lau. C: Abbott. BP. (2008). Activation of mouse and human
peroxisome proliferator-activated receptor alpha by perfluoroalkyl acids of different
functional groups and chain lengths. Toxicol Sci 106: 162-171.
http: / /dx. doi. or g/10.109 3 /toxsci /kfnl 6 6
WS (Weston Solutions Inc). (2007). Remedial investigation report. Phase 2. Fluorochemical (FC)
data assessment report for the Cottage Grove, MN site. St. Paul, MN: 3M Corporate
This document is a draft for review purposes only and does not constitute Agency policy.
R-13 PRAFT-PO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Toxicology, https://www.pca.state.mn.us/sites/default/files/pfc-cottagegrove-
remedialinvestigationreportpdf
Yoo. H: Washington. 1W: lenkins. TM: Ellington. 11. (2011). Quantitative determination of
perfluorochemicals and fluorotelomer alcohols in plants from biosolid-amended fields using
LC/MS/MS and GC/MS. Environ Sci Technol 45: 7985-7990.
http://dx.doi.org/10.1021/eslQ2972m
Zhang. T: Sun. H: Lin. Y: Oin. X: Zhang. Y: Geng. X: Kannan. K. (2013a). Distribution of poly- and
perfluoroalkyl substances in matched samples from pregnant women and carbon chain
length related maternal transfer. Environ Sci Technol 47: 7974-7981.
http://dx.doi.org/10.1021/es400937y
Zhang. Y: Beesoon. S: Zhu. L: Martin. TW. (2013b). Biomonitoring of perfluoroalkyl acids in human
urine and estimates of biological half-life. Environ Sci Technol 47: 10619-10627.
http://dx.doi.org/10.1021/es4019Q5e
Zhao. P: Xia. X: Dong. 1: Xia. N: Tiang. X: Li. Y: Zhu. Y. (2016). Short- and long-chain perfluoroalkyl
substances in the water, suspended particulate matter, and surface sediment of a turbid
river. Sci Total Environ 568: 57-65. http: //dx.doi.Org/10.1016/i.scitotenv.2016.05.221
This document is a draft for review purposes only and does not constitute Agency policy.
R-14 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
ADDENDUM A. SUMMARY OF EXISTING TOXICITY VALUE
INFORMATION FOR PERFLUOROBUTANOIC ACID (PFBA),
PERFLUOROHEXANOIC ACID (PFHXA), PERFLUOROHEXANESULFONIC
ACID (PFHXS), PERFLUORONONANOIC ACID (PFNA), AND
PERFLUORODECANOIC ACID (PFDA)
The values presented in Table A-l through Table A-3 are current as of June 2019. Readers are referred to the individual sources
for the most up-to-date information, and more recent values from agencies not listed here may be available.
Table A-l. Details on derivation of the available health effect reference values for inhalation exposure to selected
per- and polyfluoroalkyl substances (PFAS) (current as of June 2019; please consult source references for
up-to-date information)
Value name
Duration
PFAS
Value
Health effect
Point of
departure
Qualifier
Source
Uncertainty
factors
Notes on
derivation
Review
status
(mg/m3)
(ppm)
Emergency
PAC-3
lh
PFBA
3.3 x 101
3.6 x 10°
Lethality in
mice
NR
NR
NR
PAC values
derived via an
approach
developed by the
Department of
Enerev (DOE,
2016)
Final
(DOE,
2018)
This document is a draft for review purposes only and does not constitute Agency policy.
A-l DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Value
Point of
Uncertainty
factors
Notes on
Review
Value name
Duration
PFAS
(mg/m3)
(ppm)
Health effect
departure
Qualifier
Source
derivation
status
PAC-2
lh
PFBA
5.5 x 10°
6.0 x 10"1
Based on PAC-3
Based on PAC-3a
PAC-l
lh
PFBA
5.0 x 10"1
5.5 x 10"2
Based on PAC-2
Based on PAC-2b
TCEQ RfC
Chronic
PFBA
1.0 x 10"2
1.1 x 10"3
Based on TCEQ
RfD (see
Table A-2)
Based on TCEQ
RfD (route-to-
route
extrapolation)0
Final
(TCEQ,
2016)
u
!5
3
PFDA
5.3 x 10"5
2.5 x 10"6
Based on TCEQ
RfD (see
Table A-2)
Based on TCEQ
RfD (route-to-
route
extrapolation)01
TO
a>
c
ai
15
PFNA
2.8 x 10"5
1.4 x 10"6
Lung noise,
labored
breathing, and
reduced body
wt. in male rats
exposed for 4 h
67 mg/m3
0.83 mg/m3
NOAEL
NOAELhec
Kinnev
et al.
(1989)
UFC = 30,000
UFa = 3
UFh = 10
UFS= 100
UFD = 10
HEC adjusted6
PFHxS
1.3 x 10"5
7.8 x 10"7
Based on TCEQ
RfD (see
Table A-2)
Based on TCEQ
RfD (route-to-
route
extrapolation)'
aPAC-2 = PAC-3 4 6 = 33 mg/m3 4 6 = 5.5 mg/m3.
bPAC-l = PAC-1 t11 = 5.5 mg/m3 t11 = 0.5 mg/m3.
cRfC = RfD x BW t- inhalation rate = 0.0029 mg/kg-day x 70 kg -f 20 m3/day = 0.01 mg/m3.
dRfC = RfD x BW 4 inhalation rate = 0.000015 mg/kg-day x 70 kg 4 20 m3/day = 0.000053 mg/m3.
6N0AELHec = NOAEL 4 TK adjustment factor = 67 mg/m3 4 81 = 0.83 mg/m3.
fRfC = RfD x BW 4 inhalation rate = 0.0000038 mg/kg-day x 70 kg 4 20 m3/day = 0.000013 mg/m3.
This document is a draft for review purposes only and does not constitute Agency policy.
A-2 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
BW = body weight; HEC = human equivalent concentration; NOAEL = no-observed-adverse-effect level; NR = not reported; PAC = protective action criteria;
PFAS = per- and polyfluoroalkyl substances; PFBA = perfluorobutanoic acid; PFDA = perfluorodecanoic acid; PFHxS = Perfluorohexanesulfonic acid;
PFNA = perfluorononanoic acid; RfC = inhalation reference concentration; RfD = oral reference dose; TCEQ = Texas Commission on Environmental Quality;
TK = toxicokinetic; UFA = interspecies uncertainty factor; UFC = composite uncertainty factor; UFD = database uncertainty factor; UFH = intraspecies uncertainty
factor; UFL = LOAEL-to-NOAEL uncertainty factor; UFS = subchronic-to-chronic uncertainty factor.
This document is a draft for review purposes only and does not constitute Agency policy.
A-3 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table A-2. Details on derivation of the available health effect reference values for oral exposure to selected
per- and polyfluoroalkyl substances (PFAS) (current as of June 2019; please consult source references for
up-to-date information)
Value
name
Duration
PFAS
Value
(mg/kg-d)
Health effect
Point of
departure
Qualifier
Source
Uncertainty
factors
Notes on
derivation
Review
status
MDH RfD
1-30 d
PFBA
3.8 x 10"3
Decreased
cholesterol,
serum total
thyroxine, and
dialysis free
thyroxine and
increased
relative thyroid
weight in rats
3.01 mg/kg-d
0.38 mg/kg-d
BMDLisd
BMDLhed
van
Otterdiik
(2007a)
UFC = 100
UFa = 3
UFh = 10
UFd = 3
HED adjusted3
Final
(MDH,
2018)
Subchronic
2.9 x 10"3
Liver-weight
changes;
morphological
changes in the
liver and thyroid
6.9 mg/kg-d
0.86 mg/kg-d
NOAEL
NOAELhed
van
Otterdiik
(2007b)
UFc = 300
UFA = 3
UFh = 10
UFD = 10
HED adjusted15
Chronic
2.9 x 10"3
gland; and
decreased T4,
RBCs,
hematocrit, and
Hb in rats
1-30 d
PFHxS
9.7 x 10"6
Decreased free
and total T4 and
triiodothyronine
(T3), changes in
cholesterol
levels, and
increased
hepatic focal
necrosis in rats
32.4 mg/L serum
0.00292 mg/kg-d
BMDLzo
BMDLhed
NTP
(2019)
UFC = 300
UFa = 3
UFh = 10
UFd = 10
HED adjusted0
Final
(MDH,
2019)
This document is a draft for review purposes only and does not constitute Agency policy.
A-4 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Value
name
Duration
PFAS
Value
(mg/kg-d)
Health effect
Point of
departure
Qualifier
Source
Uncertainty
factors
Notes on
derivation
Review
status
NH DES
RfD
Chronic
PFHxS
4.0 x 10"6
Reduced litter
size in mice
exposed for 14 d
13,900 ng/L
serum
46.3 ng/mL
BMDL
Target
human
serum
level
Chang et
al. (2018)
UFc = 300
UFa = 3
UFh = 10
UFs = 3
UFd = 3
Target human
serum level =
BMDL-^UF
Calculatedd
Final
(New
Hampshire
DES, 2019)
PFNA
4.3 x 10"6
Increased
relative liver
weights in mice
exposed for 17 d
4,900 ng/L serum
49.0 ng/mL serum
BMDL
Target
human
serum
level
Das et al.
(2015)
UFC = 100
UFA = 3
UFh = 10
UFD = 3
Target human
serum level =
BMDL-^UF
Calculated6
TCEQ RfD
Chronic
PFBA
2.9 x 10"3
Liver-weight
changes;
morphological
changes in the
liver and thyroid
gland; and
decreased T4,
RBCs,
hematocrit, and
Hb in rats
6.9 mg/kg-d
0.86 mg/kg-d
NOAEL
NOAELhed
van
Otterdiik
(2007b)
UFC = 300
UFh = 10
UFS = 3
UFd = 10
HED adjusted'
Final
(TCEQ,
2016)
PFDA
1.5 x 10"5
Increased liver
weight in rats
dosed for 1 wk
1.2 mg/kg-d
0.015 mg/kg-d
NOAEL
NOAELhed
Kawashi
ma et al.
(1995)
UFC = 1,000
UFh = 10
UFS = 10
UFd = 10
HED adjusted8
PFHxS
3.8 x 10"6
Hematological
alterations in
male rats
0.3 mg/kg-d
0.0011 mg/kg-d
LOAEL
LOAELhed
3M
(2003)
UFC = 300
UFh = 10
UFl = 3
UFd = 10
HED adjustedh
This document is a draft for review purposes only and does not constitute Agency policy.
AS DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Value
name
Duration
PFAS
Value
(mg/kg-d)
Health effect
Point of
departure
Qualifier
Source
Uncertainty
factors
Notes on
derivation
Review
status
PFHxA
3.8 x 10"6
Adopted RfD for
PFHxS
-
-
-
-
Adopted RfD
for PFHxS
PFNA
1.2 x 10"5
Spleen cell
apoptosis in rats
1 mg/kg-d
0.012 mg/kg-d
NOAEL
NOAELhed
Fang et
al. (2010)
UFC = 1,000
UFh = 10
UFS = 10
UFd = 10
HED adjusted1
Australia
Dept. of
Health TDI
Chronic
Combined
PFOS and
PFHxS
2 x 10"5
Decreased
body-weight gain
in F0 female rats
0.1 mg/kg-d,
7.14 ng/mL
0.0006 mg/kg-d
NOAEL
NOAELhed
Luebker
et al.
(2005)
UFC = 30
UFa = 3
UFh = 10
HED adjusted'
Final
(FSANZ,
2016)
aBMDLHED = BMDLisd 4 (t1/2 Human +t1/2 Male Rat) = 3.01 mg/kg-day 4 (72 h 4 9.22 h) = 0.38 mg/kg-day.
bNOAEI_HED = NOAEL-^ (ti/2 Human + ti/2 Male Rat) = 6.9 mg/kg-day 4 (72 h 4 9.22 h) = 0.86 mg/kg-day.
cBMDLHed = BMDL x volume of distribution x (In2 -f ti/2) = 32.4 mg/L x 0.25 L/kg x (0.693 4 1,935 days) = 0.00292 mg/kg-day.
dRfD = THSL x volume of distribution x (In2 4 ti/2) = 46.3 ng/mL x 213 mL/kg x (0.693 4 1,716 days) =4.0 ng/kg-day.
eRfD = THSL x volume of distribution x (In2 4 ti/2) = 49.0 ng/mL x 200 mL/kg x (0.693 4 1,570 days) = 4.3 ng/kg-day.
'NOAELhed = NOAEL 4 TK adjustment factor = 6.9 mg/kg-day 4 8 = 0.86 mg/kg-day.
8NOAELHed = NOAEL 4 TK adjustment factor =1.2 mg/kg-day 4 81 = 0.015 mg/kg-day.
hLOAELHED = LOAEL 4 TK adjustment factor = 0.3 mg/kg-day 4 263 = 0.0011 mg/kg-day.
'NOAELhed = NOAEL 4 TK adjustment factor = 1 mg/kg-day 4 81 = 0.012 mg/kg-day.
JTDI = NOAEL x volume of distribution x (In2 4 ti/2) = 7.14 |Jg/mL x 0.23 L/kg x (0.693 4 1,971 days) = 0.0006 mg/kg-day.
BMDL = benchmark dose lower confidence limit; HED = human equivalent dose; LOAEL = lowest-observed-adverse-effect level; MDH = Minnesota Department
of Health; NH DES = New Hampshire Department of Environmental Services; NOAEL = no-observed-adverse-effect level; PFAS = per- and polyfluoroalkyl
substances; PFBA = perfluorobutanoic acid; PFDA = perfluorodecanoic acid; PFHxA = perfluorohexanoic acid; PFHxS = Perfluorohexanesulfonic acid;
PFNA = perfluorononanoic acid; PFOS = perfluorooctane sulfonate; RBC = red blood cell; RfD = oral reference dose; SD = standard deviation; TCEQ = Texas
Commission on Environmental Quality; TDI = tolerable daily intake; THSL = target human serum level; TK = toxicokinetic; UFA = interspecies uncertainty factor;
UFc = composite uncertainty factor; UFD = database uncertainty factor; UFH = intraspecies uncertainty factor; UFL = LOAEL-to-NOAEL uncertainty factor;
UFs = subchronic-to-chronic uncertainty factor.
This document is a draft for review purposes only and does not constitute Agency policy.
A-6 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table A-3. Details on derivation of PFOA and PFOS reference values which served as the basis for values for the
five per- and polyfluoroalkyl substances (PFAS) of interest (current as of June 2019; please consult source
references for up-to-date information)
Value
name
Duration
PFAS
Value
(mg/kg-d)
Health effect
Point of
departure
Qualifier
Source
Uncertainty
factors
Notes on
derivation
Review
status
EPA RfD
(OW)
Chronic
PFOA
2 x 10"5
Decreased
ossification
and
accelerated
male puberty
in Fi mice
1 mg/kg-d,
38 mg/L serum
0.0053 mg/kg-d
LOAEL
LOAELhed
Lau et al.
(2006)
UFc = 300
UFa = 3
UFh = 10
UFL= 10
UFs = 1
UFd = 1
Average serum
concentration
derived using a
PBPK model
developed by
Wambaugh et
al. (2013)
HED adjusted3
Final
(U.S. EPA,
2016b)
PFOS
2 x 10"5
Reduced body
weight in F2
rats
0.1 mg/kg-d,
6.26 Hg/mL
0.00051 mg/kg-d
NOAEL
NOAELhed
Luebker
et al.
(2005)
UFC = 30
UFA = 3
UFh = 10
UFL= 1
UFS = 1
UFd = 1
Average serum
concentration
derived using a
PBPK model
developed by
Wambaugh et
al. (2013)
HED adjusted15
Final
(U.S. EPA,
2016a)
Danish EPA
TDI
Chronic
PFOS
3 x 10"5
Liver lesions in
male rats
0.033 mg/kg-d
0.0008 mg/kg-d
BMDLio
BMDLhed
Thomford
(2002)
UFC = 30
UFa = 3
UFh = 10
Pharmaco-
kinetic
adjustments
based on
those in U.S.
EPA (2014a)
HED adjusted0
Final
(Danish
EPA, 2015)
aLOAELHED = LOAEL x volume of distribution x (In2 -f ti/2) = 38 mg/L x 0.17 L/kg x (0.693 4 839.5 days) = 0.0053 mg/kg-day.
bNOAELHED = NOAEL x volume of distribution x (In2 4 ti/2) = 6.26 ng/mL x 0.23 L/kg x (0.693 4 1,971 days) = 0.00051 mg/kg-day.
This document is a draft for review purposes only and does not constitute Agency policy.
A-7 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
cBMDLHed = BMDLio t- ([volume of distribution x (In2 -f ti/2Rat)] + [volume of distribution x (In2 4 ti/2 Human)]) = 0.033 mg/kg-day -f
([0.23 L/kg x (0.693 4 48 days)] 4 [0.23 L/kg x (0.693 4 1,971 days)]) = 0.0008 mg/kg-day.
BMDL = benchmark dose lower confidence limit; EPA = Environmental Protection Agency; HED = human equivalent dose; LOAEL = lowest-observed-adverse-
effect level; NOAEL= no-observed-adverse-effect level; OW = Office of Water; PBPK= physiologically based pharmacokinetic; PFAS = per- and polyfluoroalkyl
substances; PFOA = perfluorooctanoic acid; PFOS = perfluorooctane sulfonate; RfD = oral reference dose; TDI = tolerable daily intake; UFA = interspecies
uncertainty factor; UFC = composite uncertainty factor; UFD = database uncertainty factor; UFH = intraspecies uncertainty factor; UFL = LOAEL-to-NOAEL
uncertainty factor; UFS = subchronic-to-chronic uncertainty factor.
This document is a draft for review purposes only and does not constitute Agency policy.
A-8 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
ADDENDUM B. SEARCH AND SCREENING
STRATEGIES
Table B-l. Perfluorobutanoic acid (PFBA) database search strategy
Search
Search strategy
Dates of search
PubMed
Search
terms
375-22-4[rn] OR "Heptafluoro-l-butanoic acid"[tw] OR "Heptafluorobutanoic
acid"[tw] OR "Heptafluorobutyric acid"[tw] OR "Kyselina
heptafluormaselna"[tw] OR "Perfluorobutanoic acid"[tw] OR
"Perfluorobutyric acid"[tw] OR "Perfluoropropanecarboxylic acid"[tw] OR
"2,2,3,3,4,4,4-heptafluoro-Butanoic acid"[tw] OR "Butanoic acid,
2,2,3,3,4,4,4-heptafluoro-"[tw] OR "Butanoic acid, heptafluoro-"[tw] OR
"Perfluoro-n-butanoic acid"[tw] OR "Perfluorobutanoate"[tw] OR
"2,2,3,3,4,4,4-Heptafluorobutanoic acid"[tw] OR "Butyric acid,
heptafluoro-"[tw] OR "Fluorad FC 23"[tw] OR "H 0024"[tw] OR "NSC 820"[tw]
OR «PFBA[tw] OR "FC 23"[tw] OR HFBA[tw]) AND (fluorocarbon*[tw] OR
fluorotelomer*[tw] OR polyfluoro*[tw] OR perfluoro-*[tw] OR
perfluoroa*[tw] OR perfluorob*[tw] OR perfluoroc*[tw] OR perfluorod*[tw]
OR perfluoroe*[tw] OR perfluoroh*[tw] OR perfluoron*[tw] OR
perfluoroo*[tw] OR perfluorop*[tw] OR perfluoros*[tw] OR perfluorou*[tw]
OR perfluorinated[tw] OR fluorinated[tw] OR PFAS[tw] OR PFOS[tw] OR
PFOA[tw]))
No date
limit—7/19/2017
Literature
update
search
terms
(((375-22-4[rn] OR "Heptafluoro-l-butanoic acid"[tw] OR
"Heptafluorobutanoic acid"[tw] OR "Heptafluorobutyric acid"[tw] OR
"Kyselina heptafluormaselna"[tw] OR "Perfluorobutanoic acid"[tw] OR
"Perfluorobutyric acid"[tw] OR "Perfluoropropanecarboxylic acid"[tw] OR
"2,2,3,3,4,4,4-heptafluoro-Butanoic acid"[tw] OR "Butanoic acid,
2,2,3,3,4,4,4-heptafluoro-"[tw] OR "Butanoic acid, heptafluoro-"[tw] OR
"Perfluoro-n-butanoic acid"[tw] OR "Perfluorobutanoate"[tw] OR
"2,2,3,3,4,4,4-Heptafluorobutanoic acid"[tw] OR "Butyric acid,
heptafluoro-"[tw] OR "Fluorad FC 23"[tw] OR "H 0024"[tw] OR "NSC 820"[tw]
OR «PFBA[tw] OR "FC 23"[tw] OR HFBA[tw]) AND (fluorocarbon*[tw] OR
fluorotelomer*[tw] OR polyfluoro*[tw] OR perfluoro-*[tw] OR
perfluoroa*[tw] OR perfluorob*[tw] OR perfluoroc*[tw] OR perfluorod*[tw]
OR perfluoroe*[tw] OR perfluoroh*[tw] OR perfluoron*[tw] OR
perfluoroo*[tw] OR perfluorop*[tw] OR perfluoros*[tw] OR perfluorou*[tw]
OR perfluorinated[tw] OR fluorinated[tw] OR PFAS[tw] OR PFOS[tw] OR
PFOA[tw])) AND ("2017/08/01"[PDAT] : "2018/02/14"[PDAT])
8/1/2017-2/14/2018
Web of Science
This document is a draft for review purposes only and does not constitute Agency policy.
B-l DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Search
Search strategy
Dates of search
Search
terms
TS="Heptafluoro-l-butanoic acid" OR TS="Heptafluorobutanoic acid" OR
TS="Heptafluorobutyric acid" ORTS="Kyselina heptafluormaselna" OR
TS="Perfluorobutanoic acid" OR TS="Perfluorobutyric acid" OR
TS="Perfluoropropanecarboxylic acid" OR
TS="2,2,3,3,4,4,4-heptafluoro-Butanoic acid" OR TS="Butanoic acid,
2,2,3,3,4,4,4-heptafluoro-" ORTS="Butanoic acid, heptafluoro-" OR
TS="Perfluoro-n-butanoic acid" ORTS="Perfluorobutanoate" OR
TS="2,2,3,3,4,4,4-Heptafluorobutanoic acid" OR TS="Butyric acid,
heptafluoro-" ORTS="Fluorad FC 23" ORTS="H 0024" ORTS="NSC 820" OR
(TS=(PFBA OR "FC 23" OR HFBA) AND TS=(fluorocarbon* OR fluorotelomer*
OR polyfluoro* OR perfluoro-* OR perfluoroa* OR perfluorob* OR
perfluoroc* OR perfluorod* OR perfluoroe* OR perfluoroh* OR perfluoron*
OR perfluoroo* OR perfluorop* OR perfluoros* OR perfluorou* OR
perfluorinated OR fluorinated OR PFAS OR PFOS OR PFOA))
No date
limit-7/20/2017
Literature
update
search
terms
((TS="Heptafluoro-l-butanoic acid" OR TS="Heptafluorobutanoic acid" OR
TS="Heptafluorobutyric acid" ORTS="Kyselina heptafluormaselna" OR
TS="Perfluorobutanoic acid" OR TS="Perfluorobutyric acid" OR
TS="Perfluoropropanecarboxylic acid" OR
TS="2,2,3,3,4,4,4-heptafluoro-Butanoic acid" OR TS="Butanoic acid,
2,2,3,3,4,4,4-heptafluoro-" ORTS="Butanoic acid, heptafluoro-" OR
TS="Perfluoro-n-butanoic acid" ORTS="Perfluorobutanoate" OR
TS="2,2,3,3,4,4,4-Heptafluorobutanoic acid" OR TS="Butyric acid,
heptafluoro-" ORTS="Fluorad FC 23" ORTS="H0024" ORTS="NSC 820") OR
TS=(PFBA OR "FC 23" OR HFBA) AND TS=(fluorocarbon* OR fluorotelomer* OR
polyfluoro* OR perfluoro-* OR perfluoroa* OR perfluorob* OR perfluoroc*
OR perfluorod* OR perfluoroe* OR perfluoroh* OR perfluoron* OR
perfluoroo* OR perfluorop* OR perfluoros* OR perfluorou* OR
perfluorinated OR fluorinated OR PFAS OR PFOS OR PFOA)) AND
PY=2017-2018
2017-2018
Toxline
Search
terms
( 375-22-4 [rn] OR "heptafluoro- 1-butanoic acid" OR "heptafluorobutanoic
acid" OR "heptafluorobutyric acid" OR "kyselina heptafluormaselna" OR
"perfluorobutanoic acid" OR "perfluorobutyric acid" OR
"perfluoropropanecarboxylic acid" OR "2,2,3,3,4,4,4-heptafluoro-butanoic
acid" OR "butanoic acid 2 2 3 3 4 4 4-heptafluoro-" OR "butanoic acid
heptafluoro-" OR "perfluoro-n-butanoic acid" OR "perfluorobutanoate" OR
"2,2,3,3,4,4,4-heptafluorobutanoic acid" OR "butyric acid heptafluoro-" OR
"fluorad fc 23" OR "h 0024" OR "nsc 820" OR (( pfba OR "fc 23" OR hfba ) AND
(fluorocarbon* OR fluorotelomer* OR polyfluoro* OR perfluoro* OR
perfluorinated OR fluorinated OR pfas OR pfos OR pfoa ))) AND (ANEUPL
[org] OR BIOSIS [org] OR CIS [org] OR DART [org] OR EMIC [org] OR EPIDEM
[org] OR HEEP [org] OR HMTC [org] OR IPA [org] OR RISKUNE [org] OR
MTGABS [org] OR NIOSH [org] OR NTIS [org] OR PESTAB [org] OR PPBIB [org] )
AND NOT PubMed [org] AND NOT pubdart [org]
No date
limit-7/20/2017
This document is a draft for review purposes only and does not constitute Agency policy.
B-2 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Search
Search strategy
Dates of search
Literature
update
search
terms
@AND+@OR+("heptafluoro-l-butanoic
acid"+"heptafluorobutanoic+acid"+"heptafluorobutyric+acid"+"kyselina+hept
afluormaselna"+"perfluorobutanoic+acid"+"perfluorobutyric+acid"+"perfluor
opropanecarboxylic +acid"+"2 2 3 3 4 4
4-heptafluoro-butanoic+acid"+"butanoic+acid+2 2 3 3 4 4
4-heptafluoro-"+"butanoic+acid+heptafluoro-"+"perfluoro-n-butanoic
acid"+"perfluorobutanoate"+"2 2 3 3 4 4
4-heptafluorobutanoic+acid"+"butyric+acid+heptafluoro-"+"fluorad+fc+23"+"
h0024"+"nsc+820"+@TERM+@rn+375-22-4("pfba"+"fc+23"+"hfba"))+(
fluorocarbon*+
fluorotelomer*+polyfluoro*+perfluoro*+perfluorinated+fluorinated+pfas+pfo
s+pfoa)+@RANGE+yr+2017+2018
2017-2018
TSCATS
Search
terms
375-22-4[rn] AND tscats[org]
No date
limit-7/20/2017
This document is a draft for review purposes only and does not constitute Agency policy.
B-3 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table B-2. Perfluorodecanoic acid (PFDA) database search strategy
Search
Search strategy
Dates of search
PubMed
Search
terms
335-76-2[rn] OR "Ndfda"[tw] OR "Nonadecafluoro-n-decanoic acid"[tw] OR
"Nonadecafluorodecanoic acid"[tw] OR "Perfluoro-n-decanoic acid"[tw] OR
"Perfluorodecanoic acid"[tw] OR
"2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-nonadecafluoro-Decanoic acid"[tw]
OR "Decanoic acid,
2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-nonadecafluoro-"[tw] OR "Decanoic
acid, nonadecafluoro-"[tw] OR "Perfluorodecanoate"[tw] OR "PFDeA"[tw] OR
"PFDcA"[tw] OR ("PFDA"[tw] AND (fluorocarbon*[tw]
OR fluorotelomer*[tw] OR polyfluoro*[tw] OR perfluoro-*[tw] OR
perfluoroa*[tw] OR perfluorob*[tw] OR perfluoroc*[tw] OR perfluorod*[tw]
OR perfluoroe*[tw] OR perfluoroh*[tw] OR perfluoron*[tw] OR
perfluoroo*[tw] OR perfluorop*[tw] OR perfluoros*[tw] OR perfluorou*[tw]
OR perfluorinated[tw] OR fluorinated[tw] OR PFAS[tw] OR PFOS[tw] OR
PFOA[tw]))
No date
limit—7/26/2017
Literature
update
search
terms
((335-76-2[rn] OR "Ndfda"[tw] OR "Nonadecafluoro-n-decanoic acid"[tw] OR
"Nonadecafluorodecanoic acid"[tw] OR "Perfluoro-n-decanoic acid"[tw] OR
"Perfluorodecanoic acid"[tw] OR
"2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-nonadecafluoro-Decanoic acid"[tw]
OR "Decanoic acid,
2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-nonadecafluoro-"[tw] OR "Decanoic
acid, nonadecafluoro-"[tw] OR "Perfluorodecanoate"[tw] OR "PFDeA"[tw] OR
"PFDcA"[tw] OR ("PFDA"[tw] AND (fluorocarbon*[tw] OR fluorotelomer*[tw]
OR polyfluoro*[tw] OR perfluoro-*[tw] OR perfluoroa*[tw] OR
perfluorob*[tw] OR perfluoroc*[tw] OR perfluorod*[tw] OR perfluoroe*[tw]
OR perfluoroh*[tw] OR perfluoron*[tw] OR perfluoroo*[tw] OR
perfluorop*[tw] OR perfluoros*[tw] OR perfluorou*[tw] OR
perfluorinated[tw] OR fluorinated[tw] OR PFAS[tw] OR PFOS[tw] OR
PFOA[tw])) AND ("2017/08/01"[Date - Publication] :
"2018/03/01"[Date - Publication])
8/1/2017-2/14/2018
Web of Science
This document is a draft for review purposes only and does not constitute Agency policy.
B-4 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Search
Search strategy
Dates of search
Search
terms
TS="PFDeA" ORTS="PFDcA" ORTS="Ndfda" OR
TS="Nonadecafluoro-n-decanoic acid" OR TS="Nonadecafluorodecanoic acid"
OR TS="Perfluoro-n-decanoic acid" OR TS="Perfluorodecanoic acid" OR
TS="2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-nonadecafluoro-Decanoic acid"
OR TS="Decanoic acid,
2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-nonadecafluoro-" OR TS="Decanoic
acid, nonadecafluoro-" ORTS="Perfluorodecanoate" OR (TS=PFDA AND
TS=(fluorocarbon* OR fluorotelomer* OR polyfluoro* OR perfluoro-* OR
perfluoroa* OR perfluorob* OR perfluoroc* OR perfluorod* OR perfluoroe*
OR perfluoroh* OR perfluoron* OR perfluoroo* OR perfluorop* OR
perfluoros* OR perfluorou* OR perfluorinated OR fluorinated)) OR (TS=PFDA
AND TS=(fluorocarbon* OR fluorotelomer* OR polyfluoro* OR perfluoro-* OR
perfluoroa* OR perfluorob* OR perfluoroc* OR perfluorod* OR perfluoroe*
OR perfluoroh* OR perfluoron* OR perfluoroo* OR perfluorop* OR
perfluoros* OR perfluorou* OR perfluorinated OR fluorinated OR PFAS OR
PFOS OR PFOA
No date
limit—7/26/2017
Literature
update
search
terms
TS="PFDeA" ORTS="PFDcA" ORTS="Ndfda" OR
TS="Nonadecafluoro-n-decanoic acid" OR TS="Nonadecafluorodecanoic acid"
OR TS="Perfluoro-n-decanoic acid" OR TS="Perfluorodecanoic acid" OR
TS="2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-nonadecafluoro-Decanoic acid"
OR TS="Decanoic acid,
2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-nonadecafluoro-" OR TS="Decanoic
acid, nonadecafluoro-" ORTS="Perfluorodecanoate" OR (TS=PFDA AND
TS=(fluorocarbon* OR fluorotelomer* OR polyfluoro* OR perfluoro-* OR
perfluoroa* OR perfluorob* OR perfluoroc* OR perfluorod* OR perfluoroe*
OR perfluoroh* OR perfluoron* OR perfluoroo* OR perfluorop* OR
perfluoros* OR perfluorou* OR perfluorinated OR fluorinated)) OR (TS=PFDA
AND TS=(fluorocarbon* OR fluorotelomer* OR polyfluoro* OR perfluoro-* OR
perfluoroa* OR perfluorob* OR perfluoroc* OR perfluorod* OR perfluoroe*
OR perfluoroh* OR perfluoron* OR perfluoroo* OR perfluorop* OR
perfluoros* OR perfluorou* OR perfluorinated OR fluorinated OR PFAS OR
PFOS OR PFOA)) AND PY=2017-2018
2017-2018
Toxline
Search
terms
(335-76-2 [rn] OR
"2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-nonadecafluorodecanoic acid" OR
"2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-nonadecafluoro-decanoic acid" OR
"decanoic acid 2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,10-nonadecafluoro-" OR
"decanoic acid nonadecafluoro-" OR "nonadecafluoro-n-decanoic acid" OR
"nonadecafluorodecanoic acid" OR "perfluoro-1-nonanecarboxylic acid" OR
"perfluoro-n-decanoic acid" OR "perfluorocapric acid" OR
"perfluorodecanoate" OR "perfluorodecanoic acid" OR "ndfda" OR "PFDeA"
OR "PFDcA" OR ( pfda AND (fluorocarbon* OR fluorotelomer* OR polyfluoro*
OR perfluoro* OR perfluorinated OR fluorinated OR pfas OR pfos OR pfoa )))
AND ( ANEUPL [org] OR BIOSIS [org] OR CIS [org] OR DART [org] OR EMIC [org]
OR EPIDEM [org] OR HEEP [org] OR HMTC [org] OR IPA [org] OR RISKUNE [org]
OR MTGABS [org] OR NIOSH [org] OR NTIS [org] OR PESTAB [org] OR PPBIB
[org]) AND NOT PubMed [org] AND NOT pubdart [org]
No date
limit—7/21/2017
This document is a draft for review purposes only and does not constitute Agency policy.
B-5 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Search
Search strategy
Dates of search
Literature
update
search
terms
2017-2018
TSCATS
Search
terms
335-76-2[rn] AND TSCATS[org]
No date
limit—7/21/2017
This document is a draft for review purposes only and does not constitute Agency policy.
B-6 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table B-3. Perfluorononanoic acid (PFNA) database search strategy
Search
Search strategy
Dates of search
PubMed
Search
terms
"375-95-1"[rn] OR "2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,9-heptadecafluorononanoic
acid"[tw] OR "Nonanoic acid,
2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,9-heptadecafluoro-"[tw] OR "Nonanoic acid,
heptadecafluoro-"[tw] OR "Perfluoro-n-nonanoic acid"[tw] OR
"Perfluorononan-l-oic acid"[tw] OR "Perfluorononanoate"[tw] OR
"Perfluorononanoic acid"[tw] OR "Perfluorononanonic acid"[tw] OR
"Perfluoropelargonic acid"[tw] OR "heptadecafluorononanoic acid"[tw] OR
(("PFNA"[tw] OR "C 1800"[tw]) AND (fluorocarbon*[tw] OR fluorotelomer*[tw]
OR polyfluoro*[tw] OR perfluoro-*[tw] OR perfluoroa*[tw] OR perfluorob*[tw]
OR
perfluoroc*[tw] OR perfluorod*[tw] OR perfluoroe*[tw] OR perfluoroh*[tw]
OR perfluoron*[tw] OR perfluoroo*[tw] OR perfluorop*[tw] OR
perfluoros*[tw] OR perfluorou*[tw] OR perfluorinated[tw] OR fluorinated[tw]
OR PFAS[tw] OR PFOS[tw] OR PFOA[tw]))
No date
limit—7/26/2017
Literature
update
and
additional
PFNA
synonyms
search
terms
((("2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,9-heptadecafluorononanoic acid" [tw] OR
"Nonanoic acid, 2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,9-heptadecafluoro-" [tw] OR
"Nonanoic acid, heptadecafluoro-" [tw] OR "Perfluoro-n-nonanoic acid" [tw]
OR "Perfluorononan-l-oic acid" [tw] OR "Perfluorononanoate" [tw] OR
"Perfluorononanoic acid" [tw] OR "Perfluorononanonic acid" [tw] OR
"Perfluoropelargonic acid" [tw] OR "heptadecafluorononanoic acid" [tw] OR
"PFNA" [tw] OR "C 1800" [tw] OR "Methyl-nl-Perfluorononanoic acid" [tw] OR
"PFNA-nlCH3" [tw] OR "EINECS 206-801-3" [tw] OR
"Heptadecafluornonansaeure" [tw] OR "Heptadekafluornonansaeure" [tw] OR
"Perfluornonansaeure" [tw] OR "Perfluorononanoic acid (PFNA)" [tw] OR
"UNII-5830Z6S63M" [tw] OR "perfluoro-n-nonanoic acid" [tw] OR
"perfluorononan-l-oic acid" [tw] OR "perfluorononanoic acid" [tw] OR
"Ammonium Perfluorononanoate" [tw] OR "Ammonium perfluorononanoate"
[tw] OR "PFNA-H3N" [tw]))) AND ("2017/01/01"[Date - Publication] :
"3000"[Date - Publication])
1/2017-4/2018
Web of Science
Search
terms
((TS=PFNA OR TS="C 1800") AND TS=(fluorocarbon* OR fluorotelomer* OR
polyfluoro* OR perfluoro-* OR perfluoroa* OR perfluorob* OR perfluoroc* OR
perfluorod* OR perfluoroe* OR perfluoroh* OR perfluoron* OR perfluoroo*
OR perfluorop* OR perfluoros* OR perfluorou* OR perfluorinated OR
fluorinated OR PFAS OR PFOS OR PFOA)) OR
TS="2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,9-heptadecafluorononanoic acid" OR
TS="Nonanoic acid, 2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,9-heptadecafluoro-" OR
TS="Nonanoic acid, heptadecafluoro-" OR TS="Perfluoro-n-nonanoic acid" OR
TS="Perfluorononan-l-oic acid" ORTS="Perfluorononanoate" OR
TS="Perfluorononanoic acid" OR TS="Perfluorononanonic acid" OR
TS="Perfluoropelargonic acid" OR TS="heptadecafluorononanoic acid"
No date
limit—7/26/2017
This document is a draft for review purposes only and does not constitute Agency policy.
B-7 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Search
Search strategy
Dates of search
Literature
update
and
additional
PFNA
synonyms
search
terms
(TS="PFNA" OR TS="C 1800" OR
TS="2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,9-heptadecafluorononanoic acid" OR
TS="Nonanoic acid, 2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,9-heptadecafluoro-" OR
TS="Nonanoic acid, heptadecafluoro-" OR TS="Methyl-nl-Perfluorononanoic
acid" OR TS="PFNA-nlCH3" OR TS="EINECS 206-801-3" OR
TS="Heptadecafluornonansaeure" OR TS="Heptadekafluornonansaeure" OR
TS="Perfluornonansaeure" OR TS="Perfluorononanoic acid (PFNA)" OR
TS="UNII-5830Z6S63M" OR TS="perfluoro-n-nonanoic acid" OR
TS="perfluorononan-l-oic acid" OR TS="perfluorononanoic acid" OR
TS="Ammonium Perfluorononanoate" ORTS="Ammonium
perfluorononanoate" ORTS="PFNA-H3N") AND PY=2017-2018
2017-2018
Toxline
Search
terms
(( pfna OR "c 1800") AND (fluorocarbon* OR fluorotelomer* OR polyfluoro*
OR perfluoro* OR perfluorinated OR fluorinated OR pfas OR pfos OR pfoa ) OR
"375-95-1" [rn] OR "2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,9-heptadecafluorononanoic
acid" OR "nonanoic acid 2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,9-heptadecafluoro-" OR
"nonanoic acid heptadecafluoro-" OR "perfluoro-n-nonanoic acid" OR
"perfluorononan-l-oic acid" OR "perfluorononanoate" OR "perfluorononanoic
acid" OR "perfluorononanonic acid" OR "perfluoropelargonic acid" OR
"heptadecafluorononanoic acid")) AND (ANEUPL [org] OR BIOSIS [org] OR CIS
[org] OR DART [org] OR EMIC [org] OR EPIDEM [org] OR HEEP [org] OR HMTC
[org] OR IPA [org] OR RISKLINE [org] OR MTGABS [org] OR NIOSH [org] OR NTIS
[org] OR PESTAB [org] OR PPBIB [org] ) AND NOT PubMed [org] AND NOT
pubdart [org]
No date
limit—7/26/2017
Literature
update
and
additional
PFNA
synonyms
search
terms
@AND+@OR+(pfna+"c
1800"+fluorocarbon*+"2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,9-heptadecafluorononan
oic+acid"+"nonanoic+acid+2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,9-heptadecafluoro-"+
"nonanoic+acid+heptadecafluoro-"+"perfluoro-n-nonanoic+acid"+"perfluorono
nan-l-oic+acid"+perfluorononanoate+"perfluorononanoic+acid"+"perfluoropel
argonic+acid"+"heptadecafluorononanoic+acid"+"Methyl-nl-Perfluorononanoi
c+acid"+"PFNA-nlCH3"+"EINECS
206-801-3"+"Heptadecafluornonansaeure"+"Heptadekafluornonansaeure"+"P
erfluornonansaeure"+"Perfluorononanoic+acid
(PFNA)"+"UNII-5830Z6S63M"+"perfluoro-n-nonanoic+acid"+"perfluorononan-l
-oic+acid"+"perfluorononanoic+acid"+"Ammonium+Perfluorononanoate"+"Am
monium+perfluorononanoate"+"PFNA-H3N"+@TERM+@rn+375-95-l)+@RAN
GE+yr+2017+2018
2017-2018
TSCATS
Search
terms
"375-95-1" [rn] AND TSCATS[org]
No date
limit-7/20/2017
This document is a draft for review purposes only and does not constitute Agency policy.
B-8 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Search
Search strategy
Dates of search
Literature
@TERM+@rn+375-95-l+@RANGE+yr+2017+2018
2017-2018
update
and
additional
PFNA
synonyms
search
terms
This document is a draft for review purposes only and does not constitute Agency policy.
B-9 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table B-4. Perfluorohexanoic acid (PFHxA) database search strategy
Search
Search strategy
Dates of search
PubMed
Search
terms
((307-24-4[rn] OR "2,2,3,3,4,4,5,5,6,6,6-undecafluorohexanoic acid"[tw] OR
"2,2,3,3,4,4,5,5,6,6,6-undecafluoro-hexanoic acid"[tw] OR "hexanoic acid,
2,2,3,3,4,4,5,5,6,6,6-undecafluoro-"[tw] OR "hexanoic acid,
undecafluoro-"[tw] OR "perfluorohexanoic acid"[tw] OR
"perfluoro-l-pentanecarboxylic acid"[tw] OR "perfluorocaproic acid"[tw] OR
"perfluorohexanoate"[tw] OR "perfluorohexanoic acid"[tw] OR
"undecafluoro-l-hexanoic acid"[tw] OR "undecafluorocaproic acid"[tw] OR
"undecafluorohexanoic acid"[tw] OR "PFHxA"[tw])) AND
("2017/08/01"[PDAT] : "2018/02/28"[PDAT])
No date
limit—7/21/2017
Literature
update
search
terms
((92612-52-7[EC/RN Number]) OR 355-38-4[EC/RN Number]) OR
2062-98-8[EC/RN Number]) OR "PFHxA_ion"[tw]) OR
"Perfluorohexanoate"[tw]) OR "Hexanoyl fluoride,
2,2,3,3,4,4,5,5,6,6,6-undecafluoro-"[tw]) OR "Hexanoyl fluoride,
undecafluoro-"[tw]) OR "Perfluorohexanoyl fluoride"[tw]) OR
"Undecafluorohexanoyl fluoride"[tw]) OR
"Perfluoro(2-methyl-3-oxahexanoyl) fluoride"[tw]) OR "Propanoyl fluoride,
2,3,3,3-tetrafluoro-2-(l,l,2,2,3,3,3-heptafluoropropoxy)-" [tw]) OR
"Propanoyl fluoride, 2,3,3,3-tetrafluoro-2-(heptafluoropropoxy)-" [tw]) OR
"Propionyl fluoride, tetrafluoro-2-(heptafluoropropoxy)-" [tw]) OR
"2,2,3,3,4,4,5,5,6,6,6-Undecafluorohexanoic acid"[tw]) OR "EINECS
206-196-6"[tw]) OR "NSC 5213"[tw]) OR "Perfluoro-l-pentanecarboxylic
acid"[tw]) OR "Perfluoro-n-hexanoic acid"[tw]) OR "UNII-ZP34Q2220R"[tw])
OR "Undecafluorocaproic acid"[tw]) OR "Ammonium
Perfluorohexanoate"[tw]) OR "PFHxA-H3N"[tw]) OR "PFHxA-Na"[tw]) OR
"Sodium Perfluorohexanoate"[tw]))
8/1/2017-2/14/2018
Web of Science
Search
terms
((TS="2,2,3,3,4,4,5,5,6,6,6-undecafluorohexanoic acid" OR
TS="2,2,3,3,4,4,5,5,6,6,6-undecafluoro-hexanoic acid" OR TS="hexanoic acid,
2,2,3,3,4,4,5,5,6,6,6-undecafluoro-" OR TS="hexanoic acid, undecafluoro-"
OR TS="perfluorohexanoic acid" OR TS="perfluoro-l-pentanecarboxylic acid"
ORTS="perfluorocaproic acid" ORTS="perfluorohexanoate" OR
TS="perfluorohexanoic acid" OR TS="undecafluoro-l-hexanoic acid" OR
TS="undecafluorocaproic acid" OR TS="undecafluorohexanoic acid" OR
TS="PFHxA")) AND PY=2017-2018
No date
limit-7/24/2017
This document is a draft for review purposes only and does not constitute Agency policy.
B-10 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Search
Search strategy
Dates of search
Literature
update
search
terms
TS="PFHxA_ion" ORTS="Perfluorohexanoate" ORTS="Hexanoyl fluoride,
2,2,3,3,4,4,5,5,6,6,6-undecafluoro-" ORTS="Hexanoyl fluoride,
undecafluoro-" ORTS="Perfluorohexanoyl fluoride" OR
TS="Undecafluorohexanoyl fluoride" OR
TS="Perfluoro(2-methyl-3-oxahexanoyl) fluoride" ORTS="Propanoyl fluoride,
2,3,3,3-tetrafluoro-2-(l,l,2,2,3,3,3-heptafluoropropoxy)-" ORTS="Propanoyl
fluoride, 2,3,3,3-tetrafluoro-2-(heptafluoropropoxy)-" OR TS="Propionyl
fluoride, tetrafluoro-2-(heptafluoropropoxy)-" OR
TS="2,2,3,3,4,4,5,5,6,6,6-Undecafluorohexanoic acid" OR TS="EINECS
206-196-6" OR TS="NSC 5213" OR TS="Perfluoro-l-pentanecarboxylic acid"
ORTS="Perfluoro-n-hexanoic acid" ORTS="UNII-ZP34Q2220R" OR
TS="Undecafluorocaproic acid" OR TS="Undecafluorohexanoic acid" OR
TS="Ammonium Perfluorohexanoate" ORTS="PFHxA-H3N" OR
TS="PFHxA-Na" ORTS="Sodium Perfluorohexanoate"
2017-2018
Toxline
Search
terms
@AND+@OR+("2,2,3,3,4,4,5,5,6,6,6-undecafluorohexanoic+acid"+"2,2,3,3,4,
4,5,5,6,6,6-undecafluoro-hexanoic+acid"+"hexanoic+acid+2,2,3,3,4,4,5,5,6,6,
6-undecafluoro-"+"hexanoic+acid+undecafluoro-"+"perfluorohexanoic+acid"
+"perfluoro-l-pentanecarboxylic+acid"+"perfluorocaproic+acid"+"perfluoroh
exanoate"+"perfluorohexanoic
acid"+"undecafluoro-l-hexanoic+acid"+"undecafluorocaproic+acid"+"undec
afluorohexanoic+acid"+"pfhxa"+@TERM+@rn+(307+24+4)+@RANGE+yr+20
17+2018+@NOT+@org+"nih+reporter"
No date
limit—7/21/2017
Literature
update
search
terms
@AND+@OR+("PFHxA_ion"+"Perfluorohexanoate"+"Hexanoyl+fluoride,+2,2,
3,3,4,4,5,5,6,6,6-undecafluoro-"+"Hexanoyl+fluoride,+undecafluoro-"+"Perfl
uorohexanoyl+fluoride"+"Undecafluorohexanoyl+fluoride"+"Perfluoro(2-me
thyl-3-oxahexanoyl)+fluoride"+"Propanoyl+fluoride,+2,3,3,3-tetrafluoro-2-(l,
1,2,2,3,3,3-heptafluoropropoxy)-"+" Propanoyl+fluoride,+2,3,3,3-tetrafluoro-
2-(heptafluoropropoxy)-"+"Propionyl+fluoride,+tetrafluoro-2-(heptafluoropr
opoxy)-"+"2,2,3,3,4,4,5,5,6,6,6-Undecafluorohexanoic+acid"+"EINECS+206+l
96+6"+"NSC+5213"+"Perfluoro-l-pentanecarboxylic+acid"+"Perfluoro-n-hex
anoic+acid"+"UNII-ZP34Q2220R"+"Undecafluorocaproic+acid"+"Undecafluor
ohexanoic+acid"+"Ammonium+Perfluorohexanoate"+"PFHxA-H3N"+"PFHxA-
Na"+"Sodium+Perfluorohexanoate")+@NOT+@org+"nih+reporter"
@AND+@OR+(@TERM+@rn+92612+52+7+@TERM+@rn+355+38+4+@TER
M+@rn+2062+98+8)+@NOT+@org+"nih+reporter"
2017-2018
TSCATS
Search
terms
307-24-4[rn] AND tscats[org]
No date
limit-7/20/2017
Literature
update
search
terms
@AND+@OR+(@TERM+@rn+92612+52+7+@TERM+@rn+355+38+4+@TER
M+@rn+2062+98+8)+@org+tscat
This document is a draft for review purposes only and does not constitute Agency policy.
B-ll DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table B-5. Perfluorohexanesulfonate (PFHxS) database search strategy
Search
Search strategy
Dates of search
PubMed
Search
terms
108427-53-8[rn] OR 355-46-4[rn] OR
"1,1,2,2,3,3,4,4,5,5,6,6,6-Tridecafluorohexane-l-sulfonic acid"[tw] OR
"1,1,2,2,3,3,4,4,5,5,6,6,6-tridecafluoro-l-Hexanesulfonic acid"[tw] OR
"1-Hexanesulfonic acid, l,l,2,2,3,3,4,4,5,5,6,6,6-tridecafluoro-"[tw] OR
"1-Hexanesulfonic acid, tridecafluoro-"[tw] OR "1-Perfluorohexanesulfonic
acid"[tw] OR "Perfluoro-l-hexanesulfonate"[tw] OR "Perfluorohexane
sulfonic
acid"[tw] OR "Perfluorohexane-l-sulphonic acid"[tw] OR
"Perfluorohexanesulfonate"[tw] OR "Perfluorohexanesulfonic acid"[tw] OR
"Perfluorohexylsulfonate"[tw] OR "Tridecafluorohexanesulfonic acid"[tw] OR
"tridecafluoro-1-Hexanesulfonic acid"[tw] OR "PFHxS"[tw]
No date
limit—7/21/2017
Literature
update
search
terms
((108427-53-8[EC/RN Number]) OR 423-50-7[EC/RN Number]) OR
"1-Hexanesulfonic acid, 1,1,2,2,3,3,4,4,5,5,6,6,6-tridecafluoro-, ion(l-)"[tw])
OR "PFHxS ion(l-)"[tw]) OR "PFHxS_ion"[tw]) OR
"Perfluorohexanesulfonate"[tw]) OR "Tridecafluorohexane-l-sulfonate"[tw])
OR "perfluorohexyl sulfonate"[tw]) OR
"1,1,2,2,3,3,4,4,5,5,6,6,6-Tridecafluoro-l-hexanesulfonyl fluoride"[tw]) OR
"1-Hexanesulfonyl fluoride, l,l,2,2,3,3,4,4,5,5,6,6,6-tridecafluoro-"[tw]) OR
"1,1,2,2,3,3,4,4,5,5,6,6,6-Tridecafluoro-l-hexanesulfonic acid"[tw]) OR "EC
206-587-l"[tw]) OR "EINECS 206-587-l"[tw]) OR "PFHS"[tw]) OR
"Perfluorhexan-l-sulfonsaure"[tw]) OR "Perfluorohexane sulfonic acid
(PFHxS)"[tw]) OR "Perfluorohexane-l-sulphonic acid"[tw]) OR "acide
perfluorohexane-l-sulfonique"[tw]) OR "acido
perfluorohexano-l-sulfonico"[tw]) OR "perfluorohexane-l-sulphonic
acid"[tw]) OR "perfluorohexanesulfonic acid"[tw]) OR "Ammonium
Perfluorohexanesulfonate"[tw]) OR "Ammonium
perfluorohexanesulfonate"[tw]) OR "PFHxS-H3N"[tw]) OR "PFHxS-K"[tw]) OR
"Potassium Perfluorohexanesulfonate"[tw]) OR "Potassium
perfluorohexanesulfonate"[tw]) OR "Lithium Perfluorohexanesulfonate"[tw])
OR "Lithium perfluorohexanesulfonate"[tw]) OR "PFHxS-Li"[tw]))
8/1/2017-2/14/2018
Web of Science
Search
terms
TS="l,l,2,2,3,3,4,4,5,5,6,6,6-Tridecafluorohexane-l-sulfonic acid" OR
TS="l,l,2,2,3,3,4,4,5,5,6,6,6-tridecafluoro-l-Hexanesulfonic acid" OR
TS="l-Hexanesulfonic acid, 1,1,2,2,3,3,4,4,5,5,6,6,6-tridecafluoro-" OR
TS="l-Hexanesulfonic acid, tridecafluoro-" OR
TS="l-Perfluorohexanesulfonic acid" OR TS="Perfluoro-l-hexanesulfonate"
OR TS="Perfluorohexane sulfonic acid" OR TS="Perfluorohexane-l-sulphonic
acid" OR TS="Perfluorohexanesulfonate" OR TS="Perfluorohexanesulfonic
acid" OR TS="Perfluorohexylsulfonate" OR TS="Tridecafluorohexanesulfonic
acid" OR TS="tridecafluoro-1-Hexanesulfonic acid" OR TS="PFHxS"
No date
limit-7/24/2017
This document is a draft for review purposes only and does not constitute Agency policy.
B-12 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Search
Search strategy
Dates of search
Literature
update
search
terms
TS="l-Hexanesulfonic acid, 1,1,2,2,3,3,4,4,5,5,6,6,6-tridecafluoro-, ion(l-)"
OR TS="PFHxS ion(l-)" ORTS="PFHxS_ion" OR
TS="Perfluorohexanesulfonate" OR TS="Tridecafluorohexane-l-sulfonate"
ORTS="perfluorohexyl sulfonate" OR
TS="l,l,2,2,3,3,4,4,5,5,6,6,6-Tridecafluoro-l-hexanesulfonyl fluoride" OR
TS="l-Hexanesulfonyl fluoride, 1,1,2,2,3,3,4,4,5,5,6,6,6-tridecafluoro-" OR
TS="l,l,2,2,3,3,4,4,5,5,6,6,6-Tridecafluoro-l-hexanesulfonic acid" OR TS="EC
206-587-1" OR TS="EINECS 206-587-1" ORTS="PFHS" OR
TS="Perfluorhexan-l-sulfonsaure" OR TS="Perfluorohexane sulfonic acid
(PFHxS)" OR TS="Perfluorohexane-l-sulphonic acid" OR TS="acide
perfluorohexane-l-sulfonique" OR TS="acido perfluorohexano-l-sulfonico"
OR TS="perfluorohexane-l-sulphonic acid" OR TS="perfluorohexanesulfonic
acid" ORTS="Ammonium Perfluorohexanesulfonate" OR TS="Ammonium
perfluorohexanesulfonate" ORTS="PFHxS-H3N" ORTS="PFHxS-K" OR
TS="Potassium Perfluorohexanesulfonate" OR TS="Potassium
perfluorohexanesulfonate" ORTS="Lithium Perfluorohexanesulfonate" OR
TS="Lithium perfluorohexanesulfonate" ORTS="PFHxS-Li"
2017-2018
Toxline
Search
terms
(108427-53-8[rn] OR 355-46-4[rn] OR
"1,1,2,2,3,3,4,4,5,5,6,6,6-Tridecafluorohexane-l-sulfonic acid" OR
"1,1,2,2,3,3,4,4,5,5,6,6,6-tridecafluoro-l-Hexanesulfonic acid" OR
"1-Hexanesulfonic acid, 1,1,2,2,3,3,4,4,5,5,6,6,6-tridecafluoro-" OR
"1-Hexanesulfonic acid, tridecafluoro-" OR "1-Perfluorohexanesulfonic acid"
OR "Perfluoro-l-hexanesulfonate" OR "Perfluorohexane sulfonic acid" OR
"Perfluorohexane-l-sulphonic acid" OR "Perfluorohexanesulfonate" OR
"Perfluorohexanesulfonic acid" OR "Perfluorohexylsulfonate" OR
"Tridecafluorohexanesulfonic acid" OR "tridecafluoro-l-Hexanesulfonic acid"
OR "PFHxS") AND ( ANEUPL [org] OR BIOSIS [org] OR CIS [org] OR DART [org]
OR EMIC [org] OR EPIDEM [org] OR HEEP [org] OR HMTC [org] OR IPA [org]
OR RISKLINE [org] OR MTGABS [org] OR NIOSH [org] OR NTIS [org] OR
PESTAB [org] OR PPBIB [org]) [not] PubMed [org] [not] pubdart [org]
No date
limit—7/21/2017
Literature
update
search
terms
@AND+@OR+("l-Hexanesulfonic+acid,+l, 1,2,2,3,3,4,4,5,5,6,6,6-tridecafluor
o-,+ion(l-)"+"PFHxS+ion(l-)"+"PFHxS_ion"+"Perfluorohexanesulfonate"+"Tri
decafluorohexane-l-sulfonate"+"perfluorohexyl+sulfonate"+"l,l,2,2,3,3,4,4,
5,5,6,6,6-Tridecafluoro-l-hexanesulfonyl+fluoride"+"l-Hexanesulfonyl+fluori
de,+l, l,2,2,3,3,4,4,5,5,6,6,6-tridecafluoro-"+"l,l,2,2,3,3,4,4,5,5,6,6,6-Tridec
afluoro-l-hexanesulfonic+acid"+"EC+206-587-l"+"EINECS+206-587-l"+"PFH
S"+"Perfluorhexan-l-sulfonsaure"+"Perfluorohexane+sulfonic+acid+(PFHxS)"
+" Perfluorohexane-l-sulphonic+acid"+"acide+perfluorohexane-1-sulfonique
"+"acido+perfluorohexano-l-sulfonico"+"perfluorohexane-l-sulphonic+acid"
+"perfluorohexanesulfonic+acid"+"Ammonium+Perfluorohexanesulfonate"+
"Ammonium+perfluorohexanesulfonate"+"PFHxS-H3N"+"PFHxS-K"+"Potassi
um+Perfluorohexanesulfonate"+"Potassium+perfluorohexanesulfonate"+"Lit
hium+Perfluorohexanesulfonate"+"Lithium+perfluorohexanesulfonate"+"PF
HxS-Li")+@NOT+@org+"nih+reporter"
@OR+(@TERM+@rn+108427+53+8+@TERM+@rn+423+50+7)+@NOT+@org
+"nih+reporter"
2017-2018
This document is a draft for review purposes only and does not constitute Agency policy.
B-13 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Search
Search strategy
Dates of search
TSCATS
Search
terms
@OR+(@term+@rn+355-46-4+@term+@rn+108427-53-8)
+@AND+@org+tscats
No date
limit—7/21/2017
Literature
update
search
terms
@OR+(@TERM+@rn+"108427+53+8"+@TERM+@rn+"423+50+7")+@org+tsc
ats
2017-2018
This document is a draft for review purposes only and does not constitute Agency policy.
B-14 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table B-6. Title/abstract-level screening criteria for the initial literature
searches
Inclusion criteria
Exclusion criteria
Populations
• Humans
• Standard mammalian animal models, including
rat, mouse, rabbit, guinea pig, hamster, monkey,
dog
• Alternative animal models in standard laboratory
conditions (e.g., Xenopus, zebrafish, minipig)
• Human or animal cells, tissues, or organs (not
whole animals); bacteria, nonmammalian
eukaryotes; other nonmammalian laboratory
species
• Ecological species
Exposures
• Exposure is to a PFAS compound
• Exposure via oral, inhalation, dermal,
intraperitoneal, or intravenous injection routes
• Exposure is measured in air, dust, drinking
water, diet, gavage, injection or via a biomarker
of exposure (PFAS levels in whole blood, serum,
plasma, or breastmilk)
• Study population is not exposed to a
PFAS compound
• Exposure is to a mixture only
Outcomes
• Studies that include a measure of one or more
health effect endpoints, including but not limited
to, effects on reproduction, development,
developmental neurotoxicity, liver, thyroid,
immune system, nervous system, genotoxicity,
and cancer
• In vivo and/or in vitro studies related to toxicity
mechanisms, physiological effects/adverse
outcomes, and studies useful for elucidating
toxic modes of action (MOAs)
• Qualitative or quantitative description of
absorption, distribution, metabolism, excretion,
toxicokinetic and/or toxicodynamic models
(e.g., PBPK, PBTK, PBTK/TD)
• Studies addressing risks to infants, children,
pregnant women, occupational workers, the
elderly, and any other susceptible or
differentially exposed populations
This document is a draft for review purposes only and does not constitute Agency policy.
B-15 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Inclusion criteria
Exclusion criteria
Other
• Structure and physiochemical properties
• Not on topic, including:
• Reviews and regulatory documents
• Abstract only, inadequately reported
abstract, or no abstract and not
considered further because study was
not potentially relevant
• Bioremediation, biodegradation, or
chemical or physical treatment of
PFAS compounds, including
evaluation of wastewater treatment
technologies and methods for
remediation or contaminated water
and soil
• Ecosystem effects
• Studies of environmental fate and
transport of PFAS compounds in
environmental media
• Analytical methods for
detecting/measuring PFAS
compounds in environmental media
and use in sample preparations and
assays
• Studies describing the manufacture
and use of PFAS compounds
• Not chemical specific (studies that do
not involve testing of PFAS
compounds)
• Studies that describe measures of
exposure to PFAS compounds
without data on associated health
effects
MOA = mode of action; PBPK = physiologically based pharmacokinetic; PBTK = physiologically based toxicokinetic;
PFAS = per- and polyfluoroalkyl substance; TD = toxicodynamic.
This document is a draft for review purposes only and does not constitute Agency policy.
B-16 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
Table B-7. Example DistillerSR form questions to be used for title/abstract and full text-level screening for
literature search updates from 2019
Used in title/abstract and full-text screening
Used in full text only
Source of study
Which
If meets PECO and
if not identified
Does the article
If meets PECO,
PFAS did
endocrine outcome,
from database
meet PECO
what type of
If supplemental, what
the study
If meets PECO, which health
which endocrine
Question
search?
criteria?
evidence?
type of information?
report?
outcome(s) apply?
tags apply?
Answer
options
(can
select
multiple
options)
• Source other
than HERO
database
search
• Yes
• No
• Unclear
• Tag as
potentially
relevant
supplementa
1 information
• Human
• Animal
(mammalian
models)
• In vitro or in
silico
genotoxicity
• PBPKorPK
model
• In vivo mechanistic or
MOA studies, including
non-PECO routes of
exposure
(e.g., injection) and
populations
(e.g., nonmammalian)
• In vitro or in silico
studies
(nongenotoxicity)
• ADME/
toxicokinetic (excluding
models)
• Exposure assessment
or characterization (no
health outcome)
• PFAS Mixture Study (no
individual PFAS
comparisons)
• Human case reports or
case series
• Ecotoxicity studies
• PFBA
• PFHxA
• PFHxS
• PFNA
• PFDA
• General toxicity, including
body weight, mortality, and
survival
• Cancer
• Cardiovascular, including
serum lipids
• Endocrine (hormone)
• Gastrointestinal
• Genotoxicity
• Growth (early life) and
development
• Hematological, including
nonimmune/hepatic/
renal clinical chemistry
measures
• Hepatic, including liver
measures and serum
markers (e.g., ALT; AST)
• Immune/inflammation
• Musculoskeletal
• Adrenal
• Sex hormones
(e.g., androgen;
estrogen;
progesterone)
• Neuroendocrine
• Pituitary
• Steroidogenesis
• Thyroid
This document is a draft for review purposes only and does not constitute Agency policy.
B-17 DRAFT-DO NOT CITE OR QUOTE
-------
Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments
• Environmental fate or
occurrence (including
food)
• Manufacture,
engineering, use,
treatment,
remediation, or
laboratory methods
• Other assessments or
records with no
original data
(e.g., reviews,
editorials,
commentaries)
Nervous system, including
behavior and sensory
function
Nutrition and metabolic
Ocular
PBPK or PK model
Renal, including urinary
measures (e.g., protein)
Reproductive
Respiratory
Skin and connective tissue
effects
ADME = absorption, distribution, metabolism, and excretion; ALT = alanine aminotransferase; AST = aspartate aminotransferase; HERO = Health and
Environmental Research Online; MOA = mode of action; PBPK = physiologically based pharmacokinetic; PECO = populations, exposures, comparators, and
outcomes; PFAS = per- and polyfluoroalkyl substances; PFBA = perfluorobutanoic acid; PFDA = perfluorodecanoic acid; PFHxA = perfluorohexanoic acid;
PFHxS = Perfluorohexanesulfonic acid; PFNA = Perfluorononanoic acid; PK = pharmacokinetic.
This document is a draft for review purposes only and does not constitute Agency policy.
B-18 DRAFT-DO NOT CITE OR QUOTE
------- |